An Abstract of the Dissertation Of

Home , Agapostemon femoratus, Andrena prunorum, Anthophora urbana, Colletes compactus, Colletes fulgidus, Diadasia enavata

AN ABSTRACT OF THE DISSERTATION OF

Lauren Ashley Smith DiCarlo for the degree of Doctor of Philosophy in Wildlife Science on June 1, 2018

Title: Native Bee and Spider Community Responses to Grassland Restoration and Wildfire

Abstract approved: ______Sandra J. DeBano

Up to 99.9% of native North American grasslands have been degraded since European settlement, primarily due to agricultural conversion. Today, grasslands are a top priority for restoration as they provide essential habitat for many rare and endangered species; however, the majority of studies in grasslands have focused on vegetation or vertebrate responses to restoration while largely neglecting invertebrates. Grassland invertebrates are highly diverse and provide important ecosystem services such as pollination, nutrient cycling, food for vertebrates, and pest control. This dissertation seeks to understand the structure of spider and native bee communities within arid bunchgrass prairies and determine how grassland restoration and wildfire impact these beneficial invertebrates. In Chapter 2, I focus on spider communities in a low-elevation, arid Pacific Northwest bunchgrass prairie and compare degraded, native, and restored sites to examine how spider communities and habitat respond to arid grassland restoration. Spider communities responded strongly to invasive annual grass, litter, and biological soil crust cover. Native sites differed from those in restored and degraded sites by community composition and abundance, with fewer spiders found in native sites than degraded and restored sites. However, native and restored sites had more species than degraded sites. I also examine how responses varied with the age of the restoration project. Chronosequence data showed trends for lower abundance, higher species richness, and changing community composition as restoration projects mature.

Chapter 3 describes the unique bee community found within the same bunchgrass prairie, identifies environmental variables associated with variation in bee abundance, richness, diversity, and community structure, and assesses the effect of grassland restoration on bee communities. I identify temporal trends within the bee and floral resource community that span over several seasons and years. As with the spider community, the bee community composition at native sites differed from both the degraded and restored communities, which did not differ from each other. However, there was no statistically significant difference in bee abundance, richness, and diversity among degraded, restored, and native sites. Bee abundance was most closely associated with litter cover, bee richness was associated with maximum vegetation height and floral abundance, and bee diversity was associated with floral abundance. Chapter 4 examines spider community response to restoration at the regional scale. I compared degraded and restored communities at three separate grassland locations in eastern Oregon to determine what impact landscape context has on restoration and identify the environmental variables that underlie spider community patterns at this larger scale. Spider communities did not respond similarly to restoration among locations, indicating that landscape context may play a larger role in responses than restoration treatments. Regionally, spider abundance responded largely to changes in invasive grass and litter cover, while richness and diversity responded to changes in maximum vegetation height and forb cover. Wildfire frequency has increased across the western United States, yet it is unclear how these fires affect beneficial invertebrates in arid grasslands. In Chapter 5, I examine bee, spider, and vegetative communities one year before and one year after wildfire. Both native bee and spider community composition were significantly altered one year after the fire. The fire did not affect bee or spider abundance, or spider diversity or richness but significantly increased native bee diversity and richness. Habitat variables such as invasive annual grass and biological soil crust declined significantly, while forb abundance increased after the burn. Taken together, the findings show that invertebrate responses to grassland restoration and wildfire in inland Pacific Northwest grasslands are complex. Spider communities appear to largely respond to changes in vegetative structure (grass cover and

height) and litter, while bee communities appear to be sensitive to changes in potential nesting sites (grass and litter cover) and floral resources (floral abundance). Depending on the degree that these environmental factors are influenced by future grassland restoration or wildfire, managers may expect to see strong effects on spider and bee communities.

Native Bee and Spider Community Responses to Grassland Restoration and Wildfire

by Lauren Ashley Smith DiCarlo

A DISSERTATION

submitted to

Oregon State University

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Presented June 1, 2018 Commencement June 2018

Doctor of Philosophy dissertation of Lauren Ashley Smith DiCarlo presented on June 1, 2018.

APPROVED:

Major Professor, representing Wildlife Science

Head of the Department of Fisheries and Wildlife

Dean of the Graduate School

I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request.

Lauren Ashley Smith DiCarlo, Author

ACKNOWLEDGEMENTS

I would like to thank my advisor, Sandy DeBano, for her incredible advice, encouragement and unwavering support throughout this process. Not every advisor provides free rein over research design and spends hundreds of hours with their students in the field and for this I simply cannot express my immense gratitude. My other committee members, Andy Hulting, John Lambrinos, Taal Levi, David Pyke, and the late Don Horneck, each helped make this dissertation better and offered exceptionally helpful comments and suggestions, not only to the text, but also to the project design.

This work would not have been possible without the people at the Hermiston Agricultural Research and Extension Center including the NIFA faculty, administration, and field crew. Thank you for providing me with a fantastic opportunity by bringing me on as a PhD student at Oregon State, providing every type of assistance at all hours of the day, and fixing many flat tires. I’ve learned an unbelievable amount from all of them. Also, the other NIFA students and Fisheries and Wildlife Graduate students have been a great source of support, I’m happy to call them both friends and colleagues. In addition, thank you to all of the field and lab assistants, many of which spent their summers working long strenuous hours in 100+ degrees or bent over a lab bench separating spiders or pinning bees. The list of remarkable women include: Lauren J. Smith, Samantha Roof, Estany Campbell, Keelie Kirby, Lexie McDaniel, Mamo Waianuhea, and Briana Price.

Lastly, I would like to thank my family for their encouragement. The difficult decisions were made much easier through their support and help driving across the country (many times) and moving lots of furniture. My grandparents, in particular, have always been my biggest cheerleaders and my spirit was greatly uplifted by my grandmother’s handwritten letters each week, of which I will forever cherish. Above all, I would like to thank my husband, Michael, who has given up so much to support me. Thank you for commuting every other week to various locations around Oregon, the delicious dinners, and taking care of our two amazing dogs, Pearl and Louie. And most importantly, thank you for making my dreams into a reality.

CONTRIBUTION OF AUTHORS

Dr. Sandra DeBano contributed to all aspects of this dissertation including study design, analysis, and writing. Skyler Burrows of the USDA Bee Lab in Logan, Utah identified all bees and helped with editing of Chapters 3 and 5.

TABLE OF CONTENTS

Page

CHAPTER 1. GENERAL INTRODUCTION ...... 1 Objectives ...... 3 Literature Cited ...... 4 CHAPTER 2. Citation Information ...... 8 CHAPTER 2. SPIDER COMMUNITY RESPONSES TO GRASSLAND RESTORATION: BALANCING TRADEOFFS BETWEEN ABUNDANCE AND DIVERSITY ...... 9 Abstract ...... 9 Introduction ...... 9 Methods...... 12 Study Site ...... 12 Site Selection ...... 12 Spider Sampling ...... 13 Habitat Survey ...... 13 Analyses – Comparing Native, Restored, and Degraded Sites ...... 13 Analyses – Examining Age of Restoration on Response...... 15 Results ...... 15 Comparing Native, Restored, and Degraded Sites ...... 16 The Effect of Age of Restoration on Response ...... 17 Discussion ...... 17 Acknowledgements ...... 21 Literature Cited ...... 22 CHAPTER 3. ARID GRASSLAND BEE COMMUNITY COMPOSITION, TEMPORAL PATTERNS, AND RESPONSES TO RESTORATION ...... 33 Abstract ...... 33 Introduction ...... 33 Methods...... 36 Study Site ...... 36 Site Selection ...... 36

TABLE OF CONTENTS (Continued)

Page

Bee Sampling ...... 37 Habitat and Floral Resource Surveys ...... 37 Analyses – Characterizing bee community and its temporal variability ...... 38 Analyses – Environmental factors influencing the bee community ...... 39 Analyses – Effects of restoration on the bee community ...... 40 Results ...... 40 Characterizing bee community and its temporal variability ...... 40 Environmental factors influencing the bee community ...... 44 Effects of restoration on the bee community ...... 45 Discussion ...... 46 Acknowledgements ...... 50 Literature Cited ...... 50 CHAPTER 4. LANDSCAPE CONTEXT INFLUENCES SPIDER COMMUNITY RESPONSES TO GRASSLAND RESTORATION: A REGIONAL STUDY IN THE PACIFIC NORTHWEST, USA...... 69 Abstract ...... 69 Introduction ...... 70 Methods...... 72 Study Sites ...... 72 Site Selection ...... 73 Spider Sampling ...... 73 Habitat Survey ...... 74 Analyses ...... 74 Results ...... 76 Effect of Landscape Context and Restoration ...... 76 Environmental Variables Influencing Spider Communities at the Regional Scale ...... 78

TABLE OF CONTENTS (Continued)

Page

Discussion ...... 79 Acknowledgements ...... 83 Literature Cited ...... 84 CHAPTER 5. SHORT-TERM RESPONSE OF TWO BENEFICIAL INVERTEBRATE GROUPS TO WILDFIRE IN AN ARID GRASSLAND SYSTEM, USA ...... 100 Abstract ...... 100 Introduction ...... 100 Methods...... 103 Study Site ...... 103 Site Selection ...... 104 Spider Sampling ...... 104 Bee Sampling ...... 104 Habitat Survey ...... 105 Analyses ...... 105 Results ...... 107 Spiders...... 107 Bees ...... 108 Habitat Variables ...... 109 Discussion ...... 110 Acknowledgements ...... 115 Literature Cited ...... 115 CHAPTER 6. GENERAL CONCLUSIONS ...... 130 Literature Cited ...... 132 General Bibliography ...... 134

LIST OF FIGURES

Figure Page

2.1. Map of The Nature Conservancy Boardman Preserve sites ...... 26

2.2. Sample plot setup for spider and habitat sampling ...... 27

2.3. Average spider A) abundance and B) rarefied species richness by treatment...... 27

2.4. Average percent cover for habitat variables and average height for maximum vegetation height ...... 28

2.5. Ordination for A) Axis 1 vs. Axis 2 and B) Axis 1 vs. Axis 3 for spider families ...... 29

2.6. Average spider A) abundance and B) rarefied species richness by time since restoration ...... 30

3.1. Map of The Nature Conservancy Boardman Preserve sites ...... 55

3.2. Sample plot setup for bee and habitat sampling ...... 55

3.3. Ordination for A) sampling bouts, B) years, and C) treatments for bee species ...... 56

3.4. Regression trees for bee A) abundance, B) rarefied richness, and C) diversity...... 57

4.1. Map of three locations including The Nature Conservancy Boardman Preserve, Umatilla National Wildlife Refuge, and The Nature Conservancy Zumwalt Preserve ...... 88

4.2. Sample plot setup for spider and habitat sampling ...... 89

4.3. Average percent cover for A) invasive annual grasses, B) litter, and C) forbs, and D) maximum vegetation height for both location and restoration treatment ...... 89

4.4. Average spider A) abundance, B) richness, and C) diversity of spiders for both location and restoration treatment ...... 90

4.5. Ordination for A) restoration treatment and B) location for spider families ...91

LIST OF FIGURES (Continued)

Figure Page

4.6. Regression trees for A) spider abundance, B) rarefied richness, and C) diversity...... 92

4.7. Regression trees for A) axis 1 and B) axis 2 for spider ordination ...... 94

5.1. Map of The Nature Conservancy Boardman burned and unburned sites ...... 121

5.2. Photos of A) before wildfire, B) immediately after wildfire, and C) one year after wildfire...... 121

5.3. Sample plot setup for spider, bee, and habitat sampling...... 122

5.4. Average A) abundance, B) diversity, and C) species richness of spiders and bees before and post-fire ...... 122

5.5. Ordination for A) axis 1 and axis 2 for spider families and B) treatment centroid movement between pre- and post-fire ...... 123

5.6. Ordination for A) axis 1 and axis 2 for bee families and B) treatment centroid movement between pre- and post-fire ...... 124

5.7. Average percent cover for habitat variables and average height for maximum vegetation height pre- and post-fire ...... 125

5.8. Average forb A) abundance and B) richness pre- and post-fire ...... 125

LIST OF TABLES

Table Page

2.1. Sampling periods for spiders and habitat variables ...... 30

2.2. Total spider family abundance and relative abundance within each treatment ...... 31

2.3. Pearson correlation coefficient for spider family ordination ...... 31

2.4. Pearson correlation coefficients for spider families in restored ordination .....32

3.1. Metrics for bee sampling ...... 59

3.2. Adjusted abundance of bee taxa during each collection period ...... 60

3.3. Pearson correlation coefficients for bee species ordination ...... 63

3.4. Presence and absence of each forb species during each collection period ...... 67

4.1. Location information for each restoration site ...... 96

4.2. Two-way ANOVA results for environmental variables and spider abundance, richness, and diversity ...... 97

4.3. Pearson correlation coefficients for landscape spider family ordination including degraded and native treatments...... 98

4.4. Pearson correlation coefficients for landscape spider family ordination including degraded, native, and restored treatments ...... 99

5.1. Pearson correlation coefficients for spider family ordination ...... 126

5.2. Indicator species analysis results for spider families, bee species, and forb species ...... 127

5.3. Pearson correlation coefficients for bee species ordination ...... 128

Chapter 1. GENERAL INTRODUCTION

Up to 99.9% of native North American grasslands have been degraded or destroyed since European settlement, primarily due to agricultural conversion (Sampson & Knopf 1994; Burel et al. 1998). Today, grasslands are a top priority for restoration as they provide essential habitat for many rare and endangered plant and animal species (Kennedy et al. 2009; Rao et al. 2011; Tubbesing et al. 2014). However, the majority of studies on grassland restoration have focused either on vegetation or vertebrate responses (e.g., Wisdom et al. 2002; Knick et al. 2003; Huddleston & Young 2004; Huddleston & Young 2005; Beck et al. 2012), even though grassland invertebrates are highly diverse and provide important ecosystem services such as pollination, nutrient cycling, food for vertebrates, and pest control (Weisser & Siemann 2004; DeBano 2006; Kimoto et al. 2012 a,b; Gonzalez et al. 2013). Thus, it remains unclear how grassland restoration influences invertebrate diversity and function (Nemec et al. 2014), particularly in North American grasslands. Two beneficial groups, spiders and bees, are particularly understudied even though they provide multiple services. With approximately 40,000 species known globally, spiders (Araneae) are major contributors to biodiversity (Mirshamsi Kakhki 2005; Zamani & Rafinejad 2014). Not only are spiders diverse in form, but also in hunting strategies and they play fundamental roles in the terrestrial food web as both predators and prey (Malumbres-Olarte et al. 2013). Many spiders are considered generalist predators and can enhance natural control of agricultural pests, a service with an estimated value of $4.5 billion annually in the US (Losey & Vaughan 2006; Kovacs-Hostyanszki et al. 2013). In addition, spiders potentially respond faster than vegetation or vertebrates to environmental perturbations because of their high degree of mobility and rapid generation times (Mortimer et al. 1998). Because of this, spiders may be particularly useful indicator taxa for restoration monitoring (Wheater et al. 2000). Bees (Hymenoptera), another important beneficial invertebrate group, are one of the most diverse groups of insects with an estimated 20,000 species occurring globally

(Michener 2007). Bees are considered the most important pollinators in many ecosystems (Exeler et al. 2009), especially given that 80% of native plants and 35% of crops rely on animal pollination (Klein et al. 2007; Winfree et al. 2007; Potts et al. 2010). Like spiders, bees can be useful indicator taxa for restoration monitoring as they are closely inter-related to landscape structure and use (Sepp et al. 2004). In the last decade, documented declines in honey bees and native bees have resulted in an increase in pollinator conservation and a need to learn more about these sensitive and ecologically significant communities (Potts et al. 2010). Despite the importance of both bees and spiders in providing ecosystem services and their potential utility in monitoring, little research has focused on the impact grassland restoration has on bee and spider communities in North America. While more work has been conducted in European systems (e.g. Perner and Malt 2003; Exeler et al. 2009; Albrecht et al. 2010; Deri et al. 2011; Tarrant et al. 2013), European grasslands and their management and restoration differ in significant ways from US systems, especially those in the arid and semi-arid West. Arid grasslands of western North America are one of the most threatened and understudied grasslands in the world, with over 90% converted to agricultural land-use at one time (Tisdale 1982; Kimoto et al. 2012a). However, in the last century cropland abandonment has increased exponentially across North America due to increases in urbanization and the import of food from other areas (Cramer and Hobbs 2007). In response to this trend, there has been a growing movement in recent decades to restore abandoned fields once used for production to native grassland habitat that can provide essential wildlife habitat and conserve biodiversity (Bakker and Berendse 1999; Wilson et al. 2004; Cramer and Hobbs 2007; Öster et al. 2009; Torok et al. 2011; Porensky et al. 2014). This effort encompasses arid grasslands of North America, with multiple groups including government agencies and non-profit organizations at the forefront of restoration. Yet monitoring the effectiveness of restoration efforts in grasslands is still limited due to lack of funding and data on the response of many ecologically significant taxa (Huxel and Hastings 1999; Gerla et al. 2012).

A common problem in restoring these grasslands includes the prominence of non- native plant species, most notably the invasion of annual grasses (Kennedy et al. 2009). Non-native annual grasses, in addition to further degrading susceptible grasslands, alter disturbance regimes, especially fire frequency and intensity, and create an additional variable to consider when restoring grasslands (Pyke 1999; Davies et al. 2012). While wildfire and prescribed-burn studies are common in North America, few studies have solely focused on wildfires in arid grassland habitat and even fewer have concentrated on how they impact beneficial grassland invertebrates. Objectives. The data chapters of my dissertation examine the impact arid grassland restoration and wildfire have on invertebrates. The second chapter focuses on the spider communities within an arid grassland. Using degraded, native, and restored grassland sites I compare the spider community and investigate potential habitat variables that underlie observed responses. In addition, I examine how the same variables vary along a chronosequence of restoration. My third chapter focuses on the bee communities within this same arid grassland. Similear to the approach in the previous chapter, I examine effects of grassland restoration on the bee community by comparing degraded, native, and restored grassland sites and I investigate habitat variables that underlie observed responses. In addition, I describe this unique bee community and its seasonal and inter-annual variability. My fourth chapter focuses on the role landscape context plays in influencing spider community responses to grassland restoration. Using three grassland restoration locations I compare degraded and restored sites to determine the relative role of both landscape context and restoration on the spider communities and environmental variables. Then using a larger dataset that includes degraded, restored, and native sites I identify environmental variables associated with observed patterns of spider communities at a regional scale. My fifth chapter focuses on the effect of wildfire on native bee and spider communities within an arid grassland. Using data from both burned and unburned sites before and after fire, I assess how both communities differ one year after the burn and investigate which habitat factors are strongly associated with the observed patterns for each invertebrate group. My final chapter summarizes my major

4 findings and provides some basic recommendations for future restoration of grassland invertebrate communities.

Literature Cited

Albrecht, M., Schmid, B., Obrist, M.K., Schupbach, B., Kleijn, D., Duelli, P., 2010. Effects of ecological compensation meadows on arthropod diversity in adjacent intensively managed grassland. Biol Conserv. 143, 642-649. Bakker, J.P., Berendse, F., 1999. Constraints in the restoration of ecological diversity in grassland and heathland communities. TREE. 14, 63-68. Beck, J.L., Connelly, J.W., Wambolt, C.L. 2012. Consequences of treating Wyoming Big Sagebrush to enhance wildlife habitats. Rangeland Ecol Manag. 65, 444-455. Burel, F., Baudry, J., Butet, A., Clergeau, P., Delettre, Y., Le Coeur, D., Dubs, F., Morvan, N., Paillat, G., Petit, S., Thenail, B., Brunel, E., Lefeuvre, J., 1998. Comparative biodiversity along a gradient of agricultural landscapes. Acta Oecol. 19, 47-60. Cramer, V.A., Hobbs, R.J., eds., 2007. Old fields: dynamics and restoration of abandoned farmland. Society for Ecological Restoration International. Island Press, Washington, D.C. Davies, G.M., Bakker, J.D., Dettweiler-Robinson, E., Dunwiddie, P.W., Hall, S.A., Downs, J., Evans, J., 2012. Trajectories of change in sagebrush steppe vegetation communities in relation to multiple wildfires. Ecol Appl. 22, 1562-1577. DeBano, S.J. 2006. Effects of livestock grazing on insect communities in semi-arid grasslands of southeastern Arizona. Biodivers Conserv. 15, 2547-2564. Déri, E., Magura, T., Horvath, R., Kisfali, M., Ruff, G., Lengyel, S., Tothmeresz, B., 2011. Measuring the short-term success of grassland restoration: the use of habitat affinity indices in ecological restoration. Restor Ecol. 19, 520–528. Exeler, N., Kratochwil, A, Hochkirch, A. 2009. Restoration of riverine inland sand dune complexes: implications for the conservation of wild bees. J Appl Ecol. 46, 1097- 1105. Gerla, P.J., Cornett, M.W., Ekstein, J.D., Ahlering, M.A., 2012. Talking big: lessons learned form a 9000 hectare restoration in the northern tallgrass prairie. Sustainability. 4: 3066-3087, available online: http://www.mdpi.com/2071- 1050/4/11/3066/htm Gonzalez, N., DeBano, S.J., Kimoto, C., Taylor, R.V., Tubbesing, C., Strohm, C., 2013. Native bees associated with isolated aspen stands in Pacific Northwest Bunchgrass Prairie. Nat Areas J. 33, 374-383.

Huddleston, R.T., Young, T.P. 2004. Spacing and competition between planted grass plugs and preexisting perennial grasses in a restoration site in Oregon. Restor Ecol. 12, 546-551. Huddleston, R.T., Young, T.P. 2005. Weed control and soil amendment effects on restoration plantings in an Oregon grassland. West N Am Naturalist. 65, 507-515. Huxel, G.R., Hastings, A., 1999. Habitat loss, fragmentation, and restoration. Restor Ecol. 7, 309-315. Kennedy, P.L., DeBano, S.J., Bartuszevige, A., Lueders, A., 2009. Effects of native and nonnative grassland plant communities on breeding passerine birds: implications for restoration of Northwest bunchgrass prairie. Restor Ecol. 17, 515-525. Kimoto, C., DeBano S.J., Thorp, R.W., Rao, S., Stephen, W.P., 2012a. Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci. 12, 108. available online: http://www.insectscience.org/12.108 Kimoto, C., DeBano, S.J., Thorp, R.W., Taylor, R.V., Schmalz, H., DelCurto, T., Johnson, T., Kennedy, P.L., Rao, S., 2012b. Livestock and native bee communities: short-term responses to grazing intensity and implications for managing ecosystem services in grasslands. Ecosphere. 3, 88. http://www.esajournals.org/doi/pdf/10.1890/ES12-00118.1 Klein, A.M., Vaissie’re, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Important of pollinators in changing landscapes for world crops. P Roy Soc B-Biol Sci. 274, 303-313. Knick, S.T., Dobkin, D.S., Rotenberry, J.T., Schroeder, M,A., Vander Haegen, W.M., Van Riper III, C., 2003. Teetering on the edge or too late? Conservation and research issues for avifauna of sagebrush habitats. Condor. 105, 611-634. Kovacs-Hostyanszki, A., Elek, Z., Balazs, K., Centeri, C., Falusi, E., Jeanneret, P., Penksza, K., Podmaniczky, L., Szalkovszki, O., Baldi, A. 2013 Earthworms, spiders and bees as indicators of habitat quality and management in a low-input farming region—A whole farm approach. Ecol Indic. 33, 111-120. Losey, J.E., Vaughan, M., 2006. The economic value of ecological services provided by insects. Bioscience. 56, 311-323. Malumbres-Olarte, J., Vink, C.J., Ross, J.G., Cruickshank, R.H., Paterson, A.M., 2013. The role of habitat complexity on spider communities in native alpine grasslands of New Zealand: Habitat complexity and alpine spiders. Insect Conserv Diver. 6, 124–134. Michener, C.D., 2007. The bees of the world, 2nd edition. The John Hopkins University Press, Baltimore, USA. Mirshamski Kakhki, O., 2005. Faunistic study of spiders in Khorasan Province, Iran (Arachnida: Araneae). Int J Agric Biol. 1, 55-66.

Mortimer, S.R., Hollier, J.A., Brown, V.K., 1998. Interactions between plant and insect diversity in the restoration of lowland calcareous grasslands in southern Britain. Appl Veg Sci. 1, 101–114. Nemec, K., Allen, C.R., Danielson, S.D., Helzer, C.J., 2014. Responses of predatory invertebrates to seeding density and plant species richness in experimental tallgrass prairie restorations. Agr Ecosyst Environ. 183, 11-20. Öster, M., Ask, K., Cousins, S.A.O., Eriksson, O., 2009. Dispersal and establishment limitation reduces the potential for successful restoration of semi-natural grassland communities on former arable fields. J Appl Ecol. 46, 1266-1274. Perner, J., Malt, S., 2003. Assessment of changing agricultural land use: response of vegetation, ground-dwelling spiders and beetles to the conversion of arable land into grassland. Agr Ecosyst Environ. 98, 169-181. Porensky, L.M., Leger, E.A., Davison, J., Miller, W.W., Goergen, E.M., Espeland, E.K., Carroll-Moore, E.M., 2014. Arid old-field restoration: native perennial grasses suppress weeds and erosion, but also suppress native shrubs. Agr Ecosyst Environ. 184, 135-144. Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O., Kunin, W.E. 2010. Global pollinator declines: trends, impacts and drivers. TREE. 25, 345-353. Pyke, D.A., 1999. Invasive exotic plants in sagebrush ecosystems of the Intermountain West. In: Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., eds. Proceedings: Sagebrush Steppe Ecosystems Symposium; Boise, Idaho; June 23- 25, 1999. Boise, Idaho: Bureau of Land Management, pp. 43-54. Rao, S., Stephen, W.P., Kimoto, C., DeBano, S.J. 2011. The status of the ‘Red-Listed’ Bombus occidentalis (Hymenoptera: Apiformes) in Northeastern Oregon. Northwest Sci. 85, 64-67. Sampson, F., Knopf, F, 1994. Prairie conservation in North America. BioScience. 44, 418-421. Sepp, K., Merit, M., Mand, M., Truu, J., 2004. Bumblebee communities as indicator for landscape monitoring in the agri-environmental programme. Landscape Urban Plan. 67, 173-183. Tarrant, S., Ollerton, J., Lutfor Rahmn, M. Tarrant, J., McCollin, D., 2013. Grassland restoration on landfill sites in the east midlands, United Kingdom: an evaluation of floral resources and pollinating insects. Restor Ecol. 21, 560-568. Tisdale, E.W. 1982. Grasslands of western North America: the Pacific Northwest Bunchgrass. In: Nicholson, A.C., McLean, A., Baker, T.E., Editors. Proceedings of the 1982 Grassland Ecology and Classification Symposium. Pp. 232-245. Information Services Branch, British Columbia Ministry of Forests. Torok, P., Vida, E., Deak, B., Lengyel, S., Tothmeresz, B., 2011. Grassland restoration on former croplands in Europe: an assessment of applicability of techniques and costs. Biodivers Conserv. 20, 2311-2332.

Tubbesing, C., Strohm, C., DeBano, S.J., Gonzalez, N., Kimoto, C., Taylor, R.V. 2014. Insect visitors and pollination ecology of Spalding’s catchfly (Silene spaldingii) in the Zumwalt Prairie of northeastern Oregon. Nat Area J. 34, 200-211. Weisser, W.W., Siemann, E., 2004. The various effects of insects on ecosystem functioning. Insects Ecosyst Function. 173, 3–24. Wheater, C.P., Cullen, W.R., Bell, J.R., 2000. Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecol. 15, 401–406. Wilson, S.D., Bakker, J.D., Christian, J.M., Li, X., Ambrose, L.G., Waddington, J., 2004. Semiarid old-field restoration: is neighbor control needed? Ecol Appl, 14, 476- 484. Winfree, R., Griswold, T., Kremen, C. 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conserv Biol. 21, 213-223. Wisdom, M.J., Rowland, M.M., Wales, B.C., Hemstrom, M.A., Hann, W.J., Raphael, M.G., Holthausen, R.S., Gravenmier, R.A., Rich, T.D. 2002. Modeled effects of sagebrush-steppe restoration on Greater Sage-Grouse in the interior Columbia Basin, USA. Society for Conservation Biology 16, 1223-1231. Zamani, A., Rafinejad, J., 2014. First record of the Mediterranean recluse spider Loxosceles rufescens (Araneae: Sicariidae) from Iran. J Arthropod-Borne Di. 8, 228-231.

CHAPTER 2. CITATION INFORMATION

SPIDER COMMUNITY RESPONSE TO GRASSLAND RESTORATION: BALANCING TRADEOFFS BETWEEN ABUNDANCE AND DIVERSITY

Lauren A. Smith DiCarlo, Sandra J. DeBano

Restoration Ecology, 2018 Issue: Available online, doi:10.1111/rec.12832

CHAPTER 2. SPIDER COMMUNITY RESPONSE TO GRASSLAND RESTORATION: BALANCING TRADEOFFS BETWEEN ABUNDANCE AND DIVERSITY

Lauren A. Smith DiCarlo and Sandra J. DeBano

Abstract Spiders (Araneae) play key roles in ecosystems, not only as common and abundant generalist predators, but also as major contributors to biodiversity in many areas. In addition, due to their short generation times and high mobility, spiders respond rapidly to small changes in their environment, potentially making them useful indicators for restoration monitoring. However, few studies have focused on spider responses to grassland restoration in the US. We compared degraded, native, and restored grassland sites to examine how spider communities and habitat respond to arid grassland restoration. We also examined how responses varied with the age of the restoration project. Spider communities in native sites differed from those in restored and degraded sites in several ways: native sites had fewer spiders and a different community composition than degraded and restored sites. However, native and restored sites had more species than degraded sites. Chronosequence data showed trends for lower abundance, higher species richness, and changing community composition as restoration projects mature. Several habitat variables were closely linked to variation in spider communities including cover of invasive annual grasses, litter, and biological soil crusts. Our data suggest that spider and vegetation responses to grassland restoration efforts can be successful in the long-term—with resulting communities becoming more similar to native ones—and that spiders are useful indictors of grassland restoration. Our results also suggest that restoration may involve balancing trade-offs between ecosystem services, with potential losses in predatory control offset by increases in biodiversity with restoration effort.

Introduction

Up to 99.9% of native North American grasslands have been degraded or destroyed since European settlement, primarily due to agricultural conversion (Sampson & Knopf 1994; Burel et al. 1998). Today, grasslands are a top priority for restoration as they provide essential habitat for many rare and endangered plant and animal species (Kennedy et al. 2009; Rao et al. 2011; Tubbesing et al. 2014). However, the majority of studies on grassland restoration have focused either on vegetation or vertebrate responses (e.g., Wisdom et al. 2002; Knick et al. 2003; Huddleston & Young 2004; Huddleston & Young 2005; Beck et al. 2012), even though grassland invertebrates are highly diverse and provide important ecosystem services such as pollination, nutrient cycling, food for vertebrates, and pest control (Weisser & Siemann 2004; DeBano 2006; Kimoto et al. 2012 a,b; Gonzalez et al. 2013). Thus, it remains unclear how grassland restoration influences invertebrate diversity and function (Nemec et al. 2014), particularly in North American grasslands. With approximately 40,000 species known globally, spiders (Araneae) are the seventh most diverse taxon in the world and are major contributors to biodiversity (Mirshamsi Kakhki 2005; Zamani & Rafinejad 2014). Not only are spiders diverse in form, but also in hunting strategies and they play fundamental roles in the terrestrial food web as both predators and prey (Malumbres-Olarte et al. 2013). Many spiders are considered generalist predators and can enhance natural control of agricultural pests, a service with an estimated value of $4.5 billion annually in the US (Losey & Vaughan 2006; Kovacs-Hostyanszki et al. 2013). In addition, spiders potentially respond faster than vegetation or vertebrates to environmental perturbations because of their high degree of mobility and rapid generation times (Mortimer et al. 1998). Because of this, spiders may be particularly useful indicator taxa for restoration monitoring (Wheater et al. 2000). Different spider species require discrete habitat conditions and by sampling the entire community the presence or relative abundance of certain species can be used as indicators for habitat availability, provided that the general ecology of the different species is known (Malumbres-Olarte et al. 2013). Yet, despite their importance and potential utility in monitoring and a heightened interest in restoration within the past few decades, little research has focused on the

11 impact of grassland restoration on spider communities in North America (Samson & Knopf 1994; Nemec et al. 2014). While more work has been conducted in European systems (see Bell et al. 2001, for review), European grasslands and their management and restoration differ in significant ways from US systems, especially those in the arid and semi-arid West. Work in other grassland systems, as well as in wet heathlands and forests, demonstrate that spider communities are influenced by restoration. Habitat characteristics associated with these changes depend on the system and include vegetative structural complexity (e.g., height, cover), plant density, litter cover, temperature, and moisture (Bell et al. 2001; Borchard et al. 2003; Perner & Malt 2003; Cristofoli et al. 2010; Gollan et al 2010). Only a few studies have examined chronosequences of restoration on spiders (e.g., Perner & Malt 2003; Cristofoli et al. 2010); these have found that spider communities vary with respect to the age of the restoration project and are most likely responding to changes in the plant community. This study aims to describe not only the spider communities present in an arid grassland in the western US, but also to examine habitat variables that influence spiders in these communities. Unlike other studies that have focused on invertebrate responses to grassland/shrub-steppe restoration (e.g., Mortimore et al. 1998; Longcore 2003; Déri et al. 2011; Borchard et al. 2014; Nemec et al 2014), we focused on spiders specifically to enhance our ability to pinpoint habitat variables underlying responses of different species of spiders to restoration. We assessed three treatments including native (or reference sites), degraded, and restored grassland habitat. Restored grassland sites varied in age (2 – 12 years), providing an opportunity to examine a chronosequence of restoration. This setup allowed us to assess not only the different spider communities within each habitat but to also investigate the amount of time necessary after restoration to detect changes within the community. We had three main objectives: 1) compare spider abundance, species richness, and community composition in restored, native, and degraded treatments, 2) examine how these same variables varied along a chronosequence of restoration, and 3) investigate potential habitat variables that may underlie observed responses. The results from this study provide restoration ecologists with recommendations on which factors most heavily influence spider communities during

12 western US grassland restoration and guidance on when to expect changes within plant and spider communities in response to restoration.

Methods Study Site This study took place at The Nature Conservancy Boardman Grasslands Preserve (the Preserve) in Morrow County, Oregon, USA (45.636738°N, -119.860457°W) (Fig. 2.1). The Preserve occupies 9,163 ha of arid grassland and shrub-steppe from 120-295 m elevation. The majority of the area was grazed by cattle until 2002 but the Preserve contains extensive areas of high quality grassland (largely intact native grasses and forbs relatively uninvaded with non-native annual grasses) and degraded grassland (formerly cultivated, lacking native bunchgrasses and highly invaded with non-native annual grasses). Common invasive grasses include cheatgrass (Bromus tectorum L.) and medusahead (Taeniatherum caput-medusae (L.) Nevski), while native grasses include bluebunch wheatgrass (Pseudoroegneria spicata (Pursh) Á. Löve), Sandberg bluegrass (Poa secunda J. Prsel), bottlebrush squirreltail (Elymus elymoides (Raf.) Swezey ssp. brevifolius (J.G. Sm.) Barkworth), and needle and thread grass (Heterostipa comata (Trin. & Rupr.) Barkworth). The average precipitation is 22 cm with the majority of the precipitation falling from November to February and annual average temperatures ranging from 5–18°C with temperatures frequently reaching 32°C in the summer (30 year average, US Climate Data 2017). Site Selection In 2006 The Nature Conservancy (TNC) initiated grassland restoration at the Preserve. One large grassland area (23-42 ha) was restored each year from 2006 to 2012 except for 2007. Each of the six areas was treated with glyphosate then seeded in the fall or winter with native bunchgrasses including P. spicata, P. secunda, and E. elymoides using a range drill. All grass seed used in the restoration was initially harvested from the preserve then cultivated over several years in small plots off-site to increase the amount of available seed. Seeds were then harvested from the offspring within the small plots and used for restoration. Projects were not irrigated. We established 18 sites across the

Preserve from three treatments of interest: one site in each of the six restoration areas, six sites in native grassland, and six sites in degraded grassland. Each restoration site was located approximately in the center of the restoration area. We chose all other sites by locating native and degraded habitat that were accessible by old farm roads and were on relatively flat slopes. All sites were separated on average by 623 m. Spider Sampling At each site, eight pitfall traps were placed in a 10 m radius circle (Fig. 2.2). Pitfall traps, 470 ml plastic cups filled halfway with wildlife-friendly propylene glycol and placed flush with the soil, are well-suited for collecting ground-active spiders (Martin 1978). Propylene glycol was used instead of water or ethanol mixtures to reduce the amount of evaporation. Spiders were collected at all sites three times each year from 2014–2016 during June-July, July-August, and August-September by opening traps for one week during each time period to collect invertebrates (Table 2.1). After one week, traps from each site were collected, combined, and transported to the laboratory. Samples were then washed over a 250 μm sieve and spiders sorted from other invertebrates and debris and preserved in 70% ethanol. All juvenile spiders were identified to family and all mature spiders were identified to species, if possible. Habitat Survey To determine which habitat characteristics were related to spider abundance, richness, and species composition, each site was surveyed for environmental variables once in 2014 and three times each year after, coinciding with spider sampling (Table 2.1). Vegetative variables were estimated to the nearest 5% cover in 16 63 x 63 cm subplots located in a 50 x 50 m square around the pitfall traps (Fig. 2.2). Variables estimated included the percent cover of invasive annual grasses (cheatgrass (B. tectorum) and medusahead (T. caput-medusae)), biological soil crusts (easily visible and distinct from the sandy soil texture), litter, and forbs. We also estimated maximum vegetation height by measuring the tallest stem in each subplot. Analyses – Comparing Native, Restored, and Degraded Sites Spider abundance at each site was characterized by the average number of spiders per pitfall as not all pitfalls were present during collection due to weather or animal

14 tampering. Abundance data were averaged over the 9 sampling bouts from 2014-2016. To compare taxa richness among samples that varied in abundance, rarified species richness estimates were generated for each site by calculating the Chao1 richness estimator using EstimateS, Version 9.1.0 (Colwell et al. 2012; Colwell 2013). The Shannon-Weiner index was used to estimate spider diversity of each site. Habitat variables for each site were calculated by first averaging subplots by site and then averaging over 7 sampling bouts (1 in 2014, 3 in both 2015 and 2016). We used one-way analysis of variance (ANOVA) to compare average spider abundance, rarified richness, Shannon-Weiner diversity, and habitat variables among the three treatments. Post hoc Fisher LSD tests were used to test for pairwise differences between treatments. All univariate analyses were completed in RStudio 1.0.153 (RStudio Team 2015). PC-ORD Software version 7.287 (McCune and Mefford 2015) was used for all community analyses including summary statistics. Family level abundance was used in all multivariate analyses instead of genus or species abundance to allow for use of juvenile spiders in the analysis (juvenile spiders cannot be positively identified to genus or species without introducing error). The family dataset contained average spiders per pitfall trap (18 sites for each year (54 total) x 10 families). The environmental dataset contained habitat variables (18 sites each year (54 total) x 5 variables) including: average maximum vegetation height and average percent cover of invasive annual grasses, litter, biological soil crust, and forbs. Data met all statistical assumptions and were not transformed. Non-metric multidimensional scaling (NMS) with Sorensen distances was used to ordinate sites in the spider family space matrix and the environmental matrix. NMS does not assume linearity between family response and environmental gradients and exposes relationships between the family matrix and the environmental matrix (McCune and Grace 2002). NMS was performed with 250 random starts and ties were not penalized. A randomization procedure was included to test if solutions were stronger than those obtained by chance, resulting in p-values. R2 values were calculated to represent the percent variance explained by each axis, and relationships of each axis with spider families and habitat variables were quantified with Pearson correlation coefficients.

Multi-response Permutation Procedures (MRPP) were used with Sorensen distances to test for differences in family composition across sites among groups, where each group was a treatment (restored, native, degraded). Pairwise comparisons resulted in A-statistics, the chance-corrected within-group agreement, and p-values. Analyses – Examining Age of Restoration on Response To examine relationships between age of restoration and average spider abundance, rarified richness, and habitat variables we conducted linear regressions with the expectation that we would see a positive or negative trend (depending on the variable) from sites with the most time to the least time to recover. To analyze abundance, richness, and habitat variables relationships with time, we used the average value across all three years for the regressions (N = 6). To examine how the age of restoration affects spider community composition, we conducted a separate NMS ordination of the six restoration sites (6 sites each year, for 18 total ordination points). If spider communities are influenced strongly by grassland restoration through time, we would expect to see patterns relative to the age of restoration that would be evident in the ordination, even given year-to-year variability. Ordination was conducted as described above, except only the five most abundant spider families were used (Gnaphosidae, Salticidae, Lycosidae, Theridiidae, Thomisidae) in the analysis as not all families were collected in the restored sites. We also examined one additional environmental variable (years since restoration).

Results We collected 2,752 spiders from 2014 to 2016 consisting of 10 families and 36 identified species. Of these, 2,253 were immatures, 358 were mature, and 141 were unknown (due to body damage). We collected 811 spiders (661 immatures, 125 matures) in 2014, 1,046 (834 immatures, 126 matures) in 2015, and 895 (758 immatures, 106 matures) in 2016. Approximately 1,040 spiders were collected in the degraded sites, 661 in the native sites, and 1,051 in the restored sites (Table 2.2). Of mature spiders collected, 55% were females in 2014, 52% were females in 2015, and 49% were females in 2016.

Comparing Native, Restored, and Degraded Sites Average spider abundance significantly differed among treatments (F (2, 15) = 10.2, p = 0.002), with degraded and restored sites having more spiders than native sites (Fig. 2.3a). Average rarefied richness also differed significantly among treatments (F (2, 15) = 3.8, p = 0.05) with degraded sites having fewer species than both native and restored sites (Fig. 2.3b). Unlike abundance and rarefied richness, average Shannon- Weiner diversity did not differ between treatments (F (2, 15) = 0.3, p = 0.74). Several habitat variables differed significantly among degraded, native, and restored sites (Fig. 2.4). Restored and degraded sites had more invasive annual grass cover than native sites (F (2, 15) = 15.7, p = 0.0002) (Fig. 2.4). Litter cover and biological soil crust cover showed similar trends (F (2, 15) = 32.2, p < 0.0001 and F (2, 15) = 44.1, p < 0.0001, respectively), with no difference between restored and degraded sites but significantly higher biological soil crust cover and lower litter cover in native sites (Fig. 2.4). There was no significant difference in average percent forb cover and maximum vegetative height among treatments (F (2, 15) = 0.4, p = 0.70 and F (2, 15) = 3.2, p = 0.07, respectively; Fig. 2.4). Community composition of spiders differed among sites (Fig. 2.5); the NMS randomization procedure resulted in a stable three-dimensional solution (final stress = 9.87, final instability = 0, P = 0.02) with a cumulative R2 of 0.94. Axis 1 accounted for 64% of the variation in spider family space, axis 2 accounted for 17.5%, and axis 3 accounted for 12%. Separation among treatments types occurred primarily on Axis 1, with restored and degraded sites almost completely overlapping on the left side of the ordination (Figs. 2.5a and 2.5b), while native sites only slightly overlapped with the degraded and restored sites on the right side of the ordination. Pearson correlations between the three axes and spider taxa and environmental variables are listed in Table 2.3. Lycosidae, Theridiidae, and Thomisidae were negatively correlated with axis 1 and, thus, were more common in degraded and restored sites and less common in native sites. MRPP showed that community composition differed significantly among treatments (A = 0.09, p <0.0001). Pairwise comparisons suggested that communities in native sites were different from both degraded and restored sites (A = 0.09, P = 0.0001 and A = 0.12, P <

0.0001, respectively); however, degraded and restored communities did not differ (A = - 0.006, P = 0.66). Several environmental variables strongly correlated with axis 1 including invasive annual grasses, litter, and biological soil crust cover (Table 2.3; Figs. 2.5a and 2.5b), indicating that axis 1 represents an environmental gradient varying from sites with higher cover of invasive grass and litter to sites dominated by biological soil crust cover. Forbs had a strong negative correlation with axis 2 and a strong positive correlation with axis 3. Axis 3 was also positively correlated with invasive annual grass and litter cover and slightly negatively correlated with biological soil crust cover (Table 2.3). The Effect of Age of Restoration on Response Average spider abundance tended to decrease with time since restoration (R2 = 0.26, p = 0.07) (Fig. 2.6a) and rarified richness tended to increase, but with greater variation (R2 = 0.15, p = 0.44) (Fig. 2.6b). Of all habitat variables, only average percent forb cover showed a significant positive correlation with time since restoration (R2 = 0.41, p = 0.03). Average percent invasive annual grass and litter cover, and maximum vegetation height tended to decrease with time since restoration while biological soil crust cover tended to increase, although trends were not statistically significant (R2 = 0.02, p = 0.54; R2 = 0.13, p = 0.14; R2 = 0.11, p = 0.17, R2 = 0.13, p = 0.14, respectively). The NMS randomization procedure resulted in a stable two-dimensional solution (final stress = 10.7, final instability = 0.00, P = 0.05) with a cumulative R2 of 0.91. Axis 1 was significantly positively correlated with Salticidae and negatively correlated with Theridiidae and Axis 2 was significantly positively correlated with Lycosidae and Thomisidae but negatively correlated with Theridiidae (Table 2.4). Correlations between axis 1 and habitat variables showed a positive significant correlation with time since restoration and no variable showed a significant correlation with axis 2 (Table 2.4). Thus, the oldest restored sites had more salticid spiders and fewer theridiid spiders.

Discussion Results of this study suggest that restoration of arid grasslands in the western US can be effective in influencing ecologically significant invertebrate groups, such as

18 spiders, so that communities more closely resemble those associated with relatively intact habitat. We found that spider communities associated with native sites, while less abundant, had a greater species richness and a different community composition compared to degraded sites. Restored sites displayed characteristics that were intermediate between native and degraded sites, with a high abundance of spiders and community composition similar to degraded sites, but with higher species richness characteristic of native sites. Some studies of spider responses to restoration in different systems have found similar results, detecting differences in spider communities in restored vs. undisturbed habitat (e.g., Longcore 2003; Perner & Malt 2003; Déri et al. 2011; Borchard 2014). However, unlike our study, Déri et al. (2011) found that species richness did not differ between degraded, restored, and native grassland habitat in Hungary. We also found evidence suggesting that, with time, spider communities in restored sites are becoming more like those in native sites. The chronosequence analysis showed a decreasing trend in spider abundance but a weaker trend in species richness with restoration age; small sample sizes may contribute to this weaker trend. Species richness in the oldest sites were similar to reference state levels, although spider abundance at those sites was still higher than reference sites. Spider community composition also varied relative to age, with older sites having more jumping spiders (Salticidae) and fewer cobweb spiders (Theridiidae). While 10 years for recovery may seem slow, other researchers have found similar results in dry grasslands, with plant and invertebrate communities taking more than a decade to reach reference states (Brand & Dunn 1998; Purtauf et al. 2004; Stadler et al. 2007). Spiders may respond more quickly in other habitat types; Cristofolio et al. (2010) and Borchard et al. (2014) found changes in spider communities within five years of restoration in European heathlands, and Déri et al. (2011) found changes in two to five years in European grasslands, although restorations were in early successional stages. Our study suggests that several habitat variables can be significant drivers of spider community responses to restoration in arid grasslands. One of the most important factors appears to be the presence of invasive annual grasses, like cheatgrass and

19 medusahead. These pervasive grasses appear to play a key role in structuring spider communities through their substantive influence on the amount of ground litter. High spider abundance in degraded and restored sites may be due to greater litter cover, which benefits certain groups of spiders. Some ground dwelling spiders rely on litter for hunting and studies have shown these groups increase with greater litter depth and complexity (Uetz 1979; Bell et al. 2001; Smith et al., in review). Spider families that were abundant in degraded and restored sites were ground hunters such as wolf spiders (Lycosidae) and crab spiders (Thomisidae) which have been found to prefer higher litter cover while hunting (Rypstra et al. 1999; Bell et al. 2001). While non-native annual grasses appear to be a major driver in this system, primarily through their effect on litter production, other factors, such as biological soil crusts, may also be adding to the complexity of restoring spider communities in these systems. Native sites had much higher cover of biological soil crusts than degraded or restored sites. A positive trend between biological soil crust cover and time since restoration indicates that crusts may reestablish over time but have not fully recovered within the decade. This is expected as biological soil crusts are predicted to reestablish anywhere from 14 to 250 years after a disturbance, with 14-35 years for cyanobacteria, 45-85 years for lichens, and 20 to 250 years for mosses (Belnap & Eldridge 2000). Well- developed biological soil crusts may inhibit the colonization of exotic annual grasses (Kaltenecker 1997), subsequently creating an area that lacks dense grass and litter cover. With less annual grass cover and established biological soil crusts within the interspace between bunchgrasses, the native sites are not only more resistant to grass invasions, but may be able to support a more diverse spider community that includes web spinners in addition to ground hunters (Bell et al. 2001; Reisner et al. 2013; Condon and Pyke 2018). For example, native sites tended to have more web spinner families such as sheet weavers (Linyphiidae), and cellar spiders (Pholcidae), all of which require larger webs to collect prey (Malumbre-Olarte et al. 2013). A higher proportion of ground spiders (Gnaphosidae), jumping spiders (Salticidae), and philodromid crab spiders (Philodromidae) (only found in native sites) were also found within the native habitat indicating that these families may also benefit from less litter cover during hunting.

While the majority of families were found within each of our treatments, Theridiidae was very abundant in all three habitats, suggesting these spiders are generalists that can hunt successfully in multiple habitats. While cover of non-native grass, litter, and biological soil crusts differed among treatments, we found no similar differences relative to percent cover of forbs or vegetation height. This is most likely due to the low percent of forbs across these grasslands, their patchy distribution, and the dominance of tall bunchgrasses in all habitats; however, we did see an increase in forb cover in restoration sites with time indicating that, with time, forbs are reestablishing. However, given their rarity, recovery of forbs and any effects they may have on spiders (e.g., flower hunters) may be slow. Similar to our study, Longcore 2003 found that vegetation between newly restored vs. disturbed and mature restored vs. undisturbed did not differ in coastal sage scrub in California. Restoration focused on enhancing invertebrate-mediated ecosystem services can present several challenges when restoring grassland habitat, as restoration does not always result in increasing ecosystem services and biodiversity (Bullock et al. 2011). For example, we found higher spider abundance in degraded sites compared to the native reference sites, but lower richness. This is most likely due to differences in litter and vegetative structure that favored one group of spiders. The majority of degraded sites were homogenous due to high invasion of invasive annual grasses that provide high- quality habitat for ground-active spiders that hunt in litter. However, degraded habitat may not provide the complex vegetative structure found in native sites that provide additional open ground and diverse vegetation (shrubs, grasses, and forbs) that lead to higher species richness. Given the increased abundance of spiders in degraded areas, if the sole objective of grassland management was to enhance pest control by generalist predators such as spiders, then restoration of degraded grassland habitats in this system may be unnecessary. However, this course of action would also result in a loss of biodiversity, not only in spiders, but also in native vegetation, and potentially diversity of other species and the services they provide. Our work suggests that there are tradeoffs involved in restoring grassland services related to spider communities and the larger

21 community, and these tradeoffs must be considered in restoration aimed at enhancing multiple services, including biodiversity. Our study also illustrates the importance of examining multiple high-quality reference site such as the intact native grassland used in our study and deemed necessary by SER (2004) and Ruiz-Jaen and Aide (2005) who recommended the use of more than one reference sites. The a priori assumption is often that native conditions will have greater abundance and diversity; this was not the case in our system, in which intact grasslands had lower spider abundance albeit greater species richness. Deconstructing the impact arid grassland restoration has on spider communities highlights how complex community responses can be to changes in the environment. By comparing degraded, native, and restored sites, we examine how spider communities and habitat respond to arid grassland restoration, a community response that has not been well documented. As much of the western United States moves to restore grasslands on both small and large scales, spiders are an important group to monitor in addition to the vegetation, to evaluate the success of restoration, not only because of their sensitivity to small changes but also because of the significant role they play in the biodiversity and ecological functioning of the ecosystem.

Acknowledgements We thank Leslie Nelson at TNC Boardman Grasslands for help in planning and implementing the study, and LJ Smith, SR Roof, EM Campbell, KL Kirby, L McDaniel, LK Waianuhea, and BE Price for their help in the field and laboratory. This work was supported by a USDA NIFA National Needs Graduate Fellowship (#2012-04150) and funding from Oregon State University’s General Research Fund and the Provost’s Branch Experiment Station Experiential Learning Program. Additional funding was provided by a USDA Western Sustainable Agriculture Research and Education Graduate Student Grant (#GW16-016), a TNC Oren Pollak Memorial Student Research Grant for Grassland Science, a Soil and Water Conservation Society Kenneth E. Grant Research Scholarship, the Prairie Biotic Small Research Grants Program, and a Society for Ecological Restoration Northwest Chapter Student Research Grant.

Literature Cited

Beck, J.L., Connelly, J.W., Wambolt, C.L., 2012. Consequences of treating Wyoming Big Sagebrush to enhance wildlife habitats. Rangeland Ecol Manag. 65, 444-455. Bell, J.R., Wheater, C.P., Cullen, W.R., 2001. The implications of grassland and heathland management for the conservation of spider communities: a review. J Zool. 255, 377-387. Belnap, J., Eldridge, D.J., 2000. Impacts of disturbance on biological soil crusts. In Belnap J, Lange O editors. Biological soil crusts: structure, function, and management. Springer-Verlap, New York, New York, USA. Borchard, F., Buchholz, S., Helbing, F., Fartmann, T., 2014. Carabid beetles and spiders as bioindicators for the evaluation of montane heathland restoration on former spruce forests. Biol Conserv. 178, 185-192. Brand, R.H., Dunn, C.P., 1998. Diversity and abundance of springtails (Insects: Collembola) in native and restored tallgrass prairies. Am Midl Nat. 139, 235-242. Bullock, J.M., Aronson, J., Newton, A.C., Pywell, R.F., Rey-Benayas, J.M., 2011. Restoration of ecosystem services and biodiversity: conflicts and opportunities. Trends Ecol Evol. 26, 541-549. Burel, F., Baudry, J., Butet, A., Clergeau, P., Delettre, Y., Le Coeur, D., Dubs, F., Morvan, N., Paillat, G., Petit, S., Thenail, C., Brunel, E., Lefeuvre, J., 1998. Comparative biodiversity along a gradient of agricultural landscapes. Acta Oecologica. 19, 47-60. Colwell, R.K., 2013. EstimateS: Statistical estimation of species richness and shared species from samples, Version 9.1.0. User’s guide and application Colwell, R.K., Chao, A., Gotelli, N.J., Lin, S., Mao, C.X., Chazdon, R.L., Longino, J.T., 2012. Models and estimators linking individual-based and sample-based rarefaction, extrapolation, and comparison of assemblages. J Plant Ecol. 5, 3-21. Condon, L.A., Pyke, D.A., 2018. Fire and grazing influence site resistance to Bromus tectorum through their effects on shrub, bunchgrass and biocrust communities in the Great Basin (USA). Ecosystems. Available online: https://doi.org/10.1007/s10021-018-0230-8. Cristofoli, S., Mahy, G., Kekenbosch, R., Lambeets, K., 2010. Spider communities as evaluation tools for wet heathland restoration. Ecol Indic. 10, 773-780. DeBano, S.J., 2006. Effects of livestock grazing on insect communities in semi-arid grasslands of southeastern Arizona. Biodivers Conserv. 15, 2547-2564. Déri, E., Magura, T., Horvath, R., Kisfali, M., Ruff, G., Lengyel, S., Tothmeresz, B., 2011. Measuring the short-term success of grassland restoration: the use of habitat affinity indices in ecological restoration. Restor Ecol. 19, 520–528. Gollan, J.R., Smith, H.M., Bulbert, M., Donnelly, A.P., Wilkie, L., 2010. Using spider

web types as a substitute for assessing web-building spider biodiversity and the success of habitat restoration. Biodiver Conserv. 19, 3141-3155. Gonzalez, N., DeBano, S.J., Kimoto, C., Taylor, R.V., Tubbesing, C., Strohm, C., 2013. Native bees associated with isolated aspen stands in Pacific Northwest Bunchgrass Prairie. Nat Area J. 33, 374-383. Huddleston, R.T., Young, T.P., 2004. Spacing and competition between planted grass plugs and preexisting perennial grasses in a restoration site in Oregon. Restor Ecol. 12, 546-551. Huddleston, R.T., Young, T.P., 2005. Weed control and soil amendment effects on restoration plantings in an Oregon grassland. West N Am Naturalist. 65, 507-515. Kaltenecker, J.H., 1997. The recovery of microbiotic crusts following post-fire rehabilitation on rangelands of the western Snake River Plain. Thesis, Boise State University, Boise, Idaho, USA Kelton, J.E.H., 1978. The Insects and Arachnids of Canada Part 1: Collecting, preparing, and preserving insects, mites, and spiders. Biosystemics Research Institute, Ottawa, Ontario. Kennedy, P.L., DeBano, S.J., Bartuszevige, A., Lueders, A., 2009. Effects of native and nonnative grassland plant communities on breeding passerine birds: implications for restoration of Northwest bunchgrass prairie. Restor Ecol. 17, 515-525. Kimoto, C., DeBano, S.J., Thorp, R.W., Rao, S., Stephen, W.P., 2012a. Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci. 12, 108, 23, available online: http://www.insectscience.org/12.108 Kimoto, C., DeBano, S.J., Thorp, R.W., Taylor, R.V., Schmalz, H., DelCurto, T., Johnson, T., Kennedy, P.L., Rao, S., 2012b. Livestock and native bee communities: short-term responses to grazing intensity and implications for managing ecosystem services in grasslands. Ecosphere. 3, 1-19. http://www.esajournals.org/doi/pdf/10.1890/ES12-00118.1 Knick, S.T., Dobkin, D.S., Rotenberry, J.T., Schroeder, M.A., Vander Haegen, W.M., Van Riper III, C., 2003. Teetering on the edge or too late? Conservation and research issues for avifauna of sagebrush habitats. Condor. 105, 611-634. Kovacs-Hostyanszki, A., Elek, Z., Balazs, K., Centeri, C., Falusi, E., Jeanneret, P., Penksza, K., Podmaniczky, L., Szalkovszki, O., Baldi, A., 2013. Earthworms, spiders and bees as indicators of habitat quality and management in a low-input farming region—A whole farm approach. Ecol Indic. 33, 111-120. Longcore, T., 2003. Terrestrial arthropods as indicators of ecological restoration success in coastal sage scrub (California, USA). Restor Ecol. 11, 397–409. Losey, J.E., Vaughan, M., 2006. The economic value of ecological services provided by insects. Bioscience. 56, 311-323.

Malumbres-Olarte, J., Vink, C.J., Ross, J.G., Cruickshank, R.H., Paterson, A.M., 2013. The role of habitat complexity on spider communities in native alpine grasslands of New Zealand: Habitat complexity and alpine spiders. Insect Conserv Diver. 6, 124–134. McCune, B., Grace, J., 2002. Analysis of Ecological Communities. MjM Software, Gleneden Beach, Oregon, USA. McCune, B., Mefford, M.J., 2015. PC-ORD Version 7.287. Multivariate Analysis of Ecological Data. MjM Software, Gleneden Beach, Oregon, USA. Mirshamski Kakhki, O., 2005. Faunistic study of spiders in Khorasan Province, Iran (Arachnida: Araneae). Int J Agric Biol. 1, 55-66. Mortimer, S.R., Hollier, J.A., Brown, V.K., 1998. Interactions between plant and insect diversity in the restoration of lowland calcareous grasslands in southern Britain. Appl Veg Sci. 1, 101–114. Nemec, K., Allen, C.R., Danielson, S.D., Helzer, C.J., 2014. Responses of predatory invertebrates to seeding density and plant species richness in experimental tallgrass prairie restorations. Agr Ecosyst Environ. 183, 11-20. Perner, J., Malt, S., 2003. Assessment of changing agricultural land use: response of vegetation, ground-dwelling spiders and beetles to the conversion of arable land into grassland. Agr Ecosyst Environ. 98, 169-181. Purtauf, T., Dauber, J., Wolters, V., 2004. Carabid communities in the spatio-temporal mosaic of a rural landscape. Landscape Urban Plan. 67, 185-193. R Development Core Team, 2015. R: A Language and Environment for Statistical Computing. Version 2.11.1 R Foundation for Statistical Computing, Vienna, Austria. Rao, S., Stephen, W.P., Kimoto, C., DeBano, S.J., 2011. The status of the ‘Red-Listed’ Bombus occidentalis (Hymenoptera: Apiformes) in Northeastern Oregon. Northwest Sci. 85, 64-67. Reisner, M.D., Grace, J.B., Pyke, D.A., Doescher, P.S., 2013. Conditions favouring Bromus tectorum dominance of endangered sagebrush steppe ecosystems. J Appl Ecol. 50, 1039-1049. RStudio Team, 2015. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA. Ruiz-Jaen, M.C., Aide, T.M., 2005. Restoration success: how is it being measured? Restor Ecol. 13, 569-577. Rypstra, A.L., Carter, P.E., Balfour, R.A., Marshall, S.D., 1999. Architectural features of agricultural habitats and their impact on the spider inhabitants. J Arachnol. 27, 371-377. Sampson, F., Knopf, F., 1994. Prairie conservation in North America. BioScience. 44, 418-421.

SER (Society for Ecological Restoration International Science and Policy Working Group), 2004. The SER International Primer on Ecological Restoration. Society for Ecological Restoration International, Tucson, Arizona. Smith, L.J., Smith DiCarlo, L.A., DeBano, S.J., In review. Spider family (Thomisidae) exhibits positive response to cheatgrass invasion (Bromus tectorum L.). Submitted to Biol Invasions. Stadler, J., Trefflich, A., Brandl, R., Klotz, S., 2007. Spontaneous regeneration of dry grasslands on set-aside fields. Biodivers Conserv. 16, 621-630. Tubbesing, C., Strohm, C., DeBano, S.J., Gonzalez, N., Kimoto, C., Taylor, R.V., 2014. Insect visitors and pollination ecology of Spalding’s catchfly (Silene spaldingii) in the Zumwalt Prairie of northeastern Oregon. Nat Area J. 34, 200-211. Uetz, G.W., 1979. The influence of variation in litter habitats on spider communities. Oecologia. 40, 29–42. US Climate Data, 2017. Climate Boardman – Oregon. Your Weather Service http://www.usclimatedata.com/climate/boardman/oregon/united-states/usor0036 (accessed 15 August 2017). Weisser, W.W., Siemann, E., 2004. The various effects of insects on ecosystem functioning. Ecol Stud. 173, 3-24. Wisdom, M.J., Rowland, M.M., Wales, B.C., Hemstrom, M.A., Hann, W.J., Raphael, M.G., Holthausen, R.S., Gravenmier, R.A., Rich, T.D., 2002. Modeled effects of sagebrush-steppe restoration on Greater Sage-Grouse in the interior Columbia Basin, USA. Society for Conservation Biology. 16, 1223-1231. Wheater, C.P., Cullen, W.R., Bell, J.R., 2000. Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecol. 15, 401–406. Zamani, A., Rafinejad, J., 2014. First record of the Mediterranean recluse spider Loxosceles rufescens (Araneae: Sicariidae) from Iran. J Arthropod-Borne Di. 8, 228-231.

Figure 2.1. Map of The Nature Conservancy Boardman Grasslands Preserve with an inset of the location in Oregon, USA. Circles represent degraded sites, triangles represent native sites, and stars represent restored sites.

Figure 2.2. Sample plot configuration, with 50 m survey square around 10 m pitfall radius. Closed circles represent pitfalls, closed squares represent habitat survey subplots.

3 14 b b a A a 13 B 12 11 10 2 b 9 8 a 7 6 1 5 4

Average Spiders per Trap per Spiders Average 3 2

Average Rarefied Species Species Richness Rarefied Average 1 0 0 Degraded Restored Native Degraded Restored Native

Figure 2.3. A. Average number of spiders per trap during the study in each treatment (N = 6 degraded, N = 6 native, N = 6 restored). B. Average rarefied species richness in each treatment compared by treatments (N = 6 degraded, N = 6 native, N = 6 restored) throughout the study. Different letters denote averages that differ significantly according to a post hoc Fisher LSD test.

Figure 2.4. Average percent cover of invasive annual grasses, litter, biological soil crusts (BSC), and forbs and average maximum vegetation height in cm compared by treatment (N = 6 degraded, N = 6 native, N = 6 restored). Different letters denote averages that differ significantly according to a post hoc Fisher LSD test.

Figure 2.5. A. Non-metric multidimensional scaling ordination of sites in spider family space along with weighted average positions for spider families for axes 1 vs. 2 (A) and axes 1 vs. 3 (B). Each triangle represents a site and each circle point represents a spider family. A joint-plot from the environmental matrix is overlaid with variables of r2 > 0.20 being displayed with vector lengths corresponding to the correlation strength along the axes (shown in thick black lines). For axes 1 vs. 2, a positive 5 degree rotation was necessary to make the vectors horizontal with the axis. For axes 1 vs. 3, a negative 10 degree rotation was necessary to make the vectors horizontal with the axis Convex hulls connect each group of treatments. Sites that are closer together are more similar than sites farther away from each other.

3.5 3 A 2.5 2 1.5 1

0.5 AverageTrap perSpiders 0 16 0 2 4 6 8 10 14 B Average Time Since Restoration (years) 12 10 8 6

4 RarefiedRichness 2 0 0 2 4 6 8 10 Average Time Since Restoration (years)

Figure 2.6. A. Average spiders per trap by average time since restoration in years. Each point represents a restoration site, restored between 2006 and 2012 (N = 6). B. Rarefied species richness by time since restoration in years (since 2016). Each point represents a restoration site, restored between 2006 and 2012 (N = 6).

Table 2.1. Dates of each sampling bout from 2014 to 2016. Pitfalls were opened on the first date of each bout and collected on the last date of each bout. Dates with stars included habitat surveys.

Sampling Dates Year Sampling Bout 1 Sampling Bout 2 Sampling Bout 3 2014 June 25 – July 2 August 15 – August 22* September 5 – September 12 2015 June 9 – June 16* July 14 – July 21* August 3 – August 10* 2016 June 23 – June 30* July 27 – August 3* August 30 – September 6*

Table 2.2. Total abundance and relative abundance (% in parentheses) of each spider family within each treatment. Spiders in the unknown category were due to bodily damage.

Family Degraded Native Restored Agelenidae 1 (0.1) 0 (0) 0 (0) Corinnidae 1 (0.1) 0 (0) 1 (0.1) Gnaphosidae 136 (13.1) 143 (21.6) 110 (10.5) Linyphiidae 4 (0.4) 10 (1.5) 0 (0) Lycosidae 172 (16.5) 67 (10.1) 198 (18.8) Philodromidae 0 (0) 2 (0.3) 0 (0) Pholcidae 2 (0.2) 8 (1.2) 9 (0.9) Salticidae 17 (1.6) 17 (2.6) 10 (1.0) Theridiidae 505 (48.6) 328 (49.6) 502 (47.8) Thomisidae 186 (17.9) 61 (9.2) 206 (19.6) Unknown 16 (1.5) 25 (3.8) 15 (1.4) Total 1040 (100) 661 (100) 1051 (100)

Table 2.3. Pearson correlation coefficients between habitat variables and spider families and each axis of the 3-dimensional NMS ordination. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 Axis 3 R R2 R R2 R R2 Invasive Annual -0.58 0.34 0.15 0.02 0.37 0.14 Grasses Habitat Litter Cover -0.64 0.40 0.12 0.02 0.30 0.09 Variables BSC Cover 0.68 0.46 -0.05 0.00 -0.22 0.05 Max Veg Height -0.20 0.04 -0.18 0.03 0.23 0.06 Forbs -0.10 0.01 -0.43 0.18 0.47 0.23 Agelenidae -0.08 0.01 -0.03 0.00 -0.01 0.00 Corinnidae -0.21 0.05 0.00 0.00 -0.03 0.00 Gnaphosidae 0.20 0.04 0.45 0.20 0.68 0.46 Linyphiidae 0.24 0.06 -0.29 0.08 -0.20 0.04 Spider Lycosidae -0.50 0.25 0.62 0.38 -0.09 0.01 Families Philodromidae 0.17 0.03 -0.10 0.01 -0.11 0.01 Pholcidae 0.08 0.01 -0.26 0.07 -0.05 0.00 Salticidae 0.20 0.04 0.02 0.00 -0.13 0.02 Theridiidae -0.87 0.76 -0.27 0.07 0.29 0.09 Thomisidae -0.64 0.40 0.50 0.25 -0.26 0.07

Table 2.4. Pearson correlation coefficients between habitat variables and spider families and the 2-dimensional NMS configuration of sites in family space. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 R R2 R R2 Time Since Restoration 0.44 0.19 0.05 0.00 Invasive Annual Grasses 0.04 0.00 -0.22 0.05 Habitat Litter Cover 0.19 0.03 0.21 0.04 Variables BSC Cover 0.29 0.08 0.22 0.05 Max Veg Height 0.17 0.03 0.13 0.02 Forbs 0.38 0.15 -0.04 0.00 Gnaphosidae -0.41 0.17 0.12 0.02 Lycosidae -0.40 0.16 0.75 0.57 Spider Salticidae 0.57 0.32 -0.28 0.08 Familes Theridiidae -0.82 0.66 -0.44 0.20 Thomisidae -0.40 0.16 0.65 0.43

CHAPTER 3. ARID GRASSLAND BEE COMMUNITY COMPOSITION, TEMPORAL PATTERNS, AND RESPONSES TO RESTORATION

Lauren A. Smith DiCarlo, Sandra J. DeBano, and Skyler Burrows

Abstract

In the past several decades multiple restoration projects have been initiated in the arid grasslands of the Pacific Northwest (PNW); however, little is known about native bee communities within these systems or how restoration efforts have affected them. Native bees provide an essential ecosystem service through their pollination of both neighboring crops and native plants and thus understanding their response to restoration is a high priority. To address this issue, we conducted a three-year study in an arid PNW bunchgrass prairie with the following objectives: (1) describe the bee community of this unique PNW grassland type and its temporal variability; (2) investigate environmental factors that influence the bee community; and (3) examine the effects grassland restoration has on the bee community. We identified 62 native bee species over the course of the study, and found strong seasonal and inter-annual patterns in bee abundance, richness, and species composition. Unexpectedly, temporal trends in bee abundance, richness, and diversity did not match trends in floral resources but both forb abundance and richness did influence bee community composition. Other environmental factors that influenced the bee community included vegetation height, and the cover of invasive annual grasses, litter, and biological soil crust. We found no detectable effect of restoration on bee abundance, richness, diversity, or composition but we did find that the species composition of bees at native sites differed from both those in restored and degraded sites, which did not differ from each other. We suggest that future restoration efforts should focus not only on providing adequate floral resources for bee communities but also on enhancing environmental factors that may influence nesting resource quality.

Introduction

Since European settlement up to 99.9% of grasslands in North America have been destroyed or degraded, primarily due to agricultural conversion (Sampson and Knopf 1994; Burel et al. 1998). Today grasslands are a top priority for restoration as they support high levels of biodiversity and many threatened and endangered species (Kennedy et al. 2009; Rao et al. 2011; Tubbesing et al. 2014). However, the majority of studies on grassland restoration have focused either on vegetation or vertebrate responses (e.g., Wisdom et al. 2002; Knick et al. 2003; Huddleston & Young 2004; Huddleston & Young 2005; Beck et al. 2012), even though grassland invertebrates comprise the majority of grassland biodiversity and provide important ecosystem services such as pollination, nutrient cycling, food for vertebrates, and pest control (Weisser & Siemann 2004; DeBano 2006; Kimoto et al. 2012; Gonzalez et al. 2013). In addition, unlike plants and vertebrates, our understanding of restoration effects on these systems is hampered by a basic lack of knowledge of the species that make up invertebrate communities and how they vary temporally. Thus, there is a pressing need to improve our understanding of invertebrate communities that inhabit these grasslands and examine how grassland restoration influences their diversity and function (Nemec et al. 2014), particularly in North American grasslands. One group that is of growing conservation interest is bees. With an estimated 20,000 species occurring globally, bees are one of the most diverse groups of insects and are important ecosystem service providers that assist in pollinating approximately 80% of native plants as well as 35% of crop species, a service estimated to be worth billions of dollars each year (Klein et al. 2007; Losey and Vaughan 2006; Michener 2007; Winfree et al. 2009; Potts et al. 2010). In addition, many rare and sensitive plant species in grasslands rely on native bee pollination (Kimoto et al. 2012; Tubbesing et al. 2014); however, even though bees are the most significant pollinators in many ecosystems(Exeler et al. 2009), relatively little is known about native bee communities, especially in many North American grasslands. In the last decade, documented declines in honey bees and native bees has resulted in an increase in pollinator conservation and a need to learn more about this important group (Brown and Paxton 2009; Pettis and Delaplane 2010; Potts et al. 2010).

In the face of this need, it is increasingly important to expand the focus of grassland restoration to ensure that it benefits native bee communities; however, this requires an understanding of the environmental variables that most strongly influence bee communities and how they respond to restoration. Past research has shown that characteristics of plant communities can have large effects on bee communities, with floral abundance and diversity often positively related to native bee abundance and richness (Potts et al. 2003; Hopwood 2008; Kwaiser and Hendrix 2008). In addition, factors related to nesting availability can also affect native bee communities; for example, the amount of bare ground and soil compaction can influence ground nesters, and dead woody vegetation provides habitat for above-ground nesters (Cane 1991; Hopwood 2008; Grundel et al. 2010; Potts et al. 2010). Further, bee communities show strong temporal variation, both seasonally and annually (Williams et al. 2001; Kimoto et al. 2012), making repeated measurements essential to comprehensive descriptions of native bee communities and to understanding their responses to restoration. Several studies have examined pollinator communities of North American grasslands but the majority have taken place in the short and tallgrass prairies of the Great Plains (e.g., Reed 1995; Davis et al. 2008; Kwaiser and Hendrix 2008). Bunchgrass prairies of the Pacific Northwest are one of the most threatened grassland systems and are particularly understudied even though historically they covered over eight million hectares within the United States and Canada (Tisdale 1982). Only two studies have focused on bee communities in the bunchgrass prairie habitat of the Pacific Northwest (Kimoto et al. 2012; Rhoades et al. 2017) and these studies found key differences in the dominant genera of each community, illustrating how diverse bunchgrass prairies can be, potentially because of differences in local climate, vegetative communities, and bee communities. In addition, because both Kimoto et al. (2012) and Rhoades et al. (2017) studied grasslands at higher elevations, we still know very little about bee communities in lower elevation, drier systems that are very common in the interior of the Pacific Northwest; thus, this study aims to assess the bee community at an arid grassland at low elevation in an agriculturally dominated landscape.

We had three main objectives: (1) describe the native bee and blooming floral communities of this unique PNW grassland type (including community composition) and their temporal variability (both seasonally and year-to-year); (2) investigate environmental factors influencing the bee community; and (3) examine effects of grassland restoration on native bees. We expected to see strong temporal trends within the bee community associated with variation in floral resource abundance and diversity, as well as with environmental variables associated with nesting sites. Finally, we expected that, to the degree that restoration influenced these environmental variables, we would see a concomitant response of the bee community to restoration.

In 2006 The Nature Conservancy (TNC) initiated grassland restoration at the Preserve. One large grassland area (23-42 ha) was restored each year from 2006 to 2012 except for 2007. Each of the six areas was treated with glyphosate then seeded in the fall or winter with native bunchgrasses including P. spicata, P. secunda, and E. elymoides using a range land drill. All grass seed used in the restoration was collected from the site, increased in plots over several generations, and then seeded back at the site. Projects were not irrigated. We established 18 sites across the Preserve from three treatments of interest: one site in each of the six restoration areas, six sites in native grassland, and six sites in degraded grassland. Each restoration site was located approximately in the center of the restoration area. We chose all other sites by locating native and degraded habitat that were accessible by old farm roads and were on relatively flat slopes. All sites were separated on average by 623 m to the nearest neighbor. Bee Sampling Bees were sampled using pan traps, 236 ml white plastic cups, with one third left white and one third each painted UV reflective blue and yellow to attract different pollinators. All pan traps were elevated approximately 1 m above ground and filled with a solution of water and detergent. UV reflective pan traps are an efficient method to collect native bees in grasslands, especially if mounted at vegetation height (Westphal et al. 2008; Geroff et al. 2014). Nine pan traps were placed at each site in a 10 m radius circle, with one in the center (Fig. 3.2). Bees were collected during three sampling bouts (June-July, July-August, August-September) each year from 2014-2016, except for the June-July bout in 2014. Pan traps were collected in the field after two days and all traps from each site were combined and transported to the laboratory where bees were washed, dried, pinned, and identified. All specimens were identified to species, if possible. Habitat and Floral Resource Surveys To determine which habitat characteristics were related to bee abundance, richness, and species composition, each site was surveyed for environmental variables once in 2014 during the July-August bout and three times each year after, coinciding with bee sampling. Variables were estimated in 16 63 x 63 cm subplots located in a 50 x 50 m square around the pitfall traps (Fig. 3.2) and included percent cover (to the nearest 5%) of

38 invasive annual grasses (cheatgrass (B. tectorum) and medusahead (T. caput-medusae)), biological soil crusts, and litter. We also estimated maximum vegetation height by measuring the tallest stem in each subplot. To estimate available floral resources, every blooming stem was counted and identified in the larger 50 x 50 m square (Fig. 3.2). Analyses – Characterizing bee and floral communities and their temporal variability Bee abundance at each site was estimated by the average number of bees per pan trap as not all traps were present during collection due to weather or animal tampering. If traps dried out over the two days, traps were counted as one half trap as they most likely collected bees one out of the two days. Because sites varied in their abundance of bees, we used rarified species richness estimates by calculating a Chao1 richness estimator for each site using EstimateS, Version 9.1.0 (Colwell et al. 2012, Colwell 2013). We used the Shannon-Weiner index to describe diversity. PC-ORD Software version 7.287 (McCune and Mefford 2015) was used to characterize bee community structure. Species level abundance was used in all multivariate analyses. The species dataset contained average bees per pan trap (18 sites for each sample period (144 total) x 62 species). The environmental dataset contained habitat variables (18 sites for each sample period (144 total) x 6 variables) including average maximum vegetation height, forb abundance, forb richness and average percent cover of invasive annual grasses, litter, and biological soil crust. Non-metric multidimensional scaling (NMS) with Sorensen distances was used to ordinate sites in the bee species space matrix and the environmental matrix. NMS does not assume linearity between species response and environmental gradients and exposes relationships between the species matrix and the environmental matrix (McCune and Grace 2002). NMS was performed with 250 random starts and ties were not penalized. A randomization procedure was iused to test whether solutions were stronger than those obtained by chance, resulting in p-values. R2 values were calculated to represent the percent variance explained by each axis, and relationships of each axis with bee species and habitat variables were quantified with Pearson correlation coefficients. To examine temporal variability in abundance, richness, and diversity of the bee community, we used one-way analyses of variance (ANOVA) to compare seasons and

39 years. Post hoc Tukey HSD tests were used to test for pairwise differences between seasons or years. Multi-response Permutation Procedures (MRPP) were used with Sorensen distances to test for differences in species composition among seasons and years. Because sampling was not conducted in the early-season of 2014, we only used mid- and late-season sampling for the year-to year comparison. Pairwise comparisons resulted in A-statistics, the chance-corrected within-group agreement, and p-values. To examine temporal variability in abundance and richness of the forb community, we used one-way analyses of variance (ANOVA) to compare seasons and years. Post hoc Tukey HSD tests were used to test for pairwise differences between seasons or years. As with the bees, we only used mid- and late-season sampling for the year-to-year comparison. To investigate whether temporal trends in bee communities (abundance, richness, and diversity) were tied to floral resource availability we used linear regressions on values for each variable averaged over the 8 sampling bouts. All univariate analyses were completed in RStudio 1.0.153 (RStudio Team 2015). Analyses – Environmental factors influencing the bee community To determine which environmental factors influence the bee community, NMS was used as described above to determine which environmental variables are correlated to community structure. In addition, we used regression tree analyses to examine how habitat variables affected bee abundance, rarefied richness, and diversity. Regression tree analysis is a powerful approach suitable for examining relationships between multiple, potentially interacting, independent variables and a dependent variable (De’ath and Fabricius 2000). Regression trees repeatedly divide data into two or more groups, each as homogenous as possible, and then divides each of the resulting groups in the same manner, in an iterative process that continues until pre-specified stopping criteria are met. We used a least-squared loss function, which minimizes the sum of the squared deviation. For stopping criteria, we specified a maximum number of splits as 10, a minimum proportion reduction in error (PRE) of 0.05, a minimum split value of 0.05, and a minimum count allowed at any node of 5. We included average maximum vegetation height, forb abundance, forb richness, and average percent invasive annual grass, litter, and biological soil crust cover as potential predictor variables, and constructed separate

40 trees for bee abundance, rarefied richness, and diversity. Regression tree models were conducted using SYSTAT v. 13 (Systat Software 2009). Analyses – Effects of restoration on the bee community To investigate the effect of restoration on the bee community, we averaged abundance, richness, and diversity data over the 8 sampling bouts from 2014-2016. Habitat variables for each site were calculated by first averaging subplots by site and then averaging over 7 sampling bouts (1 in 2014, 3 in both 2015 and 2016). We used one-way analysis of variance (ANOVA) to compare average bee abundance, rarefied richness, Shannon-Weiner diversity, and floral resources (bloom abundance and richness) among the three treatments. Post hoc Tukey HSD tests were used to test for pairwise differences between treatments. NMS was used, as described above, to examine community structure relative to each restoration treatment and MRPP was used to test for differences among those treatments. Indicator Species Analysis was performed to assess species-specific associations with treatments. Resulting indicator values ranged from 0 to 100, with higher scores indicating stronger associations between species and treatments. A randomization test was used to test for statistical significance of the indicator values by using 4999 random permutations, resulting in a p-value for each indicator value. Species with high indicator values and significant p-values (alpha <0.05) were considered indicative of that treatment.

Results

Characterizing bee and floral communities and their temporal variability From 2014-2016, 12,996 bees were collected at the Preserve (4,109 in 2014, 3,831 in 2015, and 5,056 in 2016) (Table 3.1). Approximately 90% of the bees were females and 10% were males; no queens were collected. Sixty-two species were collected from 20 genera and five families (Andrenidae, Apidae, Colletidae, Halictidae, and Megachilidae), with 39 species found in 2014, 30 in 2015, and 43 in 2016 (Table 3.2). Approximately 82% of bees were halictids in the genera Agapostemon (51%) and

Lasioglossum (31%). Another 13% of the bee community were apids, with Melissodes making up 10% of all bees collected. Five species comprised 88% of all individuals collected: A. texanus (39%), unidentified taxa of Lasioglossum in the Dialictus subgenus (31%), A. femoratus (9%), M. bimatris (5%), and A. virescens (4%). The most species rich family was Apidae with 29 species that included the most diverse genus, Melissodes, with 11 species. The next most specious family was Megachilidae, with the genus Megachile having 7 species. Most species were collected during multiple years and sampling periods; however several species were rare in that they were only collected during one sampling bout. These included Perdita lingualis, Melissodes perlusa, Megachile umatillensis, Megachile wheeleri, and all Nomada, Colletes, Sphecodes, and Osmia species. The NMS ordination revealed that bee community composition varied among sites, with a stable three-dimensional solution (final stress = 12.7, final instability = 0, P = 0.004) and a cumulative R2 of 0.78. Axis 1 accounted for 42.8% of the variation in bee species space, axis 2 accounted for 19.6%, and axis 3 accounted for 15.3%. Pearson correlations between the three axes and bee taxa and environmental variables are listed in Table 3.3. Agapostemon, Andrena, Bombus, Halictus, Lasioglossum, Melissodes, and some species of Megachile were positively correlated with axis 1 and Apis, Colletes, Melissodes, Svastra and other species of Megachile were negatively correlated with axis 1 (Fig. 3.3, Table 3.3). Agapostemon, Halictus, Perdita, Triepeolus, and some species of Lasioglossum and Melissodes were positively correlated with axis 2 while Osmia and some species of Lasioglossum and Melissodes were negatively correlated with axis 2 (Fig. 3.3, Table 3.3). The bee community varied both seasonally and among years (Table 3.1). Average abundance among seasons differed significantly (F = 5.3, p = 0.008) and was 18.8 ± 5.9 in the early-sampling period (2015-2016), 30.7 ± 4.7 in the mid-sampling period (2014-2016), and 10.1 ± 1.8 in the late-sampling period (2014-2016). Pair-wise comparisons showed one significant difference in means; mid-season abundance was greater than late-season abundance. Average richness among seasons also differed significantly (F = 4.5, p = 0.02), and was 23.4 ± 4.3 species in the early-sampling period

(2015-2016), 25.5 ± 1.8 in the mid-sampling period (2014-2016), and 14.3 ± 1.3 in the late-sampling period (2014-2016), with early and mid-season having higher rarefied richness than late-sampling richness. Average diversity among seasons did not differ (F = 3.1, p = 0.06) and was 1.2 ± 0.1 in the early-sampling period (2015-2016), 1.2 ± 0.0 in the mid-sampling period (2014-2016), and 1.1 ± 0.0 in the late-sampling period (2014- 2016). While the same three genera always dominated, the relative importance of each varied with sampling bout. In the early-season sampling period, Lasioglossum was the dominant genus, comprising 63% of all individuals collected, followed by Agapostemon (22%) and Melissodes (9%). In mid-season, Agapostemon dominated (67% of all individuals), followed by Lasioglossum (26%) and Melissodes (3%). For the late- sampling period, Agapostemon continued to dominate at 36% but was closely followed by Melissodes (33%), and then Lasioglossum (18%). Species composition of the community also varied seasonally (Fig. 3.3a) and MRPP showed that community composition significantly differed relative to season (A = 0.10, p < 0.001), with the species composition of each season significantly differing from each other (all pairwise p-values < 0.001). Sampling bouts separated mainly along the axis 1 with Agapostemon, Andrena, Bombus, Halictus, Lasioglossum, some Megachile and Melissodes species, Perdita, and Sphecodes genera more common in early-sampling bouts, while Apis, Colletes, some Megachile, Melissodes, and Svastra genera were more common in late- sampling bouts. Mid-sampling bouts appeared to be a mix of the two groups (Fig. 3.3b). Abundance for each species and presence/absence over the eight sampling bouts are shown in Table 3.2. Overall year-to-year variation was also substantial for bees. When considering the average number of bees collected at each site each year for mid- and late-season sampling periods (early-season was not considered in this analysis because it was not conducted in 2014), we found abundance varied significantly (F = 14.5, p < 0.0001), with 12.7 ± 1.6 bees per pan trap in 2014, 37.8 ± 6.4 in 2015, and 10.7 ± 2.0 in 2016. Pair-wise comparisons showed that the abundance of bees in 2015 was higher compared to 2014 and 2016. However, 2015 also had many dried out traps which were counted as a half trap as they likely only collected bees one out of the two days in the field. If traps still

43 collected a large number of bees during that one day, the bee abundance would be greater than at a location that collected the same number of bees but the traps were open for two days (i.e. were not dry). Likewise, bee richness also differed significantly from year-to- year (F = 7.0, p = 0.002), with 19.4 ± 0.8 bees per site in 2014, 11.8 ± 1.6 in 2015, and 16.0 ± 1.8 in 2016, with pair-wise comparisons showing rarefied richness being significantly greater in 2014 compared to 2015. Bee diversity also differed significantly through the years (F = 14.7, p < 0.0001), with a diversity index of 1.3 ± 0.1 per site in 2014, 0.9 ± 0.1 in 2015, and 1.2 ± 0.1 in 2016. Pair-wise comparisons showed diversity was significantly greater in 2016 and 2014 compared to 2015. Species composition of the community also varied by year (Fig. 3.3b) and MRPP showed that community composition significantly differed among all years (A = 0.07, p = 0.00; all pairwise p- values < 0.001). Years separated mainly along axis 1. Agapostemon, Andrena, Bombus, Halictus, Lasioglossum, some Megachile and Melissodes species, Perdita, and Sphecodes appeared to be common in 2015 while Apis, Colletes, some Megachile, Melissodes, and Svastra were more common in 2014 and 2015 (Fig. 3.3b). Throughout the course of the study, we counted and identified 17,363 blooming stems from 28 different plant species, with 11 species in 2014, 14 in 2015, and 23 in 2016 (Table 3.4). Nineteen species were from Asteraceae, which also made up the majority of blooming stems observed (51%). The most prominent blooming species was Polygonum douglasii Greene (Douglas’ knotweed) which accounted for 47% of all blooming stems counted, but was only recorded blooming late in the season of 2015 and throughout the 2016 season. Chrysothamnus viscidiflorus (Hook.) Nutt. (yellow rabbitbrush) was the next most abundant blooming species, comprising 25% of all blooming stems counted, followed by Centaurea solstitialis L. (yellow star-thistle) at 17%. Only Centaurea solstitialis L. (yellow star-thistle) and Cirsium undulatum (Nutt.) Spreng. (wavyleaf thistle) were observed blooming during every sampling bout, with the majority of other Asteraceae species blooming during the late-sampling periods and other families blooming more sporadically but mainly during the 2016 season. Floral resources also showed substantial temporal patterns relative to season and year (Table 3.1). Average abundance (stems with blooms) among season differed

44 significantly (F = 3.1, p = 0.05) and was 37.7 ± 14.3 per site in the early-sampling period (2015-2016), 86.8 ± 27.4 in the mid-sampling period (2014-2016), and 159.5 ± 51.9 in the late-sampling period (2014-2016). Pair-wise comparisons showed one significant difference in means; late-season abundanc ewas greater than early-season abundance. Likewise, average seasonal richness also differed significantly between sampling periods (F = 5.0, p = 0.01), with 1.5 ± 0.3 in the early-sampling period (2015-2016), 1.3 ± 0.2 in the mid-sampling period (2014-2016), and 2.3 ± 0.2 in the late-sampling period (2014- 2016), with pair-wise comparisons showing two significant differences in means, rarefied richness was significantly greater in the late-season compared to both the early and mid- sampling periods. Floral resources also showed year-to-year variability (Table 3.1). When considering the average number of blooming stems counted at each site each year for mid- and late-season sampling periods (early-season was not considered in this analysis because it was not conducted in 2014), we found that average abundance varied significantly (F = 6.0, p = 0.005), with 80.1 ± 36.8 in 2014, 20.4 ± 7.9 in 2015, and 269.0 ± 84.1 in 2016. Pair-wise comparisons showed two significant differences in means; average abundance was greater in 2016 than both 2014 and 2015. Average richness varied significantly between years (F = 18.2, p < 0.0001)), with 1.2 ± 0.2 in 2014, 1.0 ± 0.2 in 2015, and 3.22 ± 0.4 in 2016. Like, average abundance, pair-wise comparisons showed two significant differences in means with average rarefied richness greater in 2016 than both 2014 and 2015. We found no statistically significant temporal relationships between floral abundance and bee abundance, richness, or diversity (r2 = 0.35, p = 0.12; r2 = 0.01, p = 0.55; r2 = 0.06, p = 0.84, respectively, N = 8 for all regressions) or floral richness (r2 = 0.48, p = 0.06; r2 = 0.01, p = 0.99; r2 = <0.001, p = 0.77, respectively, N = 8 for all regressions). Environmental factors influencing the bee community Several environmental variables strongly correlated with axis 1 including invasive annual grasses, biological soil crusts, and forb abundance (Table 3.3; Fig. 3.3), indicating that axis 1 represents an environmental gradient varying from sites with higher

45 cover of biological soil crust and forb abundance to sites dominated by invasive grass cover. Invasive annual grasses, litter, and maximum vegetation height had also had a strong negative correlation with axis 2. Axis 3 was positively correlated with invasive annual grass cover and forb richness while negatively correlated to biological soil crusts (Table 3.3). Regression tree analysis of average bee abundance produced a tree with two terminal nodes that explained approximately 46.9% of the variation (Fig. 3.4a). The first and only branching formed relative to litter cover, with sites with higher litter cover having more bees than sites with lower litter cover. Regression tree analysis of rarefied bee richness produced a tree with three terminal nodes that explained 25.9% of the variation in diversity (Fig. 3.4b). The first two branches split relative to maximum vegetation height, with sites with taller vegetation having higher species richness than sites with shorter vegetation. Sites with shorter vegetation showed additional branching relative to floral abundance; sites with higher average floral abundance had more species and sites with less forb abundance had fewer species. Regression tree analysis of average bee diversity produced a tree with two terminal nodes that explained 32.2% of the variation (Fig. 3.4c). Separation occurred relative to floral abundance, with sites with greater floral abundance having less bee diversity compared to sites with less floral abundance. Effects of restoration on the bee community Restoration treatments did not have a statistically significantly effect on average bee abundance, diversity (Shannon-Weiner Index), or rarefied species richness (F (2, 15) = 1.00, p = 0.39; F (2, 15) = 0.74, p = 0.49; F (2, 15) = 0.002, p = 1.00, respectively) or average floral abundance or richness (F (2, 15) = 3.27, p = 0.07 F (2, 15) = 0.09, p = 0.91, respectively). MRPP also showed that the community significantly differed among treatments (A = 0.01, p = 0.02) with degraded and restored sites differing from native sites (A = 0.01, p = 0.03; A = 0.01, p = 0.03, respectively) but not differing from each other (A = 0.00, p = 0.23). Treatments appeared to separate somewhat along Axis 2 with native sites spanning most of the ordination and degraded and restored sites overlapping in the center

(Fig. 3.3c). Multiple species were found to be indicators of the degraded treatment including A. texanus (IV = 48, p = 0.01), H. ligatus (IV = 13, p = 0.02), M. pallidisignatus (IV = 15, p = 0.03), and Triepeolus grinelidae (IV = 14, p = 0.05). Anthophora urbana was found to be an indicator species of the native treatment (IV = 28, p = 0.01) and M. rivalis was an indicator of the restored treatment (IV = 23, p = 0.04).

Discussion Results of this three-year study demonstrate that bee communities, even in a relatively limited region – the bunchgrass prairie of the interior PNW, can differ strongly. The bee community associated with the arid grassland in this study was not only distinctive from other bee communities in the region, it also showed significant temporal variation both seasonally and inter-annually. This temporal variability did not correspond with overall variability in floral resource availability, suggesting that other drivers underlie temporal patterns. However, we identified several environmental variables that may impact foraging or nesting that were strongly related to spatial variation in the grassland, including floral abundance, vegetative structure and the cover of invasive grass and litter. Our work also suggests that grassland restoration has not resulted in strong responses in the floral or bee communities in this area. Only two other studies have assessed bee communities within western North American bunchgrass prairie (Kimoto et al. 2012; Rhoades et al. 2017). Although we collected almost twice as many bees as Kimoto et al. (2012) and nearly as many bees as Rhoades et al. (2017), our sites had only one to two-thirds the number of species, indicating that the bee community associated with this grassland differs substantially from other bunchgrass habitat in the region. Differences are also evident relative to community composition. In our study, the most dominant genera were Agapostemon, Lasioglossum, and Melissodes, while Kimoto et al. (2012) found a community dominated by Bombus, Lasioglossum, and Melissodes and Rhoades et al. (2017) found a community dominated by Lasioglossum, Halictus, and Agapostemon. However, our total species richness is relatively similar to another study that identified only 56 species in native and degraded grasslands in a tallgrass prairie in an agriculturally dominated landscape in

Iowa (Kwaiser and Hendrix 2008). Several reasons may explain the relatively depauperate community found at this arid bunchgrass including the fact that, unlike the prairie site in Kimoto et al. (2012), the preserve has large areas of degraded grasslands that had been previously cultivated and are highly invaded with annual grasses. Yet, the prairie site studied by Rhoades et al. (2017) is also largely fragmented due to agriculture and invaded by annual grasses (Rhoades 2016). However, their sampling took place within 5 m of thickets of shrubs and small trees, while our sites were not in the vicinity of any shrub or tree thickets. In addition, our location is lower in elevation than the other two studies (200 m compared to > 1,200 m) and experiences higher summer temperatures (up to 32°C compared to 22 - 27°C) and less precipitation (22 cm compared to 25 - 76 cm). The importance of considering temporal variability in bee communities is well recognized (Williams et al. 2001, Michener 2007, Kimoto et al. 2012) and we found that communities differed greatly year-to-year and also seasonally. Year-to-year variation in species composition may be partly explained because we did not collect early-season bees in 2014 but 2015 and 2016 also differed from each other. This is most likely due to a wildfire that burned a large portion of the preserve prior to sampling in 2015. While the fire only burned five plots, it likely had a large impact on the bee community in both 2015 and 2016. For example, we found that bee community composition differed significantly in the burned and unburned sites in 2016 (Smith DiCarlo et al., in review). Seasonal variation may be driven by differences in weather and the phenologies of different bee and blooming plant species (Kimoto et al. 2012). For example, many Melissodes are well-known aster specialists and dominated in the late-season when the majority of the asters were blooming. In contrast, Agapostemon and Lasioglossum are mostly generalists and dominated each season with some variation in abundance. The relationship between floral abundance and diversity and bee abundance and diversity is often conveyed in the popular literature as being strongly and positively correlated, and indeed, a number of empirical studies have found positive relationships between some of these metrics (e.g., Tscharntke et al. 1998; Potts et al. 2003). However, the relationship between these variables are complex, and patterns vary, likely due to a

48 number of factors, including the system and whether temporal or spatial variability is examined. We did not see any statistically significant relationships beween bee abundance, richness, or diversity and floral abundance or richness through time. Floral resources undoubtedly are a key factor structuring patterns in bee communities but other studies have also found that the bee abundance, richness, and diversity are not as closely tied to floral resources as originally hypothesized, with multiple studies finding no correlation between floral and bee abundance and diversity (Grundel et al. 2010; Kimoto et al. 2012; Wood et al. 2015). This finding highlights how complex bee community responses can be to changes in the environment and necessitates the study of additional environmental variables that relate not only to foraging but also nesting. Each measured environmental factor was found to be strong driver of bee community composition including invasive annual grasses, litter, biological soil crusts, maximum vegetation height, and forb abundance. Common genera that were closely associated with invasive grass and litter cover included generalist sweat bees from Agapostemon, Halictus, and Lasioglossum, while areas of high biological soil crusts were associated with Megachile and Melissodes. Invasive grass contributes large amounts of litter in this system (Smith et al., in review) with cheatgrass and medusahead creating a thick thatch, while areas without invasive grasses have more biological soil crusts and bare ground. While the majority of ground nesters are known to prefer areas of bare ground (Hopwood 2008), some bees, mostly specialists are also known prefer areas with more litter cover (Cane 1994; Grundel et al. 2010). However, Rhoades (2016) found grass invasions reduced species richness of ground nesting bees. In contrast, our regression tree analysis showed that sites with more litter had more bees. While tall vegetation can hinder bee foraging by reducing visibility and concealing shorter blooms (Cook et al. 2011), we found higher bee richness with taller vegetation. This may be due to tall flowering plants including Douglas knotweed, sunflowers, and several species of thistles that were taller than invasive grass species and attracted more bee species. In areas with shorter vegetation, sites with more blooming flowers had more bee species. This may be because bees could forage more efficiently with shorter vegetation to navigate. In contrast, increased floral abundance showed a negative relationship with bee

49 diversity, with sites with more blooming flowers having less bee diversity. This result is consistent with other studies (Grundel et al. 2010; Wood et al. 2015) and may be due to mast flowerings of several species including Douglas’ knotweed and yellow star-thistle in several sites. While yellow star-thistle mast flowerings mainly took place in the restored sites and was present throughout the entire study, Douglas’ knotweed invaded after the 2015 fire and mainly mast-flowered within the burned sites. While these mast flowerings provided extra forage, they did not provide diverse floral resources that would attract a diverse set of bees including both generalists and specialists. Thus to greater diversify the bee community, a greater diversity of forage may be necessary to enhance bee diversity (Tscharntke et al. 1998; Potts et al. 2003; Hopwood 2008; Grundel et al. 2010). Surprisinlgy, we did not find strong patterns relative to restoration treatment. While the bee community at degraded and restored sites did differ from native sites, the bee community at degraded and restored sites did not differ from each other. In addition, restoration treatment did not impact abundance, richness, or diversity. This is contrary to findings in other grassland restoration studies that found rapid changes in response to restoration with higher richness and abundance compared to degraded sites (Hopwood 2008; Kwaiser and Hendrix 2008; Exeler et al. 2009); however, one study found no difference in richness or abundance in restored hay meadows (Forup and Memmott 2005). Given the lack of difference between the restored and degraded treatments, it appears that restoration is not currently affecting the bee community in this area. While abundance and richness of degraded and restored sites did not differ from native sites, the community composition did, with Anthophora urbana being an indicator species for native sites and Agapostemon texanus, H. ligatus, M. pallidisignatus, M. rivalis, and Triepeolus grinelidae being indicator species for degraded and restored sites. Anthophora urbana is a common generalist as are Agapostemon texanus and H. ligatus, (Roberts 1969; Sugden 1985) while M. pallidisignatus and M. rivalis are mainly oligolectic on aster species and T. grinelidae is a cleptoparasite that mainly parasitizes Melissodes and Svastra species (Hurd et al. 1980; Michener 2007). These results suggest that efforts to restore the bee community and enhance bee richness should focus on not only increasing floral resource abundance and diversity, but also focusing on planting

50 species that are used by the greatest variety of bee species throughout the entire season. One impediment to selecting appropriate flowering plant species in restoration projects is the lack of knowledge of specific bee-plant associations in a particular area. In addition, restoration efforts should seek to maximize access to nesting resources including bare ground for ground nesters and diverse vegetation for twig and cavity nesters. As restoration efforts in western North America arid grasslands continue to grow, understanding the effects of these efforts on bee communities becomes even more important. Quantifying success of restoration relative to bee communities will require identifying key taxa and taking baseline community measurements at reference sites. While little is known about the bee communities in these grasslands, even less is known about bee-floral associations. This knowledge gap poses a major challenge to restoration managers who wish to enhance bee habitat and are attempting to choose appropriate floral species to plant. Finally, restoration should focus on not only increasing floral abundance and richness but also creating diverse nesting habitat.

Literature Cited

Beck, J.L., Connelly, J.W., Wambolt, C.L., 2012. Consequences of treating Wyoming Big Sagebrush to enhance wildlife habitats. Rangeland Ecol Manag. 65, 444-455. Brown, M.J.F., Paxton, R.J., 2009. The conservation of bees: a global persepective. Apidologie. 40, 410-416. Burel, F., Baudry, J., Butet, A., Clergeau, P., Delettre, Y., Le Coeur, D., Dubs, F., Morvan, N., Paillat, G., Petit, S., Thenail, B., Brunel, E., Lefeuvre, J., 1998. Comparative biodiversity along a gradient of agricultural landscapes. Acta Oecol. 19, 47-60. Cane, J.H., 1991. Soils of ground-nesting bees (Hymenoptera: Apoidea): Texture, Moisture, Cell Depth and Climate. J Kansas Entomol Soc. 64, 406-413. Cane, J.H., 1994. Nesting biology and mating behavior of the southeastern blueberry bee, Habropoda laborisosa (Hymenoptera: Apoidea). J Kansas Entomol Soc. 67, 236- 241. Colwell, R.K., 2013. EstimateS: Statistical estimation of species richness and shared species from samples, Version 9.1.0. User’s guide and application. Colwell, R.K., Chao, A., Gotelli, N.J., Lin, S., Mao, C.X., Chazdon, R.L., Longino, J.T., 2012. Models and estimators linking individual-based and sample-based rarefaction, extrapolation, and comparison of assemblages. J Plant Ecol. 5, 3-21. Cook, S.P., Birch, S.M., Merickel, F.W., Lowe, C.C., Page-Dumroese, D., 2011. Bumble bee (Hymenoptera: Apidae) community structure on two sagebrush steppe sites in southern Idaho. Pan-Pac Entomol. 87, 161-171. Davis, J.D., Hendrix, S.D., Debinski, D.M., Hemsley, C.J., 2008. Butterfly, bee and forb community composition and cross-taxon incongruence in tallgrass prairie fragments. J Insect Conserv. 12, 69-79. De’ath, G., Fabricius, K.E., 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology. 81, 3178-3192. DeBano, S.J., 2006. Effects of livestock grazing on insect communities in semi-arid grasslands of southeastern Arizona. Biodivers Conserv. 15, 2547-2564. Exeler, N., Kratochwil, A, Hochkirch, A., 2009. Restoration of riverine inland sand dune complexes: implications for the conservation of wild bees. J Appl Ecol. 46, 1097- 1105. Forup, M.L., Memmott, J., 2005. The restoration of plant-pollinator interactions in hay meadows. Restor Ecol. 13, 265-274. Geroff, R.K., Gibbs, J., McCray, K.W., 2014. Assessing bee (Hymenoptera: Apoidea) diversity of an Illinois restored tallgrass prairie: methodology and conservation considerations. J Insect Conserv. 18, 951-964. Gonzalez, N., DeBano, S.J., Kimoto, C., Taylor, R.V., Tubbesing, C., Strohm, C., 2013. Native bees associated with isolated aspen stands in Pacific Northwest Bunchgrass Prairie. Nat Areas J. 33, 374-383.

Grundel, R., Jean, R.P., Frohnapple, K.J., Glowacki, G.A., Scott, P.E., Pavlovic, N.B., 2010. Floral and nesting resources, habitat structure, and fire influence bee distribution across an open-forest gradient. Ecol Appl. 20, 1678-1692. Hopwood, J.L., 2008. The contribution of roadside grassland restorations to native bee conservation. Biol Conserv. 141, 2632-2640. Huddleston, R.T., Young, T.P., 2004. Spacing and competition between planted grass plugs and preexisting perennial grasses in a restoration site in Oregon. Restor Ecol. 12, 546-551. Huddleston, R.T., Young, T.P., 2005. Weed control and soil amendment effects on restoration plantings in an Oregon grassland. West N Am Naturalist. 65, 507-515. Hurd, P. D., LeBerge, W. E., & Linsley, E. G., 1980. Principal sunflower bees of North America with emphasis on the Southwestern United States (Hymenoptera, Apoidea). Sm C Zool. 1-158. Kennedy, P.L., DeBano, S.J., Bartuszevige, A., Lueders, A., 2009. Effects of native and nonnative grassland plant communities on breeding passerine birds: implications for restoration of Northwest bunchgrass prairie. Restor Ecol. 17, 515-525. Kimoto, C., DeBano, S.J., Thorp, R.W., Rao, S., Stephen, W.P., 2012. Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci. 12: 108. Available online: http://www.insectscience.org/12.108 Klein, A.M., Vaissie’re, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Important of pollinators in changing landscapes for world crops. P Roy Soc B-Biol Sci. 274, 303-313. Kwaiser, K.S., Hendrix, S.D., 2008. Diversity and abundance of bees (Hymenoptera: Apiformes) in native and ruderal grasslands of agriculturally dominated landscapes. Agr Ecosyst Environ. 124, 200-204. Losey, J.E., Vaughan, M., 2006. The economic value of ecological services provided by insects. BioScience. 56, 311-323. McCune, B., Grace, J., 2002. Analysis of Ecological Communities. MjM Software, Gleneden Beach, Oregon, USA. McCune, B., Mefford, M.J., 2015. PC-ORD Version 7.287. Multivariate Analysis of Ecological Data. MjM Software, Gleneden Beach, Oregon, USA. Michener, C.D., 2007. The Bees of the World, 2nd edition. Johns Hopkins University Press. Nemec, K., Allen, C.R., Danielson, S.D., Helzer, C.J., 2014. Responses of predatory invertebrates to seeding density and plant species richness in experimental tallgrass prairie restorations. Agr Ecosyst Environ. 183, 11-20. Pettis, J.S., Delaplane, K.S., 2010. Coordinate responses to honey bee decline in the USA. Apidologie. 41, 256-263.

Potts, S.G., Bulliam, B., Dafni, A., Ne’eman, G., Willmer, P., 2003. Linking bees and flowers: how do floral communities structure pollinator communities? Ecology. 84, 2628-2642. Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O., Kunin, W.E., 2010. Global pollinator declines: trends, impacts and drivers. Trends Ecol Evol. 25, 345-353. Rao, S., Stephen, W.P., Kimoto, C., DeBano, S.J., 2011. The status of the ‘Red-Listed’ Bombus occidentalis (Hymenoptera: Apiformes) in Northeastern Oregon. Northwest Sci. 85, 64-67. Reed, C.C., 1995. Insect surveys on flowers in native and reconstructed prairies (Minnesota). Restor Manag Notes 13, 210-213. R Development Core Team., 2015. R: A Language and Environment for Statistical Computing. Version 2.11.1 R Foundation for Statistical Computing, Vienna, Austria. Rhoades, P.R., 2016. Genetic and community structure of native bees of the Palouse Prairie, PhD Dissertation, University of Idaho. Rhoades, P., Griswold, T., Ikerd, H., Waits, L., Bosque-Perez, N.A., Eigenbrode, S.D., 2017. The native bee fauna of the Palouse Prairie (Hymenoptera: Apoidea). J Melittology. 66, 1-20. Roberts, R.B., 1969. Biology of the bee genus Agapostemon (Hymenoptera: Halictidae). Kans Univ Sci Bull. 48, 689-719. Sampson, F., Knopf, F., 1994. Prairie conservation in North America. BioScience. 44, 418-421. Smith, L.J., Smith DiCarlo, L.A., DeBano, S.J., In review. Ground crab spiders (Thomisidae: Xysticus) more abundant in cheatgrass (Bromus tectorum L.) invaded grasslands. Biol Invasions. Smith DiCarlo, L.A., DeBano, S.J., 2018. Short-term response of two beneficial invertebrate groups to wildfire in an arid grassland system, USA. Restor ecol. DOI: 10.1111/rec.12832. Tisdale, E.W., 1982. Grasslands of western North America: the Pacific Northwest Bunchgrass. In: Nicholson, A.C., McLean, A., Baker, T.E., Editors. Proceedings of the 1982 Grassland Ecology and Classification Symposium. Pp. 232-245. Information Services Branch, British Columbia Ministry of Forests. Tscharntke, T., Gathmann, A., Steffan-Dewenter, I., 1998. Bioindication using trap- nesting bees and wasps and their natural enemies: community structure and interactions. J Appl Ecol. 35, 708-719. Tubbesing, C., Strohm, C., DeBano, S.J., Gonzalez, N., Kimoto, C., Taylor, R.V., 2014. Insect visitors and pollination ecology of Spalding’s catchfly (Silene spaldingii) in the Zumwalt Prairie of northeastern Oregon. Nat Areas J. 34, 200-211.

US Climate Data, 2017. Climate Boardman – Oregon. Your Weather Service http://www.usclimatedata.com/climate/boardman/oregon/united-states/usor0036 (accessed 15 April 2018). Weisser, W.W., Siemann, E., 2004. The various effects of insects on ecosystem functioning. Ecol Stud. 173, 3-24. Westphal, C., Bommarco, R., Carre, G., Lamborn, E., Morison, N., Petanidou, T., Potts, S.G., Roberts, S.P.M., Szentgyorgyi, H., Tscheulin, T., Vaissiere, B.E., Woychiechowski, M., Biesmeijer, J.C., Kunin, W.E., Settele, J., Steffan- Dewenter, I., 2008. Measuring bee diversity in different European habitats and biogeographical regions. Ecol Monogr. 78, 653-671. Williams, N.M., Minckley, R.L., Silveira, F.A., 2001. Variation in native bee faunas and its implications for detecting community changes. Conservation Ecology 5, 7. Available online: http://www.consecol.org/vol5/iss1/art7/ Winfree, R., Griswold, T., Kremen, C., 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conserv Biol. 21, 213-223. Wisdom, M.J., Rowland, M.M., Wales, B.C., Hemstrom, M.A., Hann, W.J., Raphael, M.G., Holthausen, R.S., Gravenmier, R.A., Rich, T.D., 2002. Modeled effects of sagebrush-steppe restoration on Greater Sage-Grouse in the interior Columbia Basin, USA. Conservation Biol. 16, 1223-1231. Wood, T.J., Holland, J.M., Goulson, D., 2015. Pollinator-friendly management does not increase the diversity of farmland bees and wasps. Biol Conserv. 187, 120-126.

Figure 3.1. Map of The Nature Conservancy Boardman Grasslands Preserve with an inset of the location in Oregon, USA. Circles represent degraded sites, triangles represent native sites, and stars represent restored sites.

Figure 3.2. Sample site configuration, with 50 m survey square around 10 m pan trap radius. Closed circles represent pan traps, closed squares represent habitat survey subsites.

Figure 3.3. Non-metric multidimensional scaling ordination of sites in bee species space along with weighted average positions for bee species for axes 1 vs. 2. Each triangle represents a site during one collection period (sampling bouts) and each circle point represents a bee species. A joint-plot from the environmental matrix is overlaid with variables of r2 > 0.10 being displayed with vector lengths corresponding to the correlation strength along the axes (shown in thick black lines). Sites that are closer together are more similar than sites farther away from each other. Convex hulls connect sites relative to (A) sampling bouts, (B) years, and (C) restoration treatment.

Figure 3.4. Regression trees of bee (A) average abundance; (B) rarefied richness; and (C) average diversity (Shannon-Weiner Index). Deviance explained is the amount of variability explained by the full regression tree. Variables associated with each branch are above the horizontal lines above the nodes. Each node (box) of the trees include mean response variables (bee abundance, diversity or rarefied richness); residual deviance (Dev); and sample size (n).

Table 3.1. Abundance, richness, and diversity metrics for each collection period including total traps present. Adjusted average abundance is total abundance for each of the eighteen sites (bees/traps) then averaged over the eighteen sites (N = 18). Total species richness is for the entire study (total species per sampling bout). Average diversity is total diversity for each of the eighteen sites then averages over the eighteen sites (N = 18).

2014 2015 2016 mid late early mid late early mid late Total Bee Abundance 3254 855 937 2009 885 1664 2553 839 Total Traps Present 161 162 88 92 85 137 160 160 Adjusted Bee Average 24.7 ± 20.2 ± 2.9 5.3 ± 0.7 55.8 ± 8.8 19.9 ± 4.7 14.9 ± 5.4 16.1 ± 3.1 5.3 ± 1.0 Abundance (bees/trap) 6.7 Total Bee Species Richness 30 22 22 21 13 32 26 17 Average Bee Shannon- 1.4 ± 0.1 1.2 ± 0.1 1.2 ± 0.1 1.2 ± 0.1 0.7 ± 0.1 1.2 ± 0.1 1.2 ± 0.1 1.3 ± 0.1 Weiner Diversity Average bloom abundance 14.4 ± 7.5 145.7 ± 74.5 6.1 ± 2.3 14.4 ± 7.5 26.4 ± 9.2 69.2 ± 28.0 231.6 ± 81.2 306.5 ± 81.2 Total bloom species 6 10 5 3 10 15 16 16 richness

Table 3.2. Adjusted abundance (bees/trap) of bee taxa during each collection period over the three years. Total abundance is in parentheses after genus.

2014 2015 2016 Family Genus Species mid late early mid late early mid late

Andrena (13) prunorum 2.2 0.3 aridella 3.2 0.9 0.2 0.2 0.7 0.1 Andrenidae dubia 2.7 0.4 Perdita (221) lingualis 1.4 oregonensis 0.4 4.1 10.8 exigua 0.1 Anthophora (142) urbana 1.6 0.1 2.4 1.9 0.1 2.8 9.1 Apis (59) mellifera 1.5 0.1 0.4 0.4 0.3 0.3 4.3 fervidus 0.3 0.4 Bombus (12) griseocollis 0.3 0.2 0.1 0.1 Diadasia (2) enavata 0.2 0.2 edwardsii 0.3 0.3 Eucera (7) Apidae speciosa 0.2 agilis 0.6 0.3 0.8 1.1 bimatris 1.9 50.4 0.3 0.3 17.3 lupinus 0.2 0.4 0.8 0.1 0.1 Melissodes (1316) lutulentus 4.3 0.7 0.7 0.2 0.7 0.8 0.8 0.6 pallidisignatus 2.1 0.1 0.2 0.6 0.8 0.7 0.4 perlusa 0.1 rivalis 16.1 2.8 13.1 2.8

saponellus 2.7 0.8 0.4 0.8 semilupinus 4.4 15.8 3.0 subagilis 5.3 1.8 0.1 spp. 0.3 0.3 1.3 0.7 0.6 spp. 1 0.1 spp. 2 0.1 spp. 3 0.1 Nomada (11) spp. 4 0.1 spp. 5 0.1 0.2 spp. 6 0.5 Svastra (60) obliqua 2.9 4.5 1.4 0.1 0.6 grindeliae 1.7 1.6 0.2 1.4 Triepeolus (61) paenepectoralis 1.8 spp. 1 0.1 0.2 compactus 0.1 fulgidus 0.1 Colletidae Colletes (4) gypsicolens 0.1 madibularis 0.1 femoratus 16.7 0.6 4.1 101.7 6.8 9.2 33.4 3.7 Agapostemon (6576) texanus 202.7 8.4 52.6 175.1 125.7 13.4 103.1 4.5 virescens 19.7 1.1 12.9 17.2 6.4 1.9 11.4 1.8 Halictidae farinosus 0.4 4.2 1.6 ligatus 1.2 2.8 1.1 1.0 Halictus (146) rubicundus 0.2 0.8 tripartitus 0.1 0.4 0.1 4.1 3.2

Hoplitis (1) grinnelli 0.2 Dialictus spp. 76.0 4.9 73.8 36.3 16.7 203.4 194.6 39.7 sisymbrii 0.2 Lasioglossum (3993) titusi 0.3 0.1 0.2 trizonatum 0.4 Sphecodes (2) spp. 1 0.2 pudicum 0.2 0.4 0.2 0.2 Dianthidium (7) ulkei 0.2 coquilletti 0.5 0.2 0.8 0.8 montivaga 0.2 0.2 0.1 onobrychidis 0.3 0.1 0.1 Megachilidae Megachile (55) parallela 2.0 0.7 perihirta 0.1 0.1 umatillensis 0.6 wheeleri 0.2 breva 0.1 Osmia (2) integra 0.1

Table 3.3. Pearson correlation coefficients between habitat variables and bee species and each axis of the 3-dimensional NMS ordination. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 Axis 3 R R2 R R2 R R2

Invasive Annual Grasses 0.17 0.03 0.18 0.03 0.27 0.07 Litter Cover 0.11 0.01 0.27 0.07 0.05 0.00 Habitat BSC Cover -0.22 0.05 -0.06 0.00 -0.19 0.04 Variables Max Veg Height -0.09 0.01 0.29 0.09 -0.08 0.01 Forb Abundance -0.20 0.04 0.07 0.00 0.14 0.02 Forb Richness -0.13 0.02 0.11 0.01 0.23 0.05 Agapostemon femoratus 0.38 0.14 0.21 0.04 -0.31 0.09 Agapostemon texanus 0.48 0.23 0.40 0.16 -0.30 0.09 Agapostemon virescens 0.30 0.09 0.22 0.05 -0.17 0.03 Andrena prunorum 0.18 0.03 0.09 0.01 0.14 0.02 Anthophora exigua 0.05 0.00 0.06 0.00 -0.11 0.01 Anthophora urbana -0.12 0.01 0.07 0.01 0.04 0.00 Bee Species Apis mellifera -0.20 0.04 0.12 0.02 0.07 0.01 Bombus fervidus 0.17 0.03 0.12 0.02 0.09 0.01 Bombus griseocollis 0.11 0.01 0.09 0.01 0.03 0.00 Colletes compactus -0.18 0.03 0.10 0.01 -0.10 0.01 Colletes fulgidus -0.12 0.02 -0.01 0.00 -0.06 0.00 Colletes gypsicolens -0.15 0.02 0.04 0.00 -0.13 0.02 Colletes mandibularis -0.06 0.00 0.01 0.00 -0.11 0.01

Diadasia enavata 0.09 0.01 0.04 0.00 0.03 0.00 Dianthidium pudicum 0.00 0.00 0.06 0.00 0.05 0.00 Dianthidium ulkei 0.09 0.01 -0.01 0.00 -0.08 0.01 Eucera edwarsii 0.04 0.00 0.02 0.00 0.17 0.03 Eucera speciosa -0.05 0.00 -0.06 0.00 0.20 0.04 Halictus farinosus 0.18 0.03 0.16 0.03 0.26 0.07 Halictus ligatus 0.22 0.05 0.13 0.02 -0.02 0.00 Halictus rubicundus 0.13 0.02 0.04 0.00 0.25 0.06 Halictus tripartitus 0.15 0.02 0.07 0.00 0.11 0.01 Hoplitis grinnelli -0.01 0.00 -0.06 0.00 0.06 0.00 Lasioglossum Dialictus spp. 0.31 0.10 0.35 0.12 0.30 0.09 Lasioglossum sisymbrii 0.08 0.01 0.16 0.03 0.15 0.02 Lasioglossum titusi 0.11 0.01 0.03 0.00 -0.05 0.00 Lasioglossum trizonatum 0.06 0.00 -0.19 0.04 0.06 0.00 Megachile coquilletti 0.16 0.03 -0.02 0.00 0.12 0.01 Megachile montivaga 0.03 0.00 0.09 0.01 -0.06 0.00 Megachile onobrychidis 0.04 0.00 0.01 0.00 0.03 0.00 Megachile parallela 0.15 0.02 0.12 0.02 -0.12 0.02 Megachile perihirta -0.11 0.01 0.10 0.01 0.00 0.00 Megachile umatillensis 0.10 0.01 0.04 0.00 0.19 0.03 Megachile wheeleri -0.20 0.04 0.03 0.00 -0.07 0.00 Melissodes agilis 0.07 0.00 0.15 0.02 0.11 0.01 Melissodes bimatris -0.57 0.33 0.23 0.05 -0.28 0.08

Melissodes lupinus 0.15 0.02 -0.07 0.01 -0.03 0.00 Melissodes lutulentus 0.06 0.00 0.28 0.08 -0.07 0.01 Melissodes pallidisignatus 0.18 0.03 0.20 0.04 -0.18 0.03 Melissodes perlusa 0.02 0.00 0.05 0.00 -0.07 0.01 Melissodes rivalis 0.24 0.06 -0.11 0.01 0.38 0.14 Melissodes saponellus 0.18 0.03 -0.34 0.12 0.20 0.04 Melissodes semilupinus -0.49 0.24 0.23 0.05 -0.27 0.08 Melissodes subagilis 0.02 0.00 0.14 0.02 0.19 0.04 Melissodes spp 0.00 0.00 0.21 0.04 -0.18 0.03 Nomada sp. 1 0.08 0.01 0.12 0.02 -0.07 0.00 Nomada sp. 2 -0.15 0.02 0.05 0.00 -0.11 0.01 Nomada sp. 3 0.11 0.01 0.06 0.00 -0.05 0.00 Nomada sp. 4 0.08 0.01 0.04 0.00 -0.01 0.00 Nomada sp. 5 0.09 0.01 0.07 0.01 -0.01 0.00 Nomada sp. 6 0.12 0.02 0.09 0.01 0.03 0.00 Osmia brevis 0.03 0.00 -0.06 0.00 0.02 0.00 Osmia integra -0.05 0.00 -0.17 0.03 0.11 0.01 Perdita aridella 0.06 0.00 0.21 0.05 -0.16 0.03 Perdita dubia 0.03 0.00 0.18 0.03 -0.22 0.05 Perdita lingualis 0.05 0.00 -0.05 0.00 0.23 0.05 Perdita oregonensis -0.28 0.08 0.04 0.00 0.03 0.00 Sphecodes spp. -0.28 0.08 0.05 0.00 0.01 0.00 Svastra obliqua 0.17 0.03 -0.15 0.02 -0.24 0.06

Triepeolus grindeliae -0.15 0.02 0.20 0.04 -0.14 0.02 Triepeolus paenepectoralis 0.15 0.02 0.16 0.03 -0.15 0.02 Triepeolus sp. 0.04 0.00 0.06 0.00 0.01 0.00

Table 3.4. Presence/absence of each forb species during each sampling bout over the three years.

2014 2015 2016 Family Common Name Species Total mid late early mid late early mid late

common yarrow Achillea millefolium 20   Unknown Aster spp. Aster spp. 1  yellow star-thistle Centaurea solstitialis 3004         diffuse knapweed Centaurea diffusa 90   rush skeletonweed Chondrilla juncea 4   yellow rabbitbrush Chrysothamnus viscidiflorus 4310      wavyleaf thistle Cirsium undulatum 120         bull thistle Cirsium vulgare 2  Canadian horseweed Conyza canadensis 435     Asteraceae slender hawksbeard Crepis atribarba 5   rubber rabbitbrush Ericameria nauseosa 690     threadleaf fleabane Erigeron filifolius 10   Idaho gumweed Grindelia nana 1  sunflower Helianthus spp. 23    hairy false goldenaster Heterotheca villosa 78       prickly lettuce Lactuca serriola 53      hoary tansyaster Machaeranthera canescens 46   tufted wirelettuce Stephanomeria paniculata 4   yellow salsify Tragopogon dubius 5  western tansymustard Descurainia pinnata 1  Brassicaceae tall tumblemustard Sisymbrium altissimum 64  

Chenopodiaceae Russian thistle Salsola kali 7   Fabaceae milkvetch Astragalus spp. 24   Liliaceae sagebrush mariposa lily Calochortus macrocarpus 9  Linaceae Lewis flax Linum lewisii 118    Onagraceae tall annual willowherb Epilobium brachycarpum 112    Polygonaceae Douglas' knotweed Polygonum douglasii 8115     Scrophulariaceae maiden blue eyed Mary Collinsia parviflora 12  

CHAPTER 4. LANDSCAPE CONTEXT INFLUENCES SPIDER COMMUNITY RESPONSES TO GRASSLAND RESTORATION: A REGIONAL STUDY IN THE PACIFIC NORTHWEST, USA

Lauren A. Smith DiCarlo and Sandra J. DeBano

Abstract

Grassland restoration in North America has intensified in the past few decades but it is unclear how it impacts many invertebrate communities, especially in the arid Pacific Northwest, USA. We examined spider communities at a regional scale to 1) investigate the relative roles of landscape context and restoration in structuring spider communities and 2) to identify environmental variables that may underlie community patterns. To address the first objective, we examined the effect of location, restoration, and their interaction on environmental variables and spider communities at three arid inland Pacific Northwest grassland locations, each of which had degraded sites as well as old fields that had been restored using similar methods. To address the second objective, we used a regression tree approach with a larger dataset that included not only degraded and restored sites, but also high quality native sites. We found environmental variables hypothesized to be important to spiders differed in their responses. Both location and restoration affected litter cover independently, with restored sites having less litter. Location, but not restoration, significantly affected forb cover and vegetation height, and a significant interaction between location and restoration occurred for invasive annual grass cover. For spiders, location, but not restoration treatment, had a significant effect on abundance. A significant interaction of location and restoration on spider species richness and diversity indicated that effects of restoration varied relative to location. Spider community composition also varied relative to location, rather than restoration, with structure primarily associated with elevation, vegetation height, and cover of invasive grasses, litter, and forbs. Regression tree analysis of the larger dataset indicated that invasive grass and litter cover were associated with higher spider abundance while elevation, cover of forbs and biological soil crusts impacted species richness and

70 diversity. Spider community composition appears to be influenced by elevation, and the cover of invasive grasses, forbs, and litter. This study demonstrates that even in superficially similar locations, invertebrate response to restoration may differ. To better predict spider community responses to restoration, the larger relationships driving patterns in regional communities, in addition to the location-specific attributes, must be considered.

Introduction

Since European settlement, approximately 99% of grasslands in North America have been converted to other land uses, primarily agricultural (Sampson and Knopf 1994; Burel et al. 1998). However, since the late-1800s cropland abandonment has also increased exponentially across North America due to a number of factors including increases in urbanization and the import of food from other areas (Cramer and Hobbs 2007). In response to this trend, there has been a growing movement in recent decades to restore abandoned fields once used for production to native habitat that can provide essential wildlife habitat and conserve biodiversity (Bakker and Berendse 1999; Wilson et al. 2004; Cramer and Hobbs 2007; Öster et al. 2009; Torok et al. 2011; Porensky et al. 2014). This effort encompasses arid grasslands of North America with multiple groups, including government agencies and non-profit organizations, at the forefront of restoration. Yet monitoring the effectiveness of restoration efforts in grasslands is limited, often because of lack of funding and data on the response of many ecologically significant taxa in many grassland types (Huxel and Hastings 1999; Gerla et al. 2012). This is particularly true of invertebrates, which make up the majority of grassland biodiversity and provide essential ecosystem services such as pollination, nutrient cycling, food for vertebrates, and pest control (Weisser & Siemann 2004; DeBano 2006; Kimoto et al. 2012 a,b; Gonzalez et al. 2013). One particularly significant group of beneficial invertebrates is spiders, which are the seventh largest taxon in the world with approximately 40,000 known species (Coddington and Levi 1991; Zamani and Rafinejad 2014). They not only make major

71 contributions to biodiversity, but also play sizeable roles in the food web as predators and as prey for many different types of wildlife (Malumbres-Olarte et al. 2013). In addition, they play a key role in the control of agricultural pests, a service estimated to be worth approximately $4.5 billion dollars each year in the US (Losey and Vaughan 2006). Spiders respond quickly to changes in the surrounding environment and because they are highly mobile and reproduce quickly, they are a particularly useful indicator taxon for restoration monitoring (Mortimer et al. 1998; Wheater et al. 2000). Studies assessing how grassland restoration affects spiders have mixed results. For example, while multiple studies have found that grassland restoration affects spider abundance, richness, and diversity (Richardson and Hanks 2009; Cristofoli et al. 2010; Nemec et al. 2014; Smith DiCarlo & DeBano, 2018), the direction of effects vary, with some studies finding positive effects and others finding negative ones. Still other studies have found no effect of restoration on spider abundance, richness, and diversity (Bell et al. 1998; Deri et al. 2011; Richardson and Hanks 2009; Nemec et al. 2014). Further, while many studies have found changes in spider community composition after restoration or grassland succession (Bell et al. 1998; Snazell and Clarke 2000; Perner and Malt 2003; Cristofoli et al. 2010; Deri et al. 2011; Nemec et al. 2014; Smith DiCarlo and DeBano 2018), patterns of response differ from study to study. Several factors may underlie these inconsistent responses, including characteristics of the site and restoration project itself (e.g., age, type of plantings), historical factors, and the landscape context of individual studies (Cristofoli et al. 2010; Grman et al. 2013; Smith DiCarlo and DeBano 2018). Landscape context, generally characterized by proportion, configuration, and diversity of habitat types, has been found to have a significant influence over restoration (Krauss et al. 2003; Woodcock et al. 2010; Kouki 2012). However, there is some discrepancy in how landscape context affects invertebrate communities as several studies have found a significant effect of landscape context on invertebrates (Steffan-Dewenter 2003; Schmidt et al. 2005; Woodcock et al. 2010; Kouki 2012) while others have found that local factors, such as plant and litter cover, were more important in shaping the spider communities than landscape-level factors (Jeanneret et al. 2003; Horvath et al. 2015). Landscape context can affect

72 environmental variables (e.g., plant community composition, grass and litter cover) known to influence spider abundance, richness, diversity, and community composition (Bell et al. 1998; Richardson and Hanks 2009; Nemec et al. 2014; Smith DiCarlo and DeBano 2018). Differences in spider community composition and management practices can add additional layers of complexity to responses (Bell et al. 2001; Perner and Malt 2003). Yet, it is currently unclear what role this geographic variation plays in influencing the outcome of arid grassland restoration, even among locations within regional landscapes that may be expected to respond similarly. To investigate the role of landscape context on restoration success at a regional scale and to identify environmental variables that may underlie regional community patterns, we conducted a study at three separate semi-arid bunchgrass prairies in eastern Oregon, USA that may be expected to show similar responses to restoration. Our specific objectives were twofold: 1.) compare degraded and restored sites at each location to determine the relative role of landscape context and grassland restoration affecting haitat variables and spider communities (e.g., abundance, richness, diversity, composition); and 2.) use a larger dataset of degraded, restored, and native sites to identify environmental variables that may underlie observed patterns in spider communities at a regional scale. Given the similarity of the grassland locations, we expected spider communities to respond to restoration in a similar manner and, given previous work in the region (Smith DiCarlo and DeBano 2018), that these changes would be closely associated with differences in litter cover and vegetative structure.

Methods

Study Sites The study took place at three grassland locations in Oregon, USA: The Umatilla National Wildlife Refuge (UNWR) and The Nature Conservancy Boardman Grasslands Preserve (TNC-B) in Morrow County, and The Nature Conservancy Zumwalt Prairie (TNC-Z) in Wallowa County (Fig. 4.1). Site characteristics are listed in Table 4.1. Both TNC-B and TNC-Z contain extensive areas of high quality grassland (largely intact

73 native grasses and forbs relatively uninvaded with non-native grasses) and all three locations contain degraded grassland (formerly cultivated, lacking native bunchgrasses and highly invaded with non-native grasses) and restored areas. UNWR and TNC-B, have similar invasive and native grass species, while TNC-Z has some similar native species but different invasive grass species (Table 4.1). The majority of precipitation at UNWR and TNC-B falls from November to February and the majority of precipitation at TNC-Z falls in June (Bartuszevige et al. 2012; US Climate Data 2017). Temperatures frequently reach 30°C in the summer at all locations (US Climate Data 2017). Site Selection To address our first objective, sites were established at each location from two treatments of interest: degraded and restored. In total 22 sites were selected, with six sites at UNWR (three degraded, three restored); twelve sites at TNC-B (six degraded and six restored), and four sites at TNC-Z (two degraded and two restored). We chose degraded sites by locating habitat that were accessible by old farm roads and that were on relatively flat slopes. Restored sites were placed in established restoration areas at each location. All restored sites had been first treated with herbicides (glyphosate or imazapic) to remove undesirable species, with some receiving a combination of prescribed burning and herbicides (Table 4.1). Sites were then seeded in the late fall with native bunchgrasses using a range drill. Restoration projects were not irrigated. Projects ranged in size from 4-16 ha at UNWR, 23-42 ha at TNC-B, and 20 ha at TNC-Z. To address our second objective, we added additional high-quality sites that were present at TNC-B and TNC-Z. We used six native sites at TNC-B and two native sites at TNC-Z, bringing our total number of sites for regression analyses to 30. Including native sites allowed us to examine a larger continuum of grassland variation to better understand the environmental variables underlying spider responses and increase the power of the analysis. Spider Sampling At each site, eight pitfall traps were placed in a 10 m radius circle (Fig. 4.2). Pitfall traps, 470 ml plastic cups filled with wildlife-friendly propylene glycol and placed flush with the soil, are well-suited for collecting ground-active spiders (Martin 1978).

Spiders were collected at all sites three times in 2015 during three bouts in June-July, July-August, and August-September by opening traps for one week each time period to collect invertebrates. After one week, traps from each site were collected, combined, and transported to the laboratory. Samples were then washed over a 250 μm sieve and spiders sorted from other invertebrates and debris and preserved in 70% ethanol. All juvenile spiders were identified to family and all mature spiders were identified to species, if possible. Habitat Survey To determine which habitat characteristics were related to spider abundance, richness, diversity, and species composition, each site was surveyed for environmental variables three times in 2015, coinciding with spider sampling. Variables were estimated to the nearest 5% cover in 16 63 x 63 cm subplots located in a 50 x 50 m square around pitfall traps (Fig. 4.2). Variables estimated included the percent cover of invasive grasses, biological soil crusts, litter, and forbs. We also estimated maximum vegetation height by measuring the tallest stem in each plot. Analyses Spider abundance at each site was characterized by the average number of spiders per pitfall as not all pitfalls were present during collection due to weather or animal tampering. Abundance and diversity data were averaged over the three sampling bouts from 2015. All individuals were used to quantify abundance but only mature species were used to estimate species richness and diversity. Shannon-Weiner index was used to estimate spider diversity of each site. To compare taxa richness among samples that varied in abundance, rarified species richness estimates were generated for each site by calculating the Chao1 richness estimator using EstimateS, Version 9.1.0 (Colwell et al. 2012; Colwell 2013). Environmental variables for each site were calculated by first averaging subplots for each site and then averaging over the three sampling bouts. To address our first objective, we used two-way analysis of variance (ANOVA) to compare average spider abundance, rarefied richness, Shannon-Weiner diversity, and environmental variables for two factors: restoration treatment (degraded and restored)

75 and landscape context (three locations). All univariate analyses were completed in RStudio 1.0.153 (RStudio Team 2015). To examine whether spider community composition differed relative to restoration treatment or location, and investigate the influence of environmental variables, we used PC-ORD Software version 7.287 for community analyses (McCune and Mefford 2015). Family level abundance was used in all multivariate analyses instead of genus or species abundance to allow for use of juvenile spiders in the analysis (juvenile spiders cannot be positively identified to genus or species without introducing error). The family dataset contained average spiders per pitfall trap (22 sites x 14 families). The environmental dataset contained environmental variables (22 sites x 6 variables) including: elevation, average maximum vegetation height and average percent cover of invasive grasses, litter, biological soil crusts, and forbs. Data met all statistical assumptions and were not transformed. Non-metric multidimensional scaling (NMS) with Sorensen distances was used to ordinate sites in the spider family space matrix and the environmental matrix. NMS does not assume linearity between family response and environmental gradients and exposes relationships between the family matrix and the environmental matrix (McCune and Grace 2002). NMS was performed with 250 random starts and ties were not penalized. A randomization procedure was included to test if solutions were stronger than those obtained by chance, resulting in p-values. R2 values were calculated to represent the percent variance explained by each axis, and relationships of each axis with spider families and habitat variables were quantified with Pearson correlation coefficients. Multi-response Permutation Procedures (MRPP) were used with Sorensen distances to test for differences in family composition across sites (UNWR, TNC-B, TNC-Z) and treatments (degraded, restored). Pairwise comparisons resulted in A-statistics, the chance-corrected within-group agreement, and p-values. To address our second objective, we used regression trees to investigate the relative role of environmental variables over all landscapes. Regression tree analysis is a powerful approach suitable for examining relationships between multiple, potentially interacting, independent variables and a dependent variable (De’ath and Fabricius 2000).

Regression trees repeatedly divide data into two or more groups, each as homogenous as possible, and then divides each of the resulting groups in the same manner, in an interactive process that continues until pre-specified stopping criteria are met. We used a least-squared loss function, which minimizes the sum of the squared deviation. For stopping criteria, we specified a maximum number of splits as 10, a minimum proportion reduction in error (PRE) of 0.05, a minimum split value of 0.05, and a minimum count allowed at any node of 5. We included elevation, average maximum vegetation height, and average percent invasive annual grass, litter, biological soil crusts, and forb cover as potential predictor variables, and constructed separate trees for spider abundance, rarefied richness, diversity, and community composition as quantified by ordination scores. All 30 sites were used in the analyses, including for computing ordination scores. Thus, a separate NMS ordination was conducted as above, but we used all 30 sites instead of 22. Regression tree models were conducted using SYSTAT v. 13 (Systat Software, 2009, San Jose, CA).

Results

A total of 1,596 spiders were collected from the three locations and 30 sites consisting of 14 families and 35 identified species. Out of the total individuals collected, 21% were mature, 69% were immature, and the remaining individuals’ maturity was unknown due to bodily damage. We collected 299 individuals at UNWR, 1046 at TNC- B, and 250 at TNC-Z. Effect of Landscape Context and Restoration Environmental variables differed in their pattern of response to restoration and landscape context. Both location and restoration affected litter cover independently, with restored sites having less litter than degraded sites (Fig. 4.3, Table 4.2). In contrast, a significant interaction between location and treatment for invasive annual grass cover indicated that effects of restoration depended on location (Table 4.2). Finally, maximum vegetation height and forb cover differed significantly relative to location, but not

77 restoration and neither location, treatment, nor the interaction term were significant for biological soil crust (Table 4.2). For spiders, only location had a statistically significant effect on average spider abundance. Significant interactions for rarefied species richness and diversity indicated that effects of treatment differed relative to location (Fig. 4.4, Table 4.2). The NMS randomization procedure revealed that spider community composition varied among sites, resulting in a stable three-dimensional solution (final stress = 6.98, final instability = 0, P = 0.05) with a cumulative R2 of 0.96. Axis 1 accounted for 55.3% of the variation in spider family space, axis 2 accounted for 27.5%, and axis 3 accounted for 13%. Spider community composition in the ordination did not differentiate relative to restoration treatments but did show patterns relative to location (Figs. 4.5a and b) and MRPP analysis showed community composition differed significantly among locations (A = 0.25, p < 0.00001) but not for restoration treatments (A = -0.01, p = 0.65). Pairwise comparisons suggested that communities in TNC-B and UNWR differed from TNC-Z (A = 0.25, P = 0.0001 and A = 0.19, P = 0.002, respectively), but also differed from each other (A = 0.16, P = 0.0002). Separation among locations occurred on both axes. Pearson correlations between the three axes and spider taxa and environmental variables are listed in Table 4.3. TNC-B and UNWR sites differed from TNC-Z along axis 1, with TNC-B and UNWR almost completely overlapping on the left side of the ordination (Fig. 4.5b) while TNC-Z (located on the right side of the ordination) did not overlap with either site. Corinnidae (corinnid sace spiders), Gnaphosidae (ground spiders), Hahniidae (dwarf sheet spiders), Lycosidae (wolf spiders), and Philodromidae (philodromid crab spiders) were all positively correlated with axis 1 and associated with TNC-Z sites. Theridiidae (cobweb spiders) was negatively correlated with axis 1 and were more common in TNC-B and UNWR. In addition, TNC-B and UNWR differentiated on axis 2, which was positively correlated with cobweb spiders and Thomisidae (crab spiders) (and thus associated with TNC-B sites) and negatively correlated Linyphiidae (sheet weavers) and Salticidae (jumping spiders) (and thus associated with UNWR sites).

Several environmental variables strongly correlated with axes 1 and 2. Elevation and maximum vegetation height were positively correlated with axis 1 while invasive grass cover was negatively correlated with that axis, reflecting that TNC-Z sites had higher elevation with taller vegetation and less invasive grasses relative to TNC-B and UNWR sites (Table 4.3; Figs. 4.5a and b). Invasive grass and litter cover had a strong positive correlation with axis 2, while forb cover had a strong negative correlation. Thus, TNC-B sites tended to have more invasive grass and litter and less forb cover relative to UNWR. No variable had a strong correlation with axis 3 (Table 4.3). Environmental Variables Influencing Spider Communities at the Regional Scale Regression tree analysis of average spider abundance produced a tree with four terminal nodes that explained approximately 60.3% of the variation in abundance (Fig. 4.6a). The first two branches formed relative to litter cover, with sites with high litter cover having more spiders than sites with lower litter cover. Sites with high and low levels of litter both branched into two terminal nodes based on invasive grass cover, with sites with less invasive grass cover having fewer spiders than sites with more invasive cover. Regression tree analysis of spider rarefied richness produced a tree with two terminal nodes that separated relative to elevation and explained 45.4% of the variation (Fig. 4.6b). Lower elevation sites had lower species richness than sites at higher elevations. Regression tree analysis of spider diversity produced a tree with four terminal nodes that explained 71.7% of the variation in diversity (Fig. 4.6c). The first two branches formed relative to percent forb cover, with sites with more forb cover having higher diversity. Sites with higher forb cover split into two terminal nodes relative to elevation; sites at higher elevations had higher spider diversity. Sites with less forb cover resulted in branching due to biological soil crust cover. Sites with higher biological soil crust cover had less diversity than sites with less biological soil crust cover. The NMS randomization procedure using 30 sites resulted in a stable three- dimensional solution (final stress = 7.85, final instability = 0, P = 0.01) with a cumulative R2 of 0.96. Axis 1 accounted for 51.7% of the variation in spider family space, axis 2 accounted for 33.7%, and axis 3 accounted for 10.1% (Pearson correlation scores for each axis may be seen in Table 4.4). Regression tree analysis of the spider community related

79 to axis 1 ordination scores produced a tree with three terminal nodes that explained approximately 82.4% of the variation (Fig. 4.7a). The first two branches developed relative to elevation, with higher elevation sites having larger axis 1 ordination scores, and thus being associated with more folding trapdoor spiders (Antrodiaetidae), corinnid sac spiders, ground spiders, dwarf sheet spiders, wolf spiders, and philodromid crab spiders. Lower elevation sites were more dominated by cobweb spiders, and this group branched further based on maximum vegetation height. Sites with shorter vegetation were associated with higher axis 1 ordination scores (i.e., were more dominated by non- cobweb spiders). Regression tree analysis of the spider community related to axis 2 produced a tree with four terminal nodes that explained 68.7% of the variation (Fig. 4.7b). The first two branches formed relative to litter cover, with sites with less litter cover having lower axis 2 ordination scores, indicating dominance by jumping spiders. Sites with higher litter cover further branched based on forb cover with sites with less forb cover having higher axis 2 ordination scores, indicating dominance by wolf spiders, cobweb spiders, and crab spiders. Lower forb cover branched further based on invasive grass cover, resulting in the final two nodes. Sites with lower invasive grass cover had lower axis 2 ordination scores, which were dominated by jumping spiders.

Discussion

The results of this study suggest that spider communities in Pacific Northwest semi-arid grasslands are influenced by both restoration and landscape context. The effects of restoration on spider richness and diversity were dependent on location, with restoration showing effects in some locations and not in others. In contrast, differences in abundance and community composition were associated with location only, with no detectable effect of restoration treatment. These results are surprising when compared to individual studies that assess the impact of grassland restoration on spider communities. Many of these studies find that spider communities respond to restoration with changes in abundance, richness, and/or diversity and almost all detect differences in community composition between treatments (Perner and Malt 2003; Richardson and Hanks 2009;

Cristofoli et al. 2010; Smith DiCarlo and DeBano 2018). However, none of these studies were conducted at a regional scale and did not examine the role of landscape context in modulating responses to restoration. Multiple studies have examined the effect of landscape context and management on spiders (Clough et al. 2005; Schmidt et al. 2005; Hendrickx et al. 2007; Zulka et al. 2014; Horvath et al. 2015). While one study found little evidence that landscape context affected the spider community (Horvath et al. 2015) others found, as in our study, effects at both scales (Clough 2005; Schmidt et al. 2005; Hendrix et al. 2007; Zulka et al. 2014). We found no study that specifically focused on both landscape context and the responses of spiders to restoration; however, we found two studies that focused on restoration and other invertebrates (Woodcock et al. 2010; Kouki et al. 2012). Similar to our study, both studies found that landscape context was important to restoration and recovery of invertebrates (Woodcock et al. 2010; Kouki et al. 2012). Woodcock et al. (2010) found that there was no effect of restoration treatment on beetle communities and Kouki et al. (2012) found that species richness differed between locations but did not compare restored locations to degraded locations. Not only does our study show that location strongly influences how spider communities may respond to restoration, it also identifies several environmental variables that may be driving spider community structure at a regional scale. Invasive grass and litter cover appear to be strongly correlated with spider abundance, while elevation, biological soil crust cover, and forb cover are factors that influence richness and diversity. Invasive grasses appear to be largely influencing the amount of ground litter in these systems, benefiting certain groups of ground-active spiders, such as crab spiders, that rely on litter for hunting (Bell et al. 2001; Smith et al., in review). Spider richness and diversity were both highly impacted by elevation, with higher elevations (such as TNC-Z location) having higher richness and diversity compared to the lower elevation sites, TNC-B and UNWR. This finding is similar to other studies that have found spider species richness and diversity positively correlated with elevation because higher elevation sites tend to have lower summer temperatures and higher plant richness (Uetz 1976; Finch et al. 2008). Higher forb cover was associated with higher spider diversity,

81 and like invasive grasses, forbs can add to litter accumulation. Forbs also provide complex vegetative structure that provides additional habitat for web-spinners and floral hunters, further contributing to increased diversity (Perner and Malt 2003; Richardson and Hanks 2009; Malumbres-Olarte et al. 2013; Nemec et al. 2014). A long-term study at TNC-B showed that sites with more biological soil crusts had higher diversity (Smith DiCarlo and DeBano 2018) but surprisingly, in this study we found increased biological soil crust cover, in sites with low forb cover resulted in less diversity at the regional scale. This result is most likely due to high cover of biological soil crust at one location (TNC- B) and minimal or very-patchy cover at the other two locations. Likewise, the spider community composition within the degraded and restored sites was strongly associated with several environmental variables including elevation, maximum vegetation height, litter cover, forb cover, and invasive grass cover. The higher elevation location, TNC-Z, had several common families including corrinid sac spiders, ground spiders, dwarf sheet spiders, wolf spiders, and philodromid crab spiders. As TNC-Z is at a higher elevation, it has slightly more moderate temperatures during the summer, more precipitation, and a taller grass layer. These factors most likely result in an increase of moisture that may be beneficial to these families; many spider families, especially wolf spiders, prefer areas with higher moisture (Weeks and Holtzer 2000; Oxbrough et al. 2007). TNC-B spider communities were associated with more litter and invasive annual grasses and crab spiders and cobweb spiders were more common. Crab spiders in this area are known to prefer areas of high invasive annual grasses (Smith et al., in review) and cobweb spiders are very common generalists associated with semi-arid grasslands and have been found in high quantities at TNC-B (Smith DiCarlo and DeBano 2018). The UNWR location had higher amounts of forbs but low litter cover and sheet weavers and jumping spiders were more common. Sheet weavers are known to be pioneer species and dominate areas with high disturbance and bare ground (such as after fire or in highly cultivated areas) as they are likely to tolerate larger variation in microclimate (Merrett 1976; Bell et al. 2001; Smith DiCarlo and DeBano 2018). While there are many species of jumping spiders, species within grasslands are known to prefer

82 grasslands dominated by forbs and are often collected on them (Cutler and Jennings 1985; Polchaninova 2015). Regional environmental variables closely associated with differences in spider community structure are elevation, with sites at TNC-Z having more folding trapdoor spiders, corrinid sac spiders, ground spiders, dwarf sheet weavers, wolf spiders, and philodromid crab spiders. Folding trap door spiders and dwarf sheet weavers were only collected from TNC-Z and may benefit from an area with slightly lower temperatures and more moisture. Members of the dwarf sheet weaver family are known to build webs near water to condense moisture and remain active during dry times of the day (Opell and Beatty 1976; Graham et al. 2003); however, little is known about the particular folding trapdoor spiders genus (Androdiaetus) as it has only been collected at a handful of locations in the Pacific Northwest including the Blue Mountains of eastern Oregon and species are still being identified (Hendrixson and Bond 2007). Cobweb spiders, the majority of which were western black widows, were only collected at the lower elevation sites, TNC-B and UNWR, and may benefit from the warmer temperatures and drier climate; however these spiders have been known to occupy multiple types of habitats (Levi 1959; Smith DiCarlo and DeBano 2018). These locations were also driven by maximum vegetation height, with sites with shorter vegetation containing more spider families similar to those found at TNC-Z and sites with taller vegetation associated with cobweb spiders. Differences in community was also associated with variation in litter cover, with less litter cover associated with jumping spiders and higher litter cover associated with wolf spiders, cobweb spiders, and crab spiders. A previous study at TNC-B also found that wolf spiders, cobweb spiders, and crab spiders were more common in areas of higher litter while jumping spiders were more common in areas with less litter (Smith DiCarlo and DeBano 2018). This is most likely due to preferred hunting strategies. Areas with high litter cover were further structured by forb cover with jumping spiders more common in areas with more forbs, a common pattern in grasslands (Cutler and Jennings 1985; Polchaninova 2015). Areas with less forb cover were further structured by invasive grass cover with areas with little invasive grass cover associated more with jumping spiders and areas with higher grass cover associated with wolf

83 spiders, cobweb spiders, and crab spiders. These spider families tend to prefer areas with high litter, which in turn is highly correlated with invasive grass cover (Smith et al., in review), and may benefit these spiders by providing additional cover when hunting (Rypstra et al. 1999; Bell et al. 2001). Observing spider community response to semi-arid grassland restoration of old fields at the regional scale highlights several challenges during restoration. While restoration techniques are undeniably important, landscape context may have a larger impact on the spider community. Factors to consider include site history, soils, available moisture and prey, and vegetation. Several environmental variables that appear to be important in structuring the spider community including maximum vegetation height and cover of invasive annual grasses, litter, and forbs. If restoring spider-mediated services, restoration should focus on restoring the ground litter layer to increase abundance but also vegetative structure and complexity to increase spider richness and diversity. Since landscape context appears to play a major role in determining the efficacy of old field restoration, management should seriously consider the location and tailor methods to that location to properly restore the invertebrate community. While identification of environmental drivers at the regional level can inform managers of old field restoration, management techniques must also be adapted to the site-specific characteristics.

Acknowledgements We thank Leslie Nelson at TNC Boardman Grasslands, Rob Taylor at TNC Zumwalt Prairie, and Laci Bristow at Umatilla National Wildlife Refuge, for help in planning and implementing the study, and LJ Smith, SR Roof, EM Campbell, KL Kirby, L McDaniel, LK Waianuhea, and BE Price for their help in the field and laboratory. This work was supported by a USDA NIFA National Needs Graduate Fellowship [#2012- 04150] and funding from Oregon State University’s General Research Fund and the Provost’s Branch Experiment Station Experiential Learning Program. Additional funding was provided by a USDA Western Sustainable Agriculture Research and Education Graduate Student Grant [#GW16-016], a TNC Oren Pollak Memorial Student Research Grant for Grassland Science, a Soil and Water Conservation Society Kenneth E. Grant

Research Scholarship, the Prairie Biotic Small Research Grants Program, and a Society for Ecological Restoration Northwest Chapter Student Research Grant.

Literature Cited

Bakker, J.P., Berendse, F., 1999. Constraints in the restoration of ecological diversity in grassland and heathland communities. Trends Ecol Evol. 14, 63-68. Bartuszevige, A.M., Kennedy, P.L., Taylor, R.V., 2012. Sixty-seven years of landscape change in the last, large remnant of the Pacific Northwest bunchgrass prairie. Nat Area J. 32, 166-170. Bell, J.R., Cullen, W.R., Wheater, C.P., 1998. The structure of spider communities in limestone quarry environments. In, Seldon, P., 1998. Proceedings of the 17th European Colloquium of Arachnology, Burnham Beeches: British Arachnological Society, Edinburgh. Bell, J.R., Wheater, C.P., Cullen, W.R., 2001. The implications of grassland and heathland management for the conservation of spider communities: a review. J Zool. 255, 377-387. Clough, Y., Kruess, A., Kleijn, D., Tscharntke, T., 2005. Spider diversity in cereal fields: comparing factors at local, landscape and regional scales. J. Biogeogr. 32, 2007– 2014. Coddington, J.A., Levi, H.W., 1991. Systematics and evolution of spiders (Araneae). Annu Rev Ecol Syst. 22, 565-592. Colwell, R.K., 2013. EstimateS: Statistical estimation of species richness and shared species from samples. Version 9.1.0. User’s guide and application. Colwell, R.K., Chao, A., Gotelli, N.J., Lin, S., Mao, C.X., Chazdon, R.L., Longino, J.T., 2012. Models and estimators linking individual-based and sample-based rarefaction, extrapolation, and comparison of assemblages. J Plant Ecol. 5, 3-21. Cramer, V.A., Hobbs, R.J., eds., 2007. Old fields: dynamics and restoration of abandoned farmland. Society for Ecological Restoration International. Island Press, Washington, D.C. Cristofoli, S., Mahy, G., Kekenbosch, R., Lambeets, K., 2010. Spider communities as evaluation tools for wet heathland restoration. Ecol Indic. 10, 773-780. Cutler, B., Jennings, D.T., 1985. A revision of the Metaphidippus arizonensis group (Araneae, Salticidae). J Arachnol. 13, 1-8. DeBano, S.J., 2006. Effects of livestock grazing on insect communities in semi-arid grasslands of southeastern Arizona. Biodivers Conserv. 15, 2547-2564. Déri, E., Magura, T., Horvath, R., Kisfali, M., Ruff, G., Lengyel, S., Tothmeresz, B., 2011. Measuring the short-term success of grassland restoration: the use of habitat

affinity indices in ecological restoration. Res Ecol. 19, 520–528. Finch, O.D., Blick, T., Schuldt, A., 2008. Macroecological patterns of spider species richness across Europe. Biodivers Conserv. 17, 2849-2868. Gerla, P.J., Cornett, M.W., Ekstein, J.D., Ahlering, M.A., 2012. Talking big: lessons learned form a 9000 hectare restoration in the northern tallgrass prairie. Sustainability. 4, 3066-3087, available online: http://www.mdpi.com/2071- 1050/4/11/3066/htm Gonzalez, N., DeBano, S.J., Kimoto, C., Taylor, R.V., Tubbesing, C., Strohm, C., 2013. Native bees associated with isolated aspen stands in Pacific Northwest Bunchgrass Prairie. Nat Areas J. 33, 374-383. Graham, A.K., Buddle, C.M., Spence, J.R., 2003. Habitat affinities of spiders living near a freshwater pond. Journal Arachnol. 31, 78-89. Grman, E., Bassett, T., Brudvig, L.A., 2013. Confronting contingency in restoration: management and site history determine outcomes of assembling prairies, but site characteristics and landscape context have little effect. J Appl Ecol. 50, 1234- 1243. Hendrickx, F., Maelfait, J.P., van Wingerden, W., Schweiger, O., Speelmans, M., Aviron, S., Augenstein, I., Billeter, R., Bailey, D., Bukacek, R., Burel, F., Diekotter, T., Dirksen, J., Herzog, F., Liira, J., Roubalova, M., Vandomme, V., Bugter, R., 2007. How landscape structure, land-use intensity and habitat diversity affect components of total arthropod diversity in agricultural landscapes. J Appl Ecol. 44, 340–351. Hendrixson, B.E., Bond, J.E., 2007. Molecular phylogeny and biogeography of an ancient Holarctic lineage of mygalomorph spiders (Araneae: Antrodiaetidae: Antrodiaetus). Mol Phylogenet Evol. 42, 738-755. Horvath, R., Magura, T, Szinetar, C., Eichardt, J., Kovacs, E., Tothmeresz, B., 2015. In stable, unmanaged grasslands local factors are more important than landscape- level factors in shaping spider assemblages. Agri Ecosyst Environ. 208, 106-113. Huxel, G.R., Hastings, A., 1999. Habitat loss, fragmentation, and restoration. Restor Ecol. 7, 309-315. Jeanneret, P., Schupbach, B., Luka, H., 2003. Quantifying the impact of landscape and habitat features on biodiversity in cultivated landscapes. Agr Ecosyst Environ. 98, 311-320. Kimoto, C., DeBano, S.J., Thorp, R.W., Rao, S., Stephen, W.P., 2012a. Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci. 12, 108, 23 pp., available online: http://www.insectscience.org/12.108 Kimoto, C., DeBano, S.J., Thorp, R.W., Taylor, R.V., Schmalz, H., DelCurto, T., Johnson, T., Kennedy, P.L., Rao, S., 2012b. Livestock and native bee communities: short-term responses to grazing intensity and implications for

managing ecosystem services in grasslands. Ecosphere. 3, 88. http://www.esajournals.org/doi/pdf/10.1890/ES12-00118.1 Kouki, J., Hyvarinen, E., Lappalainen, H., Martikainen, P., Smila, M., 2012. Landscape context affects the success of habitat restoration: large-scale colonization patterns of saproxylic fire-associated species in boreal forests. Divers Distrib. 18, 348-355. Krauss, J., Steffan-Dewenter, I., Tscharntke, T., 2003. How does landscape context contribute to effects of habitat fragmentation on diversity and population density of butterflies? J Biogeogr. 20, 889-900. Levi, H.W., 1959. The spider genus Latrodectus (Araneae, Theridiidae). T Am Microsc Soc. 78, 7-43. Losey, J.E., Vaughan, M., 2006. The economic value of ecological services provided by insects. Bioscience. 56, 311-323. Malumbres-Olarte, J., Vink, C.J., Ross, J.G., Cruickshank, R.H., Paterson, A.M., 2013. The role of habitat complexity on spider communities in native alpine grasslands of New Zealand: Habitat complexity and alpine spiders. Insect Conserv Diver. 6, 124–134. Martin, J.E.H., 1978. The Insects and Arachnids of Canada Part 1: Collecting, preparing, and preserving insects, mites, and spiders. Biosystemics Research Institute, Ottawa, Ontario. McCune, B., Grace, J., 2002. Analysis of Ecological Communities. MjM Software, Gleneden Beach, Oregon, USA. McCune, B., Mefford, M.J., 2015. PC-ORD Version 7.287. Multivariate Analysis of Ecological Data. MjM Software, Gleneden Beach, Oregon, USA. Merrett, P., 1976. Changes in the ground living spider fauna after heathlands fires in Dorset. Bull Br Arachnol Soc. 3, 214-221. Mortimer, S.R., Hollier, J.A., Brown, V.K., 1998. Interactions between plant and insect diversity in the restoration of lowland calcareous grasslands in southern Britain. Appl Veg Sci. 1, 101–114. Nemec, K., Allen, C.R., Danielson, S.D., Helzer, C.J., 2014. Responses of predatory invertebrates to seeding density and plant species richness in experimental tallgrass prairie restorations. Agr Ecosyst Environ. 183, 11-20. Opell, B.D., Beatty, J.A., 1976. The Nearctic Hahniidae (Arachnida: Araneae). Bull Mus Comp Zool. 147, 393.-433. Öster, M., Ask, K., Cousins, S.A.O., Eriksson, O., 2009. Dispersal and establishment limitation reduces the potential for successful restoration of semi-natural grassland communities on former arable fields. J Appl Ecol. 46, 1266-1274. Oxbrough, A.G., Gittings, T., O’Halloran, J., Giller, P.S., Kelly, T.C., 2007. Biodiversity of the ground-dwelling spider fauna of afforestation habitats. Agr Ecosyst Environ. 120, 433-441.

Perner, J., Malt, S., 2003. Assessment of changing agricultural land use: response of vegetation, ground-dwelling spiders and beetles to the conversion of arable land into grassland. Agr Ecosyst Environ. 98, 169-181. Polchaninonva, N., 2015. Recovery of spider communities after a spontaneous summer fire in the forb-bunchgrass steppe of eastern Ukraine. Hacquetia. 14, 79-96. Porensky, L.M., Leger, E.A., Davison, J., Miller, W.W., Goergen, E.M., Espeland, E.K., Carroll-Moore, E.M., 2014. Arid old-field restoration: native perennial grasses suppress weeds and erosion, but also suppress native shrubs. Agr Ecosyst Environ. 184, 135-144. Richardson, M.L., Hanks, L.M., 2009. Effects of grassland succession on communities of orb-weaving spiders. Environ Entomol. 38, 1595-1599. RStudio Team, 2015. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA. Rypstra, A.L. Carter, P.E., Balfour, R.A., Marshall, S.D., 1999. Architectural features of agricultural habitats and their impact on the spider inhabitants. J Arachnol. 27, 371-377. Schmidt, M.H., Roschewitz, I., Thies, C., Tscharntke, T., 2005. Differential effects of landscape and management on diversity and density of ground-dwelling farmland spiders. J Appl Ecol. 42, 281-287. Smith, L.J., Smith DiCarlo, L.J., DeBano, S.J., In review. Spider family (Thomisidae) exhibits positive response to cheatgrass invasion (Bromus tectorum L.). Biol Inv. Smith DiCarlo, L.A., DeBano, S.J., 2018. Spider community responses to grassland restoration: balancing tradeoffs between abundance and diversity. Restor Ecol. DOI: 10.1111/rec.12832. Snazell, R., Clark, R., 2000. The colonization of an area of restored chalk downland by spiders (Araneae). Ekol Bratisl. 19, 263-271. Steffan-Dewenter, I., 2003. Importance of habitat area and landscape context for species richness of bees and wasps in fragmented orchard meadows. Conserv Biol. 17, 1036-1044. Torok, P., Vida, E., Deak, B., Lengyel, S., Tothmeresz, B., 2011. Grassland restoration on former croplands in Europe: an assessment of applicability of techniques and costs. Biodivers Conserv. 20, 2311-2332. Uetz, G.W., 1976. Gradient analysis of spider communities in a streamside forest. Oecologia. 22, 373-385. US Climate Data, 2017. Climate Boardman – Oregon. Your Weather Service http://www.usclimatedata.com/climate/boardman/oregon/united-states/usor0036 (accessed 15 April 2018). Weeks, R.D., Holtzer, T.O., 2000. Habitat and season in structuring ground-dwelling spider (Araneae) communities in a shortgrass steppe ecosystem. Environ

Entomol. 29, 1164-1172. Weisser, W.W., Siemann, E., 2004. The various effects of insects on ecosystem functioning. Ecol Stud. 173, 3-24. Wheater, C.P., Cullen, W.R., Bell, J.R., 2000. Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecol. 15, 401–406. Wilson, S.D., Bakker, J.D., Christian, J.M., Li, X., Ambrose, L.G., Waddington, J., 2004. Semiarid old-field restoration: is neighbor control needed? Ecol Appl, 14, 476- 484. Woodcock, B.A., Vogiatzakis, I.N., Westbury, D.B., Lawson, C.S., Edwards, A.R., Brook, A.J., Harris, S.J., Lock, K.A., Maczey, N., Masters, G., Brown, V.K., Mortimer, S.R., 2010. The role of management and landscape context in the restoration of grassland phytophagous beetles. J Appl Ecol. 47, 366-376. Zamani, A., Rafinejad, J., 2014. First record of the Mediterranean recluse spider Loxosceles rufescens (Araneae: Sicariidae) from Iran. J Arthropod-Borne Di. 8, 228-231. Zulka, K.P., Abensperg-Traun, M., Milasowszky, N., Bieringer, G., Gereben-Krenn, B. A., Holzinger, W., Hölzler, G., Rabitsch, W., Reischütz, A., Querner, P., Sauberer, N., Schmitzberger, I., Willner, W., Wrbka, T., Zechmeister, H., 2014. Species richness in dry grassland patches of eastern Austria: a multi-taxon study on the role of local, landscape and habitat quality variables. Agri Ecosyst Environ. 182, 25–36.

Figure 4.1. Map of the three study locations (TNC-B, TNZ-Z, and UNWR) in northeastern Oregon, USA.

Figure 4.2. Sample site configuration, with 50 m survey square around 10 m pitfall radius. Circles represent pitfall traps, squares represent habitat survey subplots.

Figure 4.3. Average percent cover of invasive annual grasses (A), litter (B), and forbs (C) and average maximum vegetation height in cm (D) at degraded and restored sites at TNC-B, TNC-Z, and UNWR locations. TNC-B (N = 6 degraded, N = 6 restored), TNC-Z (N = 2 degraded, N = 2 restored), and UNWR (N = 3 degraded, N = 3 restored).

Figure 4.4. Average spider abundance (A), rarefied species richness (B), and Shannon- Weiner diversity (C) at degraded and restored sites at TNC-B, TNC-Z, and UNWR locations. TNC-B (N = 6 degraded, N = 6 restored), TNC-Z (N = 2 degraded, N = 2 restored), and UNWR (N = 3 degraded, N = 3 restored).

Figure 4.5. Non-metric multidimensional scaling ordination of sites in spider family space along with weighted average positions for spider families for axes 1 vs. 2. Each triangle represents a site and each circle point represents a spider family. A joint-plot from the environmental matrix is overlaid with variables of r2 > 0.25 being displayed with vector lengths corresponding to the correlation strength along the axes (shown in thick black lines). Sites that are closer together are more similar than sites farther away from each other. (A) Convex hulls connect each group of sites relative to treatment. (B) Convex hulls connect each group of sites relative to location.

Figure 4.6. Regression trees for average spider abundance (A), rarefied richness (B), and diversity (C). Deviance explained is the amount of variability explained by the full regression tree. Variables associated with each branch are above the horizontal lines above the nodes. Each node (box) of the trees include mean response variables (spider abundance, rarefied richness, or spider diversity); standard deviation (SD); and sample size (N).

Figure 4.7. Regression trees of spider community results from ordination with axis 1 (A) and axis 2 (B). Deviance explained is the amount of variability explained by the full regression tree. Variables associated with each branch are above the horizontal lines above the nodes. Each node (box) of the trees include mean response variables (spider abundance, rarefied richness, or spider diversity); standard deviation (SD); and sample size (N).

Table 4.1. Information on restoration sites from UNWR, TNC-B, and TNC-Z locations including coordinates, size (ha), total number of sites, average precipitation (cm) and average annual temperatures (°C) (30 year averages, US Climate Data, 2017), common native grasses, common invasive grasses, elevation, year seeded, and type of restoration. Site refers to a restoration project at each location.

Total Average number of Average Annual Common Common Size sites Restoration Year Type of Location Coordinates Precip Temp Native Invasive Elevation (ha) (degraded, Site Seeded Restoration (cm) (low - Grasses Grasses restored, high °C) native) Herbicide + 1 86 2007 PSSPS; Burn 45.905256°N, POSE; BRTE; UNWR 3,604 6 (3, 3, 0) 22 5 - 18 2 83 2013 Herbicide -119.584475°W ELELE; TACA8 Herbicide + HECOC8 3 84 2015 Burn 1 267 2006 Herbicide PSSPS; 2 272 2008 Herbicide POSE; 45.636738°N, BRTE; 3 274 2009 Herbicide TNC-B 9,163 18 (6, 6, 6) 19 5 – 18 ELELE; -119.860457°W TACA8 HECOC8 4 281 2010 Herbicide 5 245 2011 Herbicide 6 256 2012 Herbicide PSSPS; Herbicide + VEDU; 1 1352 2010 45.555802°N, POSE; Burn TNC-Z 13,354 6 (2, 2, 2) 48 -1 - 16 THIN6; -116.958538°W FEID; Herbicide + AGCR 2 1379 2010 KOMA Burn * AGCR: crested wheatgrass (Agropyron cristatum); BRTE: cheatgrass (Bromus tectorum); ELELE: bottlebrush squirreltail (Elymus elymoides); FEID: Idaho fescue (Festuca idahoensis); HECOC8: needle and thread grass (Heterostipa comate); KOMA: prairie Junegrass (Koeleria macrantha); POSE: Sandberg bluegrass (Poa secunda); PSSPS: bluebunch wheatgrass (Pseudoroegneria spicata); TACA8: medusahead (Taeniatherum caput-medusae); THIN6 intermediate wheatgrass (Thinopyrum intermedium); VEDU: ventenata (Ventenata dubia)

Table 4.2. Results from two-way ANOVA analyses including site, treatment, and interaction between site and treatment of all environmental variables and spider abundance, rarefied species richness, and diversity.

df F-stat p-value Invasive Annual Grasses site 2 10.78 0.001 type 1 0.05 0.83 site*type 2 5.41 0.02 Litter site 2 8.68 0.002 type 1 6.82 0.02 site*type 2 1.31 0.3 BSC site 2 2.93 0.08 type 1 1.12 0.31 site*type 2 1.67 0.22 Max Veg Height site 2 4.32 0.03 type 1 3.61 0.08 site*type 2 1.25 0.31 Forbs site 2 23.83 <0.0001 type 1 0.03 0.86 site*type 2 1.43 0.27 Abundance site 2 4.08 0.04 type 1 0.02 0.9 site*type 2 0.78 0.48 Rarefied Richness site 2 3.66 0.05 type 1 0.3 0.59 site*type 2 3.7 0.05 Diversity site 2 9.14 0.002 type 1 0.13 0.72 site*type 2 3.97 0.04

Table 4.3. Pearson correlation coefficients between environmental variables and spider families and each axis of the 3-dimensional NMS ordination. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 Axis 3 2 2 2 R R R R R R Elevation 0.77 0.59 0.19 0.03 0.33 0.11 Invasive Grass Cover -0.57 0.33 0.51 0.26 -0.04 0.00 Environmental Litter Cover -0.10 0.01 0.51 0.26 -0.34 0.12 Variables BSC Cover -0.11 0.01 -0.06 0.00 0.11 0.01 Max Veg Height 0.45 0.20 0.24 0.06 0.13 0.02 Forb Cover 0.27 0.07 -0.59 0.35 0.28 0.08 Agelenidae -0.08 0.01 -0.07 0.00 0.09 0.01 Amaurobiidae -0.08 0.01 -0.07 0.00 0.09 0.01 Antrodiaetidae 0.34 0.12 -0.12 0.02 0.77 0.60 Corinnidae 0.53 0.28 -0.09 0.01 -0.18 0.03 Gnaphosidae 0.63 0.39 -0.11 0.01 -0.58 0.34 Hahniidae 0.59 0.35 0.17 0.03 0.16 0.03 Spider Linyphiidae 0.24 0.06 -0.51 0.26 -0.04 0.00 Families Lycosidae 0.76 0.57 0.40 0.16 -0.23 0.05 Mimetidae 0.05 0.00 -0.26 0.07 0.14 0.02 Philodromidae 0.59 0.35 -0.31 0.10 -0.47 0.22 Pholcidae -0.24 0.06 -0.03 0.00 0.03 0.00 Salticidae -0.01 0.00 -0.67 0.45 -0.47 0.22 Theridiidae -0.78 0.61 0.54 0.29 -0.17 0.03 Thomisidae -0.14 0.02 0.72 0.52 -0.11 0.01

Table 4.4. Pearson correlation coefficients between environmental variables and spider families and each axis of the 3-dimensional NMS ordination. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 Axis 3 R R2 R R2 R R2 Elevation 0.82 0.66 0.16 0.03 -0.24 0.06 Invasive Grass Cover -0.50 0.25 0.55 0.31 -0.03 0.00 Habitat Litter Cover 0.08 0.01 0.64 0.40 0.20 0.04 Variables BSC Cover -0.23 0.05 -0.47 0.22 -0.08 0.01 Max Veg Height 0.30 0.09 0.46 0.21 -0.05 0.00 Forb Cover 0.36 0.13 -0.13 0.02 -0.02 0.00 Agelenidae -0.08 0.01 -0.01 0.00 -0.10 0.01 Amaurobiidae -0.08 0.01 -0.01 0.00 -0.10 0.01 Antrodiaetidae 0.39 0.15 -0.14 0.02 -0.61 0.37 Corinnidae 0.38 0.15 0.07 0.00 0.10 0.01 Gnaphosidae 0.51 0.26 -0.24 0.06 0.60 0.36 Hahniidae 0.52 0.28 -0.11 0.01 0.04 0.00 Spider Linyphiidae 0.23 0.05 -0.27 0.07 0.21 0.05 Families Lycosidae 0.75 0.56 0.52 0.27 0.08 0.01 Mimetidae -0.01 0.00 -0.13 0.02 -0.08 0.01 Philodromidae 0.65 0.43 -0.10 0.01 0.45 0.20 Pholcidae -0.19 0.04 -0.03 0.00 -0.06 0.00 Salticidae -0.01 0.00 -0.43 0.18 0.62 0.39 Theridiidae -0.76 0.58 0.53 0.28 0.10 0.01 Thomisidae -0.23 0.06 0.71 0.51 0.03 0.00

100

CHAPTER 5. SHORT-TERM RESPONSE OF TWO BENEFICIAL INVERTEBRATE GROUPS TO WILDFIRE IN AN ARID GRASSLAND SYSTEM, USA

Lauren A. Smith DiCarlo, Sandra J. DeBano, and Skyler Burrows

Abstract Wildfire frequency has increased in rangelands across the western United States, yet it is unclear how these fires affect beneficial invertebrate communities. We examined bee (Hymenoptera), spider (Araneae), and vegetative communities one year before and one year after a wildfire swept across an intact grassland in eastern Oregon. Several sites were left unburnt after the fire and a before-after-control-impact study design was used to assess changes within the communities. Both native bee and spider community composition were significantly altered one year after the fire. In addition, while the fire did not affect bee or spider abundance, or spider diversity and richness, it significantly increased native bee diversity and richness. Sheet web spiders (Linyphiidae) and several bee species (primarily large, generalist species) were associated with burned sites. Invasive annual grass and biological soil crust cover decreased significantly in burned sites but maximum vegetation height and litter cover did not differ significantly among treatments. Forb abundance increased in burned sites; however, species richness of forbs did not differ one year after the fire. Several forbs were indicative of burned areas including non-native species such as Douglas’ knotweed (Polygonum douglasii) and Russian thistle (Salsola tragus) and native species such as Canadian horseweed (Conyza canadensis), hoary tansyaster (Machaeranthera canescens) and tall willowherb (Epilobium brachycarpum). This study is unique in that we provide data on two beneficial invertebrate communities before and after wildfire, allowing us to more directly tie changes in both communities to the fire.

Introduction

101

Nearly all remaining rangeland in North America, including sagebrush steppe and arid-grassland habitat, is ecologically degraded (Sampson and Knopf, 1994; West 2000). One major contributor to this degradation is the invasion of non-native annual grasses, which alter disturbance regimes, especially fire frequency and intensity (Pyke 1999; Davies et al. 2012). The potential effects of these invasive grasses on fire regimes is predicted to be exacerbated by climate change and in the past three decades, wildfire frequency and size has increased across western North America (Schoennagel et al. 2017). While wildfire and prescribed-burn studies are common in North America, few studies have solely focused on wildfires in arid grassland habitat and even fewer have concentrated on how they impact beneficial grassland invertebrates. Grassland invertebrates are integral components of rangeland biodiversity and provide important ecosystem services such as pollination, nutrient cycling, food for vertebrates, and pest control (DeBano 2006, Weisser and Siemann 2004; Black et al. 2011; Kennedy et al. 2009; Kimoto et al. 2012 a,b; Tubbesing et al. 2014). As wildfire increases across the northwestern United States, understanding its effects on these beneficial invertebrates and their subsequent ecosystem services becomes imperative. Here, we focus on two beneficial invertebrate groups, native bees, which provide pollination to many significant rangeland plants and neighboring crops (Kremen et al. 2002), and spiders, generalist predators that enhance pest control and play an important role in rangeland food webs (Sunderland 1999; Malumbres-Olarte et al. 2013). Bees are one of the most diverse groups of insects, with an estimated 20,000 species occurring globally (Michener 2007). Given that 35% of crops and 80% of native plants rely on animal pollination (Klein et al. 2007; Winfree et al. 2007; Potts et al. 2010), declines in honey bees and some native bee species has resulted in an increase interest in pollinator conservation. While less is known about trends in spider diversity, they are the seventh most diverse taxon in the world with approximately 40,000 species globally (Mirshamsi Kakhki 2005; Zamani and Rafinejad 2014). Fire can affect both groups through mortality, injury, and displacement; however, both can survive if they have the ability to disperse (easier for larger, highly mobile bees) or find shelter underground (Bell et al. 2001; Love and Cane 2016). Bees that build shallow nests in the ground, such as cavity-

102 nesting Apidae, or aboveground nests, like some species of leafcutting and mason bees (Megachile and Osmia), are at a higher risk of mortality than species that nest deep in the soil (Cane and Neff 2011). In addition, timing of the fire in relation to the invertebrate’s location (e.g. foraging, in a nest), fire intensity, and developmental stage of the invertebrate (e.g. less-mobile bee larvae vs. mobile adult) can also influence the impact of fire on the invertebrate community (Swengel 2001). Both bee and spider communities are expected to show large responses to wildfire immediately after its occurrence, with the greatest differences in diversity, richness, and community usually evident within the first few years; after this period, a number of studies suggest that communities recover over time (Buddle et al.2000; Moretti et al. 2002; Potts et al. 2003; Pryke and Samways 2012). Several studies have shown that bees respond positively to fire, with initial increases in abundance, richness, and diversity at burned sites due to additional access to nesting resources such as charred wood cavities and more bare ground and floral resources (e.g., increased herbaceous plant growth, richness, and diversity) (Moretti et al. 2009; Grundel et al. 2010; Bogusch et al. 2015). However, Potts et al. (2001) found that bee abundance and diversity decreased in response to wildfire, most likely due to fire reducing herbaceous growth and lowering nectar standing crop. Likewise, some studies have found spider communities change drastically with increased species richness and diversity at burned sites due to quick colonization of open habitat specialists (Bell et al. 2001; Buddle et al. 2000; Koponen 2005; Langlands et al. 2012), while others have found no change in abundance or richness due to rapid vegetation recovery (Samu et al. 2010). These ambiguous results may be due to variation relative to a number of study factors, including bee and spider community composition, the duration and intensity of fire, and postfire duration. In addition, geographic area and habitat type may explain much of this variation; hence, it is important to determine how wildfire affects these invertebrate communities in Pacific Northwest arid grasslands. Another limitation of past research on wildfire is a lack of studies with prefire data. Due to the unpredictable nature of wildfires, collecting data prior to wildfire is rare and the majority of fire-impact studies only compare burned sites to control sites postfire

103

(e.g., Buddle et al. 2000; Moretti et al. 2002; Koponen 2005; Grundel et al. 2010; Bogusch et al. 2015; Love and Cane 2016). We found only one study that was able to collect invertebrate data prior to a wildfire (Bess et al. 2002) and no studies that were able to use a before-after-control-impact design (no areas were left unburnt in the Bess et al. 2002 study). Having prefire data, in addition to controls, offers the rare opportunity to better understand initial conditions and better quantify wildfire impacts on beneficial invertebrate communities and identify habitat factors associated with observed responses. The occurrence of a wildfire during the course of a multi-year study of beneficial invertebrates in a Pacific Northwest grassland allowed us to address limitations of previous work on wildfire effects on invertebrates. Because of the nature of the fire, we were able to collect bee, spider, and habitat data before and after wildfire in both burned and unburned sites. Thus, our objectives were twofold: 1) determine how wildfire affects native bee and spider communities including abundance, composition, species richness, and diversity one year after the burn and 2) investigate which habitat factors are strongly associated with observed patterns for each invertebrate group. Based on prior studies, we expected bee communities to respond strongly to fire-induced changes in floral resources (e.g. plant richness and abundance) (Grundel et al. 2010; Mola et al. 2018) and spider communities to respond to fire-induced changes in vegetation structure (e.g., invasive species cover, plant litter) (Bell et al. 2001; Samu et al. 2010; Smith et al., in review).

Methods Study Site The study took place at The Nature Conservancy Boardman Grasslands Preserve (the Preserve) in Morrow County, Oregon, USA (45.636738°N, -119.860457°W). The Preserve occupies 9,163 ha of arid grassland and shrub-steppe from 120-295 m elevation (Fig. 5.1). Common invasive grasses include cheatgrass (Bromus tectorum L.) and medusahead (Taeniatherum caput-medusae (L.) Nevski), while native grasses include bluebunch wheatgrass (Pseudoroegneria spicata (Pursh) Á. Löve), Sandberg bluegrass (Poa secunda J. Prsel), bottlebrush squirreltail (Elymus elymoides (Raf.) Swezey ssp. brevifolius (J.G. Sm.) Barkworth), and needle and thread grass (Heterostipa comata

104

(Trin. and Rupr.) Barkworth). The average precipitation is 22 cm with the majority of the precipitation falling from November to February and annual average temperatures range from 5–18°C with temperatures frequently reaching 32°C in the summer (30 year average, US Climate Data 2017). Site Selection As part of a larger study examining the effects of grassland restoration on beneficial invertebrates (Smith DiCarlo and DeBano 2018), we established 18 sites across the Preserve in 2014; all sites were accessible by old farm roads, separated on average by 623 m, and were on relatively flat slopes (Fig. 5.1). On June 1, 2015 a large wildfire started from a haystack that spontaneously combusted just outside of the preserve. In two days the fire burned across approximately 8 km and over 16,000 hectares (Plaven 2015), leaving large areas of moderate to severe surface burns on the preserve (Fig. 5.2) (Keeley 2009). In 2016, one year after the fire, we resumed sampling at the original 18 sites, five of which had burned. Spider Sampling At each site, eight pitfall traps were placed in a 10 m radius circle (Fig. 5.3). Pitfall traps, 470 ml plastic cups filled with wildlife-friendly propylene glycol and placed flush with the soil, are well-suited for collecting ground-active spiders (Kelton 1978). Spiders were collected at all sites three times each year during June-July, July-August, and August-September by opening traps for one week each time period to collect invertebrates. After one week, traps from each site were collected, combined, and transported to the laboratory. Samples were then washed over a 250 μm sieve and spiders sorted from other invertebrates and debris and preserved in 70% ethanol. All juvenile spiders were identified to family and all mature spiders were identified to species, if possible. Bee Sampling Bees were sampled using pan traps, 236 ml plastic cups painted fluorescent blue, yellow, and white to attract different pollinators. Using plant stakes, all pan traps were elevated approximately 1 m above ground to be at a similar height as the vegetation and filled with a solution of water and detergent. Nine pan traps were placed at each site in a

105

10 m radius circle, with one in the center (Fig. 5.3). Bees were collected during the same periods as spiders, except that bee sampling was not conducted during the June-July sampling bout, and pan traps were collected in the field after two days. All traps from each site were combined and transported to the laboratory where bees were washed, dried, pinned, and identified. All specimens were identified to species, if possible. Habitat Survey To determine which habitat characteristics were related to spider and bee abundance, richness, and species composition, each site was surveyed for environmental variables once in 2014 and three times in 2016, coinciding with invertebrate sampling. Variables were estimated to the nearest 5% cover in 16 63 x 63 cm subplots located in a 50 x 50 m square around the pitfall and pan traps (Fig. 5.3). Variables estimated included the percent cover of invasive annual grasses (cheatgrass (B. tectorum) and medusahead (T. caput-medusae)), biological soil crusts, and litter. We also estimated maximum vegetation height by measuring the tallest stem in each subplot. During bee sampling we also counted and identified every blooming plant within the 50 x 50 m square to determine forb density and species richness. Analyses Spider and bee abundance at each site was characterized by the average number of individuals per trap as not all traps were present during collection due to weather or animal tampering. Prefire and postfire abundance, diversity, and richness data were averaged over the 3 sampling bouts for spiders and 2 sampling bouts for bees in both 2014 and 2016. The Shannon-Weiner index was used to estimate spider and bee diversity of each site. Habitat variables for each site were calculated by first averaging subplots by site and then averaging over the number of sampling bouts per year (1 in 2014, 3 in 2016). Floral abundance and richness data were averaged over 2 sampling bouts in both 2014 and 2016. To determine if the wildfire had an impact on spider or bee abundance, diversity, richness and habitat variables a before-after-control-impact (BACI) design was used with five impact (burned) sites and thirteen control (unburned) sites with data from one year before and one year after. Because having more control sites than impact sites enhances

106 the power to detect differences (Underwood 1994), we used all control sites in analyses. With the raw data, a two-factor analysis of variance (ANOVA) was used to include the three sources of variation in the design: sampling periods within a site, site-to-site variation, and site-to-year variation (Smith 2002, table 3; Schwarz 2015). We also calculated the estimated BACI effect (the differential change between the before and after periods, Schwarz 2015) and estimated sampling period-to-sampling period variation within each site year, site-to-site variation, and the site-to-year interaction variance. Analyses resulting in significant p-values (alpha <0.05) and a small estimated interaction of site-to-year variation (indicating that the site responses over time is consistent) were considered to have had an effect due to the wildfire (Schwarz 2015). All BACI analyses were completed in RStudio 1.0.153 (RStudio Team 2015). PC-ORD Software version 7.287 (McCune and Mefford 2015) was used for all community analyses. Family level abundance was used in all multivariate analyses for spiders instead of genus or species abundance to allow for use of juvenile spiders in the analysis (juvenile spiders cannot be positively identified to genus or species without introducing error); however, species level abundance was used for bees and flowering forbs. The spider family dataset contained average spiders per pitfall trap (18 sites for each year (36 total) x 10 families), the bee species dataset contained average bees per pan trap (18 sites for each year (36 total) x 46 species), and the forb species dataset contained total stem counts (18 sites for each year (36 total) x 21 species). The environmental dataset contained habitat variables (18 sites each year (36 total) x 6 variables) including: average maximum vegetation height, forb abundance, forb richness, and average percent cover of invasive annual grasses, litter, and biological soil crust. Non-metric multidimensional scaling (NMS) with Sorensen distances was used to ordinate sites in the spider family space matrix, the bee and forb species space matrices, and the environmental matrix. NMS does not assume linearity between family or species response and environmental gradients and exposes relationships between the family matrix and the environmental matrix (McCune and Grace 2002). NMS was performed with 250 random starts and ties were not penalized. A randomization procedure was included to test if solutions were stronger than those obtained by chance, resulting in p-

107 values. R2 values were calculated to represent the percent variance explained by each axis, and relationships of each axis with spider families and habitat variables were quantified with Pearson correlation coefficients. Multi-response Permutation Procedures (MRPP) were used with Sorensen distances to test for differences in family and species composition across sites among groups, with four groups: control prefire, control postfire, burned prefire, and burned postfire. Pairwise comparisons resulted in A-statistics, the chance-corrected within- group agreement, and p-values. Indicator Species Analysis was performed to assess family-specific (spiders) and species-specific associations across groups (bees and flowering plants). Resulting indicator values ranged from 0 to 100, with higher scores indicating stronger associations between taxa and treatments. A randomization test was used to test for statistical significance of the indicator values by using 4999 random permutations, resulting in a p- value for each indicator value. Families or species with high indicator values and significant p-values (alpha <0.05) were considered indicative of that treatment.

Results We collected 811 and 895 spiders in 2014 (prefire) and in 2016 (postfire) and identified 30 species in 10 families (20 species in 2014 and 15 species in 2016). In 2014, approximately 15% of spiders were mature and in 2016 approximately 12% were mature. We collected 4,109 bees in 2014 and 3,392 bees in 2016. We identified 46 bee species in five families, with 37 species in 2014 and 31 species in 2016. Over the two years we counted 14,737 blooming plant stems and identified 11 species in 2014 and 19 species in 2016 (21 species in total). Spiders Spider abundance, diversity, and richness did not change significantly in response to the fire (F (1, 16) = 0.11, p = 0.73; F (1, 16) = 1.78, p = 0.20; F (1, 16) = 2.32, p = 0.14, respectively) (Fig. 5.4). Community composition of spiders differed among sites (Figs. 5.5a); the NMS randomization procedure resulted in a stable three-dimensional solution (final stress = 8.98, final instability = 0.00, p = 0.02) with a cumulative R2 of

108

93.6 for spiders. Axis 1 accounted for 55.8% of the variation in spider family space, axis 2 accounted for 28.5%, and axis 3 accounted for 9.3%. Pearson correlations between the three axes and spider taxa and environmental variables are listed in Table 5.1. Separation among treatments types occurred on both Axes 1 and 2 (Fig. 5.5a), and centroids for each of the four treatments is shown in Fig. 5.5b. MRPP showed that spider community composition differed significantly among treatments (A = 0.15, p < 0.00001). While pairwise comparisons showed that prefire spider communities differed significantly from postfire communities for both control and burned sites (A = 0.03, p = 0.04; A = 0.18, p = 0.002, respectively), the distance between centroids of prefire and postfire burned sites was much greater than the distance between centroids of prefire and postfire control sites (Fig. 5.5b), indicating that spider communities at burned sites changed more than those at unburned sites. In general, all spider communities became less dominated by jumping spiders (Salticidae) and more dominated by wolf spiders, (Lycosidae), cobweb spiders (Theridiidae), and crab spiders (Thomisidae) from 2014 to 2016; however, sites that burned showed more of an increase in cobweb spiders and sites that did not burn showed more of an increase in wolf and crab spiders (Fig. 5.5). Indicator analysis revealed that sheet web spiders (Linyphiidae) were strongly associated with the postfire burned sites, while wolf and crab spiders were strongly associated with postfire control sites (Table 5.2). Bees Bee abundance did not differ significantly between burned and unburned sites (F (1, 16) = 3.63, p = 0.07), but bee diversity increased in burned sites (F (1, 16) = 5.1, p = 0.04, BACI = -0.35 ± 0.15) and richness showed a similar trend (F (1, 16) = 4.27, p = 0.06, BACI = -3.71 ± 1.79) (Fig. 5.4). The NMS randomization procedure resulted in a stable three-dimensional solution (final stress = 8.36, final instability = 0.00, p = 0.004) with a cumulative R2 of 92.6. Axis 1 accounted for 41.4% of the variation in bee species space, axis 2 accounted for 40.7%, and axis 3 accounted for 10.5%. Pearson correlations between the three axes and bee taxa and environmental variables are listed in Table 5.3. Separation among treatments types occurred on both Axes 1 and 2 (Fig. 5.6a), and centroids for each of the four treatments are shown in Fig. 5.6b. MRPP showed that bee

109 community composition differed significantly among treatments (A = 0.18, p < 0.00001). Pairwise comparisons showed that prefire bee communities differed significantly from postfire communities for both control and burned sites (A = 0.15, p < 0.0001; A = 0.11, p = 0.01); however, unlike spiders, the distance between centroids of prefire and postfire control sites was much greater than the distance between centroids of prefire and postfire burned sites (Fig. 5.6b), indicating that bee communities at control sites changed more than those at burned sites. In general, bee communities at burned and unburned sites were similar prior to the burn, but after the fire, unburned communities were dominated by sweat bees (Halictus and Lasioglossum), and burned communities were dominated by a number of other species in six different genera (Fig. 5.6b). Indicator species analysis revealed significant indicator species for all four treatments (Table 5.2). Species indicative of prefire burned sites were small mining bees (Perdita). In contrast, indicator species for postfire burned sites included large bees, many of which are known generalists (e.g., Apis mellifera) (Table 5.1). Indicators species for control sites, both prefire and postfire, were primarily species of long-horned bees (Melissodes). Habitat Variables Invasive annual grass cover and biological soil crust cover significantly decreased in burned sites (F (1, 16) = 34.53, p < 0.0001, BACI = 38.30 ± 6.52; F (1, 16) = 18.92, p = 0.0005, BACI = 28.73 ± 6.60) but there was no change in litter cover or maximum vegetation height (F (1, 16) = 0.01, p = 0.93; F (1, 16) = 0.09, p = 0.77, respectively) (Fig. 5.7). There was also a significant effect of the fire on forb abundance (F (1, 16) = 4.99, p = 0.04, BACI = 331.42 + 148); but the wildfire did not significantly affect forb richness (F (1, 16) = 0.22, p = 0.64) (Fig. 5.8). A NMS randomization procedure for forbs did not result in a stable solution. Indicator analysis resulted in eight blooming plant species for all treatments except for control prefire (Table 5.1). Burned postfire indicator species included Canadian horseweed (Conyza canadensis (L.) Cronquist), tall willowherb (Epilobium brachycarpum C. Presl), hoary tansyaster (Machaeranthera canescens (Pursh) A. Gray), Douglas’ knotweed (Polygonum douglasii Greene), and Russian thistle (Salsola tragus L.).

110

Discussion This study is the first to document responses of two beneficial invertebrate groups to wildfire in a Pacific Northwest grassland. Wildfire affected both spider and bee communities one year after the burn. For spiders, responses were strongest relative to community composition, with dominant families differing between burned and control sites. Bee communities, on the other hand, responded to wildfire not only with compositional changes, but also with increases in diversity and richness. Although direct mortality associated with fire can affect invertebrate communities, changes are more commonly tied to indirect effects of wildfire on habitat characteristics that can both positively and negatively affect certain taxa. In our system, habitat responses associated with wildfire included decreases in invasive annual grass and biological soil crust cover. Biological soil crusts are known to burn in high-intensity fires and can be lost if fire intervals are shorter than the recovery period (Green et al. 1990; Whisenant 1990; Johanesen 1993; Belnap 2000). However, the decrease in invasive annual grass cover observed in our system differs from other studies that have found increases in cheatgrass within the first year after burn, especially at low elevation sites (West and Hassan 1985; Davies et al. 2012). Invasive annual grass cover may have decreased in our system due to the early season timing of the fire (Meyer et al. 2000) or because of the relatively low coverage of invasive grasses within burned sites before the burn. Several other habitat variables did not show significant responses to wildfire, including maximum vegetation height. Vegetation height may not have been strongly affected because of the potentially rapid regeneration of bunchgrasses (Uresk 1980; West and Hassan 1985) or because of the presence of pockets of unburned tall bunchgrasses within burned sites. While perennial bunchgrasses can reestablish cover at preburn levels (West and Hassan 1985), some studies have found that the grass layer may be reduced and not completely recover until 2.5 years after the fire (Britton et al. 1990; Samu et al 2010). Litter cover also did not respond to wildfire, a result also found by Niwa and Peck (2002) in a forested system. This was somewhat surprising because the fire burned with moderate severity at our study site, and both litter and the majority of grasses in burned areas combusted completely (Fig. 5.2). Thus, we would have expected

111 to find a decrease in litter one year after the fire, given the destruction of the existing litter layer and the removal of grasses that would have contributed to a new layer. One explanation is that litter may have blown into the area from surrounding intact habitats (the area is known for its strong wind gusts and dust storms). In addition, we did not measure changes in litter biomass or complexity, which were most likely reduced after the fire and are known to have large impacts on the spider communities (Uetz 1979; Bell et al. 2001). Surprisingly, observed changes in habitat variables to wildfire did not translate to responses in spider abundance, diversity, and richness, a result that differs from other studies that found significant responses in one or more of these variables (Bell et al. 2001; Bess et al. 2002; Koponen 2005; Larrivee et al. 2002; Langlands et al. 2012; Pryke and Samways 2012). Declines in spider abundance after wildfire found in other studies may be due to direct mortality or to changes in plant communities and litter (Bess et al. 2002; Moretti et al. 2002; Richardson and Hanks 2009). In contrast to trends in abundance, both spider richness and diversity have been found to increase over time after wildfire (Merrett 1976; Haskins and Shaddy 1986; Buddle et al. 2000, Kopenen 2005, Langlands et al. 2012), potentially because of the survival of species present before the fire and the addition of colonizing species that specialize in open habitats. The lack of response observed in spider abundance in our study may be due to the fact that there were no detectable effects of fire on litter; previous work in this system showed that spider abundance is closely tied to litter cover (Smith DiCarlo and DeBano 2018; Smith et al., in review). In addition, we found that all families except for funnel weavers (Agelenidae) were present at burned sites, indicating that many spider species may have survived underground or in unburned patches and then quickly recolonized after the fire. While we did see some early colonizers, they were not particularly abundant or diverse. Similar to our results, Samu et al. (2010) found no responses in spider abundance or richness to wildfire and suggested that this was due to a relatively quick recovery of vegetation and a lack of pioneer species. In contrast to spider abundance, diversity, and richness, spider community composition did respond to wildfire. Families such as wolf spiders (Lycosidae), cobweb

112 spiders (Theridiidae), and crab spiders (Thomisidae) were abundant at all sites, including burned ones, which may be explained by the fact that successful recolonization is often tied to prior dominance, with more abundant arthropod species more likely to be successful at recolonization after fire (Bess et al. 2002). Individuals in these families may have survived the fire or quickly colonized from adjacent or unburned habitat, especially if the fire did not burn uniformly and left pockets of surviving individuals (Swengel 2001; Moretti et al. 2002). However, despite this, community composition at burned and unburned sites diverged. Communities at burned sites became more dominated by cobweb spiders and sheet weavers (linyphyiids), while control sites were more dominated by wolf spiders and crab spiders. Several studies have found similar results in grasslands, with cobweb and sheet weaver spiders dominating burned areas, most likely due to these families’ high tolerance of changes in microclimate (Riechert and Reeder 1972; Merrett 1976; Bell et al. 2001). Sheet weavers may also be indicative of burned treatments due to their lack of dependence on vegetative structure and their ability to disperse through ballooning throughout their entire lives (Bell et al. 2001). However, other studies have shown different responses in community composition. Koponen (2005) found wolf spiders dominated in burned sites and sheet weavers dominated in non-burned sites in a forested system. These differences may be due to species-specific preferences. The dominant wolf spider found in Koponen’s study (Xerolycosa nemoralis) is known to colonize open areas while the species in our study belong to the Schizocosa genus. In our study area, both Schizocosa and crab spiders are more abundant in heavily vegetated areas dominated by invasive annual grasses (Smith DiCarlo and DeBano 2018; Smith et al., in review). In addition, Schizocosa are also known to respond positively to high soil moisture and full litter cover, which may be reduced in burned areas (Stratton 1991). In contrast to spiders, bees did show responses in diversity and richness to wildfire, with both variables increasing. Like spiders, however, abundance was unaffected. Other studies found similar results with higher bee species richness and diversity after fires, with both peaking several years after the burn (1-3 years) due to increases in nesting resources, including charred wood, and in floral resources (plant

113 richness and abundance) (Potts et al. 2003; Moretti et al. 2009; Grundel et al. 2010; Bogusch et al. 2015). After these peaks, bee diversity and richness may steadily decrease with time as vegetation recovers and pioneering forbs are outcompeted and bare ground available for nesting decreases (Potts et al. 2003; Bogusch et al. 2015). These longer-term responses differ from very short-term responses; for example, Love and Cane (2016) found fewer active species 14 to 21 days after the fire, most likely due to mortality. In our study, higher bee richness and diversity after the fire may have been driven by increased nesting habitat for some species of bees. The fire significantly decreased invasive annual grass cover, potentially providing ground-nesting bees with additional nesting resources. In addition, while the fire intensity was high, a majority of aboveground vegetation was burned leaving some charred stems from larger sagebrush and juniper plants, potentially providing additional nesting resources for aboveground nesters. Increases in forb abundance may have also provided additional nectar and pollen resources although forb richness did not significantly differ due the burn. Bee communities diverged in composition after the fire and several species were found to be indicative of burned sites that were not collected at the sites before the fire, including Halictus tripartitus, Megachile coquilletti and two Nomada species. One of these species nest in the soil (H. tripartitus, Kim et al. 2006) and may benefit from decreased invasive annual grass species cover, potentially opening up additional nesting sites. M. coquilletti is the only species that nests above-ground in cavities (Sheffield et al. 2011) and may take advantage of the recently burned woody material. The two unknown Nomada species are cuckoo bees that parasitize many other bee species nests including Agapostemon species (Eickwort and Abrams 1980), which were more dominant in sites that burned compared to sites that did not. Shifts in bee community composition after fire may not only be due to changes in soil and vegetative structural characteristics that affect nesting habitat, but also to changes in forb diversity and richness (Ne’eman et al. 2000; Moretti et al. 2009; Grundel 2010; Pryke and Samways 2013; Love and Cane 2016). Several studies have found that wildfire increases forb abundance and/or richness (Moretti et al. 2009; Grundel et al. 2010; Mola and Williams 2018). We found forb abundance increased after the fire and

114 that fire influenced species composition. Several forb species were indicative of burned sites, including Douglas’ knotweed and Russian thistle (both invasive), and Canadian horseweed, tall willowherb, and hoary tansyaster (all native). Canadian horseweed and hoary tansyaster are both from the Asteraceae family while Douglas’ knotweed, Russian thistle, and tall willowherb are from Polygonaceae, Chenopodiaceae, and Onagraceae, respectively. Unfortunately, bee-floral associations are not well known in this area, but bee species that were indicators for burned sites are known to visit each plant family. Agapostemon, Anthophora, Apis, Halictus, and Megachile visit members of the Asteraceae family, with Anthophora urbana a known pollinator of hoary tansyaster (U.S. National Pollinating Insects Database, 2014). While fewer genera are known to pollinate Chenopodiaceae plants, two genera associated with burned sites (Agapostemon and Halictus) are pollinators of species within this plant family, with Agapostemon femoratus recorded on Russian thistle specifically (U.S. National Pollinating Insects Database, 2014). Agapostemon, Anthophora, Apis, Halictus, and Megachile are all known pollinators of species in Onagraceae and Polygonaceae (U.S. National Pollinating Insects Database, 2014). Thus, changes in the dominance of these forb species after wildfire may also contribute to observed changes in bee community composition, richness, and diversity. This study is also unusual in that we had the rare opportunity to collect data before and after wildfire in both burned and control sites; the vast majority of studies are only able to collect data after wildfire. Using a BACI design allowed us to make inferences about how the fire (rather than temporal differences) affected bee, spider, and forb communities. However, there are also limitations with BACI designs, including using them with sites that show highly variable responses over time, decreasing the tests power (Underwood 1991; Schwarz 2015). While this was not an issue with bee and spider communities in our study, floral abundance and richness were highly variable, both spatially and temporally, and several of our sites had mass-flowering species (1000+ flowering stems from one species) while others had no blooming stems during a sampling period.

115

Invertebrate response to wildfire can vary widely with respect to many factors including fire timing, intensity, season of sampling, and geographical location. This study extends our understanding of geographical patterns of wildfire effects on beneficial invertebrates in rangelands by providing the first study of a Pacific Northwest grassland focusing on both native bee and spider communities. In addition, using a BACI design gave us an opportunity to focus directly on community responses and habitat variables that effect change within these communities. This knowledge is imperative, especially as wildfire continues to increase in frequency and intensity across western rangelands.

Acknowledgements We thank Leslie Nelson at TNC Boardman Grasslands for help in planning and implementing the study, and LJ Smith, SR Roof, EM Campbell, KL Kirby, L McDaniel, LK Waianuhea, and BE Price for their help in the field and laboratory. This work was supported by a USDA NIFA National Needs Graduate Fellowship [grant number 2012- 04150] and funding from Oregon State University’s General Research Fund and the Provost’s Branch Experiment Station Experiential Learning Program. Additional funding was provided by a USDA Western Sustainable Agriculture Research and Education Graduate Student Grant [grant number GW16-016], a TNC Oren Pollak Memorial Student Research Grant for Grassland Science, a Soil and Water Conservation Society Kenneth E. Grant Research Scholarship, the Prairie Biotic Small Research Grants Program, and a Society for Ecological Restoration Northwest Chapter Student Research Grant.

Literature Cited

Bell, J.R., Wheater, C.P., Cullen, W.R., 2001. The implications of grassland and heathland management for the conservation of spider communities: a review. J Zool. 255, 377–387. Belnap, J., 2000. Structure and function of biological soil crusts. In Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., compilers, 2000. Proceedings: Sagebrush Steppe Ecosystem Symposium. Bureau of Land Management Publication No. BLM/ID/PT-001001+1150, Boise, Idaho, USA.

116

Bess, E.C., Parmenter, R.R., McCoy, S., Molles Jr., M.C., 2002. Responses of a riparian forest-floor arthropod community to wildfire in the Middle Rio Grande Valley, New Mexico, Env Entomol, 31, 774–784. Black, S.H., Shepherd, M., Vaughan, M., 2011. Rangeland management for pollinators. Rangelands. 33, 9-13. Bock, C.E., Bock, J.H., 1991. Response of grasshoppers (Orthoptera: Acrididae) to wildfire in a southeastern Arizona grassland. Am Midl Nat. 1, 162-167. Bogusch, P., Blazej, L., Milos, T., Heneberg, P., 2015. Forgotten role of fires in Central European forests: critical importance of early postfire successional stages for bees and wasps (Hymenoptera: Aculeata). Eur J Forest Res. 134, 153-166. Britton, C.M., McPherson, G.R., Sneva, F.A., 1990. Effects of burning and clipping on five bunchgrasses in eastern Oregon. Great Basin Nat. 50, 115-120. Buddle, C.M., Spence, J.R., Langor, D.W., 2000. Succssion of boreal forest spider assemblages following wildfire and harvesting. Ecography. 23, 424-436. Burel, F., Baudry, J., Butet, A., Clergeau, P., Delettre, Y., Le Coeur, D., Dubs, F., Morvan, N., Paillat, G., Petit, S., Thenail, B., Brunel, E., Lefeuvre, J., 1998. Comparative biodiversity along a gradient of agricultural landscapes. Acta Oecol. 19, 47-60. Campbell, J.W., Hanula, J.L., Waldrop, T.A., 2007. Effects of prescribed fire and fire surrogates on floral visiting insects of the Blue Ridge province in North Carolina. Biol Conserv. 134, 393-404. Cane, J.H., Neff, J.L., 2011. Predicted fates of ground-nesting bees in soil heated by wildfire: Thermal tolerances of life stages and a survey of nesting depths. Biol Conserv. 144, 2631-2636. Cane, J.H., Love, B., 2016. Floral guilds of bees in sagebrush steppe: Comparing bee usage of wildflowers available for postfire restoration. Nat Area J. 36, 377-391. Davies, G.M., Bakker, J.D., Dettweiler-Robinson, E., Dunwiddie, P.W., Hall, S.A., Downs, J., Evans, J., 2012. Trajectories of change in sagebrush steppe vegetation communities in relation to multiple wildfires. Ecol Appl. 22, 1562-1577. DeBano, S.J. 2006. Effects of livestock grazing on insect communities in semi-arid grasslands of southeastern Arizona. Biodivers Conserv. 15, 2547-2564. Eickwort, G.C., Abrams, J., 1980. Parasitism of sweat bees in the genus Agapostemon by cuckoo bees in the genus Nomada (Hymenoptera: Halictidae, Anthophoridae). Pan-Pac Entomol. 56, 144-152. Greene, R.S.B., Chartres, C.J., Hodgkinson, K.C., 1990. The effects of fire on the soil in a degraded semiarid woodland. I. Crypotogam cover and physical and micromorphological properties. Aust J Soil Res. 28, 755-777. Grundel, R., Jean, R.P., Frohnapple, K.J., Glowacki, G.A., Scott, P.E., Pavlovic, N.B., 2010. Floral and nesting resources, habitat structure, and fire influence bee

117

distribution across an open-forest gradient. Ecol Appl. 20, 1678-1692. Haskins, M.F., Shaddy, J.H., 1986. The ecological effects of burning, mowing, and plowing on ground-inhabiting spiders (Araneae) in an old-field ecosystem. J Arachnol, 14, 1-14. Johansen, J.R., 1993. Cryptogamic crusts of semiarid and arid lands of North America. J Phycol. 29, 140-147. Keeley, J.E. 2009. Fire intensity, fire severity and burn severity: a brief review and suggested usage. Int J Wildland Fire. 18, 116-126. Kelton, L.A., 1978. The Insects and Arachnids of Canada. Biosystemics Research Institute, Ottawa, Ontario. Kennedy, P.L., DeBano, S.J., Bartuszevige, A., Lueders, A., 2009. Effects of native and nonnative grassland plant communities on breeding passerine birds: implications for restoration of Northwest bunchgrass prairie. Restor Ecol. 17, 515-525 Kim, J., Williams, N., Kremen, C. 2006. Effects of cultivation and proximity to natural habitat on ground-nesting native bees in California sunflower fields. J Kansas Entomol Soc. 79: 309-320. Kimoto, C., DeBano S.J., Thorp, R.W., Rao, S., Stephen, W.P., 2012a. Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci. 12, 108. available online: http://www.insectscience.org/12.108 Kimoto, C., DeBano, S.J., Thorp, R.W., Taylor, R,V., Schmalz, H., DelCurto, T., Johnson, T., Kennedy, P.L., Rao, S., 2012b. Livestock and native bee communities: short-term responses to grazing intensity and implications for managing ecosystem services in grasslands. Ecosphere. 3, 88. available online: http://www.esajournals.org/doi/pdf/10.1890/ES12-00118.1 Klein, A.M., Vaissie’re, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Important of pollinators in changing landscapes for world crops. P Roy Soc B-Biol Sci. 274, 303-313. Koponen, S., 2005. Early succession of a boreal spider community after forest fire. J Arachnol. 33, 230-235. Kremen, C., Williams, N.M., Thorp, R.W., 2002. Crop pollination from native bees as risk from agricultural intensification. PNAS. 99, 16812-16816. Langlands, P.R., Brennan, K.E.C., Ward, B., 2012. Is the reassembly of an arid spider assemblage following fire deterministic?: Successional trajectory of a spider assemblage. Austral Ecol. 37, 429-439. Love, B.G., Cane, J.H., 2016. Limited direct effects of a massive wildfire on its sagebrush steppe bee community. Ecol Entomol. 3, 317-326. Malumbres-Olarte, J., Vink, C.J., Ross, J.G., Cruickshank, R.H., Paterson, A.M., 2013. The role of habitat complexity on spider communities in native alpine grasslands

118

of New Zealand: Habitat complexity and alpine spiders. Insect Conserv Diver. 6, 124–134. McCune, B., Grace, J., 2002. Analysis of Ecological Communities. MjM Software, Gleneden Beach, Oregon, USA. McCune, B., Mefford, M.J., 2015. PC-ORD Version 7.287. Multivariate Analysis of Ecological Data. MjM Software, Gleneden Beach, Oregon, USA. Merrett, P., 1976. The phenology of spiders on heathland in Dorest. Families Lycosidae, Pisauridae, Agelenidae, Mimetidae, Theridiidae, Tetragnatidae, Argiopidae. J Zool. 156, 239-256. Meyer, S.E., Nelson, D.L., Clement, S., Paulsen, A. 2000. Exploring the potential for biocontrol of cheatgrass with the head smut pathogen. In Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., compilers, 2000. Proceedings: Sagebrush Steppe Ecosystem Symposium. Bureau of Land Management Publication No. BLM/ID/PT-001001+1150, Boise, Idaho, USA. Michener, C.D., 2007. The bees of the world, 2nd edition. The John Hopkins University Press, Baltimore, USA. Mirshamski Kakhki, O., 2005. Faunistic study of spiders in Khorasan Province, Iran (Arachnida: Araneae). Int J Agric Biol. 1, 55-66. Mola, J.M., Williams, N.M., 2018. Fire-induced change in floral abundance, density, and phenology benefits bumble bee foragers. Ecosphere. 9:e02056. Moretti, M., Conedera, M., Duelli, P. Edwards, P.J., 2002. The effects of wildfire on ground-active spiders in deciduous forests on the Swiss southern slope of the Alps. J Appl Ecol. 39, 321-336. Niwa, C.G., Peck, R.W., 2002. Influence of prescribed fire on carabid beetle (Carabidae) and spider (Araneae) assemblages in forest litter in southwestern Oregon. Environ Entomol. 5, 785-796. Plaven, G. 2015. Update: Firefighters back on attack of Boardman blaze. http://www.eastoregonian.com/eo/local-news/20150602/haystack-combusts- starts-large-fire-near-boardman (accessed 21 February 2018). Potts, S. G., Dafni, A., Ne’eman, G., 2001. Pollination of a core flowering shrub species in Mediterranean phrygana: variation in pollinator diversity, abundance and effectiveness in response to fire. Oikos. 92, 71-80. Potts, S.G., Vulliamy, B., Dafni, A., Ne’eman, G., Willmer, P., 2003. Response of plant- pollinator communities to fire: changes in diversity, abundance and floral reward structure. Oikos. 101, 103-112. Pryke, J.S., Samways, M.J., 2012. Differential resilience of invertebrates to fire. Austral Ecol. 37, 460-469. Pyke, D.A., 1999. Invasive exotic plants in sagebrush ecosystems of the Intermountain West. In: Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., eds.

119

Proceedings: Sagebrush Steppe Ecosystems Symposium; Boise, Idaho; June 23- 25, 1999. Boise, Idaho: Bureau of Land Management, pp. 43-54. Richardson, M.L., Hanks, L.M., 2009. Effects of grassland succession on communities of orb-weaving spiders. Environ. Entomol. 38, 1595-9. Riechert, S.E., Reeder, W.G., 1972. Effects of fire on the spider distribution in southwestern Wisconsin prairies. In Proceedings 2nd Midwest Prairie Conference: 73-90. Zimmermann, J.H. (Ed.). Maddison: University of Wisconsin. Sampson, F., Knopf, F, 1994. Prairie conservation in North America. BioScience. 44:418-421. Samu, F., Kadar, F., ONodi, G., Kertesz, M., Sziranyi, A., Szita, E., Fetyko, K., Neidert, D., Botos, E., Altbacker, V., 2010. Differential ecological responses of two generalist arthropod groups, spiders and carabid beetles (Araneae, Carabidae), to the effects of wildfire. Community Ecol. 11, 129-139. Schoennagel, T., Balch, J.K., Brenkert-Smith, H, Dennison, P.E., Harvey, B.J., Krawchuk, M.A., Mietkiewicz, N., Morgan, P., Moritz, M.A., Rasker, R., Turner, M.G., Whitlock, C., 2017. Adapt to more wildfire in western North American forests as climate changes. PNAS. 114, 4582-4590. Schwarz, C.J., 2015. Analysis of BACI experiments. In Course Notes for Beginning and Intermediate Statistics. http://www.stat.sfu.ca/~cschwarz/CourseNotes (accessed 25 February 2018). Sheffield, C.S., Ratti, C., Packer, L., Griswold, T. 2011. Leafcutter and mason bees of the genus Megachile Latreille (Hymenoptera: Megachilidae) in Canada and Alaska. Can J Arthrop Identification. 18, 1-107. Smith, E.P., 2002. BACI design, in: El-Shaarawi, A.H., Piegorsch, W.W., (Eds.), Encyclopedia of Environmetrics. John Wiley and Sons, Ltd, Chichester, pp. 141- 148. Smith, L.J., Smith DiCarlo, L.A., DeBano, S.J., In review. Ground crab spiders (Thomisidae: Xysticus) more abundant in cheatgrass (Bromus tectorum L.) invaded grasslands. Biol Invasions. Smith DiCarlo, L.A., DeBano, S.J., 2018. Spider community responses to grassland restoration: balancing tradeoffs between abundance and diversity. Restor Ecol. DOI: 10.1111/rec.12832. Stratton, G.E., 1991. A new species of wolf spider, Schizocosa stridulans (Araneae, Lycosidae). J Arach. 19, 2 9-39. Sunderland, K., 1999. Mechanisms underlying the effects of spiders on pest populations. J Arachnol. 27, 308-316. Swengel, A.B., 2001. A literature review of insect responses to fire, compared to other conservation managements of open habitat. Biodivers Conserv. 10, 1141-1169. Tubbesing, C., Strohm, C., DeBano, S,J., Gonzalez, N,, Kimoto, C., Taylor, R.V., 2014.

120

Insect visitors and pollination ecology of Spalding’s catchfly (Silene spaldingii) in the Zumwalt Prairie of northeastern Oregon. Nat Areas J. 34, 200-211. Uetz, G.W., 1979. The influence of variation in litter habitats on spider communities. Oecologia. 40, 29-42. Underwood, A.J., 1991. Beyond BACI: experimental designs for detecting human environmental impacts on temporal variations in natural populations. Aust J Mar Freshwater Res. 42, 569-587. Underwood, A.J., 1994. On beyond BACI—sampling designs that might reliably detect environmental disturbances. Ecol Appl. 4, 3-15. Uresk, D.W., Rickard, W.H., Cline, J.F., 1980. Perennial grasses and their response to a wildfire in south-central Washington. J Range Manage. 33, 111-114. [database] U.S. National Pollinating Insects Database. USDA-ARS Pollination Insects – Biology, Management and Systematics Research Unit Museum Records. Bee Biology and Systematics Laboratory, Logan, Utah (Accessed 2018-15-4, through DiscoverLife http://www.discoverlife.org). Weisser, W.W., Siemann, E., 2004. The various effects of insects on ecosystem functioning. Insects Ecosyst Function. 173, 3–24. West, N.E., Hassan, M.A., 1985. Recovery of sagebrush-grass vegetation following wildfire. J Range Manage. 38, 131-134. West, N.E. 2000. Synecology and disturbance regimes of sagebrush steppe ecosystems, p. 15-26. In P.G. Entwistle, et al. (compilers), Proceedings: sagebrush steppe ecosystems symposium. USDI Bureau of Land Management Publication BLM/ID/PT-001001+1150, Boise, ID. Whisenant, S.G., 1990. Changing fire frequencies on Idaho’s Snake River Plains: ecological and management implications. Pages 4-10 in McArthur, E.D., Romney, E.M., Smith, S.D., Tueller, P.T., editors. Proceedings—symposium on cheatgrass invasion, shrub die-off, and other aspects of shrub biology and management. USDA Forest Service General Technical Report INT-276, Ogden, Utah, USA. Williams, N.M., et al., 2010. Ecological and life-history traits predict bee species responses to environmental disturbances. Biol Conserv. 143, 2280-2291. Winfree, R., Griswold, T., Kremen, C. 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conserv Biol. 21, 213-223. Zamani, A., Rafinejad, J., 2014. First record of the Mediterranean recluse spider Loxosceles rufescens (Araneae: Sicariidae) from Iran. J Arthropod-Borne Di. 8, 228-231.

121

Figure 5.1. Map of The Nature Conservancy Boardman Grasslands Preserve with an inset of the location in Oregon, USA. Stars represent burned sites and circles represent control sites.

Figure 5.2. A. Photograph of a burned site in 2014, one year before the fire. B. Photograph of a burned site showing moderate to severe surface burn in 2015, seven days after the fire. C. Photograph of vegetation regrowth at a site in 2016, one year after the fire.

122

Figure 5.3. Sample plot configuration, with 50 m survey square around 10 m plot radius. Closed circles represent pitfalls, open circles represent pan traps, and closed squares represent habitat survey subplots.

20 2 20 A Burned B Burned C Burned Control Control * Control * 15 15

A Weiner Ind Ind Weiner

10 - 1 10

5 5

AverageAbundance 0 0 0

PreFire PostFire PreFire PostFire PreFire PostFire PreFire PostFire PreFire PostFire PreFire PostFire

AverageShannon AverageRichnessSpecies Spiders Bees Spiders Bees Spiders Bees

Figure 5.4. A. Average individual per trap prefire and postfire for both spiders and bees. B. Average Shannon-Weiner Index prefire and postfire for both spiders and bees. C. Average species richness prefire and postfire for both spiders and bees. Stars denote significant BACI results. N = 5 burned sites, N = 13 control sites.

123

Figure 5.5. A. Non-metric multidimensional scaling ordination of sites in spider family space along with weighted average positions for spider families for axes 1 vs. 2. Each triangle represents a site and each circle point represents a spider family. A joint-plot from the environmental matrix is overlaid with variables of r2 > 0.15 being displayed with vector lengths corresponding to the correlation strength along the axes (shown in thick black lines). Convex hulls connect each group of treatments. Sites that are closer together are more similar than sites farther away from each other. B. Arrows from each treatment’s centroid in spider family space indicate pre- to postfire for axes 1 vs. 2. Spider families that have significant positive and negative correlations with each axis are indicated. See Table 5.2 for correlations of each axis with habitat variables and spider families.

124

Figure 5.6. A. Non-metric multidimensional scaling ordination of sites in bee species space along with weighted average positions for bee species for axes 1 vs. 2. Each triangle represents a site and each circle point represents a bee species. A joint-plot from the environmental matrix is overlaid with variables of r2 > 0.15 being displayed with vector lengths corresponding to the correlation strength along the axes (shown in thick black lines). Convex hulls connect each group of treatments. Sites that are closer together are more similar than sites farther away from each other. B. Arrows from each treatment’s centroid in bee species space indicate pre- to postfire for axes 1 vs. 2. Bee genera that have significant positive and negative correlations with each axis are indicated. See Table 5.3 for correlations of each axis with habitat variables and bee species.

125

100 Burned 80 Control

40 * * 20

Average % Cover (height in cm for max veg)Average formax Covercm(heightin%

PreFire PreFire PreFire PreFire

PostFire PostFire PostFire PostFire Invasive Litter BSC Max Veg Annual Grasses Height

Figure 5.7. Average percent cover of invasive annual grasses, litter, biological soil crusts (BSC) and average maximum vegetation height in cm compared by prefire and postfire (N = 5 burned sites, N = 13 control sites). Stars denote significant BACI results.

1200 Burned 6 Burned A Control B Control 1000 5

800 4

600 3

400 * 2 200

AverageForbRichness 1 AverageForbAbundance 0 0 PreFire PostFire PreFire PostFire

Figure 5.8. A. Average forb abundance prefire and postfire. B. Average forb richness prefire and postfire. Stars denote significant BACI results. N = 5 burned sites, N = 13 control sites.

126

Table 5.1. Pearson correlation coefficients between habitat variables and spider families and each axis of the 3-dimensional NMS ordination. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 Axis 3 R R2 R R2 R R2 Invasive Annual Grasses -0.55 0.30 0.36 0.13 -0.14 0.02 Litter Cover -0.69 0.47 0.28 0.08 -0.09 0.01 Habitat BSC Cover 0.73 0.54 -0.07 0.01 0.28 0.08 Variables Max Veg Height 0.12 0.01 0.02 0.00 0.37 0.14 Forb Abundance 0.39 0.15 -0.15 0.02 -0.39 0.15 Forb Richness -0.07 0.01 -0.27 0.07 -0.50 0.25 Agelenidae -0.12 0.01 -0.05 0.00 0.02 0.00 Corinnidae -0.30 0.09 -0.04 0.00 0.04 0.00 Gnaphosidae 0.22 0.05 0.10 0.01 0.79 0.63 Linyphiidae 0.27 0.07 -0.18 0.03 -0.43 0.19 Spider Lycosidae -0.60 0.36 0.51 0.26 0.09 0.01 Families Philodromidae 0.22 0.05 -0.20 0.04 -0.39 0.15 Pholcidae 0.16 0.03 -0.28 0.08 -0.12 0.01 Salticidae 0.34 0.11 -0.14 0.02 0.09 0.01 Theridiidae -0.82 0.68 -0.50 0.25 0.16 0.02 Thomisidae -0.62 0.38 0.66 0.43 0.05 0.00

127

Table 5.2. Indicator values (IV) and p-values for spider families, bee species, and forb species. Only significant families or species are reported.

IV p IV p Burned Pre-Fire Burned Post-Fire None Linyphiidae 49 0.01 Spider Control Pre-Fire Control Post-Fire Families None Thomisidae 46 0.002 Lycosidae 41 0.03 Burned Pre-Fire Burned Post-Fire Perdita aridella 69 0.0004 Anthophora urbana 77 0.003 Tripeolus paenepectoralis 49 0.005 Apis mellifera 73 0.0006 Perdita dubia 39 0.0008 Agapostemon femoratus 68 0.0002 Nomada spp. 4 40 0.03 Halictus tripartitus 38 0.04 Bee Nomada spp. 5 36 0.03 Species Megachile coquilletti 31 0.04 Control Pre-Fire Control Post-Fire Melissodes subagilis 80 0.0002 Melissodes rivalis 48 0.02 Melissodes semilupinus 65 0.0002 Melissodes bimatris 48 0.01 Triepeolus grindeliae 44 0.03 Burned Pre-Fire Burned Post-Fire Chrysothamnus 69 0.02 Polygonum douglasii 100 0.0002 viscidiflorus Conyza canadensis 77 0.0004 Forb Epilobium brachycarpum 76 0.0004 Species Salsola tragus 60 0.002 Machaeranthera canescens 46 0.05 Control Pre-Fire Control Post-Fire None Ericameria nauseosa 91 0.0002 Centaurea solstitialis 52 0.02

128

Table 5.3. Pearson correlation coefficients between habitat variables and bee species and each axis of the 3-dimensional NMS ordination. Cells with significant correlation coefficients are in bold.

Axis 1 Axis 2 Axis 3 R R2 R R2 R R2 Invasive Annual Grasses 0.61 0.38 -0.21 0.05 0.42 0.18 Litter Cover 0.31 0.10 -0.09 0.01 0.67 0.45 Habitat BSC Cover -0.32 0.10 0.05 0.00 -0.42 0.18 Variables Max Veg Height -0.46 0.21 0.13 0.02 0.30 0.09 Forb Abundance 0.07 0.01 0.27 0.07 -0.67 0.45 Forb Richness 0.47 0.22 0.21 0.05 -0.30 0.09 Agapostemon femoratus -0.04 0.00 0.38 0.15 -0.52 0.27 Agapostemon texanus -0.41 0.17 0.75 0.57 0.14 0.02 Agapostemon virescens -0.16 0.03 0.34 0.12 0.48 0.23 Anthophora exigua -0.19 0.04 0.18 0.03 -0.23 0.05 Anthophora urbana 0.14 0.02 0.13 0.02 -0.62 0.39 Apis mellifera -0.02 0.00 -0.01 0.00 -0.63 0.39 Bombus fervidus 0.12 0.01 0.27 0.08 0.05 0.00 Bombus griseocollis -0.36 0.13 0.21 0.04 0.23 0.05 Colletes gypsicolens -0.06 0.00 -0.07 0.01 0.09 0.01 Colletes mandibularis -0.03 0.00 -0.02 0.00 -0.01 0.00 Colletes compactus -0.10 0.01 0.03 0.00 0.04 0.00 Bee Colletes fulgidus -0.20 0.04 0.23 0.05 0.29 0.09 Species Dianthidium pudicum -0.04 0.00 0.14 0.02 -0.13 0.02 Halictus farinosus 0.33 0.11 0.31 0.10 -0.03 0.00

Halictus ligatus -0.04 0.00 0.27 0.07 -0.08 0.01 Halictus tripartitus -0.15 0.02 0.17 0.03 -0.10 0.01 Lasioglossum Dialictus spp. 0.45 0.21 0.54 0.29 0.32 0.11 Lasioglossum titusi -0.16 0.03 0.00 0.00 0.17 0.03 Megachile parallela -0.28 0.08 0.11 0.01 0.09 0.01 Megachile coquilletti 0.10 0.01 0.12 0.01 -0.32 0.10 Megachile onobrychidis -0.17 0.03 -0.01 0.00 -0.10 0.01 Megachile montivaga -0.01 0.00 0.21 0.04 -0.29 0.08 Megachile wheeleri -0.16 0.03 0.05 0.00 -0.03 0.00 Megachile perihirta 0.08 0.01 -0.23 0.05 0.08 0.01 Melissodes lupinus 0.05 0.00 0.24 0.06 0.28 0.08

129

Melissodes agilis 0.09 0.01 0.42 0.17 0.04 0.00 Melissodes bimatris -0.66 0.44 0.12 0.01 0.22 0.05 Melissodes lutulentus -0.40 0.16 0.23 0.05 -0.04 0.00 Melissodes pallidisignatus -0.28 0.08 0.17 0.03 -0.19 0.04 Melissodes perlusa -0.18 0.03 0.03 0.00 -0.04 0.00 Melissodes semilupinus -0.40 0.16 0.10 0.01 0.28 0.08 Melissodes subagilis -0.47 0.22 0.18 0.03 0.29 0.08 Melissodes rivalis 0.29 0.09 -0.01 0.00 0.18 0.03 Nomada sp. 1 -0.20 0.04 0.23 0.05 0.29 0.09 Nomada sp. 2 -0.15 0.02 -0.05 0.00 0.00 0.00 Nomada sp. 3 0.13 0.02 0.13 0.02 -0.48 0.23 Nomada sp. 4 0.08 0.01 0.12 0.01 -0.11 0.01 Nomada sp. 5 0.04 0.00 0.22 0.05 0.11 0.01 Nomada sp. 6 -0.06 0.00 0.21 0.04 -0.14 0.02 Perdita aridella -0.48 0.23 0.23 0.05 -0.16 0.03 Perdita dubia -0.51 0.26 0.14 0.02 -0.02 0.00 Perdita oregonensis -0.10 0.01 0.15 0.02 -0.57 0.33 Sphecodes 1 -0.10 0.01 -0.03 0.00 -0.24 0.06 Svastra obliqua -0.14 0.02 0.02 0.00 0.15 0.02 Triepeolus grindeliae -0.35 0.13 0.42 0.18 0.33 0.11 Triepeolus paenepectoralis -0.45 0.20 0.14 0.02 0.04 0.00

130

CHAPTER 6: GENERAL CONCLUSIONS

Spiders and bees are critically important to ecosystems as they contribute greatly to biodiversity and are vital to ecosystem services such as pest control and pollination; furthermore both respond quickly to changes in the surrounding habitat, making them ideal taxa for restoration monitoring (Wheater et al. 2000; Klein et al. 2007; Exeler et al. 2009; Malumbres-Olarte et al. 2013). I examined each group’s response, in conjunction with multiple associated habitat variables, to arid grassland restoration and wildfire in the Pacific Northwest, USA. In addition to describing these unique and understudied communities, I found each group varied in their responses to restoration and wildfire and I identified several underlying environmental variables associated with these responses. At a local scale (Chapter 2) spider community composition and abundance in degraded and restored sites differed from native sites, with fewer spiders found in native sites than degraded and restored sites. However, native and restored sites had more species than degraded sites. Habitat variables that were closely associated with spider community composition included invasive annual grass, litter, and biological soil crust. A chronosequence of restoration sites indicated that with time, spider abundance decreases, species richness increases, and composition changes, indicating that restoration is altering the community so that it is more similar to those found in native sites. However, my work suggests that the process may take a decade or more to recover invertebrate and vegetation within these arid grasslands, a result that is consistent with other studies in arid systems (Brand & Dunn 1998; Purtauf et al. 2004; Stadler et al. 2007). To restore these communities to resemble native communities (e.g., less abundance but higher species richness), management should focus on decreasing invasive grass and litter cover and reestablish native bunchgrasses and biological soil crusts. I found no evidence that grassland restoration impacted the bee community at the local scale (Chapter 3). As with the spider community, bee composition at native sites differed from restored and degraded sites but there was no difference in abundance, richness, or diversity among restoration treatments. Many studies cite variation in floral

131 resources as a key driver of temporal and spatial patterns in bee communities (Potts et al. 2003; Hopwood 2008; Kwaiser and Hendrix 2008); however, I found that even though the bee community had strong temporal patterns in abundance, richness, and diversity, these did not correspond to temporal trends in floral resources. Furthermore, spatial variation in floral abundance among sites was actually negatively associated with bee diversity, indicating that increasing floral abundance alone, without considering the type or diversity of flowering species planted, is not an effective restoration strategy for enhancing bee habitat. Other variables that greatly influenced the bee community included maximum vegetation height, and invasive annual grass, litter, and biological soil crust cover. Sites with greater litter cover had more bees, while sites with taller vegetation had more bee species. These factors, especially factors that may influence efficient bee foraging (e.g. maximum vegetation height, invasive annual grasses) and nesting site quality (e.g. litter cover, soils, woody vegetation), must be considered when developing restoration plans that aim to enhance bee habitat in arid grasslands. At the regional scale (Chapter 4) location, not restoration, influenced spider abundance and community composition; however, both location and restoration played a role in influencing bee richness and diversity. Variables with strong influences on the regional bee community included invasive grass, litter, and biological soil crust cover, in addition to elevation, forb cover, and maximum vegetation height. Location affected forb cover and maximum vegetation height, while both location and restoration affected litter cover. An interaction between location and restoration impacted invasive grass cover. Sites with more litter cover had more spiders, while sites at higher elevation with greater forb cover had greater spider richness and diversity. Location and site-specific characteristics, in addition to restoration treatment, should be considered during restoration planning as spider communities within similar locations may respond differently. Wildfire, a common occurrence that has increased in arid and semi-arid grassland ecosystems (Schoennagel et al. 2017), had a significant impact on spider and bee communities one year after the burn (Chapter 4). The fire affected both groups’ composition but also increased bee diversity and richness. In addition, the fire decreased

132 invasive annual grass and biological soil crust cover but increased forb abundance. As invasive annual grasses greatly impact the disturbance regime, especially fire frequency and intensity (Pyke 1999; Davies et al. 2012), invasive annual grass reduction should be a top priority during restoration. As much of the western United States considers restoration of grassland habitat, it is important to consider its impact on beneficial invertebrates that contribute substantially to ecosystem function. Yet, very little is still known about these important arid grassland communities and the underlying environmental variables they are associated with. While additional research should continue to focus on beneficial invertebrates within these grassland systems, in the interim, managers should focus on creating complex grassland habitats that can support the most diverse invertebrate communities to maximize subsequent ecosystem services. Restored grassland habitat for spiders should prioritize vegetative complexity and decrease litter, while restored grassland habitat for bees should prioritize access to nesting sites.

Literature Cited

Brand, R.H., Dunn, C.P., 1998. Diversity and abundance of springtails (Insects: Collembola) in native and restored tallgrass prairies. Am Midl Nat. 139, 235-242. Davies, G.M., Bakker, J.D., Dettweiler-Robinson, E., Dunwiddie, P.W., Hall, S.A., Downs, J., Evans, J., 2012. Trajectories of change in sagebrush steppe vegetation communities in relation to multiple wildfires. Ecol Appl. 22, 1562-1577. Exeler, N., Kratochwil, A, Hochkirch, A. 2009. Restoration of riverine inland sand dune complexes: implications for the conservation of wild bees. J Appl Ecol. 46, 1097- 1105. Hopwood, J.L. 2008. The contribution of roadside grassland restorations to native bee conservation. Biol Conserv. 141, 2632-2640. Klein, A.M., Vaissie’re, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Important of pollinators in changing landscapes for world crops. P Roy Soc B-Biol Sci. 274, 303-313. Kwaiser, K.S., Hendrix, S.D. 2008. Diversity and abundance of bees (Hymenoptera: Apiformes) in native and ruderal grasslands of agriculturally dominated landscapes. Agr Ecosyst Environ. 124, 200-204. Malumbres-Olarte, J., Vink, C.J., Ross, J.G., Cruickshank, R.H., Paterson, A.M., 2013. The role of habitat complexity on spider communities in native alpine grasslands

133

of New Zealand: Habitat complexity and alpine spiders. Insect Conserv Diver. 6, 124–134. Potts, S.G., Vulliamy, B., Dafni, A., Ne’eman, G., Willmer, P., 2003. Response of plant- pollinator communities to fire: changes in diversity, abundance and floral reward structure. Oikos. 101, 103-112. Purtauf, T., Dauber, J., Wolters, V., 2004. Carabid communities in the spatio-temporal mosaic of a rural landscape. Landscape Urban Plan. 67, 185-193. Pyke, D.A., 1999. Invasive exotic plants in sagebrush ecosystems of the Intermountain West. In: Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., eds. Proceedings: Sagebrush Steppe Ecosystems Symposium; Boise, Idaho; June 23- 25, 1999. Boise, Idaho: Bureau of Land Management, pp. 43-54. Schoennagel, T., Balch, J.K., Brenkert-Smith, H, Dennison, P.E., Harvey, B.J., Krawchuk, M.A., Mietkiewicz, N., Morgan, P., Moritz, M.A., Rasker, R., Turner, M.G., Whitlock, C., 2017. Adapt to more wildfire in western North American forests as climate changes. PNAS. 114, 4582-4590. Stadler, J., Trefflich, A., Brandl, R., Klotz, S., 2007. Spontaneous regeneration of dry grasslands on set-aside fields. Biodivers Conserv. 16, 621-630. Wheater, C.P., Cullen, W.R., Bell, J.R., 2000. Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecol. 15, 401–406.

134

General Bibliography

Albrecht, M., Schmid, B., Obrist, M.K., Schupbach, B., Kleijn, D., Duelli, P., 2010. Effects of ecological compensation meadows on arthropod diversity in adjacent intensively managed grassland. Biol Conserv. 143, 642-649. Bakker, J.P., Berendse, F., 1999. Constraints in the restoration of ecological diversity in grassland and heathland communities. Trends Ecol Evol. 14, 63-68. Bartuszevige, A.M., Kennedy, P.L., Taylor, R.V., 2012. Sixty-seven years of landscape change in the last, large remnant of the Pacific Northwest bunchgrass prairie. Nat Area J. 32, 166-170. Beck, J.L., Connelly, J.W., Wambolt, C.L., 2012. Consequences of treating Wyoming Big Sagebrush to enhance wildlife habitats. Rangeland Ecol Manag. 65, 444-455. Bell, J.R., Cullen, W.R., Wheater, C.P., 1998. The structure of spider communities in limestone quarry environments. In, Seldon, P., 1998. Proceedings of the 17th European Colloquium of Arachnology, Burnham Beeches: British Arachnological Society, Edinburgh. Bell, J.R., Wheater, C.P., Cullen, W.R., 2001. The implications of grassland and heathland management for the conservation of spider communities: a review. J Zool. 255, 377-387. Belnap, J., 2000. Structure and function of biological soil crusts. In Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., compilers, 2000. Proceedings: Sagebrush Steppe Ecosystem Symposium. Bureau of Land Management Publication No. BLM/ID/PT-001001+1150, Boise, Idaho, USA. Belnap, J., Eldridge, D.J., 2000. Impacts of disturbance on biological soil crusts. In Belnap J, Lange O editors. Biological soil crusts: structure, function, and management. Springer-Verlap, New York, New York, USA. Bess, E.C., Parmenter, R.R., McCoy, S., Molles Jr., M.C., 2002. Responses of a riparian forest-floor arthropod community to wildfire in the Middle Rio Grande Valley, New Mexico, Env Entomol, 31, 774–784. Black, S.H., Shepherd, M., Vaughan, M., 2011. Rangeland management for pollinators. Rangelands. 33, 9-13. Bock, C.E., Bock, J.H., 1991. Response of grasshoppers (Orthoptera: Acrididae) to wildfire in a southeastern Arizona grassland. Am Midl Nat. 1, 162-167. Bogusch, P., Blazej, L., Milos, T., Heneberg, P., 2015. Forgotten role of fires in Central European forests: critical importance of early postfire successional stages for bees and wasps (Hymenoptera: Aculeata). Eur J Forest Res. 134, 153-166. Borchard, F., Buchholz, S., Helbing, F., Fartmann, T., 2014. Carabid beetles and spiders as bioindicators for the evaluation of montane heathland restoration on former spruce forests. Biol Conserv. 178, 185-192.

135

Brand, R.H., Dunn, C.P., 1998. Diversity and abundance of springtails (Insects: Collembola) in native and restored tallgrass prairies. Am Midl Nat. 139, 235-242. Britton, C.M., McPherson, G.R., Sneva, F.A., 1990. Effects of burning and clipping on five bunchgrasses in eastern Oregon. Great Basin Nat. 50, 115-120. Brown, M.J.F., Paxton, R.J., 2009. The conservation of bees: a global persepective. Apidologie. 40, 410-416. Buddle, C.M., Spence, J.R., Langor, D.W., 2000. Succssion of boreal forest spider assemblages following wildfire and harvesting. Ecography. 23, 424-436. Bullock, J.M., Aronson, J., Newton, A.C., Pywell, R.F., Rey-Benayas, J.M., 2011. Restoration of ecosystem services and biodiversity: conflicts and opportunities. Trends Ecol Evol. 26, 541-549. Burel, F., Baudry, J., Butet, A., Clergeau, P., Delettre, Y., Le Coeur, D., Dubs, F., Morvan, N., Paillat, G., Petit, S., Thenail, C., Brunel, E., Lefeuvre, J., 1998. Comparative biodiversity along a gradient of agricultural landscapes. Acta Oecologica. 19, 47-60. Campbell, J.W., Hanula, J.L., Waldrop, T.A., 2007. Effects of prescribed fire and fire surrogates on floral visiting insects of the Blue Ridge province in North Carolina. Biol Conserv. 134, 393-404. Cane, J.H., 1991. Soils of ground-nesting bees (Hymenoptera: Apoidea): Texture, Moisture, Cell Depth and Climate. J Kansas Entomol Soc. 64, 406-413. Cane, J.H., 1994. Nesting biology and mating behavior of the southeastern blueberry bee, Habropoda laborisosa (Hymenoptera: Apoidea). J Kansas Entomol Soc. 67, 236- 241. Cane, J.H., Neff, J.L., 2011. Predicted fates of ground-nesting bees in soil heated by wildfire: Thermal tolerances of life stages and a survey of nesting depths. Biol Conserv. 144, 2631-2636. Cane, J.H., Love, B., 2016. Floral guilds of bees in sagebrush steppe: Comparing bee usage of wildflowers available for postfire restoration. Nat Area J. 36, 377-391. Clough, Y., Kruess, A., Kleijn, D., Tscharntke, T., 2005. Spider diversity in cereal fields: comparing factors at local, landscape and regional scales. J. Biogeogr. 32, 2007– 2014. Coddington, J.A., Levi, H.W., 1991. Systematics and evolution of spiders (Araneae). Annu Rev Ecol Syst. 22, 565-592. Colwell, R.K., 2013. EstimateS: Statistical estimation of species richness and shared species from samples, Version 9.1.0. User’s guide and application Colwell, R.K., Chao, A., Gotelli, N.J., Lin, S., Mao, C.X., Chazdon, R.L., Longino, J.T., 2012. Models and estimators linking individual-based and sample-based rarefaction, extrapolation, and comparison of assemblages. J Plant Ecol. 5, 3-21.

136

Condon, L.A., Pyke, D.A., 2018. Fire and grazing influence site resistance to Bromus tectorum through their effects on shrub, bunchgrass and biocrust communities in the Great Basin (USA). Ecosystems. Available online: https://doi.org/10.1007/s10021-018-0230-8. Cook, S.P., Birch, S.M., Merickel, F.W., Lowe, C.C., Page-Dumroese, D., 2011. Bumble bee (Hymenoptera: Apidae) community structure on two sagebrush steppe sites in southern Idaho. Pan-Pac Entomol. 87, 161-171. Cramer, V.A., Hobbs, R.J., eds., 2007. Old fields: dynamics and restoration of abandoned farmland. Society for Ecological Restoration International. Island Press, Washington, D.C. Cristofoli, S., Mahy, G., Kekenbosch, R., Lambeets, K., 2010. Spider communities as evaluation tools for wet heathland restoration. Ecol Indic. 10, 773-780. Cutler, B., Jennings, D.T., 1985. A revision of the Metaphidippus arizonensis group (Araneae, Salticidae). J Arachnol. 13, 1-8. Davies, G.M., Bakker, J.D., Dettweiler-Robinson, E., Dunwiddie, P.W., Hall, S.A., Downs, J., Evans, J., 2012. Trajectories of change in sagebrush steppe vegetation communities in relation to multiple wildfires. Ecol Appl. 22, 1562-1577. Davis, J.D., Hendrix, S.D., Debinski, D.M., Hemsley, C.J., 2008. Butterfly, bee and forb community composition and cross-taxon incongruence in tallgrass prairie fragments. J Insect Conserv. 12, 69-79. De’ath, G., Fabricius, K.E., 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology. 81, 3178-3192. DeBano, S.J., 2006. Effects of livestock grazing on insect communities in semi-arid grasslands of southeastern Arizona. Biodivers Conserv. 15, 2547-2564. Déri, E., Magura, T., Horvath, R., Kisfali, M., Ruff, G., Lengyel, S., Tothmeresz, B., 2011. Measuring the short-term success of grassland restoration: the use of habitat affinity indices in ecological restoration. Restor Ecol. 19, 520–528. Eickwort, G.C., Abrams, J., 1980. Parasitism of sweat bees in the genus Agapostemon by cuckoo bees in the genus Nomada (Hymenoptera: Halictidae, Anthophoridae). Pan-Pac Entomol. 56, 144-152. Exeler, N., Kratochwil, A, Hochkirch, A., 2009. Restoration of riverine inland sand dune complexes: implications for the conservation of wild bees. J Appl Ecol. 46, 1097- 1105. Finch, O.D., Blick, T., Schuldt, A., 2008. Macroecological patterns of spider species richness across Europe. Biodivers Conserv. 17, 2849-2868. Forup, M.L., Memmott, J., 2005. The restoration of plant-pollinator interactions in hay meadows. Restor Ecol. 13, 265-274. Gerla, P.J., Cornett, M.W., Ekstein, J.D., Ahlering, M.A., 2012. Talking big: lessons learned form a 9000 hectare restoration in the northern tallgrass prairie.

137

Sustainability. 4, 3066-3087, available online: http://www.mdpi.com/2071- 1050/4/11/3066/htm Geroff, R.K., Gibbs, J., McCray, K.W., 2014. Assessing bee (Hymenoptera: Apoidea) diversity of an Illinois restored tallgrass prairie: methodology and conservation considerations. J Insect Conserv. 18, 951-964. Gollan, J.R., Smith, H.M., Bulbert, M., Donnelly, A.P., Wilkie, L., 2010. Using spider web types as a substitute for assessing web-building spider biodiversity and the success of habitat restoration. Biodiver Conserv. 19, 3141-3155. Gonzalez, N., DeBano, S.J., Kimoto, C., Taylor, R.V., Tubbesing, C., Strohm, C., 2013. Native bees associated with isolated aspen stands in Pacific Northwest Bunchgrass Prairie. Nat Area J. 33, 374-383. Graham, A.K., Buddle, C.M., Spence, J.R., 2003. Habitat affinities of spiders living near a freshwater pond. Journal Arachnol. 31, 78-89. Greene, R.S.B., Chartres, C.J., Hodgkinson, K.C., 1990. The effects of fire on the soil in a degraded semiarid woodland. I. Crypotogam cover and physical and micromorphological properties. Aust J Soil Res. 28, 755-777. Grman, E., Bassett, T., Brudvig, L.A., 2013. Confronting contingency in restoration: management and site history determine outcomes of assembling prairies, but site characteristics and landscape context have little effect. J Appl Ecol. 50, 1234- 1243. Grundel, R., Jean, R.P., Frohnapple, K.J., Glowacki, G.A., Scott, P.E., Pavlovic, N.B., 2010. Floral and nesting resources, habitat structure, and fire influence bee distribution across an open-forest gradient. Ecol Appl. 20, 1678-1692. Haskins, M.F., Shaddy, J.H., 1986. The ecological effects of burning, mowing, and plowing on ground-inhabiting spiders (Araneae) in an old-field ecosystem. J Arachnol, 14, 1-14. Hendrickx, F., Maelfait, J.P., van Wingerden, W., Schweiger, O., Speelmans, M., Aviron, S., Augenstein, I., Billeter, R., Bailey, D., Bukacek, R., Burel, F., Diekotter, T., Dirksen, J., Herzog, F., Liira, J., Roubalova, M., Vandomme, V., Bugter, R., 2007. How landscape structure, land-use intensity and habitat diversity affect components of total arthropod diversity in agricultural landscapes. J Appl Ecol. 44, 340–351. Hendrixson, B.E., Bond, J.E., 2007. Molecular phylogeny and biogeography of an ancient Holarctic lineage of mygalomorph spiders (Araneae: Antrodiaetidae: Antrodiaetus). Mol Phylogenet Evol. 42, 738-755. Horvath, R., Magura, T, Szinetar, C., Eichardt, J., Kovacs, E., Tothmeresz, B., 2015. In stable, unmanaged grasslands local factors are more important than landscape- level factors in shaping spider assemblages. Agri Ecosyst Environ. 208, 106-113. Hopwood, J.L., 2008. The contribution of roadside grassland restorations to native bee conservation. Biol Conserv. 141, 2632-2640.

138

Huddleston, R.T., Young, T.P., 2004. Spacing and competition between planted grass plugs and preexisting perennial grasses in a restoration site in Oregon. Restor Ecol. 12, 546-551. Huddleston, R.T., Young, T.P., 2005. Weed control and soil amendment effects on restoration plantings in an Oregon grassland. West N Am Naturalist. 65, 507-515. Hurd, P. D., LeBerge, W. E., & Linsley, E. G., 1980. Principal sunflower bees of North America with emphasis on the Southwestern United States (Hymenoptera, Apoidea). Sm C Zool. 1-158. Huxel, G.R., Hastings, A., 1999. Habitat loss, fragmentation, and restoration. Restor Ecol. 7, 309-315. Jeanneret, P., Schupbach, B., Luka, H., 2003. Quantifying the impact of landscape and habitat features on biodiversity in cultivated landscapes. Agr Ecosyst Environ. 98, 311-320. Johansen, J.R., 1993. Cryptogamic crusts of semiarid and arid lands of North America. J Phycol. 29, 140-147. Kaltenecker, J.H., 1997. The recovery of microbiotic crusts following post-fire rehabilitation on rangelands of the western Snake River Plain. Thesis, Boise State University, Boise, Idaho, USA Keeley, J.E. 2009. Fire intensity, fire severity and burn severity: a brief review and suggested usage. Int J Wildland Fire. 18, 116-126. Kelton, L.A., 1978. The Insects and Arachnids of Canada. Biosystemics Research Institute, Ottawa, Ontario. Kennedy, P.L., DeBano, S.J., Bartuszevige, A., Lueders, A., 2009. Effects of native and nonnative grassland plant communities on breeding passerine birds: implications for restoration of Northwest bunchgrass prairie. Restor Ecol. 17, 515-525. Kim, J., Williams, N., Kremen, C. 2006. Effects of cultivation and proximity to natural habitat on ground-nesting native bees in California sunflower fields. J Kansas Entomol Soc. 79, 309-320. Kimoto, C., DeBano, S.J., Thorp, R.W., Rao, S., Stephen, W.P., 2012a. Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci. 12, 108, 23, available online: http://www.insectscience.org/12.108 Kimoto, C., DeBano, S.J., Thorp, R.W., Taylor, R.V., Schmalz, H., DelCurto, T., Johnson, T., Kennedy, P.L., Rao, S., 2012b. Livestock and native bee communities: short-term responses to grazing intensity and implications for managing ecosystem services in grasslands. Ecosphere. 3, 1-19. http://www.esajournals.org/doi/pdf/10.1890/ES12-00118.1 Klein, A.M., Vaissie’re, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Important of pollinators in changing

139

landscapes for world crops. P Roy Soc B-Biol Sci. 274, 303-313. Knick, S.T., Dobkin, D.S., Rotenberry, J.T., Schroeder, M.A., Vander Haegen, W.M., Van Riper III, C., 2003. Teetering on the edge or too late? Conservation and research issues for avifauna of sagebrush habitats. Condor. 105, 611-634. Koponen, S., 2005. Early succession of a boreal spider community after forest fire. J Arachnol. 33, 230-235. Kouki, J., Hyvarinen, E., Lappalainen, H., Martikainen, P., Smila, M., 2012. Landscape context affects the success of habitat restoration: large-scale colonization patterns of saproxylic fire-associated species in boreal forests. Divers Distrib. 18, 348-355. Kovacs-Hostyanszki, A., Elek, Z., Balazs, K., Centeri, C., Falusi, E., Jeanneret, P., Penksza, K., Podmaniczky, L., Szalkovszki, O., Baldi, A., 2013. Earthworms, spiders and bees as indicators of habitat quality and management in a low-input farming region—A whole farm approach. Ecol Indic. 33, 111-120. Krauss, J., Steffan-Dewenter, I., Tscharntke, T., 2003. How does landscape context contribute to effects of habitat fragmentation on diversity and population density of butterflies? J Biogeogr. 20, 889-900. Kremen, C., Williams, N.M., Thorp, R.W., 2002. Crop pollination from native bees as risk from agricultural intensification. PNAS. 99, 16812-16816. Kwaiser, K.S., Hendrix, S.D., 2008. Diversity and abundance of bees (Hymenoptera: Apiformes) in native and ruderal grasslands of agriculturally dominated landscapes. Agr Ecosyst Environ. 124, 200-204. Langlands, P.R., Brennan, K.E.C., Ward, B., 2012. Is the reassembly of an arid spider assemblage following fire deterministic?: Successional trajectory of a spider assemblage. Austral Ecol. 37, 429-439. Levi, H.W., 1959. The spider genus Latrodectus (Araneae, Theridiidae). T Am Microsc Soc. 78, 7-43. Longcore, T., 2003. Terrestrial arthropods as indicators of ecological restoration success in coastal sage scrub (California, USA). Restor Ecol. 11, 397–409. Losey, J.E., Vaughan, M., 2006. The economic value of ecological services provided by insects. Bioscience. 56, 311-323. Love, B.G., Cane, J.H., 2016. Limited direct effects of a massive wildfire on its sagebrush steppe bee community. Ecol Entomol. 3, 317-326. Malumbres-Olarte, J., Vink, C.J., Ross, J.G., Cruickshank, R.H., Paterson, A.M., 2013. The role of habitat complexity on spider communities in native alpine grasslands of New Zealand: Habitat complexity and alpine spiders. Insect Conserv Diver. 6, 124–134. McCune, B., Grace, J., 2002. Analysis of Ecological Communities. MjM Software, Gleneden Beach, Oregon, USA.

140

McCune, B., Mefford, M.J., 2015. PC-ORD Version 7.287. Multivariate Analysis of Ecological Data. MjM Software, Gleneden Beach, Oregon, USA. Merrett, P., 1976. The phenology of spiders on heathland in Dorest. Families Lycosidae, Pisauridae, Agelenidae, Mimetidae, Theridiidae, Tetragnatidae, Argiopidae. J Zool. 156, 239-256. Meyer, S.E., Nelson, D.L., Clement, S., Paulsen, A. 2000. Exploring the potential for biocontrol of cheatgrass with the head smut pathogen. In Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., compilers, 2000. Proceedings: Sagebrush Steppe Ecosystem Symposium. Bureau of Land Management Publication No. BLM/ID/PT-001001+1150, Boise, Idaho, USA. Michener, C.D., 2007. The bees of the world, 2nd edition. The John Hopkins University Press, Baltimore, USA. Mirshamski Kakhki, O., 2005. Faunistic study of spiders in Khorasan Province, Iran (Arachnida: Araneae). Int J Agric Biol. 1, 55-66. Mola, J.M., Williams, N.M., 2018. Fire-induced change in floral abundance, density, and phenology benefits bumble bee foragers. Ecosphere. 9:e02056. Moretti, M., Conedera, M., Duelli, P. Edwards, P.J., 2002. The effects of wildfire on ground-active spiders in deciduous forests on the Swiss southern slope of the Alps. J Appl Ecol. 39, 321-336. Mortimer, S.R., Hollier, J.A., Brown, V.K., 1998. Interactions between plant and insect diversity in the restoration of lowland calcareous grasslands in southern Britain. Appl Veg Sci. 1, 101–114. Nemec, K., Allen, C.R., Danielson, S.D., Helzer, C.J., 2014. Responses of predatory invertebrates to seeding density and plant species richness in experimental tallgrass prairie restorations. Agr Ecosyst Environ. 183, 11-20. Niwa, C.G., Peck, R.W., 2002. Influence of prescribed fire on carabid beetle (Carabidae) and spider (Araneae) assemblages in forest litter in southwestern Oregon. Environ Entomol. 5, 785-796. Opell, B.D., Beatty, J.A., 1976. The Nearctic Hahniidae (Arachnida: Araneae). Bull Mus Comp Zool. 147, 393-433. Öster, M., Ask, K., Cousins, S.A.O., Eriksson, O., 2009. Dispersal and establishment limitation reduces the potential for successful restoration of semi-natural grassland communities on former arable fields. J Appl Ecol. 46, 1266-1274. Oxbrough, A.G., Gittings, T., O’Halloran, J., Giller, P.S., Kelly, T.C., 2007. Biodiversity of the ground-dwelling spider fauna of afforestation habitats. Agr Ecosyst Environ. 120, 433-441. Perner, J., Malt, S., 2003. Assessment of changing agricultural land use: response of vegetation, ground-dwelling spiders and beetles to the conversion of arable land into grassland. Agr Ecosyst Environ. 98, 169-181.

141

Pettis, J.S., Delaplane, K.S., 2010. Coordinate responses to honey bee decline in the USA. Apidologie. 41, 256-263. Plaven, G. 2015. Update: Firefighters back on attack of Boardman blaze. http://www.eastoregonian.com/eo/local-news/20150602/haystack-combusts- starts-large-fire-near-boardman (accessed 21 February 2018). Polchaninonva, N., 2015. Recovery of spider communities after a spontaneous summer fire in the forb-bunchgrass steppe of eastern Ukraine. Hacquetia. 14, 79-96. Porensky, L.M., Leger, E.A., Davison, J., Miller, W.W., Goergen, E.M., Espeland, E.K., Carroll-Moore, E.M., 2014. Arid old-field restoration: native perennial grasses suppress weeds and erosion, but also suppress native shrubs. Agr Ecosyst Environ. 184, 135-144. Potts, S. G., Dafni, A., Ne’eman, G., 2001. Pollination of a core flowering shrub species in Mediterranean phrygana: variation in pollinator diversity, abundance and effectiveness in response to fire. Oikos. 92, 71-80. Potts, S.G., Bulliam, B., Dafni, A., Ne’eman, G., Willmer, P., 2003a. Linking bees and flowers: how do floral communities structure pollinator communities? Ecology. 84, 2628-2642. Potts, S.G., Vulliamy, B., Dafni, A., Ne’eman, G., Willmer, P., 2003b. Response of plant- pollinator communities to fire: changes in diversity, abundance and floral reward structure. Oikos. 101, 103-112. Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O., Kunin, W.E., 2010. Global pollinator declines: trends, impacts and drivers. Trends Ecol Evol. 25, 345-353. Pryke, J.S., Samways, M.J., 2012. Differential resilience of invertebrates to fire. Austral Ecol. 37, 460-469. Purtauf, T., Dauber, J., Wolters, V., 2004. Carabid communities in the spatio-temporal mosaic of a rural landscape. Landscape Urban Plan. 67, 185-193. Pyke, D.A., 1999. Invasive exotic plants in sagebrush ecosystems of the Intermountain West. In: Entwistle, P.G., DeBolt, A.M., Kaltenecker, J.H., Steenhof, K., eds. Proceedings: Sagebrush Steppe Ecosystems Symposium; Boise, Idaho; June 23- 25, 1999. Boise, Idaho: Bureau of Land Management, pp. 43-54. R Development Core Team, 2015. R: A Language and Environment for Statistical Computing. Version 2.11.1 R Foundation for Statistical Computing, Vienna, Austria. Rao, S., Stephen, W.P., Kimoto, C., DeBano, S.J., 2011. The status of the ‘Red-Listed’ Bombus occidentalis (Hymenoptera: Apiformes) in Northeastern Oregon. Northwest Sci. 85, 64-67. Reed, C.C., 1995. Insect surveys on flowers in native and reconstructed prairies (Minnesota). Restor Manag Notes 13, 210-213.

142

Reisner, M.D., Grace, J.B., Pyke, D.A., Doescher, P.S., 2013. Conditions favouring Bromus tectorum dominance of endangered sagebrush steppe ecosystems. J Appl Ecol. 50, 1039-1049. Rhoades, P.R., 2016. Genetic and community structure of native bees of the Palouse Prairie, PhD Dissertation, University of Idaho. Rhoades, P., Griswold, T., Ikerd, H., Waits, L., Bosque-Perez, N.A., Eigenbrode, S.D., 2017. The native bee fauna of the Palouse Prairie (Hymenoptera: Apoidea). J Melittology. 66, 1-20. Richardson, M.L., Hanks, L.M., 2009. Effects of grassland succession on communities of orb-weaving spiders. Environ. Entomol. 38, 1595-9. Riechert, S.E., Reeder, W.G., 1972. Effects of fire on the spider distribution in southwestern Wisconsin prairies. In Proceedings 2nd Midwest Prairie Conference: 73-90. Zimmermann, J.H. (Ed.). Maddison: University of Wisconsin. RStudio Team, 2015. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA. Roberts, R.B., 1969. Biology of the bee genus Agapostemon (Hymenoptera: Halictidae). Kans Univ Sci Bull. 48, 689-719. Ruiz-Jaen, M.C., Aide, T.M., 2005. Restoration success: how is it being measured? Restor Ecol. 13, 569-577. Rypstra, A.L., Carter, P.E., Balfour, R.A., Marshall, S.D., 1999. Architectural features of agricultural habitats and their impact on the spider inhabitants. J Arachnol. 27, 371-377. Sampson, F., Knopf, F., 1994. Prairie conservation in North America. BioScience. 44, 418-421. Samu, F., Kadar, F., ONodi, G., Kertesz, M., Sziranyi, A., Szita, E., Fetyko, K., Neidert, D., Botos, E., Altbacker, V., 2010. Differential ecological responses of two generalist arthropod groups, spiders and carabid beetles (Araneae, Carabidae), to the effects of wildfire. Community Ecol. 11, 129-139. Schmidt, M.H., Roschewitz, I., Thies, C., Tscharntke, T., 2005. Differential effects of landscape and management on diversity and density of ground-dwelling farmland spiders. J Appl Ecol. 42, 281-287. Schoennagel, T., Balch, J.K., Brenkert-Smith, H, Dennison, P.E., Harvey, B.J., Krawchuk, M.A., Mietkiewicz, N., Morgan, P., Moritz, M.A., Rasker, R., Turner, M.G., Whitlock, C., 2017. Adapt to more wildfire in western North American forests as climate changes. PNAS. 114, 4582-4590. Schwarz, C.J., 2015. Analysis of BACI experiments. In Course Notes for Beginning and Intermediate Statistics. http://www.stat.sfu.ca/~cschwarz/CourseNotes (accessed 25 February 2018).

143

Sepp, K., Merit, M., Mand, M., Truu, J., 2004. Bumblebee communities as indicator for landscape monitoring in the agri-environmental programme. Landscape Urban Plan. 67, 173-183. SER (Society for Ecological Restoration International Science and Policy Working Group), 2004. The SER International Primer on Ecological Restoration. Society for Ecological Restoration International, Tucson, Arizona. Sheffield, C.S., Ratti, C., Packer, L., Griswold, T. 2011. Leafcutter and mason bees of the genus Megachile Latreille (Hymenoptera: Megachilidae) in Canada and Alaska. Can J Arthrop Identification. 18, 1-107. Smith, E.P., 2002. BACI design, in: El-Shaarawi, A.H., Piegorsch, W.W., (Eds.), Encyclopedia of Environmetrics. John Wiley and Sons, Ltd, Chichester, pp. 141- 148. Smith, L.J., Smith DiCarlo, L.A., DeBano, S.J., In review. Spider family (Thomisidae) exhibits positive response to cheatgrass invasion (Bromus tectorum L.). Submitted to Biol Invasions. Smith DiCarlo, L.A., DeBano, S.J., 2018. Spider community responses to grassland restoration: balancing tradeoffs between abundance and diversity. Restor Ecol. DOI: 10.1111/rec.12832. Snazell, R., Clark, R., 2000. The colonization of an area of restored chalk downland by spiders (Araneae). Ekol Bratisl. 19, 263-271. Stadler, J., Trefflich, A., Brandl, R., Klotz, S., 2007. Spontaneous regeneration of dry grasslands on set-aside fields. Biodivers Conserv. 16, 621-630. Steffan-Dewenter, I., 2003. Importance of habitat area and landscape context for species richness of bees and wasps in fragmented orchard meadows. Conserv Biol. 17, 1036-1044. Stratton, G.E., 1991. A new species of wolf spider, Schizocosa stridulans (Araneae, Lycosidae). J Arach. 19, 29-39. Sunderland, K., 1999. Mechanisms underlying the effects of spiders on pest populations. J Arachnol. 27, 308-316. Swengel, A.B., 2001. A literature review of insect responses to fire, compared to other conservation managements of open habitat. Biodivers Conserv. 10, 1141-1169. Tarrant, S., Ollerton, J., Lutfor Rahmn, M. Tarrant, J., McCollin, D., 2013. Grassland restoration on landfill sites in the east midlands, United Kingdom: an evaluation of floral resources and pollinating insects. Restor Ecol. 21, 560-568. Tisdale, E.W., 1982. Grasslands of western North America: the Pacific Northwest Bunchgrass. In: Nicholson, A.C., McLean, A., Baker, T.E., Editors. Proceedings of the 1982 Grassland Ecology and Classification Symposium. Pp. 232-245. Information Services Branch, British Columbia Ministry of Forests.

144

Tscharntke, T., Gathmann, A., Steffan-Dewenter, I., 1998. Bioindication using trap- nesting bees and wasps and their natural enemies: community structure and interactions. J Appl Ecol. 35, 708-719. Torok, P., Vida, E., Deak, B., Lengyel, S., Tothmeresz, B., 2011. Grassland restoration on former croplands in Europe: an assessment of applicability of techniques and costs. Biodivers Conserv. 20, 2311-2332. Tubbesing, C., Strohm, C., DeBano, S.J., Gonzalez, N., Kimoto, C., Taylor, R.V., 2014. Insect visitors and pollination ecology of Spalding’s catchfly (Silene spaldingii) in the Zumwalt Prairie of northeastern Oregon. Nat Area J. 34, 200-211. Uetz, G.W., 1979. The influence of variation in litter habitats on spider communities. Oecologia. 40, 29–42. Underwood, A.J., 1991. Beyond BACI: experimental designs for detecting human environmental impacts on temporal variations in natural populations. Aust J Mar Freshwater Res. 42, 569-587. Underwood, A.J., 1994. On beyond BACI—sampling designs that might reliably detect environmental disturbances. Ecol Appl. 4, 3-15. Uresk, D.W., Rickard, W.H., Cline, J.F., 1980. Perennial grasses and their response to a wildfire in south-central Washington. J Range Manage. 33, 111-114. US Climate Data, 2017. Climate Boardman – Oregon. Your Weather Service http://www.usclimatedata.com/climate/boardman/oregon/united-states/usor0036 (accessed 15 August 2017). U.S. National Pollinating Insects Database. USDA-ARS Pollination Insects – Biology, Management and Systematics Research Unit Museum Records. Bee Biology and Systematics Laboratory, Logan, Utah (Accessed 2018-15-4, through DiscoverLife http://www.discoverlife.org). Weeks, R.D., Holtzer, T.O., 2000. Habitat and season in structuring ground-dwelling spider (Araneae) communities in a shortgrass steppe ecosystem. Environ Entomol. 29, 1164-1172. Weisser, W.W., Siemann, E., 2004. The various effects of insects on ecosystem functioning. Insects Ecosyst Function. 173, 3–24. West, N.E., Hassan, M.A., 1985. Recovery of sagebrush-grass vegetation following wildfire. J Range Manage. 38, 131-134. West, N.E. 2000. Synecology and disturbance regimes of sagebrush steppe ecosystems, p. 15-26. In P.G. Entwistle, et al. (compilers), Proceedings: sagebrush steppe ecosystems symposium. USDI Bureau of Land Management Publication BLM/ID/PT-001001+1150, Boise, ID. Westphal, C., Bommarco, R., Carre, G., Lamborn, E., Morison, N., Petanidou, T., Potts, S.G., Roberts, S.P.M., Szentgyorgyi, H., Tscheulin, T., Vaissiere, B.E., Woychiechowski, M., Biesmeijer, J.C., Kunin, W.E., Settele, J., Steffan-

145

Dewenter, I., 2008. Measuring bee diversity in different European habitats and biogeographical regions. Ecol Monogr. 78, 653-671. Williams, N.M., Minckley, R.L., Silveira, F.A., 2001. Variation in native bee faunas and its implications for detecting community changes. Conservation Ecology 5, 7. Available online: http://www.consecol.org/vol5/iss1/art7/ Winfree, R., Griswold, T., Kremen, C., 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conserv Biol. 21, 213-223 Wisdom, M.J., Rowland, M.M., Wales, B.C., Hemstrom, M.A., Hann, W.J., Raphael, M.G., Holthausen, R.S., Gravenmier, R.A., Rich, T.D., 2002. Modeled effects of sagebrush-steppe restoration on Greater Sage-Grouse in the interior Columbia Basin, USA. Society for Conservation Biology. 16, 1223-1231. Wheater, C.P., Cullen, W.R., Bell, J.R., 2000. Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecol. 15, 401–406. Whisenant, S.G., 1990. Changing fire frequencies on Idaho’s Snake River Plains: ecological and management implications. Pages 4-10 in McArthur, E.D., Romney, E.M., Smith, S.D., Tueller, P.T., editors. Proceedings—symposium on cheatgrass invasion, shrub die-off, and other aspects of shrub biology and management. USDA Forest Service General Technical Report INT-276, Ogden, Utah, USA. Williams, N.M., et al., 2010. Ecological and life-history traits predict bee species responses to environmental disturbances. Biol Conserv. 143, 2280-2291. Wilson, S.D., Bakker, J.D., Christian, J.M., Li, X., Ambrose, L.G., Waddington, J., 2004. Semiarid old-field restoration: is neighbor control needed? Ecol Appl, 14, 476- 484. Winfree, R., Griswold, T., Kremen, C. 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conserv Biol. 21, 213-223. Wood, T.J., Holland, J.M., Goulson, D., 2015. Pollinator-friendly management does not increase the diversity of farmland bees and wasps. Biol Conserv. 187, 120-126. Woodcock, B.A., Vogiatzakis, I.N., Westbury, D.B., Lawson, C.S., Edwards, A.R., Brook, A.J., Harris, S.J., Lock, K.A., Maczey, N., Masters, G., Brown, V.K., Mortimer, S.R., 2010. The role of management and landscape context in the restoration of grassland phytophagous beetles. J Appl Ecol. 47, 366-376. Zamani, A., Rafinejad, J., 2014. First record of the Mediterranean recluse spider Loxosceles rufescens (Araneae: Sicariidae) from Iran. J Arthropod-Borne Di. 8, 228-231. Zulka, K.P., Abensperg-Traun, M., Milasowszky, N., Bieringer, G., Gereben-Krenn, B. A., Holzinger, W., Hölzler, G., Rabitsch, W., Reischütz, A., Querner, P., Sauberer, N., Schmitzberger, I., Willner, W., Wrbka, T., Zechmeister, H., 2014. Species richness in dry grassland patches of eastern Austria: a multi-taxon study on the role of local, landscape and habitat quality variables. Agri Ecosyst Environ.

146

182, 25–36.