Views See Kats and Dill; 1998; Dick and Grostal, 2001), While Predators Use Chemical Cues to Locate Prey (Koivula and Korpimaki, 2001)

Miami University

The Graduate School

Certificate for Approving the Dissertation

We hereby approve the Dissertation

Kerri M. Wrinn

Candidate for the Degree:

Doctor of Philosophy

______

Ann L. Rypstra, Advisor

______

Michelle D. Boone, Reader

______

Thomas O. Crist, Reader

______

Maria J. Gonzalez, Reader

______

David L. Gorchov

Graduate School Representative ABSTRACT

IMPACTS OF AN HERBICIDE AND PREDATOR CUES ON A GENERALIST PREDATOR IN AGRICULTURAL SYSTEMS

by Kerri M. Wrinn

Animals use chemical cues for signaling between species. However, anthropogenic chemicals can interrupt this natural chemical information flow, affecting predator- prey interactions. I explored how a glyphosate-based herbicide influenced the reactions of Pardosa milvina, a common wolf spider in agricultural systems, to its predators, the larger wolf spider, Hogna helluo and the carabid beetle, Scarites quadriceps. First, I tested the effects of exposure to herbicide and chemical cues from these predators on the activity, emigration, and survival of P. milvina in laboratory and mesocosm field experiments. In the presence of H. helluo cues in the laboratory, P. milvina always decreased activity and increased time to emigration. However, in the presence of S. quadriceps cues, these spiders only decreased activity and increased time to emigration when herbicide was also present. Presence of predator cues and herbicide did not affect the emigration of P. milvina from field mesocosms, but survival was highest for spiders exposed to S. quadriceps cues alone and lowest for those exposed to herbicide alone. Secondly, I tested the effects of predator cues, herbicide and prey availability on foraging and reproduction in female P. milvina. Spiders offered more prey captured and consumed more, while those exposed to H. helluo cues consumed less. Availability of prey and exposure to predator cues and herbicide in foraging trials had interactive effects on P. milvina’s subsequent reproductive success. In the low prey treatments, exposure to predator cues and herbicide each reduced reproductive success. In the high prey treatments, exposure to herbicide reduced reproductive success for spiders also exposed to S. quadriceps cues, but increased reproductive success for spiders also exposed to H. helluo cues. Finally, I exposed juvenile P. milvina to S. quadriceps cues and herbicide but found no effect of either on the spider’s growth and development. Together, these results indicate that predation risk and herbicide application likely interact in complex ways to affect the movement, reproduction and survival of a major arthropod predator in agricultural systems, and thus may have complex effects on the food web.

IMPACTS OF AN HERBICIDE AND PREDATOR CUES ON A GENERALIST PREDATOR IN AGRICULTURAL SYSTEMS

A DISSERTATION

Submitted to the Faculty of

Miami University in partial

Fulfillment of the requirements

for the degree of

Doctor of Philosophy

Department of Zoology

Kerri M. Wrinn

Miami University

Oxford, Ohio

2010

Advisor: Ann L. Rypstra

Table of Contents Table of contents ii. List of Tables iv. List of Figures v. Acknowledgements vi. Chapter 1: General Introduction 1 Literature cited 4 Chapter 2: Predator cues and an herbicide impact activity, emigration and survival 10 in an agrobiont wolf spider Introduction 10 Methods 13 Collection and maintenance of animals 13 Herbicide preparation 13 Laboratory experiment 14 Field mesocosms 15 Results 17 Laboratory experiment 17 Field mesocosms 18 Discussion 18 Literature cited 23 Chapter 3: Effects of predator cues, prey level and an herbicide on reproduction in 35 Pardosa milvina (Araneae Lycosidae) Introduction 35 Methods 38 Animal collection and maintenance 38 Herbicide preparation 38 Collection of predator cues and application of herbicide 39 Exposure to chemical cues and foraging trials 39 Reproduction 39 Data analysis 40 Results 40

Foraging trials: Prey capture and consumption 40 Reproduction 41 Survival 42 Discussion 43 Foraging trials: Prey capture and consumption 43 Reproduction 44 Survival 46 Conclusions 48 Literature cited 50 Chapter 4: Exposure to herbicide and predator cues has no effect on growth, 62 development time, fluctuating asymmetry or survival in Pardosa milvina (Araneae: Lycosidae). Introduction 62 Methods 64 Animal care 64 Experimental design 65 Preparation and exposure to predator cues and herbicide 65 Data collection and analysis 66 Results 67 Growth, development time and survival 67 Fluctuating asymmetry 68 Discussion 68 Literature cited 72 Chapter 5: General conclusions and synthesis 81 Literature cited 86 Appendices 90 Appendix 90

iii

List of Tables

Table 2.1: The impacts of predator cues and/or herbicide on laboratory tests of 28 activity variables Table 2.2: Mean ± standard error of activity measures for P. milvina in response 29 to predator cues and/or herbicide or water Table 3.1: Mean ± standard error for number of crickets killed across treatments 55 Table 3.2: The effects of predator cues, herbicide and prey level on prey capture 56 and consumption Table 3.3: Tests for impacts of treatment on reproduction and survival in 57 P. milvina Table 4.1: Means ± standard error and Kruskal-Wallis results for measures of 76 growth between molts 3-4 Table 4.2: Means ± standard error and Kruskal-Wallis results for measures of 77 growth between molts 4-5 Table 4.3: Means ± standard error and Kruskal-Wallis results for measures of 78 growth between molts 5-6 Table 5.1: Effects of the stressors: predator cues, herbicide and prey availability 88 on the behavior and life history of P. milvina (Summary of Chapters 2-4)

List of Figures Fig. 1.1: Presence of P. milvina and its intraguild predators H.helluo and 7 S. quadriceps in A. soy and B. corn fields. Fig. 1.2: Number of individuals captured in pitfall traps for three orders of prey 8 likely consumed by P. milvina in A. Soy and B. corn fields. Fig. 1.3: Outline of data chapters: Potential impacts of herbicide and predator 9 cues on different behaviors and life stages of P. milvina. Fig. 2.1: Laboratory arena for exposing Pardosa milvina to herbicide and/or 30 predator cues. Fig. 2.2: Field mesocosm for testing emigration of Pardosa milvina in response 31 to herbicide and/or S. quadriceps cues. Fig. 2.3: Activity of Pardosa milvina when exposed to herbicide or water and/or 32 predator cues in 15 minute laboratory trials. Fig. 2.4: Proportion of spiders remaining in the container over time for P. milvina 33 exposed to predator cues and/or herbicide or water in 15 minute laboratory trials. Fig. 2.5: Survival over a 60 day period of P. milvina after being subjected to 34 S. quadriceps cues, herbicide, neither, or both for 24 hours. Fig. 3.1: Fully factorial experimental design for the effects of Predator cues, 58 herbicide, and prey availability on foraging and reproduction in P. milvina. Fig. 3.2: Abdomen width change in P. milvina after foraging for two hours on 59 crickets during exposure to predator cues and/or herbicide. Fig. 3.3: Reproduction by P. milvina after exposure to predator cues and 60 herbicide. Fig. 3.4: Proportion of adult P. milvina surviving over a 60 day period after 61 exposure to predator cues and herbicide. Fig. 4.1: Diagram of the container in which juvenile Pardosa milvina were exposed 79 to Scarites quadriceps cues and/or glyphosate-based herbicide for 24 hours. Fig. 4.2: Relationship between the absolute value of fluctuating asymmetry (FA) of 80 Patella-tibia length and Cephalothorax width in adult Pardosa milvina. Fig. 5.1: Number of adult female, adult male and juvenile P. milvina collected in 89 dry pitfall traps over a 24 hour period from A. soy and B. corn fields.

Acknowledgements

I would never have been able to complete my dissertation without the support of many people. I would like to thank my advisor, Ann Rypstra for guiding me through the doctoral process from start to finish with plenty of advice and encouragement. I would also like to express my gratitude to my committee, Michelle Boone, Tom Crist, Maria Gonzalez, and Dave Gorchov for their aid in designing, analyzing and writing up my work. I am grateful to my lab mates Chad Hoefler, Jen Riem, Shawn Wilder, Jason Schmidt, and Michael Sitvarin for their support in helping me with fieldwork, reading drafts of my chapters or just giving me an encouraging word. I appreciate the assistance of the multitude of undergraduate students that contributed to my research through their help in the lab and field and by providing comments on my papers during lab meetings over the years. I would like to express special thanks to Sam Evans, Chris Carter and Sandra Rittman; I definitely couldn’t have completed my experiments without them! I would also like to acknowledge the other graduate students and post docs who offered their friendship and support including: Makiri Sei, Gary Gerald, Lisette Torres, Molly Steinwald, Adrian Chesh, Rick Seidel, and many others. I am grateful to my family and non science friends who never failed to support me in my obsession with arachnids. Finally, I want to thank my husband Todd Levine, for supporting me both personally and professionally throughout the Ph.D process and Cassie, our daughter to be, for giving me a reason to finish my dissertation on time! For financial support I thank Miami University and the Zoology department.

Chapter 1: General Introduction

The process of detection and response to chemical cues plays a critical role in the lives of animals across various ecosystems. Chemical cues are particularly important in predator- prey interactions; prey recognize inter-specific chemical cues and avoid them to evade predators (for reviews see Kats and Dill; 1998; Dick and Grostal, 2001), while predators use chemical cues to locate prey (Koivula and Korpimaki, 2001). Therefore, failure of animals to detect these natural chemical cues could disrupt predator-prey interactions, with impacts at both the population and community levels. Evidence suggests anthropogenic chemicals can act to indirectly harm non- target organisms by interrupting the flow of natural chemical information (reviewed in Lurling and Scheffer, 2007; Klaschka, 2008). However, little is known about how anthropogenic chemicals influence natural chemical signaling in terrestrial systems (Komeza et al., 2001; Salerno et al., 2002), perhaps because cues do not disseminate as well as in aquatic systems. Agricultural systems are terrestrial habitats in which animals are likely to encounter cues from both natural and anthropogenic chemicals simultaneously. Pressure is often high in these systems from predators that might be detected through their chemical cues (Marshall and Rypstra 1999; Marshall et al., 2000). Furthermore, anthropogenic chemicals commonly sprayed there to manage common insects, fungi, weeds, and bacteria might also affect the behavior and survival of non-target animals (Pimental et al.,1991; Freemark and Boutin, 1994; reviewed in Desneux, et al., 2007). Animals in these systems also encounter a variety of other stressors that impact their behavior, life history, and survival (Freemark, 1995 and references therein) and might serve to amplify or nullify the interactions between natural and anthropogenic chemical cues. For example, agricultural systems are strongly seasonal, with wide variation in habitat and food availability (Harwood et al., 2001; Wise, 2006), both which could conceivably affect an animal’s reactions to encountering chemical cues. Spiders are generalist predators likely to be heavily impacted by the presence of both natural and anthropogenic chemical cues (Persons and Rypstra, 2001; Persons et al., 2001; 2002; Benamú, et al., 2010). Spiders are ecologically important terrestrial arthropods often useful in agricultural systems because they exert top down pressure on “pest” species (Carter and Rypstra, 1995; Snyder and Wise, 2001; Hlivko and Rypstra, 2003; Nyffeler and Sunderland, 2003) and these effects are not limited to direct consumption (see Sunderland, 1999 for a review). Some

1 spider species may reach very high densities in agricultural fields and are able to re-colonize quickly after a disturbance (Marshall et al., 2002; Buddle and Rypstra, 2003), making them an integral part of any predator assemblage. Studying the effects of natural and anthropogenic chemical cues on these predators and their interactions with other species should increase our understanding of food webs in these systems. Throughout the growing season, the wolf spider Pardosa milvina is the most abundant epigeal arthropod predator in many agro-ecosystems across the eastern United States (Marshall and Rypstra, 1999; Marshall et al., 2002), making this a good species with which to test the impacts of multiple stressors there. Because of their high abundance in these systems, changes in their behavior and population dynamics due to exposure to these stressors could have large effects on the ecosystem. I began in Summer 2006 by documenting the timing of stressors including presence of predators, herbicide and prey in agricultural systems across the season (see appendix 1 for methods). Specifically, I observed activity density of P. milvina and its arthropod predators (the wolf spider Hogna helluo and the carabid beetle, Scarites quadriceps), as well as the timing of herbicide spraying. Furthermore, I looked at the availability of prey across the season to determine periods where P. milvina might experience food limitation. First, Pardosa milvina co-occur with two other abundant intraguild predators, the wolf spider Hogna helluo, and the carabid beetle, Scarites quadriceps across the entire growing season in agricultural fields (Fig. 1.1). This supports previous data on the presence of H. helluo and S. quadriceps, in these specific fields (Marshall et al., 2002; Dan Pavuk, personal communication), as well as in similar systems elsewhere (Snyder and Wise, 1999; Pavuk et al., 1997; Clark et al., 2006). There appears to be no direct effect of the presence of predators, even at their highest densities, on the numbers of P. milvina captured in this system, although it is difficult to tell what other factors could be impacting the presence of these species as their phenology is variable across years (Fig. 1.1; Marshall et al., 2002). Previous research shows that H. helluo preys upon P. milvina and that P. milvina respond to chemical cues from this predator (Persons and Rypstra, 2001; Persons et al., 2001, 2002). Scarites quadriceps might also prey upon or compete with P. milvina (Snyder and Wise, 1999, 2001), but it is unknown how P. milvina responds to the chemical cues from this predator. Second, these systems were planted with herbicide resistant crops, allowing commercially formulated glyphosate-based herbicides to be sprayed at several times across the season.

Furthermore, the spray dates overlapped with periods when P. milvina and its two intraguild predators were at relatively high abundances in the field, indicating P. milvina likely encounters both predators and herbicide simultaneously (Fig. 1.1). Commercially formulated glyphosate- based herbicides (e.g. Roundup®) are one common group of anthropogenic chemicals that might impact receipt of chemical signals (Tierney et al., 2006). Although some studies have addressed the impacts of this type of herbicide on survival in terrestrial arthropods (see Giesy et al., 2000 for a review), little is known about its direct impacts on their behavior and life history (Brust, 1990). Furthermore, to date, no studies have addressed the interaction between this herbicide and predator chemical cues in terrestrial systems. Finally, abundances of three orders of prey known to be consumed commonly by spiders in the field (Collembola, Diptera, and Homoptera- Nyffeler and Breener, 1990) fluctuated considerably across the season, indicating that there might be periods of low food availability (Fig. 1.2). Based on the information I gained from the initial surveys, I developed several experiments to test the impacts of predator cues, herbicide, and food limitation on the activity and life history of P. milvina. In the following chapters I report the results of experiments aiming to test the effects of exposure to a commercially formulated glyphosate-based herbicide on the reactions of P. milvina to cues from two intraguild predators (H. helluo and S. quadriceps) (Fig. 1.3). Pardosa milvina coexist throughout their life-cycles with their predators, so I test the impacts of predator cues and herbicide on P. milvina at different life stages. To begin, in chapter 2, I focus on whether female P. milvina at the penultimate molt or adulthood change activity and emigration in response to predator cues and herbicide in the laboratory. In a separate experiment, I look at changes in emigration by this species in response to herbicide and/or S. quadriceps cues in field mesocosms. Finally, after bringing these experimental animals back to the laboratory, I observe patterns in survival of these spiders over the next two months. In Chapter 3, I address foraging and reproduction in females and if and how these behaviors are impacted by herbicide, predator cues (H. helluo and S. quadriceps), and food limitation. In Chapter 4, I consider how growth, development and survival are impacted by herbicide and predator cues (S. quadriceps only). Finally, in Chapter 5, I discuss my overall findings and ultimately how anthropogenic and natural stressors might interact to impact the life history and abundance of P. milvina and, by extension, its role in arthropod food webs in agricultural systems.

Literature Cited

Benamú, MA., Schneider, MI., Sánchez, NE. 2010. Effects of the herbicide glyphosate on biological attributes of Alpaida veniliae (Araneae, Araneidae), in laboratory. Chemosphere. 78: 871-876.

Brust, GE. 1990. Direct and indirect effects of four herbicides on the activity of carabid beetles (coleopteran: Carabidae). Pesticide Science. 30: 309-320.

Buddle C. N., A. L. Rypstra. 2003. Factors initiating emigration of two wolf spider species (Araneae : Lycosidae) in an agroecosystem. Environmental Entomology 32:88-95.

Carter P. E., A. L. Rypstra. 1995. Top-Down Effects in Soybean Agroecosystems - Spider Density Affects Herbivore Damage. Oikos 72:433-439.

Clark, S., Szlavecz, K., Cavigelli, MA., Purrington, F. 2006. Ground beetle (Coleoptera: Carabidae) assemblages in organic, no-till and chisel-till cropping systems in Maryland. Environmental Entomology. 35: 1304-1312.

Dicke, M., Grostal, P. 2001. Chemical detection of natural enemies by arthropods: An ecological prospective. Annual Review of Ecology and Systematics. 32: 1-23.

Desneux, N., Decourtye, A., Delpeuch J. 2007. The sublethal effects of pesticides on beneficial arthropods. Annual Review of Entomology. 52: 81-106.

Freemark, K., Boutin, C. 1994. Impacts of agricultural herbicide use on terrestrial wildlife in temperate landscapes: A review with special reference to North America. Agriculture, Ecosystems and Environment. 52:67-91

Freemark, K. 1995. Assessing effects of agriculture on terrestrial wildlife: developing a hierarchical approach for the US EPA. Landscape and Urban Planning. 31:99-115

Giesy, JP., Dobson, S., Soloman, KR., 2000. Ecotoxicological risk assessment for Roundup® herbicide. Reviews in Environmental Contamination and Toxicology. 167: 35–120

Harwood, J.D., Sunderland, K.D., Symondson, WOC. 2001. Living where the food is: web location by linyphiid spiders in relation to prey availability in winter wheat. Journal of Applied Ecology. 38:88-99.

Hlivko J. T., A. L. Rypstra. 2003. Spiders reduce herbivory: Nonlethal effects of spiders on the consumption of soybean leaves by beetle pests. Annals of the Entomological Society of America 96:914-919.

Kats, LB., Dill, LM. 1998. The scent of death: Chemosensory assessment of predation risk by prey animals. Ecoscience. 5: 361-394.

Klaschka, U. 2008. The infochemical effect- a new chapter in ecotoxicology. Environmental Science and Pollution Research. 15:452-462.

Koivula, M, Korpimaki, E. 2001. Do scent marks increase predation risk of microtine rodents? Oikos. 95:275-281.

Komeza, N, Fouillet, P, Bouletreau, M, Delpuech, JM. 2001. Modification, by the insecticide chlorpyrifos, of the behavioral response to kairomones of a parasitoid wasp, Leptopilina boulardi. Archives of Environmental Contamination and Toxicology. 41: 436-442.

Lurling, M, Scheffer, M. 2007. Info-disruption: pollution and the transfer of chemical information between organisms. Trends in Ecology and Evolution. 22:374-379.

Marshall, SD, Rypstra, AL. 1999. Spider competition in structurally simple ecosystems. Journal of Arachnology. 27:343-350.

Marshall, S.,Walker, S., Rypstra A. 2000. A test for differential colonization and competitive ability in two generalist predators. Ecology. 81:3341–3349.

Marshall S. D., D. M. Pavuk, and A. L. Rypstra. 2002. A comparative study of phenology and daily activity patterns in the wolf spiders Pardosa milvina and Hogna helluo in soybean agroecosystems in southwestern Ohio (Araneae, Lycosidae). Journal of Arachnology 30:503-510.

Nyffeler, M., Breener, R.G. 1990. Evidence of low daily food consumption by wolf spiders in meadowland and comparison with other cursorial hunters. Zeitschrift für Angewandte Entomologie. 110:73-81.

Nyffeler, M., Sunderland, D.K. 2003. Composition, abundance and pest control potential of spider communities in agroecosystems: a comparison of European and US studies. Agriculture, Ecosystems and the Environment 95:579-612

Pavuk, DM., Purrington, FF., Williams, CE., Stinner, BR. 1997. Ground beetle (Coleoptera: Carabidae) activity density and community composition in vegetationally diverse corn agroecosystems. American Midland Naturalist. 138: 14-28.

Persons M. H., A. L. Rypstra. 2001. Wolf spiders show graded antipredator behavior in the presence of chemical cues from different sized predators. Journal of chemical ecology 27:2493-2504.

Persons MH., Walker, SE., Rypstra, AL., Marshall, SD. 2001. Wolf spider predator avoidance tactics and survival in the presence of diet-associated predator cues (Araneae: Lysociade). Animal Behaviour. 61: 43-51.

Persons, MH., Walker, SE., Rypstra, AL. 2002. Fitness costs and benefits of antipredator behavior mediated by chemotactile cues in the wolf spider Pardosa milvina (Araneae: Lycosidae). Behavioral Ecology. 13: 386-392

Pimental D., Acquay, H., Biltonen, M., Rice, P., et al.,1991. Environmental and economic costs of pesticide use. Bioscience. 42: 750-760.

Salerno, G; Colazza, S; Conti, E. 2002. Sub-lethal effects of deltamethrin on walking behaviour and response to host kairomone of the egg parasitoid Trissolcus basalis. Pest Management Science. 58:663-668

Snyder W. E., D. H. Wise. 2001. Contrasting trophic cascades generated by a community of generalist predators. Ecology 82:1571-1583.

Snyder W. E., D. H. Wise. 1999. Predator interference and the establishment of generalist predator populations for biocontrol. Biological Control 15:283-292.

Sunderland, K. 1999. Mechanisms underlying the effects of spiders on pest populations. Journal of Arachnology. 27: 308-316.

Tierney, KB., Singh, CR., Ross, PS., Kennedy, CJ. 2006. Relating olfactory neurotoxicology to altered olfactory-mediated behavior in rainbow trout exposed to three currently-used pesticides. Aquatic Toxicology. 81; 55-64.

Wise, D.2006. Cannibalism,food limitation, intraspecific competition and the regulation of spider populations. Annual Review of Entomology. 51:441-465.

Fig 1.1: Presence of P. milvina and its intraguild predators H.helluo and S. quadriceps in A. soy and B. corn fields across the agricultural season from pre-planting through post harvest in 2006. Arrows indicate times during the season when the fields were sprayed with a glyphosate-based herbicide. Number of individuals per field was pooled across the five traps set for each field and then averaged across the three replicate fields of each type where they were collected.

Fig 1.2: Number of individuals captured in pitfall traps across the agricultural season in 2006 for three orders of prey likely consumed by P. milvina in A. Soy and B. corn fields. Number of individuals per field was pooled across the five traps set for each field and was then averaged across the three replicate fields of each type. * Indicates number of homoptera actually present for the second time period, as they were well above the regular scale for the other groups.

Fig. 1.3: Outline of data chapters: Potential impacts of herbicide and predator cues on different behaviors (activity, etc) and life stages (Chapter 2 and 3-adult and Chapter 4-juvenile) of P. milvina. Solid arrows indicate possible direct effects of factors whereas dotted arrows indicate possible interactive effects between factors.

Chapter 2: Predator cues and an herbicide impact activity, emigration, and survival in an agrobiont wolf spider

Introduction: Chemical cues are important for signaling both within and between species. Intra- specific chemical cues (pheromones) are used for advertising territories (Sillero-Zubiri and Macdonald, 1998), locating mates (Byrne and Keogh, 2007) and recognizing and communicating with kin (Howard and Blomquist, 2005). Inter-specific chemical cues (kairomones) are used by prey to avoid predators (Grostal and Dicke, 1999) or conversely by predators to locate prey (Koivula and Korpimaki, 2001). Therefore, factors that affect the abilities of animals to detect these natural chemical cues could have impacts at both population and community levels. There is a growing body of evidence that suggests anthropogenic chemicals often interrupt this natural chemical information flow between animals (reviewed in Lurling and Scheffer, 2007; Klaschka, 2008). For example, Saglio et al. (1996) found that goldfish exposed to the insecticide carbofuran were less able to detect a prey extract compared to unexposed individuals. Anthropogenic chemical cues might also impact the ability of prey to detect predators via either alarm cues from conspecifics or cues from predators themselves. For instance, Scott et al. (2003) found that in the presence of cadmium (a common trace metal contaminant) juvenile rainbow trout are unable to sense chemical alarm cues from injured conspecifics. Furthermore, Mandrillon and Saglio (2007) found that toad tadpoles did not respond to cues from a crayfish predator when those cues were combined with greater than 0.01 mg/l of the herbicide amitrole. Although scientists have begun to examine the impacts of anthropogenic chemicals on natural chemical signaling in more detail, these studies are overwhelmingly biased towards aquatic systems (over 80% of studies conducted- Lurling and Scheffer, 2007). Only a few studies (Komeza et al., 2001; Salerno et al., 2002) have addressed the impacts of anthropogenic chemicals on natural chemical signaling in terrestrial systems, despite their potential importance in these systems. Modern agriculture involves the application of chemicals to manage insects, fungi, weeds and bacteria (Pimentel et al., 1991). Commercially formulated glyphosate-based herbicides comprise one widely used group of such chemicals. These herbicides have increased rapidly in agricultural use worldwide with the development of glyphosate-resistant crops (Woodburn,

2000; Lundgren et al., 2009). The chance of non-target organisms contacting these herbicides is high; so the herbicides’ direct impacts on survival have been studied for a variety of taxa (Hassan, et al., 1988; see Giesey et al., 2000 for a review; Haughton et al., 2001). At standard field application rates, herbicides have only minimal direct impacts on survival of most of the terrestrial species tested in laboratory studies (less than 25% mortality) (Hassan, et al., 1988; Giesey et al. 2000, Haughton et al. 2001). Therefore, researchers explain changes in terrestrial arthropod activity and density after exposure to these herbicides in the field as indirect effects resulting from changes in plant structure (Giesey et al., 2000, Haughton et al., 1999; 2001). Little is known, however, about the direct effects of these herbicides on the behavior of terrestrial arthropods (Brust, 1990; Michalkova and Pekar, 2009; Evans et al., in review). There has also been a lack of studies conducted on how commercially formulated glyphosate-based herbicides might interact with natural chemical cues. This is potentially important as many of the interactions between terrestrial arthropod predators and prey are mediated by chemical cues (Freund and Olmstead, 2000; for a review see Dicke and Grostal, 2001). For example, spider mites (Tetranychidae) emigrate from surfaces exposed to chemical cues from predatory mites (Phytoseiidae) (Grostal and Dicke, 1999). Recent studies in other taxa suggest that glyphosate can inhibit olfaction, and might therefore interfere with the detection or reaction of animals to their natural pheromones or kariomones (Tierney et al., 2006; 2007). Chemical interference of this type can result in decreased individual success as well as changes in food web interactions (Lurling and Scheffer, 2007). The goal of this study was to determine if a glyphosate based herbicide would affect the manner in which a generalist predator responded to chemical information from its intraguild predators. To accomplish this, I quantified activity and the propensity of this predator to move away from an area in response to chemical cues. A change in activity level is important as this can affect an animal’s susceptibility to predation (Persons et al., 2001). Furthermore, immigration and emigration are both key to the success of animals living in highly disturbed agroecosystems where use of herbicides is common (Brust, 1990; Bishop and Riechert, 1990; Halaj et al., 2000; Thorbek and Bilde, 2004). Study system The focal species for these experiments, Pardosa milvina (Araneae: Lycosidae), is the numerically dominant, epigeal generalist arthropod predator in agricultural fields throughout

11 eastern North America (Marshall and Rypstra, 1999; Marshall et al., 2002). I also chose to use two intraguild predator species that prey upon P. milvina and overlap in range and habitat usage: the wolf spider, Hogna helluo (Araneae: Lycosidae), and the ground beetle Scarites quadriceps (Coleoptera: Carabidae) (Marshall et al., 2002; Snyder and Wise, 1999). Pardosa milvina is very chemically aware; it uses both air- and substrate-borne cues to find mates and detect predators (Searcy et al., 1999; Persons and Rypstra, 2001; Rypstra et al., 2003; Schonewolf et al., 2006). Specifically, these spiders display effective anti-predator behavior when placed in an area previously occupied by the large wolf spider, Hogna helluo (Persons et al 2001; 2002). In fact, P. milvina is so sensitive to the silk feces and other excreta deposited on a substrate by H. helluo (hereafter referred to as chemical cues) they show a risk appropriate response to the size, sex and hunger level of H. helluo that produced the cues (Persons and Rypstra, 2001; Bell et al., 2006). Scarites quadriceps occur on the soil surface of agroecosystems at densities similar to H. helluo (about 1/m2 (Snyder and Wise, 1999)) and are likely predators of P. milvina as well. Given the sensitivity of P. milvina to substrate borne chemical cues from H. helluo and a mantid predator (Persons et al., 2001; Wilder and Rypstra, 2004), I expect that these spiders will be able to detect chemical cues through feces or other excreta from S. quadriceps as well. However, I do not know if or how P. milvina might respond to this predator’s cues. I was interested in how a commercially formulated herbicide with the active ingredient glyphosate affected the activity of P. milvina and if the herbicide interfered with the spider’s reaction to substrate borne chemical cues from either H. helluo or S. quadriceps. A previous laboratory study demonstrated that exposure to this type of herbicide alone via the substrate can decrease activity in P. milvina (Evans et al., in review). However, this study did not address the herbicide’s potential interaction with predator chemical cues. To address this, in the laboratory I explored the effect of chemical cues from both predators, alone and accompanied by herbicide, on activity level and the propensity for Pardosa milvina to leave a closed arena. Furthermore, in a field mesocosm study I documented the effects of S. quadriceps, and herbicide on propensity of P. milvina to emigrate within 24 hours. These animals were subsequently maintained in the laboratory to document any treatment effects on survival. I did not run a parallel mesocosm study with H. helluo partly because Buddle and Rypstra (2003) had already addressed the impacts of this predator’s presence on the emigration of P. milvina in the field. More importantly however, my decision was also based on the results of the laboratory experiment which

12 suggested that presence of S. quadriceps combined with herbicide might have far more interesting effects than the combination of H. helluo and herbicide. Methods: Collection and maintenance of animals: I conducted the following experiments between May and September of 2007. I collected animals for all the experiments from corn and soybean fields at the Miami University Ecology Research Center (Oxford, OH). I maintained all animals in a laboratory environmental room at 25C, 55-60%, RH and an 13:11 L:D cycle. I housed all animals in opaque plastic containers (8 cm diameter x 5 cm high for each P. milvina and 12 cm diameter x 10 cm high for each H. helluo and S. quadriceps) with a moistened layer of 50:50 peat moss and soil mix covering the bottom (2 cm deep for P. milvina and H. helluo, and 4 cm deep for S. quadriceps, as this species burrows). I fed all animals a weekly diet of two appropriately sized crickets (Acheta domesticus). I only used female H. helluo at the penultimate molt or adulthood because they are significantly larger than males, differ in morphology and tend to consume more prey (Walker and Rypstra, 2002); however male and female S. quadriceps are similar in size and morphology so I did not sort them by sex. Herbicide preparation For both experiments, I used the commercially formulated herbicide Buccaneer® Plus, also known as Roundup original®, created by the Monsanto Company, St. Louis, Missouri, USA. This herbicide contains the active ingredient glyphosate (480g/L) in the form of isopropylamine salt and an added polyethoxylated tallowamine (POEA) surfactant. Before applying the herbicide to the substrate I diluted it to 2.5% for both the laboratory and field experiments, which was within the manufacturer’s recommended levels of 0.625%-5%. For the laboratory experiment I used a spray rate of 127.4 mL/m2 (or 15.3 kg a.i. ha-1 of glyphosate). For the field experiment, I used a spray rate approximately half of that used in the laboratory of 60.2 mL/m2 (or 7.2 kg a.i. ha-1 of glyphosate), as this was closer to the spray rate normally used in the field and I did not need as much liquid to get even coverage on soil as I did on filter paper. Although both of these spray rates exceeded the recommended field coverage rate of 18.7 ml/m2 (2.25 kg a.i. ha-1 of glyphosate) and were thus unrealistic as to what a spider might be exposed to under normal conditions, they were the minimum necessary to gain a complete and uniform coverage of the areas for the laboratory container with filter paper and field mesocosm container

13 with soil respectively. Furthermore, pooling in low areas in the field, or improper application could lead to higher spray rates than recommended, making our application rates indicative of a worst case scenario. Laboratory experiment Predators leave substrate-borne cues behind as they occupy an area. I measured the impacts of these cues separately and combined with herbicide on the activity and emigration of P. milvina in the laboratory. I employed a full factorial design of six treatments, each with a sample size of n=20. Pardosa milvina were randomly assigned to one of the six following treatments: 1) Hogna helluo cues/water, 2) H. helluo cues/herbicide, 3) Scarites quadriceps cues/water, 4) S. quadriceps cues/herbicide, 5) No cues/water, or 6) No cues/herbicide. Arena preparation I conducted each trial in a cylindrical plastic test arena 15 cm in diameter with filter paper on the bottom. The arena was divided into quadrants: alternating wet papers (herbicide or water) with dry papers (predator cues or blank paper), such that for each treatment two quadrants were covered with water or herbicide and two quadrants were covered with no cues or predator cues (Fig 2.1). Each arena had four 1 cm diameter escape holes (one per quadrant) that were opened during the second part of each trial. Collection of predator cues and application of herbicide For the treatments with predator cues, I allowed H. helluo or S. quadriceps to deposit cues on a piece of filter paper for 24 hours prior to running the trials where I exposed P. milvina to these cues. To control for differences in cue signatures by hunger, I fed each of these predators one 16 mm cricket 15 minutes prior to beginning their deposition of cues. Directly before the trials I removed each predator from the deposition container, cut its cue paper into four pieces and placed two of these in non adjacent quadrants within the test arena. I added herbicide or water using an airbrush sprayer to another piece of filter paper, cut it into four pieces and added two of these to the other two non-predator cues quadrants (Fig. 2.1). Trials For each trial I placed a female P. milvina under a clear vial in the center of the arena and allowed her to acclimate for five minutes, then I removed the vial and the trial began. I recorded each spider from above during the 15 minute activity trial using a video camera, and measured activity levels using an automated digital data collection system (VideoMex V, Columbus

Instruments, Columbus, OH, USA). The VideoMex summarizes the time spent ambulatory (time during which the movement is more than one body length per second), time spent moving in place (which includes smaller motions such as waving legs or rotating but not relocating) and total distance travelled. In addition I calculated average speed as time spent in locomotion divided by distance traveled. After the 15 minute activity trial, I recaptured the spider under a vial and allowed her 5 minutes to re-acclimate. I then removed paper barriers from the emigration holes, allowing the spider to leave the arena. I then monitored the spiders for another 15 minutes and recorded if and when they left the container through the emigration holes. Data Analysis I used either a natural log or square root transformation to make the distributions more normal for each activity measure quantified by the VideoMex (time spent ambulatory, time spent moving in place, distance traveled and average speed). I then used a two-way ANOVA (analysis of variance) with predator cues and herbicide as factors to compare the above activity variables between treatments. I additionally addressed the question of whether spiders showed a preference that differed by treatment for the “wet” parts (covered with water or herbicide) vs. the “dry” parts of the arena (blank or covered with S. quadriceps or H. helluo cues). To do this, I first subtracted the time spent in behaviors on the wet parts of the arena from the time spent on the dry parts of the arena. Then, I used a two-way ANOVA with predator cues and herbicide as factors to compare these dry- wet side activity variables between treatments. I next explored the effects of treatment on whether or not P. milvina emigrated in the time allotted for a trial (yes or no) using a Logit model (Menard, 2002) with presence or absence of predator cues and presence of herbicide or water as factors. Finally, I measured time to emigration and compared it between treatments using a Proportional Hazards analysis (Cox, 1972), where I censored individuals that did not emigrate within the 15 minute trials. All statistical analyses were conducted using JMP 7.0 (SAS institute Inc.). Field mesocosms In the field I measured the impacts of Scarites quadriceps separately and combined with herbicide on the propensity of P. milvina to emigrate from enclosures. I employed a full factorial design of four treatments, each with a sample size of n=10: S. quadriceps accompanied by water or herbicide, or no predator with water or herbicide.

Mesocosm setup I ran each trial in a 46.5 cm diameter circular plastic mesocosm with 15 cm of soil and a thin layer of straw placed on the bottom to provide cover and act as a preferred substrate of P. milvina (Rypstra et al., 2007) (Fig. 2.2). Each mesocosm had four 1 cm diameter escape holes that led into 9 cm diameter x 12 cm high plastic cups with lids to prevent the spiders from escaping. I also placed a moistened, crumpled paper towel within each cup to provide spiders that had emigrated with water and shade. After introduction of the test animals and spraying water or herbicide, I covered each mesocosm with bridal veil mesh cloth to allow airflow and light to penetrate, but prevent escape of the animals and protect them from external predation. Animal preparation/introduction Both spiders and beetles were collected from the surrounding agricultural fields within 24 hours prior to the experiment and held in the laboratory until use with access to water but no food. Once in the field, I began the experiment by placing three beetles in each mesocosm and allowing them one hour to burrow and acclimate. Therefore, in this experiment spiders were potentially exposed to beetles themselves as well as their cues. Then, I placed 12 spiders in each mesocosm along with 24 three mm long laboratory-reared crickets (Acheta domesticus) to act as prey for the spiders and beetles. Finally, I sprayed each container with 10 mL of Buccaneer® Plus herbicide or distilled water as appropriate using a backpack sprayer. Data collection/analysis Prior to the trials, I measured abdomen width in all spiders to the nearest 0.01 mm using a Wild 5 microscope and a digital micrometer. I calculated abdomen width to determine body condition in all spiders as it changes with food intake (Anderson, 1974; Jakob et al., 1996; Persons et al., 2002; Rypstra et al, 2007). I compared the body condition of spiders between treatments using a two way ANOVA with presence/absence of predator and herbicide as factors. After the 24 hour period that the animals were in the mesocosms, I counted the number of spiders that had emigrated through the escape holes into the pitfall traps. I compared emigration among treatments using a two-way ANOVA with presence of S. quadriceps and herbicide as factors. After the experiment, I placed the spiders in individual cups and maintained them in the laboratory for 60 days checking them weekly. I compared survival over a 60-day period using a proportional hazards analysis (Cox, 1972) with predator and herbicide as factors. Although I recorded the date and treatment for each spider, I failed to record the specific container from

16 which I retrieved each spider before I returned to the laboratory. To determine if including container in my analysis of survival would alter the results, I conducted a simulation in which spiders from a given treatment on a given day were randomly assigned to containers and container was nested in treatment in ANOVAs. I repeated the randomization 1,000 times. I reasoned that if the ANOVA was significant more than 95% of the time then the likelihood of the container affecting my results would be less than 5% (or the standard p-value). Results: Laboratory experiment Predator cues significantly influenced the distance P. milvina traveled, their time spent and their average speed, whereas herbicide alone only impacted motion in place (Table 2.1, Fig. 2.3). Pardosa milvina spent significantly less time ambulatory, spent less time moving in place, traveled more slowly, and covered less distance when exposed to H. helluo cues, regardless of the presence of herbicide. In fact, these spiders nearly halved their average speed and time spent ambulatory and travelled about a third of the distance compared to control spiders not exposed to predator cues (Fig. 2.3). In contrast, when exposed to S. quadriceps cues alone, P. milvina showed no statistically significant change in activity compared to control spiders. However, there was a tendency for these spiders to spend more time ambulatory and travel farther (Fig. 2.3 A,C). Interestingly, when P. milvina were exposed to S. quadriceps cues and herbicide together, time spent ambulatory decreased as evidenced by the significant predator cues x herbicide interaction (Table 2.1, Fig. 2.3 A). Although not statistically significant, this same pattern held for distance traveled, with spiders exposed to both S. quadriceps cues and herbicide traveling shorter distances (Fig. 2.3 C). Finally, P. milvina were less ambulatory on wet parts of the arena (those with herbicide or water) compared to dry parts (those with predator cues or no cues) and this was consistent across treatments (Table 2.2; ANOVA: model: F5,117= 1.07; p=0.3799). Likewise, P. milvina spent more time moving in place, and traveled farther and faster on dry than on wet substrate (Table 2.2). Furthermore, this preference for dry substrate was consistent across treatments for movement in place (ANOVA model: F5,117=0.1319; p=0.9848), distance travelled

(ANOVA model: F5,98=1.3073; p=0.2677) and speed (ANOVA model: F5,95= 0.8868; p=0.4934). Whether or not a spider emigrated from the arena was significantly impacted by 2 treatment (Model: χ 5 = 24.51, p<0.0002; Table 2.3). Predator cues, herbicide and the interaction 2 between the two were all significant factors affecting whether a spider emigrated or not (χ 2 =

2 2 18.53, p<0.0002; χ 1 = 4.11, p<0.0427; χ 5 = 7.59, p<0.0224 respectively). Spiders in the H. helluo and herbicide combined treatment only emigrated 43% of the time, whereas spiders in the S. quadriceps and water treatment emigrated 100% of the time. The other treatments were intermediate (Fig. 2.4). Out of the spiders that emigrated, there was a significant difference 2 among treatments for time to emigration (Model: χ 5 = 17.33, p=0.0039). According to Likelihood ratio tests these differences among treatments were driven by presence of predator 2 cues (χ 2 = 14.53, p=0.0007). Spiders in both of the treatments exposed to H. helluo cues took the longest to emigrate, spiders exposed to herbicide and S. quadriceps cues combined were intermediate and the other three groups emigrated much more quickly (Fig. 2.4). There was no 2 significant impact of herbicide (χ 1 = 0.89, p=0.3431) on time to emigration or an interaction 2 between predator cues and herbicide (χ 2 = 2.83, p=0.2425). Field mesocosm experiment

Prior to trials, I found no difference in abdomens width between treatments (F3,521= 0.53, p=0.66), indicating that spiders in all treatments began the experiment in similar body condition. After 24 hours, on average, 7 ± 0.6 spiders had emigrated from each mesocosm and this was similar for all treatments. Therefore, neither presence of predator nor herbicide influenced this emigration in P. milvina and there was no interaction between the two (Model: F3,36= 0.2640, p= 2 0.8509). Interestingly, there was a significant impact of treatment on survival (Model: Χ3 =7.89, p= 0.0482) where S. quadriceps cues increased survival significantly (Χ2=4.14, p=0.0419), and herbicide caused a slight decrease (Χ2=3.64, p=0.0565), while the control group and the predator cues and herbicide combined group were intermediate (Fig. 2.5). The inclusion of container as a nested factor in our analyses of simulated groups did not change the importance of our primary treatments. Of 1,000 randomly assignations of spiders to containers, 961 still revealed a significant affect of treatment on survival. Thus the probability that the significant result is due to our ignoring a strong container effect is only 3.9 % Discussion: The results from these experiments confirm that Pardosa milvina wolf spiders are very sensitive to their chemical environment. Most importantly, although herbicide did not detectably affect the strong response of P. milvina to chemical cues from the co-occurring wolf spider, Hogna helluo, it did seem to alter the response to cues from Scarites quadriceps beetles. Specifically, when S. quadriceps information was presented along with herbicide there was a

18 significant reduction in time spent ambulatory and a slight, but nonsignificant reduction in the distance traveled, reducing the probability that they escaped from laboratory containers. Intriguingly, survival of animals exposed to S. quadriceps in field mesocosms seemed to be enhanced whereas survival of individuals in treatments with herbicide present was reduced. These results underscore the need to explore the reactions of animals to the combinations of stressors they may experience in a natural situation. We might have concluded that Pardosa milvina did not detect or respond to S. quadriceps cues at all had we not looked for the interaction with herbicide and we might have concluded that this spider did not respond to herbicide at all had we only explored its response to H. helluo cues. The shift in activity in the presence of S. quadriceps cues and herbicide could make P. milvina more vulnerable to predation. Thus, even if the herbicide does not cause direct mortality, it might have potentially large impacts on the population dynamics of these intraguild predators. Pardosa milvina responded to predator cues alone in a way that was dependent on the predator to which they were being exposed. In the presence of cues from the wolf spider H. helluo, they exhibited a very strong reduction in activity and subsequently emigrated from the container less often. In contrast, in the presence of cues from S. quadriceps, they responded by a slight increase in activity and emigrated more often. The difference in the strength of P. milvina’s reaction to its two intraguild predators could be explained by the silk laid down by H. helluo as it travels. This silk contains chemical cues and gives an additional tactile cue that is not present from S. quadriceps. Alternatively, H. helluo is likely a larger threat to P. milvina in nature than S. quadriceps is, and its cues would thus be expected to induce a stronger response. I was intrigued by the fact that P. milvina was less active on cues from H. helluo but more active on cues from S. quadriceps. I suspect that these spiders have evolved different strategies to deal with different kinds of predators. Hogna helluo are sit-and-wait predators, so much of their time is spent stationary (Walker et al., 1999). Furthermore, it has been suggested that wolf spiders might have motion-based vision and cannot detect other animals that are stationary (Rovner, 1996). Therefore, by remaining mostly stationary and moving quickly between resting spots, P. milvina might avoid contact with and/or detection by H. hello (Persons et al., 2001). However, although they share some of the same prey as H. helluo (Snyder and Wise, 1999), S. quadriceps appear to be much slower, more cumbersome hunters (personal observation, K.W.). Pardosa milvina might be able to escape this predator simply by moving

19 around more and leaving the area when signs of the predator are present. This method of predator specific behavior by prey is common in different taxa, including insects (Ferris and Rudolf, 2007), amphibians (Relyea, 2001), and mammals (Sefarth et al., 1980). Herbicide alone had little impact on the behavior of P. milvina, but there were significant interactions between predator cues and herbicide with subsequent effects on activity and emigration in the laboratory. In the presence of H. helluo cues, P. milvina decreased their activity levels similarly with or without herbicide. However, spiders exposed to S. quadriceps cues significantly decreased their activity only when those cues were paired with herbicide. Furthermore, when herbicide was combined with cues from either predator, P. milvina had a reduced chance of emigrating from the arena (Table 2.3). Recent chemical cue research has begun to address the potential of anthropogenic cues to interfere with animals’ detection of natural cues (reviewed in Lurling and Scheffer, 2007). In this study, it appears that there is an interactive effect of S. quadriceps cues and herbicide that changes the direction of the impacts of S. quadriceps cues alone on activity and emigration. Perhaps the herbicide is interfering with P. milvina’s recognition of S. quadriceps cues and as a result the spider misinterprets these cues and reacts in a way that is adaptive to avoiding other predators such as H. helluo. Alternatively, it is possible that P. milvina are becoming more active to avoid “dangerous” predator cue areas in the S. quadriceps only treatments because there are also “safe” water covered filter paper areas to travel to. However, in the herbicide and S. quadriceps cues treatment, the spider does not encounter any “safe” areas and thus may react by decreasing movement. Interactions between herbicide and predator cues might be generalized to other animals as well. For example, Relyea (2005b) found that the herbicide Roundup® significantly decreased survival in wood frogs when combined with predator cues. Further studies of predator cues and herbicide combined would therefore be valuable for a variety of taxa that encounter these chemical signals simultaneously. Not surprisingly, depressed activity in the presence of H. helluo cues reduced the number of spiders that escaped laboratory containers. Buddle and Rypstra (2003) found no effect of the presence of H. helluo on the emigration of P. milvina from a 7.5 m diameter patch. However, even if P. milvina only decrease activity and emigration in the field in response to H. helluo cues at a smaller scale, this might still ultimately increase their exposure to herbicide or decrease their ability to escape another environmental stressor. Although propensity to emigrate was impacted by S. quadriceps with and without herbicide in the laboratory, we did not see these effects in the

20 field mesocosm experiment. This is probably due to the fact that in the laboratory we measured time to emigration over a 15 minute period whereas in the field we measured number of individuals emigrating after 24 hours. Differences in number of spiders emigrating by treatment might have occurred in less than 24 hours and we were not able to detect them. Furthermore, the application rate in the laboratory experiments was double that of the field application rates, which could have unknown impacts on the spider’s behavior. Thus, future experiments should address emigration over shorter time spans (e.g. 1-2 hrs), over larger areas (e.g. larger plots) and with different spray rates. I also found that survival was influenced by S. quadriceps and herbicide. Surprisingly, P. milvina had a higher survival rate when they had been exposed to S. quadriceps for 24 hours in the field mesocosms. In contrast, 24 hours of exposure to herbicide alone had a slight negative impact on survival (Fig. 2.5). These results support previous experiments in the laboratory that suggest that this herbicide can decrease survival in P. milvina after only 30 minutes of exposure (Evans et al., in review). Interestingly, glyphosate-based herbicides have had minimal impacts on the survival of most arthropods tested in past studies, with the exception of a few species (Hassan, et al., 1988; Giesey et al. 2000, Haughton et al. 2001). However, as in many ecotoxicology studies, the aforementioned studies only followed survival over a short period of time (1-5 days usually), whereas our study followed the individuals for 60 days after exposure. Changes in survival over a longer time period could be important for P. milvina at both the individual and population levels, because over a period of 60 days reproductive females could produce two or even three eggsacs (see chapter 3). It is puzzling why P. milvina actually had higher survival after exposure to S. quadriceps cues alone and further experimentation is needed to determine the mechanism behind this. Research suggests that glyphosate (the active ingredient in the herbicide) might not be the cause of reduced survival in animals exposed to the herbicide. Glyphosate targets plant-specific enzymes and is thus supposed to be non-toxic to animals (Williams et al., 2000). However, several other studies have shown decreased survival in response to exposure from glyphosate- based herbicides for several aquatic organisms (Relyea, 2005 a; Folmar et al., 1979; Achiorno et al., 2008). Many glyphosate-based herbicides, including the one we used, Buccaneer® Plus, also include an inactive ingredient, the surfactant POEA (polyethyoxylated tallowamine) which helps the glyphosate enter the plant tissues. This surfactant contributes largely to lowered survival in

21 some animals exposed to glyphosate-based herbicides including frogs, daphnia, mussels and fish (Mann and Bidwell, 1999; Brausch et al., 2007; Bringolf et al., 2007; Folmar et al., 1979). However, Achiorno et al. (2008) found that technical grade glyphosate and the herbicide containing a surfactant both equally decreased survival in exposed Nematomorph worms. My goal was to explore the effects of stressors experienced by spiders in the field, so we did not try to uncouple the glyphosate from the surfactant. Nevertheless, in future studies it would be interesting to look at the impacts on survival separately for each of the components. Taken together, these results suggest that a commercially formulated glyphosate-based herbicide and predator cues collectively change the movement and survival of an important arthropod predator. Changes in P. milvina’s survival and behavior induced by these chemical stressors might have important impacts at both the population and community levels. For example, Pardosa milvina are a vagile species able to track habitat quality (e.g. prey abundance) and colonize areas accordingly (Walker, et al., 1999; Marshall et al., 2000). Therefore, any factors that lead to changes in movement in this species could impact its ability to forage on insect prey as well as colonize new habitat patches. Ultimately, these results contribute to growing knowledge of how anthropogenic and natural chemical cues might interact to mediate behavior in an understudied terrestrial system. Furthermore, the results of these experiments suggest directions for further research. To maximize the chance of producing an effect and to provide even coverage, I applied herbicide at levels significantly higher than those recommended for use on crops. These concentrations could be indicative of worst case scenarios where herbicide was improperly applied or pooled in lower areas. However, further tests should be conducted to verify these results at different herbicide concentrations. Furthermore, some interesting questions are raised for additional studies on animals dealing with chemical stressors in terrestrial systems: What mechanisms lead to different impacts of anthropogenic chemicals on an animal’s responses to different predators? And do these chemicals change an animal’s reactions to natural chemical cues other than those from predators (e.g. prey cues, or cues from conspecifics)?

Literature Cited

Achiorno, CL., Villalobos, CD., Ferrari, L. 2008. Toxicity of the herbicide glyphosate to Chordodes nobilii (Gordiida, Nematomorpha). Chemosphere. 71:1816-1822.

Anderson, J.F. 1974. Responses to starvation in the spiders Lycosa lenta: Hentz and Filistata hibernalis. Ecology. 55(3): 576-585

Bell, RD, Rypstra, AL, Persons, MH. 2006. The effect of predator hunger on chemically mediated antipredator responses and survival in the wolf spider Pardosa milvina (Araneae: Lycosidae). Ethology. 112:903-910.

Bishop, L, Riechert, SE, 1990. Spider colonization of agroecosystems: mode and source. Environmental Entomology. 19: 1738-1745.

Brausch, JM., Beall, B., Smith, PN. 2007. Acute and sub-lethal toxicity of three POEA surfactant formulations to Daphnia magna. Bulletin of Environmental Contamination and Toxicology. 78: 510-514.

Bringolf, RB, Cope, WG, Mosher, S., Barnhart MC, Shea, D. 2007. Acute and chronic toxicity of technical-grade pesticides to glochidia and juveniles of freshwater mussels (Unionidae).” Enviromental Toxicology and Chemistry. 26: 2086-2093.

Brust, GE. 1990. Direct and indirect effects of four herbicides on the activity of carabid beetles (coleopteran: Carabidae). Pesticide Science. 30: 309-320.

Buddle, CM., Rypstra, AL. 2003. Factors initiating emigration of two wolf spider species (Araneae: Lycosidae) in an agroecosystem. Environmental Entomology. 32: 88-95.

Byrne, PG, Keogh, JS. 2007. Terrestrial toadlets use chemosignals to recogonize conspecifics, locate mates and strategically adjust calling behaviour. Animal Behaviour. 74: 1155-1162.

Cox, DR. 1972. Regression models and life-tables. Journal Of The Royal Statistical Society Series B-statistical Methodology. 34: 187-220.

Dicke, M, Grostal, P. 2001. Chemical detection of natural enemies by arthropods: An ecological perspective. Annual Review of Ecology and Systematics. 32: 1-23.

Ferris, G., Rudolf, VHW. 2007. Responses of larval dragonflies to conspecific and heterospecific predator cues. Ecological Entomology. 32: 283-288.

Folmar, LC., Sanders, HO., Julin, AM. 1979. Toxicity of the herbicide glyphosate and several of its formulations to fish and aquatic invertebrates. Archives of Environmental Contamination and Toxicology. 8:269-278. 23

Freund, R; Olmstead, K. 2000. Role of vision and antennal olfaction in habitat and prey location by three predatory heteropterans. Enviromental Entomology. 29:721-732.

Giesy, JP., Dobson, S., Soloman, KR., 2000. Ecotoxicological risk assessment for Roundup® herbicide. Reviews in Environmental Contamination and Toxicology. 167: 35–120

Grostal, P; Dicke, M. 1999. Direct and indirect cues of predation risk influence behavior and reproduction of prey: a case for acarine interactions. Behavioral Ecology. 10: 422- 427.

Halaj, J., Cady, AB., Uetz, GW. 2000. Modular habitat refugia enhance generalist predators and lower plant damage in soybeans. Environmental Entomology. 29:383-393.

Hassan, SA., Bigler, F., Bogenschutz, H., Boller, E., et al. 1988. Results of the 4th joint pesticide testing program carried out by the IOBC/WPRS-working group pesticides and beneficial organisms. Journal of Applied Entomology- Zeitschrift Fur Angewandte Entomologie. 105: 321-329.

Haughton, AJ., Bell, JR., Boatman, ND., Wilcox, A.1999. The effects of different rates of the herbicide glyphosate on spiders in arable field margins. Journal of Arachnology. 27: 249-254

Haughton, AJ., Bell, JR., Wilcox, A., Boatman, ND. 2001. The effect of the herbicide glyphosate on non-target spiders: Part 1. Direct effects on Lepthyphantes tenuis under laboratory conditions. Pest Management Science. 57:1033-1036

Howard, R; Blomquist, G. 2005. Ecological, behavioral, and biochemical aspects of insect hydrocarbons. Annual Review of Entomology. 50:371-393.

Jakob, E.M., Marshall, S.D., and Uetz, G.W. 1996. Estimating fitness: a comparison of body condition indices. Oikos . 77(1): 61-67

Klaschka, U. 2008. The infochemical effect- a new chapter in ecotoxicology. Environmental Science and Pollution Research. 15: 452-462.

Koivula, M, Korpimaki, E. 2001. Do scent marks increase predation risk of microtine rodents? Oikos. 95:275-281. Komeza et al., 2001

Komeza, N. et al. 2001. Modification, by the insecticide chlopyrifos, of the behavioural response to kairomones of the parasitoid wasp, Leptopilina boulardi. Archives of Environmental Contamination and Toxicology. 41, 436–442

Lundgren, JG., Gassmann, AJ., Bernal, J., Duan, JJ., Ruberson, J. 2009. Ecological compatibility of GM crops and biological control. Crop Protection. 28: 1017-1030.

Lurling, M, Scheffer, M. 2007. Info-disruption: pollution and the transfer of chemical information between organisms. Trends in Ecology and Evolution. 22:374-379.

Mandrillon, A, Saglio, P. 2007. Herbicide exposure affects the chemical recognition of a non native predator in common toad tadpoles (Bufo bufo). Chemoecology. 17: 31-36.

Mann, RM., Bidwell, JR. 1999. The toxicity of glyphosate and several glyphosate formulations to four species of southwestern Australian frogs. Archives of Environmental Contamination and Toxicology. 36: 193-199.

Marshall, SD, Rypstra, AL. 1999. Spider competition in structurally simple ecosystems. Journal of Arachnology. 27:343-350.

Marshall, S.,Walker, S., Rypstra A. 2000. A test for differential colonization and competitive ability in two generalist predators. Ecology. 81:3341–3349.

Marshall, SD., Pavuk, DM., Rypstra, AL. 2002. A comparative study of phenology and daily activity patterns in the wolf spiders Pardosa milvina and Hogna helluo in soybean agroecosystems in southwestern Ohio (Araneae, Lycosidae). Journal of Arachnology. 30:503-510.

Menard, SW. 2002. Applied logistic regression analysis. Sage University Papers Series on Quantitative Applications in the Social Sciences, 07-106. Thousand Oaks, CA: Sage.

Michalkova, V., Pekar, S. 2009. How glyphosate altered the behaviour of agrobiont spiders (Araneae: Lycosidae) and beetles (Coleoptera: Carabidae). Biological Control. 51: 444-449.

Persons MH., Rypstra, A.L. 2001. Wolf spiders show graded antipredator behavior in the presence of chemical cues from different sized predators. Journal of Chemical Ecology. 27: 2493-2504.

Pimentel D., Acquay, H., Biltonen, M., Rice, P., et al.,1991. Environmental and economic costs of pesticide use. Bioscience. 42: 750-760.

Relyea, R; Mills, N. 2001. Predator-induced stress makes the pesticide carbaryl more deadly to gray treefrog tadpoles (Hyla versicolor). Proceedings of the National Academy of Sciences United States of America. 98: 2491-2496.

Relyea, RA. 2005a. The lethal impact of roundup on aquatic and terrestrial amphibians. Ecological Applciations. 15:1118-1124.

Relyea, RA. 2005b. The lethal impacts of roundup and predatory stress on six species of North American tadpoles. Archives of Environmental Contamination and Toxicology. 48: 351-357.

Rovner, J. 1996. Conspecific interactions in the lycosid spider Rabidosa rabida: The roles of different senses. Journal of Arachnology. 24:16-23.

Rypstra, AL., Wieg , C., Walker, S E., Persons , MH. 2003. Mutual mate assessment in wolf spiders: Differences in the cues used by males and females. Ethology 109:297–306.

Rypstra, AL., Schmidt, JM., Reif, BD., DeVito, J., Persons, MH. 2007. Tradeoffs involved in site selection and foraging in a wolf spider: effects of substrate structure and predation risk. Oikos. 116: 853-863

Saglio, P; Trijasse, S; Azam, D. 1996. Behavioral effects of waterborne carbofuran in goldfish. Archives of Environmental Contamination and Toxicology. 31: 232-238.

Schonewolf, KW.,Bell, R., Rypstra, AL., Persons, MH. 2006. Field evidence of an airborne enemy-avoidance kairomone in wolf spiders. Journal of Chemical Ecology 32:1565- 1576

Scott, G; Sloman, K; Rouleau, C; Wood, C. 2003. Cadmium disrupts behavioural and physiological responses to alarm substance in juvenile rainbow trout (Oncorhynchus mykiss). Journal of Experimental Biology. 206:1779-1790.

Searcy, L.E.; Rypstra, A.L., Persons, M.H. 1999. Airborne chemical communication in the wolf spider Pardosa milvina. Journal of Chemical Ecology. 25:2527-2533.

Seyfarth, RM., Cheney, DL., Marler, P.1980. Monkey responses to three different alarm calls: evidence of predator classification and semantic communication. Science. 210: 801-803.

Sillero-Zubiri, C; Macdonald, D. 1998. Scent-marking and territorial behaviour of Ethiopian wolves Canis simensis.” Journal of Zoology. 245: 351-361.

Snyder, WE., Wise, DH. 1999. Predator interference and the establishment of generalist predator populations for biocontrol. Biological Control 15:283–292.

Thorbek, P; Bilde, T. 2004. Reduced numbers of generalist arthropod predators after crop management. Journal of Applied Ecology. 41:526-538.

Tierney, KB., Ross, PS., Jarrard, HE., Delaney, KR., Kennedy, CJ. 2006. Changes in juvenile coho salmon electro-olfactogram during and after exposure to some current-use pesticides. Environmental Toxicology and Chemistry. 25: 2809– 2817.

Tierney, KB., Singh, CR., Ross, PS., Kennedy, CJ. 2007. Relating olfactory neurotoxicity to altered olfactory-mediated behaviours in rainbow trout exposed to three currently-used pesticides. Aquatic toxicology. 81: 55-64.

Walker, SE., Marshall, SD., Rypstra, AL., Taylor, DH. 1999. The effects of hunger on locomotory behaviour in two species of wolf spider (Araneae, Lycosidae). Animal Behaviour. 58: 515-520.

Walker, SE., Rypstra, AL. 2002. Sexual dimorphism in trophic morphology and feeding behavior of wolf spiders (Araneae: Lycosidae) as a result of differences in reproductive roles. Canadian Journal of Zoology. 80:679-688.

Wilder, SM., Rypstra, AL. 2004. Chemical cues from an introduced predator (Mantoedea, Mantidae) reduce the movement and foraging of a native wolf spider (Araneae, Lycosidae) in the laboratory. Environmental Entomology. 33: 1032-1036.

Woodburn, A. 2000. Glyphosate: production, pricing and use worldwide. Pest Management Science. 56:309-312.

Table 2.1: Results of ANOVAs measuring the impacts of predator cues and/or herbicide on laboratory tests of transformed activity variables (time spent in ambulatory and stereotypical movement, distance traveled and average speed) for P. milvina.

Ambulatory (s)* Movement in place (s)* Distance (cm)** Avg. Speed- (cm/s** (p-value) (p-value) (p-value) (p-value)

Model (F, p-value) F5,119=7.83; p<0.0001 F5, 119=2.54; p=0.0321 F=7.345,114, p< 0.0001 F5, 119; p<0.0001 Predator <0.0001 0.0476 <0.0001 <0.0001 Herbicide 0.1052 0.0144 0.5121 0.7724 Predator x Herbicide 0.0254 0.8710 0.0559 0.0961 significant p values <0.05 are in bold *square root transformation of data ** natural log transformation of data

Table 2.2: Mean ± standard error of activity measures for P. milvina in response to predator cues (H. helluo, S. quadriceps or no cues) and/or herbicide or water. Ambulatory (s) Movement in place (s) Distance (cm) Speed (cm/s)

Treatment Wet Dry Wet Dry Wet Dry Wet Dry

No cues/water 53.3±10.5 69.2±16.6 72.5±35.4 339.7±44.6 152.9±40.7 157.4±45.4 0.61±0.13 1.17±0.39 No cues/herbicide 57.8±10.2 81.1±14.3 40.7±24.2 252.8±40.5 169.9±31.5 176.6±34.0 0.48±0.07 0.63±0.07 H. helluo/water 13.6±4.2 26.2±7.3 45.4±41.2 268.6±56.9 19.9±6.4 24.4±7.5 0.76±0.11 1.32±0.10 H. helluo/herbicide 18.7±7.5 21.5±6.8 31.9±20.8 220.2±52.4 25.1±11.5 21.6±9.5 0.76±0.11 1.38±0.13 S. quadriceps/water 69.5±11.5 96.5±13.5 32.8±16.4 239.8±24.4 206.9±44.7 215.0±47.5 0.46±0.07 0.77±0.10 S. quadriceps/herbicide 27.9±7.6 52.6±12.7 12.8±7.4 197.4±44.1 92.9±34.2 106.8±36.4 0.39±0.04 0.89±0.15

Fig. 2.1: Laboratory arena for exposing Pardosa milvina to herbicide and/or predator cues. I alternated filter paper pieces with herbicide or water by those with predator cues (Hogna helluo or Scarites quadriceps) or blank paper. For example, in the H. helluo/herbicide treatments, there were two pieces of filter paper with H. helluo cues alternating with two that were covered with herbicide. Four exit holes located around the arena were blocked during the activity part of the trials but were opened afterwards to allow emigration to be measured.

Fig. 2.2: Field mesocosm for testing emigration of Pardosa milvina in response to herbicide and/or S. quadriceps cues. Dirt with a thin covering of straw was placed on the bottom of each container to simulate early season field conditions. Four exit holes leading into plastic solo cups were located around the mesocosm. Mesocosms were covered with mesh cloth (not shown here) to prevent spider escape or external predation.

Fig. 2.3: Activity of Pardosa milvina when exposed to herbicide or water and/or predator cues (Hogna helluo, Scarites quadriceps or no cues) in 15 minute laboratory trials. Bars indicate ±SE. A. Time spiders spent ambulatory in seconds. B. Average speed across the trial. C. Distance spiders traveled in centimeters. Letters on the figures show comparisons between treatments according to Tukey post hoc tests, where treatments with the same letter are not significantly different.

Fig. 2.4: Proportion of spiders remaining in the container over time for P. milvina when exposed to predator cues (H. helluo, S. quadriceps, or none) and/or herbicide or water in 15 minute laboratory trials.

Fig. 2.5: Survival over a 60 day period of P. milvina after being subjected to S. quadriceps cues, herbicide, neither, or both for 24 hours. Spiders subjected to predator cues only survived the longest. Those subjected to herbicide only survived the shortest, and the other two groups were intermediate.

Chapter 3: Effects of predator cues, prey level and an herbicide on reproduction in Pardosa milvina (Araneae: Lycosidae). Introduction: Generalist predators play an important role in the food web both directly and indirectly in many terrestrial systems (Hairston et al., 1960; Symondson, et al., 2002; Ripple and Beschta, 2004; Schmitz, 2006). The fitness of these predators is dependent on their ability to acquire enough resources to survive and reproduce, but this is often influenced by human activities, especially in managed environments (Stinner and House, 1990; Chiverton and Sotherton, 1991; Thorbeck and Bilde, 2004). Unfortunately, it is often hard to predict how anthropogenic activities will affect different species, as activities that positively affect one species might have negative effects on another species. For example, Haughton et al., (2003) found that the management practices of genetically modified herbicide resistant crops reduced numbers of bees and butterflies, but increased numbers of collembolans in agricultural fields, when compared to more traditional crop management. Predatory arthropods are an important part of the food web in highly managed agricultural systems, often exerting top down effects on insect pests, reducing herbivory and frequently increasing crop yield (Carter and Rypstra, 1995; Sunderland, 1999; Snyder and Wise, 2001; Nyffeler and Sunderland, 2003). A large part of the management in these systems is through chemical pest control (Pimentel, 1991), but it is often difficult to predict how chemical pest control influences the behavior and population dynamics of individual predator species. For instance, predators respond to pesticide application with either increased (Chiverton, 1984; Prasifka, et al., 2008), or decreased activity or density (Chiverton and Sothern, 1991; Baatrup and Bayley, 1993; Haughton et al., 1999). Even in the absence of direct effects, anthropogenic chemicals can indirectly affect animals by disrupting the natural flow of chemical information that these animals use to detect the world around them (reviewed in Lurling and Scheffer, 2007; Klaschka, 2008). Many arthropod predators and prey detect one another using chemical cues (see Kats and Dill, 1998; Dicke and Grostal, 2001 for reviews). If exposure to anthropogenic chemicals interferes with an animal’s ability to detect these signals, or elicits an inappropriate response, the effects could be important (for a review see Lurling and Scheffer, 2007). For example Saglio et al., (1998) discovered that when goldfish were exposed to the herbicides diuron or atrazine along with

35 conspecific chemical alarm cues, their normal predator avoidance behaviors were reduced. Moreover, anthropogenic chemicals might have a synergistic effect, decreasing survival in animals also exposed to predator cues, as has been demonstrated for wood frogs (Relyea, 2005a). Interactions between anthropogenic chemicals and predator chemical cues that influence key life history components in exposed individuals could in turn have effects on the population or community (Lurling and Scheffer, 2007; Klaschka, 2008). For instance, arthropods often change their foraging behavior when exposed to predator chemical cues (see Dicke and Grostal, 2001 for a review), so if exposure to agricultural chemicals alters the detection of or response of animals to these predator cues, impacts on the food web could be amplified. Furthermore, if a chemical changes reproduction there could be long term consequences for the persistence of the population (Chen et al., 2004; Tietjen, 2006; Soso et al., 2007). Here I explore how an anthropogenic chemical (herbicide) interacts with predator chemical cues to affect prey capture and consumption and ultimately reproduction in a generalist arthropod predator common in agricultural systems. Large differences in prey density across the season are common in the field for predatory arthropods (Wise, 2006). This variation in prey abundance might be particularly true in agricultural systems where disturbance events (planting, spraying, etc.) are common and vegetation changes rapidly across the season (see chapter 1). Furthermore, foraging history can greatly influence reproduction in predatory arthropods in any system (Morse & Stephens, 1996; Wilder and Rypstra, 2008). Therefore, I examined the single and interactive effects of herbicide and predation risk on prey capture and consumption and on fitness at two different prey levels (low or high). Study System: The small wolf spider Pardosa milvina (Araneae, Lycosidae) is an ideal species to test for interactions among anthropogenic and natural infochemicals. These spiders are the numerically dominant epigeic arthropod predator in agricultural fields throughout the midwestern United States, often reaching densities of more than 5 per m2 (Marshall and Rypstra, 1999; Marshall et al., 2002). As such, they likely have important impacts on the arthropod food webs in these systems. Two other arthropod predators are common inhabitants of the soil surface of agricultural fields, the wolf spider Hogna helluo and the carabid beetle Scarites quadriceps, each found in densities of about 1/m2 (Marshall et al., 2002; Snyder and Wise,

1999). Both H. helluo and S. quadriceps potentially prey upon P. milvina (Persons et al, 2001; Snyder and Wise, 1999). Previous work demonstrated that Pardosa milvina is very sensitive to its chemical environment (Persons et al., 2001; Wilder and Rypstra, 2004; Rypstra et al., 2003). Specifically, it can detect the size, hunger level and sex of Hogna helluo from the silk, feces and other excreta that this predator leaves behind as it occupies an area (Persons and Rypstra, 2001; Bell et al., 2006). These chemical cues left by H. helluo can negatively impact reproduction and foraging in P. milvina (Persons et al., 2002; Rypstra et al., 2007). The chemical cues of Scarites quadriceps in the form of feces and other excreta can also affect movement in P. milvina (Chapter 2), but it is unknown how these cues might impact other factors such as foraging and reproduction. Finally, P. milvina detects and responds to herbicide with changes in activity as well, but these responses can also depend on the presence of predator chemical cues (Evans et al., in review; Chapter 2). I used a glyphosate- based herbicide as this is one of the most common type of herbicide sprayed in agricultural systems (Woodburn, 2000; Lundgren et al., 2009). This herbicide, is used on crops genetically engineered to be resistant to glyphosate and can be sprayed at any time during the growing season (Woodburn, 2000). As P. milvina co-occur with their predators across the growing season (Marshall et al., 2002; Snyder and Wise, 1999, Chapter 1), simultaneous exposure to predator cues and herbicide likely occurs in the field. These herbicides are supposed to be harmless to animals, as glyphosate targets an enzyme pathway found specifically in plants (Williams et al., 2000). However, studies show that some of the commercial formulations of this herbicide, specifically those that include the surfactant Polyethoxylated tallowamine (POEA), can influence behavior and survival in some animals (Abdelghani, et al., 1997; Mann and Bidwell, 1999; Relyea, 2005 a,b; Achiorno et al., 2008). Furthermore, glyphosate-based herbicides can decrease reproduction in arthropods, fish, and amphibians (Chen et al. 2004; Soso et al., 2007; Schneider, 2009). My first goal was to determine the separate and combined effects of predator cues (H. helluo and S. quadriceps) and herbicide on prey capture and consumption by female Pardosa milvina foraging at different prey densities. To accomplish this, I exposed female P. milvina to predator cues (H. helluo or S. quadriceps) in the presence or absence of herbicide and allowed them to forage on prey at either high or low availability. My second goal was to determine

37 whether exposure to these stressors during the foraging trials ultimately had effects on the reproduction of these spiders. To achieve this goal, after their foraging trials, I maintained the spiders in the laboratory on the same feeding regime and quantified treatment effects on egg sac production and spiderling emergence. Methods: Animal collection and maintenance During June and July 2008 experimental animals were collected from agricultural fields at the Miami University Ecology Research Center, Oxford, OH. For the focal species, P. milvina, only females carrying egg sacs were collected to ensure they were reproductive (Persons et al., 2002). Adults of the beetle predator, S. quadriceps and large females of the wolf spider predator, H. helluo that were adult or at their penultimate molt were also collected and all animals were placed in separate plastic containers (8 cm diameter x 5 cm high for P. milvina and 12 cm diameter x 10 cm high for H. helluo and S. quadriceps) with a 50:50 peat moss: potting soil mix on the bottom. All animals were maintained in the laboratory under a temperature of 25C, a humidity level of 55-60%, and an 13:11 light:dark cycle and provided them with water ad libitum. Animals were fed weekly with appropriately sized domestic crickets, Acheta domesticus. I assigned P. milvina to one of twelve treatments based on presence of predator cues, herbicide and prey availability using a factorial design (Fig. 3.1). Initial sample size was 22 spiders per treatment. However, a few spiders produced an egg sac or died during the initial foraging trial and were subsequently removed from further analysis, making the final sample sizes n=17-22 per treatment. Herbicide preparation I used the commercially formulated herbicide Buccaneer Plus®, which is made of 41% active ingredient (glyphosate in the form of isopropylamine salt, 480g/L) and 59 % inactive ingredients including the surfactant polyethoxylated tallowamine (POEA) (Monsanto Company, St. Louis, Missouri, USA). I diluted the commercially formulated herbicide with double distilled water to 2.5%, which is within the manufacturer’s recommended range of (0.625%-5%). I then applied this dilution to the substrate using a spray rate of 0.013 mL/cm2 (or 15.3 kg a.i. ha-1 of glyphosate), which was the minimum needed to obtain complete coverage of the substrate. Furthermore, previous experiments showed that the herbicide at this concentration and spray rate can change activity and emigration in P. milvina (Evans et al. in review; Chapter 2).

Collection of predator cues and application of herbicide Prior to trials, I introduced predator cues to the appropriate treatments by placing an individual S. quadriceps or H. helluo for 24 hours in a circular plastic container (15 cm w x 8 cm h) containing a 50:50 potting soil: peat moss mix that was 1 cm deep. After that time, I removed the predator and evenly applied herbicide or an equal amount of double distilled water to the soil surface using an airbrush applicator. Exposure to chemical cues and foraging trials To ensure that hunger levels in the foraging trials were constant among groups and across the experiment, I fed each female P. milvina two 10 day old crickets (0.3 cm in length) exactly one week before it was scheduled to be used in foraging trials. I also gently removed each spider’s egg sac 48 hours prior to experimentation by pinching the small silk connection between the spider’s abdomen and the egg sac with a pair of forceps. Finally, just prior to experimentation, I measured each spider’s cephalothorax width and abdomen width to the nearest 0.01 mm using a digital micrometer attached to a Wild 5 microscope. Cephalothorax width is a measure of fixed size as it does not change with feeding, but abdomen width increases with food intake and can therefore any change in its dimensions provide evidence of recent prey consumption (Anderson, 1974; Jakob et al., 1996; Persons et al., 2002; Rypstra et al., 2007). After measuring each spider, I placed it in a container for a total of 24 hours of exposure to predator cues and/or herbicide. After 22 hours of exposure, I added either 5 (simulating low prey availability) or 20 (high prey availability) domestic crickets (<0.2 cm) to each arena. After 2 hours, I counted the crickets remaining and presumed that any missing were killed and eaten by the spider. I also re-measured the abdomen width of each spider directly following a trial and calculated the change (abdomen width after the trial minus abdomen width before) as a measure of prey consumption (Rypstra et al., 2007). Reproduction After the cue exposure/foraging trials I returned spiders to their home containers and maintained them in the laboratory over a 60 day period. Spiders from all treatments were fed weekly with two or three 0.6 cm long domestic crickets, Acheta domesticus. During that period, I checked the spiders weekly and recorded if they were alive, if and when an eggsac was produced, and if and when spiderlings emerged from the eggsac.

Data Analysis I compared crickets killed/missing among treatments using an ordinal logistic regression (Menard, 2002). I further compared abdomen width between treatments used a three-way ANOVA with prey level, herbicide and predator cues as factors, and including all interaction terms. If the model was significant, I further conducted Tukey post hoc tests to determine which treatments were significantly different. I compared reproductive success among groups (e.g. whether or not an eggsac was produced) via a logit model (Menard, 2002) with prey level, herbicide and predator cues as factors and included all interactions in the initial model. However, I subsequently removed non-significant three-way interactions and the most non- significant two-way interactions. Finally, I compared survival among groups using a Proportional Hazards survival analysis (Cox, 1972). Prey (high or low) had a significant effect on all measures either by itself or through interactions with other factors. Therefore, I divided the groups by prey treatment (high or low) on all figures and often discuss the prey treatments separately in the results and discussion. Results: Foraging trials: Prey capture and consumption As expected, P. milvina provided with 20 crickets killed significantly more during the two hour foraging trials than those provided with only 5 crickets (9.4 ± 3 vs. 3.9 ± 1; Table 3.1, 3.2). Herbicide had no effect on the number of crickets killed, but predator cues had a slight but non-significant effect (p=0.067), with P. milvina killing fewer crickets in the H. helluo cues treatments (Table 3.1, 3.2). Prior to foraging trials, spiders did not differ across treatments in cephalothorax width (ANOVA: F11,238 = 0.9533, p = 0.4901) or abdomen width (ANOVA:

F11,238 = 0.9717, p = 0.4730), indicating size of the spiders and initial body condition was similar across treatments. All spiders consumed prey and subsequently mean abdomen width increased in all treatments during the trial (Table 3.2; Fig 3.2). According to abdomen width change, spiders provided with 20 crickets consumed much more than those provided with only five crickets (Table 3.2, Fig 3.2). Interestingly, however, pairwise comparisons suggest that the differences between consumption in high and low prey availability treatments were caused by the water only and the S. quadriceps only treatments. When herbicide was added, there was no longer a difference in consumption between spiders foraging on high vs. low levels of available prey (Tukey post hoc tests; Fig. 3.2). Furthermore, spiders exposed to Hogna helluo cues did

40 not consume as much as spiders in some of the other treatments (had a smaller abdomen width after the trial). However, the addition of herbicide seemed to ameliorate this effect (Table 3.2, Fig. 3.2). Reproduction Overall, 74% of the spiders produced an egg sac after treatment; however, spiderlings emerged from only 54% of those egg sacs. Both reproductive variables (egg sac production and success of spiderling emergence) were affected by treatment to some degree (Table 3.3; Fig 3.3). 2 The initial overall model for egg sac production was not significant (χ 11 =16.06, p=0.1388), but this was influenced by the non-significant three-way interaction (predator cues x herbicide x prey availability; p=0.9216) and the non-significant two-way interaction between predator cues x herbicide (p=0.5846). After these two interaction terms were removed, the model was then 2 significant (χ 7 = 14.75, p=0.0393). Although predator cues and herbicide alone did not impact egg sac production, there was a trend for prey availability to have an effect (p=0.0666; table 3.3). Surprisingly this trend appeared to be driven by a higher proportion of spiders producing egg sacs in the low prey, no predator cues treatments compared to those in the high prey, no predator cues treatments (90-95% vs. 65-72% respectively; Fig. 3.3 A,C). Furthermore, data suggest that exposure to predator cues in the low prey treatments reduced egg sac production by about 15- 30%, but no effect of predator cues was evident in high prey treatments (Fig. 3.3 A, C). This explains the significant prey x predator cues interaction in the model for egg sac production (Table 3.3). In contrast to predator cues, herbicide had no impact on egg sac production in low prey treatments. However, exposure to herbicide in high prey treatments lead to a 10-25% decrease in egg sac production, with the exception of the group also exposed to H. helluo cues, for which there was no effect (Fig. 3.3 A, C). This is the result that likely produced the slight, but non-significant interaction between herbicide and prey (p=0.0812; Table 3.3). 2 The overall model for spiderling emergence was initially not significant (χ 11 =20.76, p=0.0525), but after the removal of the non-significant three-way interaction (predator cues x 2 herbicide x prey availability; p=0.4273), the model became significant (χ 9= 17.81; p=0.0374). Predator cues, herbicide and prey availability alone had no effects on spiderling emergence, but there were some significant or moderate but non-significant interactive effects between these (Table 3.3). In the low prey treatments there was a strong effect of herbicide on spiderling emergence. Specifically, 15-40% fewer spiders exposed to herbicide in these treatments

41 produced egg sacs from which spiderlings emerged (Fig 3.3 B). Herbicide also modestly decreased the probability of spiderling emergence (by about 10%) in the high prey treatments where spiders were exposed to no cues or S. quadriceps cues (Fig 3.3 D). Surprisingly however, spiders in the high prey treatment exposed to both herbicide and H. helluo cues produced egg sacs that were 40% more likely to have successful spiderling emergence (Fig 3.3 D). These results are likely driving the significant or slight, but non-significant prey availability x herbicide (p= 0.0192) and cues x herbicide (p= 0.0857) interactions in the spiderling emergence model. Survival Treatment played a role in the length of time that the spiders survived over a 60 day period after exposure. Although the overall model was for survival was not significant (Proportional Hazards test: χ2=18.53, p=0.07, Table 3.2), some notable trends did occur in the data. For example, exposure to predator cues impacted survival of P. milvina in the low prey availability treatments such that on average those spiders exposed to S. quadriceps cues alone had the longest survival time and those exposed to H. helluo cues alone had the shortest survival time (Fig 3.4 A). The difference in the two groups was considerable, as after 60 days over 60% of the spiders in the S. quadriceps treatment were still alive whereas only about 5% of spiders in the H. helluo group were alive (Fig. 3.4 A). This is likely driving the significant cues x prey level effect in the model (p= 0.0284), because these differences in survival were not observed in the high prey treatments (fig 3.4 B). The slight but non-significant predator cues x herbicide interaction (p=0.0619) is probably due to differences in survival between spiders exposed to predator cues alone vs. those exposed to predator cues combined with herbicide, especially in the low prey treatment groups (Table 3.2). To elaborate, spiders exposed to herbicide and S. quadriceps cues combined exhibited slightly lower survival than those exposed to S. quadriceps cues alone, whereas those exposed to herbicide and H. helluo cues combined exhibited slightly higher survival than those exposed to H. helluo cues alone (Fig. 3.4 A) Discussion Overall, exposure to herbicide and perceived risk of predation along with prey availability can combine to stress Pardosa milvina, a chemically-aware, generalist predator, in ways that might affect its fitness. Not surprisingly, the number of prey spiders were provided influenced the number they killed and their consumption of that prey. Interestingly, these differences in a single feeding event also affected the spider’s subsequent survival and

42 reproductive success. Likewise, presence of predation risk and/or herbicide influenced the same parameters in complex ways, especially for those spiders provided with only five prey items during exposure. Additionally, spiders from both feeding levels responded differentially to the cues from the two predator species in ways that were enhanced or mitigated by herbicide. For example, even when provided with plentiful prey, spiders exposed to S. quadriceps cues exhibited reduced reproductive success whether or not herbicide was present. However, although exposure to H. helluo cues alone had negative effects on reproduction, these effects seemed to be negated by simultaneous exposure to herbicide in the high prey treatment. Taken together, these results suggest that exposure to herbicide and predator cues affects the reproduction and survival of P. milvina, but that these effects can be mediated by variation in prey capture by the spider during exposure. Foraging trials: prey capture and consumption Pardosa milvina are active foragers, so differences in overall movement could translate to changes in prey capture or consumption (Walker et al., 1999; Persons, et al., 2002; Wilder and Rypstra, 2004). In fact, it is likely that the effects of predator cues on prey consumption, or increase in the size of the abdomen observed here (Fig 3.2), are due to changes in P. milvina’s overall activity level. For example, P. milvina react to H. helluo cues with strong reductions in movement, which can increase their chance of survival (Persons et al., 2001). Reductions in activity in the presence of H. helluo cues, might have caused P. milvina to subsequently encounter fewer prey, explaining the lower prey consumption in those treatments (Fig. 3.2). This idea is supported by previous studies that found reduced foraging success for P. milvina exposed to cues from either H. helluo or another predator, the praying mantis Tenodera aridifolia sinensis (Persons, et al., 2002; Wilder and Rypstra, 2004). Pardosa milvina appear to have a different strategy for dealing with S. quadriceps, this means increasing activity, presumably to get away (Chapter 2). Pardosa milvina running away from S. quadriceps might have increased encounters with prey. However, without additional time to consume this prey, it is not surprising that consumption was similar in the S. quadriceps and no cues treatments (Fig 3.2). These changes in activity and the related changes in prey consumption have implications for P. milvina in the field where they are likely to be food stressed (Wise 2006). The negative effects of an encounter with H. helluo cues could be higher during times of food stress, if the individual’s chance of successful foraging was further decreased (Wise, 2006; Chapter 1).

There were no significant effects of herbicide on prey consumption, but there was an overall tendency in most treatments for the spiders to increase in abdomen width with herbicide exposure (Fig. 3.2). Preliminary results from another experiment suggest that cricket prey are easier to detect and capture when exposed to herbicide on the substrate (unpublished data). If this is the case, it would explain the slight increase in prey consumption for P. milvina in most of the herbicide treatments. In contrast, Benamú, et al. (2010) found that after 24 hours, the web- spinning spider Alpaida veniliae consumed a lower percentage of prey that had been dipped in a glyphsate-based herbicide compared to that consumed by spiders foraging on untreated prey. The fact that in my study the prey was only exposed to herbicide indirectly (via contact with the exposed substrate) could explain the difference between these results and those of Benamú , et al. (2010). Likewise, other studies have also failed to find negative effects of glyphosate-based herbicide on prey capture and consumption for rainbow trout, wolf spiders and carabid beetles (Morgan and Kiceniuk, 1992; Michalkova and Pekar; 2009). According to these results, this herbicide is unlikely to have direct negative effects on foraging for P. milvina in the field. However, prey in the field could be exposed to either direct or residual spray in the field depending on whether they were in a protected area (e.g. under a plant) when spraying occurred. Therefore, in light of the results of Benamú , et al., (2010), it would be constructive to conduct foraging trials where prey were exposed to direct spray, rather than just residual. Reproduction It is well documented that access to food influences reproductive success in spiders (Wise, 2006), but I was intrigued to observe that differences in access to food in just one feeding bout had such strong effects on the reproductive parameters I measured, particularly when herbicide and predator cues were present. For example, in low prey treatments exposure to cues from either predator significantly decreased the likelihood of spiders producing an egg sac; an effect eliminated when more prey were available (Fig 3.3 A,C). Thus, for P. milvina in the field where access to food is chronically low (Wise, 2006), encountering a predator could have damaging effects on reproduction. Not surprisingly, presence of predator cues can also negatively affect egg production in other animals (Grostal and Dicke, 1999; Montserrat et al., 2007). However, these studies have not considered food availability as a factor. Exposure to a glyphosate-based herbicide can have negative, neutral or positive effects on reproduction in other animals (Folmar et al, 1979; Yokoyama and Pritchard; 1984; Soso et al.,

2007; Schneider, et al., 2009; Benamú, et al., 2010). This makes it hard to generalize the influence of glyphosate-based herbicides on reproduction, which can be further complicated by various factors (herbicide formulation, pH, feeding levels, etc.). In this study, exposure to exposure to herbicide lead to a trend for decreased egg sac production in the high prey treatments only, but decreased hatching success in both prey treatments (Fig 3 A-D). These results indicate that the amount of food consumed during exposure could play a role in the impacts of this herbicide later on. Pardosa milvina likely ingested herbicide through grooming (running their legs through their mouth parts), consuming exposed crickets and drinking fluid from the soil during their 24 hour exposure. If P. milvina in the high prey treatments ingested more herbicide from exposed prey than spiders in low prey treatments, this might explain the decreased egg sac production in high prey treatments only. Ingestion of a glyphosate-based herbicide is also known to decrease reproduction in a species of neuropteran (Schneider et al., 2009) and a species of web-spinning spider (Benamú, et al., 2010). For example, Benamú et al., (2010) found that females of the web-spinning spider Alpaida veniliae fed herbicide-dipped prey produced fewer eggs and out of those produced, fewer hatched successfully. These negative effects of herbicide on reproduction, could have implications for animals in systems planted with genetically engineered, herbicide-resistant crops where spraying could occur at any time. Specifically, for P. milvina, a mid-season spray during June-July could reduce reproduction of this species at their highest population peak (Marshall et al., 2002; Chapter 1- Fig. 1.1). Food availability during exposure was not the only factor that interacted with herbicide to affect hatching success. In this study, herbicide also had a slight but non-significant influence on hatching success which differed depending on the exposure to predator cues (Fig 3.3 B, D). These differences might be a result of the way P. milvina changes movement in response to each predator (Chapter 2). Increased movement by P. milvina in response to S. quadriceps cues (Chapter 2) could also have increased their exposure to herbicide, maximizing its negative effects on reproduction. In contrast, decreased activity by P. milvina in response to H. helluo cues (Chapter 2; Persons et al., 2001) could have reduced herbicide exposure, minimizing its negative effects on reproduction. However, this doesn’t explain why individuals exposed to H. helluo cues and herbicide simultaneously in the high prey treatments were more likely to produce hatching egg sacs than control spiders not exposed to either stressor. One possible mechanism for this is that of hormesis/hormoligosis, where exposure to low levels of a toxic substance

45 actually has stimulatory affects on an organism with resulting positive effects on growth, reproduction or survival (Luckey, 1968; Croft and Brown, 1975; Calabrese and Baldwin, 2003). In fact, Yokoyama and Pritchard, (1984) found a trend for increased egg viability of the big- eyed bug, Geocoris pallens after exposure to a glyphosate-based herbicide. Pardosa milvina were exposed to levels of herbicide higher than those in field applications. Nevertheless, reduced activity in response to H. helluo cues (Chapter 2; Persons et al., 2001) could have made P. milvina’s actual exposure to herbicide considerably lower. It is unknown, however, whether this exposure would be low enough to induce hormesis/hormoligosis or if such effects even occur in spiders, but further study of this would be interesting. For this experiment I collected females with egg sacs from the field, removed them and examined the impact of predator cues and herbicide on the production of a new one. Since a female would have already invested considerable energy and resources in the first eggsac, this protocol increased the likelihood that I would see an impact of prey consumption and eggsac production. Having established that there are strong influences, it would be interesting to mate females in the laboratory and test the influence of a subset of these factors (predator cues, herbicide and food availability) on initial egg sac production and hatching. Failure to produce the initial egg sac (or loss of its viability) in response to predator cues or herbicide would have even more drastic fitness consequences for the spider, as the initial egg sac tends to produce the most spiderlings (personal observation, KW). Survival: There was a trend for a short (24 hour) exposure to predator cues to ultimately affect the survival of P. milvina, but only in the low prey availability treatments. Spiders exposed to S. quadriceps cues alone survived longer and those exposed to H. helluo cues alone died more quickly than those in the other treatments (Fig 3.4 A). Previously, I found that exposure to S. quadriceps (the beetles themselves, not just cues) over a 24 hour period in field mesocosms also had a positive effect on survival in P. milvina (Chapter 2). Although exposure to S. quadriceps cues can increase activity in P. milvina (Chapter 2), this did not seem to increase prey capture or consumption in the current experiment (Fig. 3.2 A). Therefore, it is still unclear what the mechanism is behind the increased survival in response to S. quadriceps cues, so this would require further testing. Unlike their response to S. quadriceps cues, the response of P. milvina to H. helluo cues was not unexpected. Exposure to predator cues can also reduce survival in other

46 animals, both with or without the presence of additional stressors (Stoks, 2001; Relyea, 2003). The lower survival of P. milvina in the presence of H. helluo cues could be linked to their decreased prey consumption during exposure. If this is the case, exposure to H. helluo cues could be particularly detrimental to P. milvina in the field during times of food limitation. There was a trend for exposure to herbicide alone to have a neutral to slightly positive influence on survival in P. milvina (Fig 3.4). This is in contrast to previous studies that found exposure to herbicide to be detrimental to the survival of P. milvina (Chapter 2; Evans, Shaw and Rypstra, in review). This discrepancy could be due to differences in experimental design as environmental variables are important to the toxicity of glyphosate (Tsui and Chu, 2003). However, because the study here differed substantially in several variables from that in Chapter 2 and that of Evans et al. (in review), it is difficult to pin down what might be responsible. For example, length of exposure and substrate differed between the three studies. Exposure was for 30 minutes in the laboratory on filter paper for Evans et al., in review, but for 24 hours on soil in the laboratory or mesocosms for the other two studies. Furthermore, Evans et al. (in review) did not involve foraging, whereas the other two studies did. Finally, the application rate differed among the three studies, being lowest in the mesocosm study and highest in the current laboratory study. Therefore, further studies looking at application rate and length of exposure, substrate and foraging would be necessary to determine what led to the differences between these three studies. Survival has been studied fairly extensively with respect to glyphosate-based herbicides, with mixed results (see Giesy, et al. 2000 for a review; Relyea, 2005a,b; Achiorno et al., 2008). However, one thing common to many of these studies is they have included the surfactant polyethoxylated tallowamine (POEA), which is more toxic than the glyphosate itself (Abdelghani, et al., 1997; Mann and Bidwell, 1999; Relyea, 2005 a,b; but see Achiorno et al., 2008). My goal was to address how the stressor experienced by the spiders in the field might influence their response and so I chose to use the commercial formulation used at our field site which includes the surfactant. Additional study would be required to identify the agent or agents more responsible for the responses I observed. The 2.5% herbicide concentration I used was well within the manufacturer’s recommended range, but to get full coverage of the substrate, I used an application rate higher than what would normally be sprayed on a field. The higher experimental application could

47 simulate a worst case scenario (e.g. where herbicide pooled in a low area, or was applied improperly). Nevertheless, having found specific effects in this experimental design, future studies should look at the influence of this herbicide on survival and reproduction in P. milvina over a range of concentrations. There was a trend for herbicide to interact with predator cues to affect survival. Adding herbicide to S. quadriceps cues had a negative effect on survival, whereas adding herbicide to H. helluo cues had a neutral to positive effect. A previous study conducted in mesocosms showed similar results in that while S. quadriceps cues alone increased survival, this effect was negated by adding herbicide (Chapter 2). Relyea (2005a) likewise discovered that combining a commercially formulated glyphosate-based herbicide and predator cues decreased amphibian survival, making it likely that this pattern is more widely applicable. The neutral to positive effects of adding herbicide to H. helluo cues could be linked to the slight increase in consumption during the foraging trials by spiders also exposed to herbicide, or to the possibility of hormec/hormoligosic effects discussed above. Interestingly, these results for survival and reproduction suggest that for P. milvina in the field, encountering herbicide might be detrimental if S. quadriceps densities are also high, but beneficial if H. helluo are abundant. In 2006, I found that in soybean fields, S. quadriceps tended to have higher activity density than H. helluo, especially around the time of the mid-season herbicide spray in early July (Chapter 1- fig1.1). If these results are true for other years, then P. milvina might encounter S. quadriceps in the field more often than H. helluo during herbicide application, with resulting negative impacts on reproduction and survival. Conclusions Overall, I discovered that changes in reproduction and survival of this important arthropod predator occur due to exposure to cues from its own predators and/or herbicide, but that these changes are further influenced by the availability of prey around the time of exposure. These results emphasize the need to conduct studies with multiple environmental stressors presented both combined and separately. This is highlighted by the result that something as simple as the availability of prey in one foraging bout when combined with another stressor was enough to significantly change subsequent reproductive success. Furthermore, I found that exposure to predator cues combined with herbicide can either be positive (H. hello cues and herbicide cause higher survival and higher egg sac viability) or negative (combination of S.

48 quadriceps cues and herbicide cause lower survival and lower egg sac viability). Changes in survival and reproductive success after exposure to herbicide or to predator cues could lead to population level effects on P. milvina, with implications for their role as predators in agricultural systems

Literature Cited Abdelghani, A A., Tchounwou, P B., Anderson, AC., Sujono, H., et al. 1997. Toxicity Evaluation of Single and Chemical Mixtures of Roundup, Garlon-3A, 2,4-D, and Syndets Surfactant to Channel Catfish (Ictalurus punctatus), Bluegill Sunfish (Lepomis microchirus), and Crawfish (Procambarus spp.). Environmental Toxicology and Water Quality. 12:237-243.

Achiorno CL., Villalobos, CD., Ferrari, L. 2008. Toxicity of the herbicide glyphosate to Chordodes nobilii (Gordiida, Nematomorpha). Chemosphere. 71:1816-1822.

Anderson, J.F. 1974. Responses to starvation in the spiders Lycosa lenta: Hentz and Filistata hibernalis. Ecology. 55(3): 576-585.

Baatrup, E., Bayley, M. 1993. Effects of the pyrethroid insecticide cypermethrin on the locomotor activity of the wolf spider Pardosa amendata: Quantitative analysis employing computer-automated video tracking. Ecotoxicology and Environmental Safety. 26: 138- 152.

Benamú, MA., Schneider, MI., Sánchez, NE. 2010. Effects of the herbicide glyphosate on biological attributes of Alpaida veniliae (Araneae, Araneidae), in laboratory. Chemosphere. 78: 871-876.

Bell, R.D., Rypstra, AL., Persons, MH. The effect of predator hunger on chemically mediated antipredator responses and survival in the wolf spider Pardosa milvina (Araneae: Lycosidae).

Calabrese, EJ., Baldwin, LA. 2003. Inorganics and hormesis. Critical Reviews in Toxicology. 33: 215-304.

Carter, PE., Rypstra, AL.1995. Top-down effects in soybean agroecosystems: spider density affects herbivore damage. OIKOS. 72: 433-439.

Chen, C.Y., Hathaway, K.M., Folt, C.L., 2004. Multiple stress effects of Visions herbicide, pH, and food on zooplankton and larval amphibian species from forest wetlands. Environmental Toxicolology and Chemistry. 23:823–831.

Chiverton, P.A. 1984. Pitfall-trap catches of the carabid beetle Pterosuchus melanarius in relation to gut contents and prey densities in insecticide treated and untreated spring barley. Entomologia experimentalis et Applicata. 36: 23-30.

Chiverton, PA. Sotherton, NW. 1991. The effects on beneficial arthropods of the exclusion of herbicides from cereal crop edges. Journal of Applied Ecology. 28:1027- 1039.

Cox, DR. 1972. Regression Models And Life-tables. Journal Of The Royal Statistical Society Series B-statistical Methodology. 34: 187-220.

Croft, BA., Brown, AWA. 1975. Responses of arthropod natural enemies to insecticides. Annual Review of Entomology. 20:285-335.

Dicke, M, Grostal, P. 2001. Chemical detection of natural enemies by arthropods: An ecological perspective. Annual Review of Ecology and Systematics. 32: 1-23.

Folmar LC, Sanders HO, Julin AM (1979) Toxicity of the herbicide glyphosate and several of its formulations to fish and aquatic invertebrates. Arch Environ Contam Toxicol 8:269-278.

Giesy, JP., Dobson, S, Solomon, KR 2000. Ecotoxicological risk assessment for Roundup ® herbicide. Review of Environmental Contamination and Toxicology.

Grostal, P., Dicke, M. 1999. Direct and indirect cues of predation risk influence behavior and reproduction of prey: a case for acarine interactions. Behavioral Ecology. 10: 422-427.

Hairston, NG. Smith, FE., Slobodkin, LB. 1960. Community structure, population control and competition. American Naturalist. 94: 421-425.

Haughton AJ., Bell, JR., Boatman, ND., Wilcox, A. 1999. The effects of different rates of the herbicide glyphosate on spiders in arable field margins. Journal of Arachnology. 27: 249-254.

Haughton, AJ., Champion, GT., Hawes, C., Heard, MS. et al., 2003. Invertebrate responses to the management of genetically modified herbicide-tolerant and conventional spring crops. II. Within-field epigeal and aerial arthropods. Philosophical Transactions of the Royal Society of London B. 358: 1863-1877.

Jakob, E.M., Marshall, S.D., and Uetz, G.W. 1996. Estimating fitness: a comparison of body condition indices. Oikos . 77(1): 61-67.

Klaschka, U. 2008. The infochemical effect- a new chapter in ecotoxicology. Environmental Science and Pollution Research. 15:452-462.

Luckey, TD., 1968. Insect hormoligosis. Journal of Economic Entomology. 61: 7-12.

Lundgren, JG., Gassmann, AJ., Bernal, J., Duan, JJ., Ruberson, J. 2009. Ecological compatibility of GM crops and biological control. Crop Protection. 28: 1017- 1030.

Lurling, M, Scheffer, M. 2007. Info-disruption: pollution and the transfer of chemical information between organisms. Trends in Ecology and Evolution. 22:374-379.

Marshall, SD, Rypstra, AL. 1999. Spider competition in structurally simple ecosystems. Journal of Arachnology. 27:343-350.

Menard, SW. 2002. Applied logistic regression analysis. Sage University Papers Series on Quantitative Applications in the Social Sciences, 07-106. Thousand Oaks, CA: Sage.

Michalkova, V., Pekar, S. 2009. How glyphosate altered the behavior of agrobiont spiders (Araneae: Lycosidae) and beetles (Coleoptera: Carabidae). Biological Control. 51: 444-449.

Montserrat, M., Bas, C., Magalhaes, S. et al. 2007. Predators induce egg retention in prey. Oecologia. 150: 699-705.

Morgan, M., Kiceniuk, J. 1992. Response of Rainbow-trout to a 2 month exposure to Vision ®, a Glyphosate herbicide. Bulletin of Environmental Contamination and Toxicology. 48: 772-780.

Morse, D; Stephens, E. 2006. The consequences of adult foraging success on the components of lifetime fitness in a semelparous, sit and wait predator. Evolutionary Ecology, 10: 361-373.

Nyffeler, M., Sunderland, KD. 2003. Composition, abundance and pest control potential of spider communities in agroecosystems: a comparison of European and US studies. Agriculture, Ecosystems and Environment. 95: 579-612

Persons, MH., Rypstra, AL. 2001. Wolf spiders show graded antipredator behavior in the presence of chemical cues from different sized predators. Journal of Chemical Ecology. 27: 2493-2504.

Pimentel D., Acquay, H., Biltonen, M., Rice, P., et al.,1991. Environmental and economic costs of pesticide use. Bioscience. 42: 750-760.

Prasifka, JR., Lopez, MD., Hellmich, RL., Prasifka, PL. 2008. Effects of insecticide exposure on movement and population size estimates of predatory ground beetles (Coleoptera: Carabidae). Pest Management Science. 64:30-36.

Ripple, WJ., Beschta, RL. 2004. Wolves and the ecology of fear; can predation risk structure ecosystems? Bioscience. 54: 755-766.

Relyea, RA. 2003. Predator cues and pesticides: A double dose of danger for amphibians. Ecological Applications. 13: 1515-1521.

Relyea, RA. 2005a. The lethal impacts of Roundup and predatory stress on six species of North American tadpoles. Archives of Environmental Contamination and Toxicology. 48: 351-357.

Relyea, RA. 2005b. The lethal impact of roundup on aquatic and terrestrial amphibians. Ecological Applications. 15: 118-1124.

Rypstra, AL., Schmidt, JM., Reif, BD., DeVito, J, Persons, MH. 2007. Tradeoffs involved in site selection and foraging in a wolf spider: effects of substrate structure and predation risk. Oikos. 116: 853-863.

Saglio, P. and Trijasse, S. (1998) Behavioral responses to atrazine and diuron in goldfish. Archives of Environmental Contamination and Toxicology. 35: 484–491.

Schmitz, JO. 2006. Predators have large effects on ecosystem properties by changing plant diversity, not pant biomass. Ecology: 87: 1432-1437.

Schneider, MI., Sanchez, N., Pineda, S., Chi, H., Ronco, A. 2009. Impact of glyphosate on the development, fertility and demography of Chrysoperla externa (Neuroptera: Chrysopidae): Ecological approach. Chemosphere. 76: 1451-1455.

Snyder, WE.,Wise, DH. 1999. Predator interference and the establishment of generalist predator populations for biocontrol. Biological Control 15:283–292.

Snyder, WE., Wise, D.H. 2001. Contrasting trophic cascades generated by a community of generalist predators. Ecology. 82: 1571-1583.

Soso, AB., Barcellos, LJG., Ranzani-Paiva, MJ., Kreutz, LC., et al. 2007. Chronic exposure to sub-lethal concentration of a glyphosate-based herbicide alters hormone profiles and affects reproduction of female Jundia (Rhamdia quelen). Environmental Toxicology and Pharmacology. 23: 308-313.

Stinner, BR., House, GJ. 1990. Arthropods and other invertebrates in conservation-tillage agriculture. Annual Reviews in Entomology. 35: 299-318.

Stoks, R. 2001. Food stress and predator-induced stress shape developmental performance in a damselfly. Oecologia. 127: 222-229.

Sunderland, K. 1999. Mechanisms underlying the effects of spiders on pest populations. Journal of Arachnology. 27: 308-316.

Symondson, WOC., Sunderland, KD., Greenstone, MH. 2002. Can generalist predators be effective biocontrol agents? Annual Review of Entomology. 47: 561-594.

Thorbeck, P., Bilde, T. 2004. Reduced numbers of generalist arthropod predators after crop management. Journal of Applied Ecology. 41: 526-538.

Tietjen, WJ. 2006. Pesticides affect the mating behavior of Rabidosa rabida (Araneae, Lycosidae). Journal of Arachnology. 34: 285-288.

Tsui, MTK., Chu, LM. 2003. Aquatic toxicity of glyphosate-based formulations: Comparison between different organisms and the effects of environmental factors. Chemosphere. 52: 1189-1187.

Walker, S.E., Marshall, SD., Rypstra, AL., and Taylor DH. 1999. The effects of hunger on locomotory behavior in two species of wolf spider (Araneae, Lycosidae). Animal Behaviour. 58: 512-520.

Wilder, S; Rypstra, A. 2004. Chemical cues from an introduced predator (Mantodea, Mantidae) reduce the movement and foraging of a native wolf spider (Araneae, Lycosidae) in the laboratory. Environmental Entomology. 33: 1032-1036.

Wilder, S; Rypstra, A. 2008. Diet quality affects mating behaviour and egg production in a wolf spider. Animal Behaviour. 76: 439-445.

Wise, D.2006. Cannibalism,food limitation, intraspecific competition and the regulation of spider populations. Annual Review of Entomology. 51:441-465.

Woodburn, A. 2000. Glyphosate: production, pricing and use worldwide. Pest Management Science. 56:309-312.

Yokoyama, VY., Pritchard, J. 1984. Effect of pesticides on the mortality, fecundity and egg viability of Geocoris pallens (Hemiptera: Lygaeidae). Journal of Economic Entomology. 77:876-879

Table 3.1: Mean ± standard error for number of crickets killed across treatments

Treatment Number of crickets killed

Low prey `No cues/water 4.25 ± 0.2

S. quadriceps/water 3.74 ± 0.2

H. helluo/water 3.60 ± 0.2

No cues/herbicide 3.86 ± 0.2 S. quadriceps/herbicide 4.06 ± 0.2

H. helluo/herbicide 3.58 ± 0.2 High prey

No cues/water 9.63 ± 0.7 S. quadriceps/water 10.00 ± 0.8

H. helluo/water 8.75 ± 0.8

No cues/herbicide 9.82 ± 0.8

S. quadriceps/herbicide 9.00 ± 0.8 H. helluo/herbicide 9.05 ± 0.8

Table 3.2: The effects of predator cues, herbicide and prey level on prey capture and consumption Prey capturea Consumptionb (p- value) (p- value) 2 2 c Model ( χ /F, p-value) χ 11 =197.92, p=0.0001 F11,228= 5.55, p=0.0001 Prey (Low/High) 0.0001 0.0001 Predator cues 0.0661 0.0019 Herbicide 0.7168 0.3232 Prey x predator cues 0.7201 0.9284 Prey x herbicide 0.9622 0.1938 Predator cues x herbicide 0.8016 0.3993 Predator cues x herbicide x prey 0.2453 0.3628 a Number of crickets found dead or missing compared between treatments using an ordinal logistic regression b Change in abdomen width across the 2 hour foraging trial compared between treatments using a three-way ANOVA. cSignificant p values are shown in bold

Table 3.3: Tests for impacts of treatment (prey availability= high or low, predator cues = none, H. helluo, or S. quadriceps, and herbicide= herbicide or water) on reproduction (egg sac produced and spiderlings emerged) and survival in P. milvina. Egg sac produced?a Spiderlings emerged?a Survivalb (p-value) (p-value) (p-value) 2 2 2 2 Model (χ , p-value) χ 7=14.75; p=0.0393 χ 9=17.81; p=0.0374 χ ,11 =18.53; p=0.0700 Prey (High or Low) 0.0666 0.9880 0.5471 Predator cues 0.1190 0.4263 0.2997 Herbicide 0.8941 0.2952 0.3183 Prey x predator cues 0.0347 0.1175 0.0284 Prey x herbicide N/A 0.0192 0.6618 predator cues x herbicide 0.0812 0.0857 0.0619 Prey x cues x herbicide N/A N/A 0.1836

a Logit models and b Proportional hazards tests were used respectively to analyze the effects of all three factors on reproduction and survival of adult spiders; values significant at p<0.05 are in bold

Fig. 3.1: Fully factorial experimental design for the effects of Predator cues (none, H. helluo, or S. quadriceps), Herbicide (or water), and Prey availability (high or low) on foraging and reproduction in P. milvina. Sample sizes (n) are indicated after each group.

Fig. 3.2: Abdomen width change in P. milvina after foraging for two hours on crickets during exposure to predator cues and/or herbicide. High prey availability treatment spiders were given 20 crickets each and in the Low prey availability treatment spiders were given 5 crickets each. Symbols represent the means ± standard error. Letters are from Tukey post hoc tests where shared letters indicate no significant difference between treatments.

Fig. 3.3: Reproduction by P. milvina after exposure to predator cues (H. helluo, S. quadriceps, or none) and/or herbicide or water. A. Proportion of spiders from low prey availability treatments producing egg sacs and B. Proportion of those with egg sacs from which spiderlings emerged. C. Proportion of spiders from high prey availability treatments producing egg sacs and D. Proportion of those with egg sacs from which spiderlings emerged. Error bars indicate binomial error of the mean.

Fig. 3.4: Proportion of adult P. milvina surviving over a 60 day period after exposure to predator cues (H. helluo, S. quadriceps, or none) and/or herbicide or water. A. Spiders from the low prey treatments and B. Spiders from the high prey treatments.

Chapter 4: Exposure to herbicide and predator cues has no impact on growth, development time, fluctuating asymmetry or survival in Pardosa milvina (Araneae: Lycosidae) Introduction: Exposure to environmental stressors such as hunger, predation, temperature and chemicals can affect juvenile animals in ways that influence them throughout their lives (see Nylin and Gotthard, 1998 for a review; Danner and Joern, 2004; Cauble and Wagner, 2005). Responses to stressors during development are commonly measured using three interacting factors: size, growth rate and development time, and these can ultimately impact an individual’s fitness (Nylin and Gotthard, 1998; Danner and Joern, 2004). For example, female grasshoppers exposed to predation risk as juveniles exhibit prolonged development time and, as a result, have reduced egg production as adults (Danner and Joern, 2004). In addition to influencing growth rate and development time, exposure to environmental stressors might also decrease developmental stability, an animal’s ability to correct for errors during development, often resulting in increased fluctuating asymmetry (differences in left-right side symmetry of body parts) (Valentine et al., 1973; Eeva et al., 2000; Stoks, 2001; Chang et al., 2007). Increases in fluctuating asymmetry can then affect an individual’s performance via decreased competitive ability or reduced attractiveness to potential mates (see Polak 2008 for a review). For instance, male Japanese scorpionflies with more asymmetrical forewings are less likely to win mating contests and attract females (Thornhill, 1992 a,b). Animals that complete their development and live in association with humans, particularly in agricultural systems, are subject to stressors in the form of anthropogenic chemicals (Freemark and Boutin, 1995). Anthropogenic chemicals can harm non-target organisms by negatively affecting growth, development time, fluctuating asymmetry and survival (Bridges, 2000; Greulich and Pflugmacher, 2003; Rohr et al., 2004). For example, the herbicide Atrazine, common in runoff from agricultural fields, reduces survival, decreases development time and reduces size at metamorphosis in salamanders (Rohr et al., 2004). In addition to anthropogenic chemical cues, another common stressor that impacts growth and development in juveniles is the threat of predation; even in the absence of direct contact with the predator (Stoks, 2001; Smith et al., 2005; Beketov and Liess, 2007). For example, Stoks (2001)

62 found that exposure to predator cues, especially when combined with another stressor, decreased size and growth rate and increased fluctuating asymmetry in larval damselflies. Although individuals often encounter natural chemical cues from predators coincident to anthropogenic chemical cues, it can be difficult to predict their combined effects (Klaschka, 2008; Lurling and Scheffer, 2007; Relyea, 2003; 2005b). This is because when combined, natural chemical cues and anthropogenic chemical cues might act synergistically so that together they are worse than each alone (Relyea, 2003; 2005). Alternatively, the anthropogenic chemical might change an animal’s ability to detect a natural chemical cue, inhibiting the effects of this natural cue (Klaschka, 2008; Lurling and Scheffer, 2007). Furthermore, despite their potential importance, studies of exposure to both types of cues are lacking in terrestrial systems (Lurling and Scheffer, 2007). Here I tested the separate and combined impacts of an herbicide and predator cues on the growth, development time, fluctuating asymmetry (FA) and survival of a common arthropod predator in agroecosystems. I hypothesized that exposure to herbicide would interact with exposure to predator cues such that together their impacts on growth, development time, FA and survival are different than that of each stressor alone. Study System The wolf spider Pardosa milvina (Hentz, 1844) is a numerically dominant predator in agricultural systems throughout the eastern United States, found at densities of 5 -100 individuals/m2 (Young, 1989; Draney, 1997; Marshall and Rypstra, 1999; Marshall et al., 2002). The chemical sensitivity of this species has been well documented, as individuals detect substrate-borne chemicals in the form of silk and feces from both predators and mates (Persons et al., 2001; Wilder and Rypstra, 2004; Rypstra et al., 2003). The chemical awareness of P. milvina and its abundance in agricultural systems makes it an ideal species to explore the impacts of herbicide and predator chemical cues on growth, developmental stability and survival. Pardosa milvina co-occurs in agricultural systems with the predatory carabid beetle Scarites quadriceps (Chaudoir, 1843) (Chapter 1; Synder and Wise, 1999). Scarites quadriceps is found at densities of about 1 individual/m2 and likely acts as a predator of P. milvina in these systems (Snyder and Wise, 1999). Additionally, P. milvina in agricultural fields are often exposed to a glyphosate-based herbicide (Chapter 1). Although glyphosate targets an enzyme pathway specific to plants (Williams et al., 2000), some studies have found that exposure to this chemical, surfactants, or a combination of both, found in the commercially formulated herbicide

63 mixture can decrease survival in several animal groups (Abdelghani, et al., 1997; Mann and Bidwell, 1999; Relyea, 2005 a,b; Achiorno et al., 2008). Additionally, exposure to these herbicides can change growth rate, time to adulthood and size in some taxa, including annelids and amphibians (Springett and Gray, 1992; Howe et al., 2004; Cauble and Wagner, 2005). Interestingly, although glyphosate-based herbicides are common (Woodburn, 2000; Lundgren et al., 2009) and therefore likely to be encountered simultaneously with predator cues, the combined effects of the two stressors have only previously been examined in one aquatic system (Relyea, 2005 b). Previous experiments demonstrate that this herbicide alone decreases survival, activity and reproduction in adult P. milvina (Evans et al., in review; Chapters 2, 3). Additionally, exposure to chemical cues from S. quadriceps in the form of feces and other excreta can impact the survival, activity and reproductive success of adult female P. milvina (Chapters 2, 3). However, these effects can change when the spider is simultaneously exposed to S. quadriceps cues and herbicide (Chapters 2, 3). This is relevant to spiders in the field because both adults and juveniles are exposed to herbicide at discrete times during the season when S. quadriceps are also present (Chapter 1). However, nothing is known about how these stressors act either separately or together to effect juvenile P. milvina. Therefore, I tested how exposure to a commercially-formulated, glyphosate- based herbicide and cues from the arthropod predator S. quadriceps affected growth, survival and developmental stability of P. milvina. Methods: Animal care During July and August of 2008, experimental subjects were collected in corn and soybean fields at the Miami University Ecology Research Center (Oxford, OH, USA). Female P. milvina with egg sacs were collected, and the offspring of these females were used in the experiment. Adult S. quadriceps were also collected and used to deposit chemical cues to represent predation risk. Female spiders were housed individually in the laboratory in round plastic containers (8 cm w x 5 cm h) with a two cm deep substrate of 50:50 peat moss: potting soil covering the bottom. Scarites quadriceps were kept in larger containers (12 cm w x 8 cm h) and were given deeper substrate (4 cm) to allow burrowing. All animals were maintained in controlled environmental chambers (25C, 55-60% relative humidity, 13:11h light:dark cycle). Animals were fed weekly with two-three domestic crickets, Acheta domesticus (Order:

Orthoptera, Family: Gryllidae), 0.3 cm in length for adult P. milvina and 0.6 cm in length for S. quadriceps. At each feeding, water was additionally provided for each animal by moistening the substrate in each container with distilled water. Spiderlings from the egg sacs of five field caught females were used for the experiment. Within 7-10 days of hatching, spiderlings dispersed from their mother’s abdomen. Dispersed spiderlings were placed in a plastic container like those of the adult spiders (see above). However, each of these containers had been inoculated two weeks previously with 15-20 Sinella curviseta (Order: Collembola, Family: Entomobryidae) and a potato to create largely self sustaining prey populations. Every 48 hours I remoistened the containers with distilled water, added S. curviseta if fewer than 10 adult prey were observed, and recorded any spider molts. Spiderlings were sustained on S. curviseta until their third molt. Thereafter, they were fed 2-3 appropriately sized A. domesticus twice a week. An appropriately-sized prey was defined as one that had a body length shorter than the body length of the spiderling. Experimental design Spiders were exposed to predator cues, herbicide, both or neither in a factorial design. There were initially 25 spiders per treatment, although some spiders died prior to experimentation, creating final sample sizes of 22-25. Previous experiments indicated that P. milvina completes 5-8 molts before adulthood, with the majority of spiders taking 6-7 molts (personal observation). I attempted to expose the majority of spiders to their appropriate treatment twice during development. To do this, I exposed spiderlings once within 48 hours of the third molt (mid-way through development) and once within 48 hours after the fifth molt (close to maturity). However, almost half of the spiders matured at the fifth molt and were therefore only exposed once. Preparation and exposure to predator cues and herbicide Fifteen minutes prior to deposition of predator cues, each S. quadriceps was fed and allowed to fully consume one cricket (0.6 cm in length), to control for any effects of hunger. The predator was then placed in a circular plastic container (12.5 cm w x 8 cm h) for 24 hours to deposit chemical cues on filter paper that was used for the experiment. I used the commercially formulated herbicide Buccaneer Plus®, also known as Roundup original® (Monsanto Company, St. Louis, Missouri, USA) for herbicide exposures. This herbicide contains the active ingredient glyphosate (480g/L) in the form of isopropylamine salt,

65 as well as other inactive ingredients. I diluted the herbicide to 2.5% with double distilled water and applied it to the substrate using a spray rate equivalent to 17.6 kg a.i. ha-1 of glyphosate. Exposures were conducted in cylindrical plastic containers (12.5 cm w x 8 cm h) with filter paper covering the bottom. The filter paper was divided into four equal quadrants (Fig 1). Two alternate quarters were either blank (i.e. untreated) or contained cues left by the beetle predator while the other two were both treated with approximately 0.45 mL of herbicide solution or distilled water (Fig 1). After 24 hrs of exposure, spiderlings were measured for size and mass (see below) and were then returned to their original containers. Data collection and analysis: Growth and Survival I calculated the number of days between each molt after exposure to stressors for each spider. Additionally, I used a Wild 5 microscope and digital micrometer to measure the size, cephalothorax width (the widest point across the carapace), of each spider to the nearest 0.01 mm. The animals were also weighed to the nearest 0.1 mg to determine mass, and growth was determined as change in mass/days between molts. Finally, I recorded the frequency with which individuals survived to adulthood. According to Shapiro-Wilk tests, the distributions of days between molts, size and growth rate were often non-normal (p<0.05) and could not be made normal with transformation. Therefore, I used the nonparametric Kruskal-Wallis test (Zar, 1999) to determine differences between treatments for days between molts, size and growth rate. Furthermore, to determine my ability to detect differences between treatments using current sample sizes, I conducted a post hoc power analysis on any of these measurements that were not significant in the Kruskal-

Wallis test. Treatments were compared for differences in survival to adulthood (yes or no) using a Logit model (Menard, 2002) with predator cues and herbicide as factors. I conducted all statistical analyses using JMP 7.0 (SAS Institute, Inc.). Development- Fluctuating Asymmetry Upon reaching adulthood, spiders were preserved in 70% ethanol for quantification of fluctuating asymmetry (FA), as a measure of stability of development (Palmer and Stroebeck, 1986; Clark, 1995). I tested for developmental stability in P. milvina using FA measurements of the forelegs. On the first pair of legs, the femur and combined length of the patella and tibia

66 were measured on the left and right sides of each spider to the nearest 0.01 mm. Each leg section was measured three times and legs were repositioned between measurements to quantify measurement error (Palmer, 1994). According to a nested ANOVA (measure nested within side) there was no significant measurement error for patella tibia length (F4=1.68, p=0.155). Therefore, I used an average of the three measurements per individual for each patella tibia segment on each side to calculate FA measures. However, there was a significant effect of measure within side for femur length (F4=3.08, p=0.017), indicating significant measurement error. Therefore, this trait was not suitable for FA calculations, and was removed from further analysis. Fluctuating asymmetry was calculated using the absolute value of the difference between left and right side measurements (Palmer, 1994). This measure of FA was compared between treatments using a three-way ANOVA with herbicide, predator cues and number of exposures (1 or 2) as factors. Fluctuating asymmetry was further analyzed for evidence of differences between males and females. Finally, I conducted a post hoc power analysis to determine my ability to detect differences in FA between treatments using my current sample sizes. Results: Growth, development time and survival Following exposure, spiders spent an average (±SE) of 13.6 ± 1.0 days between molts 3-4, an average of 26.6 ± 1.6 days between molts 4-5 and an average of 20.9 ± 1.8 days between molts 5-6 (following second exposure). However, treatment had no effect on days between molts for any of these periods (Tables 4.1-4.3). Spiders gained an average of 0.26 ± 0.09 mm in size (cephalothorax width) between molts 3-4, gained an average of 0.28 ± 0.10 mm between molts 4-5, and gained 0.31 ± 0.10 mm between molts 5-6. However, treatment had no impact on change in size (cephalothorax width) during any of these time periods (Tables 4.1-4.3). Spiders grew an average of 1.7 x 10-4 ± 8.4 x 10-6 mg/day during molts 3-4, grew 2.3 x 10-4 ± 1.9 x 10-5 mg/day during molts 4-5, and grew 3.6 x 10-4 ± 3.2 x 10-5 mg/day during molts 5-6. As before, treatment did not affect growth rate in the spiders at any of the time periods (Tables 4.1-4.3). Post hoc power analyses demonstrated that power for each of these tests was low (<0.50), but that sample sizes would have to be more than doubled to correct for this. Across treatments, 77% of spiders survived to reach adulthood and there was no evidence that either herbicide or 2 predation risk influenced survival (Χ3 = 1.60, p=0.66).

Fluctuating Asymmetry Mean signed FA is an indication of the absolute value of the differences in symmetry between an organism’s left and right sides. For P. milvina, mean signed FA of the patella-tibia was normally distributed (Shapiro-Wilk test: W= 0.99; p= 0.83). Furthermore, this mean did not differ significantly from a mean of 0 (t-test: t=0.13, p=0.89) indicating “true” FA. Average FA of the combined patella-tibia-segments was a 0.23 ± 0.18 mm difference between sides.

However, there was no significant impact of treatment on patella-tibia FA (model: F6,71=0.16, p=0.98). A post hoc power analyses demonstrated that power for these tests was low (0.07), but that sample size would have to be raised by a factor of about 20 to correct for this. When spiders were divided by sex, there were differences in the relationship between FA and body size (Fig. 4.2). Males showed a negative relationship between size at maturity (cephalothorax width) and combined patella-tibia segment FA, where larger spiders exhibited lower levels of FA (ANOVA:

F1,23 = 11.55; p<0.01; Fig 2a). This pattern was not present for females (ANOVA: F1,46 = 2.63, p=0.11; Fig. 2b), and disappeared when males and females were combined. Due to an unbalanced sex ratio or lack of survival to adulthood, there were too few males (n = 4- 7/treatment) to run a separate analysis, but females (n=10-13) were tested separately for impacts of treatment on FA. For females, there was no impact of treatment on FA (F3,46= 0.69, p=0.56). Discussion: Previous studies have shown exposure to predator or anthropogenic chemical cues can induce changes in growth rate or size at maturity and increase developmental instability (fluctuating asymmetry) (Bridges, 2000; Eeva et al., 2000; Stoks, 2001; Rohr et al., 2004). However, I found no effect of exposure to a glyphosate-based herbicide or to chemical cues of the beetle predator Scarites quadriceps on Pardosa milvina’s survival, growth or development. For this experiment, I applied herbicide at a rate 7-8 x higher than that recommended for use in the field, to ensure complete coverage of the substrate during exposure and to maximize chances of observing an effect. This rate of herbicide application could simulate a worst case scenario in the field (where the herbicide is improperly applied or pooled in a low area). However, as I saw no effects at this elevated rate, this indicates that acute exposure in the field to residual herbicide on the substrate is unlikely to result in disruptions to survival, growth or development in P. milvina.

Glyphosate has a strong tendency to adsorb to soil particles and is rapidly broken down by microbes, quickly reducing the amount of herbicide that is bioavailable (Rueppel, et al., 1977). However, differences in soil type and other environmental conditions allow glyphosate to have extremely variable residence times in soil, anywhere from a couple of days to several months (Giesey et al., 2000). Therefore, although a 24 hour exposure time is realistic under environmental conditions where glyphosate breaks down quickly, it would also be worthwhile to test the survival, growth and development of P. milvina under exposure to herbicide over a much longer time period. In fact, chronic exposure to this herbicide does impact growth or development in other terrestrial organisms including earthworms, neuropterans and spiders (Springett and Gray, 1992; Schneider et al., 2009; Benamú et al., 2010). Twenty-four hours of exposure to a high application rate of herbicide did not noticeably affect the survival, growth or development of P. milvina, but short- term exposure to high levels of this herbicide can be detrimental to juvenile organisms in aquatic systems (Folmar, 1979; Mann and Bidwell, 1999; Lajmanovich et al., 2003). For example, Folmar (1979) found that exposure to a glyphosate-based herbicide for only 6 hours reduced survival in larval fish. Furthermore, acute exposure to high levels of this herbicide (24-96 hours) can also decrease survival and lead to deformations in several species of larval amphibian (Mann and Bidwell, 1999; Lajmanovich et al., 2003; Edginton et al., 2004). It is likely that the detrimental effects of acute applications of this herbicide to larval aquatic organisms is due to their susceptibility to the POEA (polyethoxylated tallowamine) surfactant included in the commercially-formulated mixture (Folmar, 1979; Mann and Bidwell, 1999; Lajmanovich et al., 2003; Edginton et al., 2004). For example, gill damage in larval amphibians can be associated with exposure to herbicide mixtures which include this surfactant (Lajmanovich et al., 2003). Terrestrial animals would obviously not be susceptible in the same way, and in fact, adult frogs can be less vulnerable to this herbicide than tadpoles (Mann and Bidwell, 1999; but see Relyea, 2005a). Additionally, in the case of P. milvina and other arthropods, their hard exoskeleton could protect against direct penetration of the herbicide and its surfactant. This is supported by a previous study in which P. milvina did not exhibit reduced survival after topical exposure of this herbicide, but did die sooner after residual exposure to herbicide on the substrate (Evans et al., in review). In light of this, spiders might be more likely affected via ingestion of the herbicide either from the substrate (Evans et al., in review) or from

69 exposed prey (Schneider, et al., 2009; Bernamu, et al., 2010). In the current experiment, juveniles were not fed during the exposure trials, so if they did not directly ingest the herbicide through drinking it from the substrate or grooming it off of their legs, their contact could have been minimized. As a result, feeding juvenile P. milvina during exposure could increase the chance of seeing effects in further experiments. I found no impact of predator cues either alone or combined with herbicide on the growth, development or survival of P. milvina. Previous studies have shown that exposure to predator cues can affect growth and development in other organisms (Stoks, 2001; Smith et al., 2005; Beketov and Liess, 2007). Furthermore, Relyea (2005b) found that a glyphosate-based herbicide was even more deadly to amphibian survival when combined with predator cues. However, these studies (with the exception of Smith et al., 2005) were conducted using chronic exposure to the stressors over a considerable portion of the organism’s developmental period. Perhaps chronic exposure to S. quadriceps cues either with or without herbicide could lead to effects on survival, growth or development that were not seen here and this should be addressed in further experiments. Interestingly, in previous experiments, exposure to S. quadriceps cues over a 24 hour period did influence survival and reproduction in adult P. milvina (Chapters 2, 3). Differences in detection of chemical cues between juveniles and adults are possible. For example, juveniles of one species of lizard are able to detect and avoid chemical cues from predators, whereas adults are not; likely because adults are too large to be attacked by these predators (Head et al., 2002). This could be reversed for P. milvina. If Scarites quadriceps have difficulty capturing small P. milvina and are thus not a danger, juveniles might be less sensitive to this predator’s cues than are adults. Alternatively, in the current experiment juveniles were frequently seen climbing the walls of the containers (personal observation, K.W) and thus might have avoided excessive exposure to predator cues. Climbing has previously been documented for this species to avoid cues from another predator, the wolf spider, Hogna helluo (Persons et al., 2002; Folz, et al., 2006). Further experiments are needed to determine if avoidance of predator cues by juvenile P. milvina then leads to indirect impacts on survival, growth and development. For example, although climbing might help the spider avoid predator cues in the short term, it could potentially have negative effects on foraging over a longer term exposure (but see Folz et al., 2006).

Treatment had no significant impact on fluctuating symmetry (FA), but there were some interesting differences in how the relationship between FA and body size varied by sex. Larger males tended to have a smaller degree of FA than smaller males, whereas females showed no relationship between size and leg segment FA. It would be valuable to measure a larger sample of adult males from field populations to determine if general FA levels in leg segments follow this pattern. If this pattern were supported, it could indicate that larger males have potential fitness advantages that come with greater symmetry. For example, more symmetrical males might have greater success in competing with other males, or in attracting females (Thornhill, 1992a,b). Planned sample sizes were reduced by the time spiders reached adulthood due to mortality (n= 17-19/treatment), making them considerably lower than the minimum recommended for FA measures (Palmer, 1994). However, post hoc power analyses indicated that significant differences between treatments were unlikely to be detected without raising the sample size by a factor of about 20. Therefore, I am confident that there are in fact no significant differences in FA between treatments for patella-tibia length and trying to detect them by increasing sample size would be ineffective. However, future studies could look at FA in other traits such as the length of mouth parts (chelicerae or pedipalps) or that of the segments of other legs besides just the first pair. Conclusions: To conclude, acute exposure to predator cues and to a commercially-formulated glyphosate- based herbicide during development had no significant effect on survival, growth or developmental stability in Pardosa milvina. I did not measure the impacts of exposure to these stressors during earlier instars however, and this could conceivably make a difference. Additionally, measuring FA in more than two traits at adulthood might give a better quantification of how exposure to these stressors impacts developmental stability (Palmer, 1994). Finally, although these stressors did not influence growth and development, they might affect the behavior of these spiders during development, as they have been shown to do in adults (Chapter 2).

Literature Cited

Abdelghani, A A., Tchounwou, P B., Anderson, AC., Sujono, H., et al. 1997. Toxicity Evaluation of Single and Chemical Mixtures of Roundup, Garlon-3A, 2,4-D, and Syndets Surfactant to Channel Catfish (Ictalurus punctatus), Bluegill Sunfish (Lepomis microchirus), and Crawfish (Procambarus spp.). Environmental Toxicology and Water Quality. 12:237-243.

Achiorno, CL., Villalobos, CD., Ferrari, L. 2008. Toxicity of the herbicide glyphosate to Chordodes nobilii (Gordiida, Nematomorpha). Chemosphere. 71:1816-1822.

Beketov, MA., Liess, M. 2007. Predation risk perception and food scarcity induce alterations of life-cycle traits of the mosquito Culex pipiens. Ecological Entomology. 32: 405-410.

Benamú, MA., Schneider, MI., Sánchez, NE. 2010. Effects of the herbicide glyphosate on biological attributes of Alpaida veniliae (Araneae, Araneidae), in laboratory. Chemosphere. 78: 871-876.

Bridges, CM. 2000. Long term effects of pesticide exposure at various life stages of the southern leopard frog (Rana sphenocephala). Archives of Environmental Contamination and Toxicology. 39:91-96.

Cauble, K., Wagner, RS. 2005. Sublethal effects of the herbicide glyphosate on amphibian metamorphosis and development. Bulletin of Environmental Contamination and Toxicology. 75: 429-435.

Chang, X., Zhai, B., Liu, X., Wang, M. 2007. Effects of temperature stress and pesticide exposure on fluctuating asymmetry and mortality of Copera annulata (selys) (Odonata: Zygoptera) larvae. Ecotoxicology and Environmental Safety. 67: 120- 127.

Clark, GM. 1995. Relationships between developmental stability and fitness: application for conservation biology. Conservation biology. 9: 18-24.

Danner, BJ., Joern, A. 2004. Development, growth and egg production of Ageneotettix deorum (Orthoptera: Acrididae) in response to spider predation risk and elevated resource quality. Ecological Entomology. 29: 1-11

Draney, M. 1997. Ground-layer spiders (Araneae) of a Georgia piedmont floodplain agroecosystem: species list, phenology and habitat selection. The Journal of Arachnology 25:333–35.

Edginton AN, Sheridan PM, Stephenson GR, Thompson DG, Boermans HJ (2004) Comparative effects of pH and Vision® herbicide on two life stages of four anuran species. Environmental Toxicology and Chemistry 23:815-822.

Eeva, T., Tanhuanpaa, S, Rabergh, C, et al., 2000. Biomarkers and fluctuating asymmetery as indicators of pollution-induced stress in two hole-nesting passerines. Functional Ecology. 14: 235-243.

Folz, HC., Wilder, SM., Persons, MH., Rypstra, AL. 2006. Effects of predation risk on vertical habitat use and foraging of Pardosa milvina. Ethology. 112: 1152-1158.

Freemark, K, Boutin, C. 1995. Impacts of agricultural herbicide use on terrestrial wildlife in temperate landscapes: A review with special reference to North America. Agriculture, Ecosystems and Environment. 52: 67-91.

Giesey, JP., Dobson, S., Soloman, KR., 2000. Ecotoxicological risk assessmentfor Roundup® herbicide. Reviews in Environmental Contamination and Toxicology. 167: 35–120

Greulich, K, Pflugmacher ,S. 2003. Differences in susceptibility of various life stages of amphibians to pesticide exposure. Aquatic Toxicology 65:329–336.

Head, ML. Keogh, JS., Doughty, P. 2002. Experimental evidence of an age-specific shift in chemical detection of predators in a lizard. Journal of Chemical Ecology. 28:541-554.

Howe, CM., Berrill, M., Pauli, BD et al., 2004. Toxicity of glyphosate based herbicides to four North American frog species. Environmental Toxicology and Chemistry. 23: 1928-1938.

Klaschka, U. 2008. The infochemical effect- a new chapter in ecotoxicology. Environmental Science and Pollution Research. 15: 452-462.

Lajmanovich, RC., Sandoval, MT., Pelzer, PM. 2003. Induction of mortality and malformation in Scinax nasicus tadpoles exposed to glyphosate formulations. Bulletin of Environmental Contamination and Toxicology. 70: 612-618.

Lundgren, JG., Gassmann, AJ., Bernal, J., Duan, JJ., Ruberson, J. 2009. Ecological compatibility of GM crops and biological control. Crop Protection. 28: 1017- 1030.

Lurling, M, Scheffer, M. 2007. Info-disruption: pollution and the transfer of chemical information between organisms. Trends in Ecology and Evolution. 22:374-379.

Marshall, SD, Rypstra, AL. 1999. Spider competition in structurally simple ecosystems. Journal of Arachnology. 27:343-350.

Menard, SW. 2002. Applied logistic regression analysis. Sage University Papers Series on Quantitative Applications in the Social Sciences, 07-106. Thousand Oaks, CA: Sage.

Nylin, S, Gotthard, K. 1998. Plasticity in life history traits. Annual Review of Entomology. 43: 63-83.

Palmer, AR. 1994. Fluctuating asymmetry analyses: a primer. Pp.335-364. In TA. Markow (ed.), Developmental Instability: Its origins and evolutionary implications. Kluwer, Dordrecht, Netherlands.

Palmer, AR., Stroebeck, C. 1986. Fluctuanting asymmetery: measurement, analysis, paterns. Annual Review of Ecology and Systematics. 17: 392-421.

Polak, M. 2008. The developmental instability- sexual selection hypothesis: A general evaluation and case study. Evolutionary Biology. 35: 208-230.

Relyea, RA. 2005a. The lethal impact of roundup on aquatic and terrestrial amphibians. Ecological Applications. 15:1118-1124.

Relyea, RA. 2005b. The lethal impacts of roundup and predatory stress on six species of North American tadpoles. Archives of Environmental Contamination and Toxicology. 48: 351- 357.

Rohr, JR., Elskus, AA., Shepherd, B.S., et al., 2004. Multiple Stressors and salamanders: effects of an herbicide, food limitation and hydroperiod. Ecological Applications. 14: 1028- 1040.

Rueppel, ML., Brightwell, BB., Schaefer, J., Marvel, JT. 1977. Metabolism and degredation of glyphosate in soil and water. Journal of Agricultural Food and Chemistry. 25: 517-528.

Rypstra, AL., Weig, C., Walker, SE., Persons, MH. 2003. Mutual mate assessment in wolf spiders: Differences in the cues used by males and females. Ethology. 109: 315-325.

Snyder, WE.,Wise, DH. 1999. Predator interference and the establishment of generalist predator populations for biocontrol. Biological Control 15:283–292.

Smith, MJ., Drew, MM., Peebles, M., Summers, K. 2005. Predator cues during the egg stage affect larval development in the gray treefrog, Hyla versicolor (Anura: Hylidae). Copeia. 1: 169-173.

Springett, JA., Gray, RA. 1992. Effect of repeated low doses of biocides on the earthworm Aporrectodea caliginosa in laboratory culture. Soil Biology and Biochemistry. 24:1739- 1744.

Stoks, R. 2001. Food stress and predator-induced stress shape developmental performance in a damselfly. Oecologia. 127: 222-229.

Thornhill, R. 1992a. Fluctuating asymmetry, interspecific aggression and male mating tactics in 2 species of Japanese scorpionflies. Behavioral Ecology and Sociobiology. 30: 357-363.

Thornhill, R. 1992b. Female preference for the pheromone of males with low fluctuating asymmetry in the Japanese scorpionfly (panorpa-japonica, mecoptera). Behavioral Ecology. 3: 277-283.

Valentine , DW., Soule, ME., Samollow, P. 1973. Asymmetry analysis in fishes- possible statistical indicator of environmental stress. Fishery Bulletin. 71: 357-370.

Williams, GM., Kroes, R., Munroe, IC. 2000. Safety evaluation and risk assessment of the herbicide roundup and its active ingredient, glyphosate for humans. Regulatory Toxicology and Pharmacology. 31: 117-165. Woodburn, A. 2000. Glyphosate: production, pricing and use worldwide. Pest Management Science. 56:309-312.

Young, O.P., Lockley, T.C., Edwards, G.B. 1989. Spiders of Washington County, Mississippi. Journal of Arachnology. 17 :27- 41

Zar, JH. 1999. Biostatistical Analysis. 4th ed. Prentice Hall, Inc. Upper Saddle River, NJ.

Table 4.1: Kruskal-Wallis results and means ± standard error for measures of growth between molts 3-4 (directly after exposure to predator cues and/or herbicide)

Daysa Change in Size Growth rate (mm)b (mg/day)c

Model (χ2, p-value) χ2 =1.50, p = 0.22 χ2 =1.37, p = 0.26 χ2 = 0.49, p = 0.69 No cues, water (n=24) 12.70 ± 1.68 0.25 ± 0.02 1.8 x 10-4 ± 1.7 x 10-5 No cues, herbicide (n=23) 15.43 ± 1.72 0.25 ± 0.02 1.8 x 10-4 ± 1.7 x 10-5 Cues, water (n=25) 17.28 ± 1.65 0.29 ± 0.02 1.8 x 10-4 ± 1.6 x 10-5 Cues, herbicide (n=25) 15.68 ± 1.65 0.25 ± 0.02 1.6 x 10-4 ± 1.7 x 10-5 a Number of days between molts 3 and 4 for each spider b Change in size of the cephalothorax (carapace) width (mm) for each spider between molts 3 and 4. c Growth rate as measured by change in mass of the spider (mg) per day between molts 3 and 4.

Table 4.2: Kruskal-Wallis results and means ± standard error for measures of growth between molts 4-5 (two molts after exposure to predator cues and/or herbicide)

Daysa Change in Size Growth rate (mm)b (mg/day)c

Model( χ2 , p-value) χ2 = 1.13, p = 0.34 χ2 = 0.78, p = 0.51 χ2 = 0.51, p = 0.68 No cues water (n= 24) 22.70 ± 3.15 0.27 ± 0.02 2.3 x 10-4 ± 3.8 x 10-4 No cues herbicide (n= 22) 28.00 ± 3.15 0.31 ± 0.02 2.5 x 10-4 ± 3.8 x 10-5 Cues water (n=25) 24.77 ± 3.15 0.27 ± 0.02 2.1 x 10-4 ± 3.7 x 10-5 Cues Herbicide (n=24) 30.72 ± 3.15 0.28 ± 0.02 2.1 x 10-4 ± 3.8 x 10-5 a Number of days between molts 4 and 5 for each spider b Change in size of the cephalothorax (carapace) width (mm) for each spider between molts 4 and 5. c Growth rate as measured by change in mass of the spider (mg) per day between molts 4 and 5.

Table 4.3: Kruskal-Wallis results and means ± standard error for measures of growth between molts 5-6 (molts directly after second exposure to predator cues and/or herbicide)

Daysa Change in Size Growth rate (mm)b (mg/day)c

Model (χ2 , p-value) χ2 = 2.73, p = 0.29 χ2 = 1.78, p = 0.62 χ 2 = 2.07, p = 0.56 No cues water (n=15) 17.53 ± 3.18 0.32 ± 0.03 3.1 x 10-4 ± 5.8 x 10-5 No cues herbicide (n=12) 21.42 ± 3.55 0.32 ± 0.03 3.8 x 10-4 ± 7.0 x 10-5 Cues water (n=10) 16.60 ± 3.89 0.30 ± 0.03 4.2 x 10-4 ± 7.0 x 10-5 Cues Herbicide (n=13) 27.78 ± 3.41 0.28 ± 0.03 3.5 x 10-4 ± 6.3 x 10-5 a Number of days between molts 5 and 6 for each spider b Change in size of the cephalothorax (carapace) width (mm) for each spider between molts 5 and 6. c Growth rate as measured by change in mass of the spider (mg) per day between molts 5 and 6.

Fig. 4.1: Diagram of the container in which juvenile Pardosa milvina were exposed to Scarites quadriceps cues and/or glyphosate-based herbicide for 24 hours. Filter paper split into quadrants covered the bottom, and a lid was placed on the container to prevent the spider from escaping.

Fig. 4.2: Relationship between the absolute value of fluctuating asymmetry (FA) of Patella-tibia length and Cephalothorax width in adult A) males and B) females of Pardosa milvina.

Chapter 5: General Conclusions and Synthesis My experiments encompass the first comprehensive attempt to quantify the impacts of exposure to a commercially formulated, glypohsate-based herbicide on the reactions of a terrestrial arthropod to cues from its predators. In the previous chapters I examined the impacts of herbicide and predator cues on activity and emigration (chapter 2), survival (chapters 2 & 4), foraging and reproduction (chapter 3- also includes effects of prey availability) and finally growth and development (chapter 4) of the agrobiont wolf spider, Pardosa milvina. From these results I draw some general conclusions and speculate on the larger impacts of herbicide and predator cues on populations of P. milvina and their role in the food web in agricultural systems. First, although P. milvina likely recognizes both Hogna helluo and Scarites quadriceps as predators, it reacts in a different manner to the cues from each (Table 1). Exposure to H. helluo cues alone is almost always detrimental for adult P. milvina; reducing activity and emigration, and lowering foraging and reproductive success. In contrast, exposure to S. quadriceps cues can be helpful or detrimental depending on the behavior or life history trait being measured. For example, after exposure to cues, activity and emigration is increased and long term survival is higher, but reproductive success is decreased. Ultimately, exposure to H. helluo cues alone is likely to be much more detrimental for adult P. milvina in the field than is exposure to S. quadriceps cues alone, unless the exposed spider is a reproductive female. Secondly, exposure to herbicide had a variety of effects (Table 1). For example, it led to a slight increase in prey consumption, but decreases in survival (in one experiment) and reproduction. Exposure to herbicide alone only decreased survival in the experiment where exposure had occurred in mesocosms and the spiders were then brought back into the laboratory (Chapter 2). It is possible that this reduction in survival was due to herbicide exposure in the outdoor mesocosms interacting with an uncontrolled variable (e.g. high temperatures) that was not present in the laboratory during exposure. In other experiments, the impacts of herbicide on survival will sometimes depend on environmental conditions (Folmar et al., 1979; Wan et al., 1989; Tsui and Chu, 2003). For example, water quality, temperature and pH can change the toxicity of herbicide-exposed water to fish and other aquatic organisms (Folmar et al., 1979; Wan et al., 1989; Tsui and Chu, 2003). However, another experiment did show reduction in survival for P. milvina exposed to herbicide for only 30 minutes in the laboratory, where the temperature was controlled (Evans et al., in review), so other variables are probably playing a

role as well. Finally, the POEA (polyethoxylated tallowamine) surfactant used with this type of herbicide to increase penetration of the plant tissues has often been implicated as more toxic than the active ingredient glyphosate (Folmar et al., 1979; Mann and Bidwell, 1999; Howe et al., 2004). I did not differentiate between the surfactant and the glyphosate as both are included in the commercial formulation. Therefore, it would be valuable to expose P. milvina to glyphosate and POEA separately in future experiments to determine which ingredient is responsible for the changes in survival seen here. In addition to affecting survival, exposure to herbicide also decreased the reproductive success of P. milvina. However, these results do not seem indicative of a larger pattern as studies in other taxa have had varied results. For instance, exposure to this type of herbicide positively impacted reproduction in big eyed bugs (Yokoyama and Pritchard; 1984) but had neutral or negative effects on reproduction in fish, frogs, neuropterans and spiders (Folmar et al., 1979; Soso et al., 2007; Takahashi, 2007; Quassinti et al., 2009; Schneider, et al., 2009; Benamú, et al., 2010). Third, exposure to herbicide and predator cues combined had different effects depending on the behavior being measured and on the predator species (Table 1). More specifically, exposure to herbicide seemed to negate positive impacts and exacerbate negative impacts of S. quadriceps cues. For example, P. milvina exposed to herbicide and S. quadriceps cues together exhibited decreased activity, survival and reproductive success. In contrast, additional exposure to herbicide seemed to alleviate some of the negative impacts of exposure to H. helluo cues alone. For instance, spiders experiencing herbicide and H. helluo cues together slightly increased survival in the low prey treatment and substantially increased reproductive success (spiderling emergence) in the high prey treatment. In 2006, activity density of H. helluo was much lower than that of S. quadriceps in soybean fields during the two periods when herbicide spraying occurred (although these differences were not present in corn fields) (Fig. 1 A,B- Chapter 1). If this pattern is true in other years as well, it is probable that P. milvina encounter S. quadriceps cues more often than H. helluo cues in the soybean fields, so the overall effects of exposure to predator cues and herbicide may be more negative than positive. Simultaneous exposure to a glyphosate-based herbicide and predator cues may be detrimental to other species as well. For example, Relyea (2005) found that herbicide became twice as toxic to wood frogs also exposed to predator cues. Other common herbicides such as amitrol and atrazine can also interact with predator cues to affect the behavior and life history of

animals (Mandrillion and Saglio, 2007; 2009; LaFiandra et al., 2008). However, none of these studies has looked at the impacts of herbicides on predator cues in terrestrial systems. Furthermore, despite the common usage of glyphosate-based herbicides (Woodburn, 2000), only the current study and that of Relyea (2005) have dealt with how they interact with predator cues to affect survival, behavior and life history of animals in any system. As in the Relyea (2005) study, I too established that this herbicide can have synergistic effects on survival and make the presence of predator cues (from S. quadriceps) more lethal. However, I also found the opposite effect for exposure to the other predator H. hello, and furthermore I found that other factors (i.e. food availability) can help determine the negative effects of this herbicide. Therefore, this leaves a large and important area of research to be addressed. Interestingly, the effects of herbicide and predator cues both separately and combined were often determined by the amount of prey that P. milvina was offered during exposure. For example, survival was impacted by herbicide and predator cues in the low prey but not high prey treatments (Table 1). Likewise, Chen et al. (2008) found that a different herbicide, triclopyr, interacted with food level and another factor (pH) to decrease survival in zooplankton and amphibians. Food availability is a factor that could interact with herbicide and predator cues to impact a wide range of taxa and should therefore be addressed in further experiments. Finally, although acute exposure to herbicide and predator cues influenced behavior, reproduction and survival in adult P. milvina, it did not affect survival growth or development in juvenile P. milvina (Table 1). First, it is possible that P. milvina are less sensitive to herbicide as juveniles than as adults. Conversely though, Folmar et al. (1979) found that trout were more susceptible to this type of herbicide as juveniles than as eggs or adults. Second, juvenile P. milvina might be less sensitive than adults to chemical cues from large predators. Head et al. (2002) found that juvenile lizards detected and avoided chemical cues from predators, whereas adults did not. The authors attributed this partly to the fact that juvenile lizards were small enough to be attacked by invertebrate predators, but adults were too large. This could be reversed for P. milvina where juveniles are too small to often be preyed upon by adult predators, and therefore are less affected by their chemical cues. There is some indication that newly hatched spiderlings can detect fresh cues laid down by H. helluo, because they dismount from the mother’s abdomen less frequently in the presence of these cues (Persons and Lynam, 2004). However, aside from this evidence, we do not know how juvenile P. milvina react to predator

cues from adult predators, particularly not those of S. quadriceps that were used in the experiment here. Therefore, it would be valuable to look into this further. Another possibility is that the parameters being measured in juveniles (growth and development) are less sensitive to being affected by herbicide and predator cues in this species. However, this hypothesis contrasts studies of other animals, including amphibians, annelids arthropods , for which herbicide (Springett and Gray, 1992; Howe et al., 2004; Cauble and Wagner, 2005 ) or predator cues (Peckarsky et al., 2002; Smith et al., 2005; Beketov and Liess, 2007) can negatively affect growth and development. One major difference between the current study and those other studies is the exposure of P. milvina was over a very short period (only 24 hours) whereas most of the other studies had chronic exposure times lasting the length of the developmental period. In comparison, exposure of these spiders to these cues in the field would likely be more variable but also last longer than 24 hours (Giesy et al., 2000). It is possible that under longer term exposure a change in the growth and development of P. milvina in response to herbicide or predator cues would be noticeable. Furthermore, I did not measure the effects of these stressors on activity and foraging in juveniles, which would be interesting to address with further research. In the current study, survival is the only parameter that was also measured in adults. In some cases, adult survival was influenced by herbicide and predator cues. It is possible that differences in the way the adult and juvenile spiders were exposed to these stressors contributed to the lack of effects on juvenile spiders. For example, my methods differed because juveniles were not allowed to forage during exposure (Chapter 4) whereas adults were; and furthermore, feeding influenced survival in exposed adults (Chapter 3). Taken together, the results of these experiments indicate that predator cues and a commercially formulated glyphosate-based herbicide have the potential to impact P. milvina at both the population and community levels in the field. Based on the results of chapter 4, pre- emergence spraying early in the season, when the population is mostly juveniles and adult males, (Marshall et al., 2002; Fig 5.2) could have lower impacts, as it is unlikely to change the juvenile spider’s growth, development and survival (Table 1). However, glyphosate-based herbicides have a soil residence time that varies widely from days to months depending on soil and other environmental conditions (Giesy et al., 2000). Therefore, once again, chronic exposure to herbicide could affect these juvenile spiders differently and should thus be tested. A mid- season spray is likely to occur during peak reproductive times in this system as adult males and

reproductive females tend to peak in July (Marshall et al., 2002; Fig 2). In fact, when I collected initial data in 2006, spraying had occurred in early July right before this peak (Fig 1- chapter 1). Therefore, female spiders are likely to be exposed prior to producing egg sacs, with potential negative consequences. However, H. helluo and S. quadriceps are also common during that time, so exposure to herbicide along with cues from one of these predators could become either more beneficial or more harmful (Chapter 1-Fig.1 A,B). Although food availability was not low during that time period as compared to other times of the year (Chapter 1: Fig 2 A, B), I only gathered field data for one year and weather could play a role in the density of edible prey present in other years. Further research could test whether a mid-season spray translates into lower reproduction and whether there are any population-level effects in P. milvina in the field. Because P. milvina is in the middle of the food web in these systems as an important predator of insect prey but also prey itself to other arthropod predators, changes in populations could ultimately cascade up or down to affect the community. Testing the hypotheses corresponding to these ideas could be accomplished by conducting a multi-year study where some plots receive a typical mid-season spray and others receive no spray, but weeds are manually removed. Activity density of P. milvina and its major prey and predators could then be measured through pitfall trapping and hand collection in each plot type to determine if there are effects of spraying on population densities in the field. Ultimately, these experiments comprise a critical first look at how this commonly used herbicide interacts with predator cues to impact a generalist predator in terrestrial systems. Even after only a short term exposure to these stressors, complex relationships were uncovered between herbicide, predator cues and food level that affected activity, reproduction and survival. This indicates the importance of measuring multiple factors when attempting to determine the impacts of anthropogenic chemicals on animals in these systems. Furthermore, these results encourage further studies both with P. milvina and with other terrestrial species in this understudied area of research.

Literature Cited

Beketov, MA., Liess, M. 2007. Predation risk perception and food scarcity induce alterations of life-cycle traits of the mosquito Culex pipiens.

Benamú, MA., Schneider, MI., Sánchez, NE. 2010. Effects of the herbicide glyphosate on biological attributes of Alpaida veniliae (Araneae, Araneidae), in laboratory. Chemosphere. 78: 871-876.

Cauble, K., Wagner, RS. 2005. Sublethal effects of the herbicide glyphosate on amphibian metamorphosis and development. Bulletin of Environmental Contamination and Toxicology. 75: 429-435.

Folmar LC, Sanders HO, Julin AM (1979) Toxicity of the herbicide glyphosate and several of its formulations to fish and aquatic invertebrates. Arch Environ Contam Toxicol 8:269-278.

Giesy, JP., Dobson, S., Solomon, KR. 2000. Ecotoxicological risk assessment for Roundup® Herbicide. Review of Environmental Contamination and Toxicology. 167: 35-120.

Head, ML. Keogh, JS., Doughty, P. 2002. Experimental evidence of an age-specific shift in chemical detection of predators in a lizard. Journal of Chemical Ecology. 28:541-554.

Howe, CM., Berrill, M., Pauli, BD et al., 2004. Toxicity of glyphosate based herbicides to four North American frog species. Environmental Toxicology and Chemistry. 23: 1928-1938.

LaFiandra, EM., Babbitt, KJ., Sower, SA. 2008. Effects of atrazine on anuran development are altered by the presence of a nonlethal predator. Journal of Toxicology and Environmental Health Part A- Current Issues. 71: 505-511.

Mandrillon, A., Saglio, P. 2007. Herbicide exposure affects the chemical recognition of non native predator in common toad tadpoles (Bufo bufo). Chemoecology. 17: 31-36.

Mandrillon, A., Saglio, P. 2009. Effects of single and combined embryonic exposure to herbicide and conspecific chemical alarm cues and larval traits in the common frog (Rana temporaria). Archives of Environmental Contamination and Toxicology. 56: 566-576.

Marshall, SD., Pavuk, DM., Rypstra, AL. 2002. A comparative study of phenology and activity patterns in the wolf spiders Pardosa milvina and Hogna helluo in soybean agroecosystems in southwestern Ohio (Araneae: Lycosidae). Journal of Arachnology. 30: 503-510.

Peckarsky, BL., McIntosh, AR., Taylor, BW., Dahl, J. 2002. Predator chemicals induce changes in mayfly life history traits: A whole-stream manipulation. Ecology. 83: 612-618.

Persons, MH., Lynam, EC. Pardosa milvina (Araneae, Lycosidae) spiderling movement in the presence of conspecific and heterospecific silk and excreta. Journal of Arachnology. 32: 341-344.

Quassinti, L; Maccari, E; Murri, O; Bramucci, M. 2009. Effects of paraquat and glyphosate on steroidogenesis in gonads of the frog Rana esculenta in vitro. Pesticide Biochemistry and Physiology. 93: 91-95.

Relyea, RA. 2005. The lethal impacts of Roundup and predatory stress on six species of North American tadpoles. Archives of Environmental Contamination and Toxicology. 48: 351-357.

Smith, MJ., Drew, MM., Peebles, M., Summers, K. 2005. Predator cues during the egg stage affect larval development in the gray treefrog, Hyla versicolor (Anura: Hylidae). Copeia. 1: 169-173.

Springett, JA., Gray, RA. 1992. Effect of repeated low doses of biocides on the earthworm Aporrectodea caliginosa in laboratory culture. Soil Biology and Biochemistry. 24:1739- 1744.

Tsui, MTK., Chu, LM. 2003. Aquatic toxicology of glyphosate-based formulations: comparison between different organisms and the effects of environmental factors. Chemosphere. 52: 1189-1197.

Takahashi, M. 2007. Oviposition site selection: Pesticide avoidance by gray treefrogs. Enivronmental Toxicology and Chemistry. 26: 1476-1480.

Wan, MT., Watts, RG., Moul, DJ. 1989. Effects of different dilution water types on the acute toxicity of juvenile Pacific salmonids and rainbow trout of glyphosate and its formulated products. Bulletin of Environmental Contamination and Toxicology. 43: 378-385.

Woodburn, A. 2000. Glyphosate: production, pricing and use worldwide. Pest Management Science. 56:309-312.

Table 5.1: Effects of the stressors: predator cues (H. helluo and S. quadriceps), herbicide and prey availability on the behavior and life history of P. milvina (Summary of results from Chapters 2-4).

Effects on Effects on Effects on Effects on Effects on Growth and Stressor(s) Activity/emigration Survival Foraging success Reproductive success Development present (Chapter 2) (Chapter 2, 3) (Chapter 3) (Chapter 3) (Chapter 4)

H. helluo Less activity and - N/A (Chapter 2) Lower prey capture and - Lower in low prey treatments N/A cues emigration - Lower survival in low less consumption (fewer egg sacs) prey treatments (Chapter 3) - Lower in high prey treatments (fewer with spiderling emergence)

S. quadriceps More activity and - Higher survival: (in No effect - Lower in low prey treatments No effect cues emigration Chapter 2 and low prey (fewer egg sacs) treatments in Chapter 3) - Lower in high prey treatments - No effect in developing (fewer with spiderling spiders (Chapter 4) emergence)

Herbicide - No change in forward - Lower survival (Chapter No significant effect, but - Lower in high prey treatments No effect movement or emigration 2) there was a tendency for (both egg sac and spiderling - Tendency toward less - No effect (Chapter 3, increased consumption in emergence) non-forward movement Chapter 4) low prey treatments. - Lower in low prey treatments (spiderling emergence)

Cues + - H. helluo cues + - Intermediate survival No significant - Lower for S. quadriceps + No Herbicide herbicide = no interaction. (Chapter 2) interactions herbicide in both low and high interactions - S. quadriceps cues + - Lower survival of S. prey treatments (spiderling herbicide = less activity quadriceps + herbicide (low emergence) and emigration. prey treatments- Chapter 3) - Higher for H. helluo cues + - Higher survival for H. herbicide in high prey treatment helluo + herbicide (low prey (spiderling emergence). treatments- Chapter 3) - No effect Chapter 4)

Prey N/A No effect (Chapter 3) Fewer prey available = No Effect N/A availability lower capture and less consumption

Fig. 5.1: Number of adult female, adult male and juvenile P. milvina collected in dry pitfall traps over a 24 hour period from A. soy and B. corn fields. The number of individuals is an average of three replicate fields of each type. Arrows indicate dates when a glyphosate-based herbicide was sprayed in the fields.

Appendicies

Appendix 1: Methods for survey and sampling of P. milvina, its predators, and the stressors it undergoes across the agricultural season. In late April through mid-November of 2006, I collected animals in six agricultural fields 60 x 70 m in size and separated from one another by 15m wide strips of grass. On May 24th three of the fields were planted in Roundup® Ready corn (Zea mays) and the others were planted with Roundup® Ready soybeans (Glycine max), at a distance of 76 cm between each crop row in a field. Traps were set up in the fields initially on April 20th, about a month before planting. I set up pitfall traps in groups of five at each field with one in the center and the other four at 10m x 10m distances in from each corner of the field. Each trap opening consisted of an 18 cm piece of PVC pipe sunk into the ground to protect the trap and maintain the shape of the hole in which the trap was placed. I inserted a 0.5 L plastic cup in each PVC sleeve so that the top of the cup was level with the ground. I filled each cup one third full of a 1:1 water and propylene glycol (nontoxic antifreeze) mix, and placed a 0.2m X 0.2m plywood square over the top of each trap, with 25 mm long wooden dowels attached to hold it above the ground. Finally, I placed hexagonal steel wire mesh “chicken wire” over the entire apparatus and staked it down to prevent disturbance from vertebrates. I emptied traps once every two weeks and reset them one week after being emptied. In the lab, I removed invertebrates from the traps and placed them in 70% ethanol, for later sorting, counting and identification to order (insects) or family (spiders). I enumerated P. milvina, H. helluo and S. quadriceps, separately. Once during each week between resetting traps, I set out dry cups for 24 hours to catch additional P. milvina. The following day, any trapped P. milvina were recorded as juvenile, adult male or adult female and released. I also recorded any anthropogenic disturbance events in the fields including planting, herbicide spray, and harvest. A glyphosate-based herbicide (Bucaneer® Ultra) was sprayed at two occasions in the fields. One spray event occurred, prior to planting (May 6th) and the other occurred at mid- season (June 29th). This herbicide is made of 41% active ingredient (glyphosate in the form of isopropylamine salt, 480g/L) and 59 % inactive ingredients including the surfactant polyethoxylated tallowamine (POEA) (Monsanto Company, St. Louis, Missouri, USA). The commercial herbicide formulation was diluted at a rate of 0.95 L of herbicide per 75.7 L of water to form a concentration of 1.25%. This diluted formulation was then sprayed in the field at a rate of 191.5 L/ha. Harvest occurred in the fields on November 3rd and 4th.