University of Cincinnati

! "# $ % & % ' % ! "# !' "$ % &'()! *+'

,, , ,

"# ' '% $$('

Behavioral research on wolf spiders (Araneae: Lycosidae): Assessing common assumptions

and methods

A dissertation for submission to the Graduate School

Of the University of Cincinnati

In partial fulfillment of the requirements for the degree of

Doctor of Philosophy

In the Department of Biological Sciences

Of the McMicken College of Arts and Sciences

Jenai M. Rutledge

B.S., Animal Behavior

Bucknell University, Lewisburg, PA, May 2003

Committee Chair: Dr. George W. Uetz ABSTRACT.

Ecological, behavioral, and evolutionary theory and research is based on a network of assumptions that simplify the otherwise complex physical and natural systems of life.

Assumptions are a necessary part of conducting research because they provide a framework from which predictions about these systems can be made and tested through the interpretation of statistical analyses. However, the validity of conclusions drawn from any empirical study is only as good as the assumptions upon which the research design and interpretations were made. The research presented here addresses and tests a number of specific assumptions commonly made in research studies conducted on invertebrate animals, as applied to Schizocosa wolf spiders, an emerging animal model in animal behavior. My research focuses on two species, S. ocreata and

S. rovneri well-known in studies of communication and mate choice. In two studies, I examined the traditional assumption that in invertebrates, flexibility of female mate choice behavior is minimal. From the first study it is clear that behavioral plasticity of these invertebrate animals in response to experience is greater than previous recognized. Exposure of female S. rovneri as juveniles to altered male phenotypes resulted in avoidance of familiar and preference for novel phenotypes as adults. However, these studies also show that certain types of experience

(chemical vs. visual cues) may affect female mate preferences more than others, and that manipulation of male phenotypes (e.g., with nail polish) can sometimes have unintended consequences. In a second study with both S, ocreata and S. rovneri, the mechanisms that underlie species-level mate recognition (e.g., female mate preference) are more rigidly defined and do not appear to be influenced by social experience. A third study examined how well traditional measurements of body condition of spiders are able to separate out differences in feeding and/or hydration histories. This study provides evidence that S. ocreata may be able to

iii modulate their apparent body condition (calculated by traditional body condition measures) when faced with short-term food and/or water deprivation. Specifically, the morphological body condition of spiders deprived of food or water remained remarkably similar for up to eight days.

This suggests that current measures of body condition may not provide an accurate estimate of current body condition. Reults of these studies suggest that in designing experiments, the assumptions underlying commonplace experimental methods/techniques often go unrecognized, and if assumptions are inaccurate or wrongly applied, the validity of research can be jeopardized.

It is therefore important that assumptions be evaluated in the light of novel data to ensure they are logical and current.

ACKNOWLEDGEMENTS

I would like to thank my research advisor, George Uetz for his constant encouragement, feedback, and support. Without his dedication to my success and continued confidence in my abilities as a researcher and teacher, I would have ‘thrown-in the towel’ long ago. His down-to- earth personality and sincere interest in helping his students succeed (both in and outside of the lab) has made him great adviser and a valued friend. My committee members Elke Buschbeck,

Eric Maurer, Ken Petren, and Ann Rypstra for valuable discussion and feedback, time, patience and support throughout the many many reinventions of my dissertation research. Ken Petren for all of the advice, supplies and lab space during the two years I spent trying to develop microsatellite markers for the cursed, Californian, colonial web-building spider, Metepeira spinipes. My husband Matt, my sister Jessica, and my Mom, Debbe, and Dad, Charlie for hours of field and lab assistance, professional guidance, and much emotional support (here too I thank my son, Brayden who always knows how to make me smile). To Kitty Uetz, for being a person I could turn to for help, guidance, support, encouragement, good artwork, and pretty much anything else I’ve needed.

Thanks to many past and present graduate students for helpful feedback, advice, good discussion, and social pursuits; especially Julianna Johns (I would have lost my mind long ago if it weren’t for your terrific sense of humor and friendship), Brian and Christine Moskalik, and Shira

Gordon. To the undergraduate researchers who have helped over the years with collection of spiders and maintenance of lab animals (especially Justin Allen and Melita Skelton). Funding for this research was provided by the American Arachnological Society, the University of Cincinnati

Research Council, the National Science Foundation (IBN 0239164 to G.Uetz), the

Wiemen/Wendel/Benedict Student Research Fund, and the Department of Biological Sciences.

TABLE OF CONTENTS

Behavioral research on wolf spiders (Araneae: Lycosidae): Assessing common assumptions and methods...... i ABSTRACT...... iii ACKNOWLEDGEMENTS...... vi LIST OF TABLES...... ix LIST OF FIGURES...... xi Chapter 1: Introduction and General Overview...... 1 References...... 9 Chapter 2: Exposure to multiple sensory cues as a juvenile affects adult female mate preferences in wolf spiders...... 11 Abstract...... 12 Introduction...... 13 Methods...... 15 Study Species...... 15 Collection and Housing...... 17 Experimental Methods...... 17 Results...... 23 Discussion...... 26 Acknowledgements...... 31 References...... 32 Chapter 3: Effects of Juvenile Experience on Adult Female Mating Preferences in Two Closely Related Sympatric Wolf Spider Species...... 48 Abstract...... 49 Introduction...... 50 Methods...... 52 Study Species...... 52 General Methods...... 53 Juvenile Exposure ...... 54 Adult Mate Recognition Trials ...... 55 Statistical Analyses ...... 56 Results...... 57 Discussion...... 60 Acknowledgements...... 65

vii

Chapter 4: Testing the limits of body condition indices: a comparison of approaches to estimating condition during short-term starvation in wolf spiders ...... 83 Abstract...... 84 Introduction...... 85 Materials and Methods...... 89 Experiment 1: Juveniles...... 89 Subjects and Housing...... 89 Experimental Methods...... 90 Calculating Body Condition and Statistics ...... 91 Results...... 93 Experiment 2: Adult males ...... 94 Subjects and Housing...... 95 Experimental Methods...... 96 Results...... 98 Discussion...... 101 References...... 106 Chapter 5: General Conclusions ...... 124 References...... 132 Appendix I: Behavioral assays for predicting mating success in two species of wolf spiders: rates of general behaviors, male signals, and female receptivity ...... 136 References...... 142

viii

LIST OF TABLES.

Chapter 2

Table 1. 2-Way ANOVA for the effects of juvenile and adult treatment group on female receptivity…………………………………………………………………………………….. 39

Table 2. Nominal logistic fit of juvenile/adult treatment on mating outcome (Y/N)…………. 40

Table 3. Summary of mass spectrophotometry results for three different different colors of nail polish…………………………………………………………………………………… 41

Chapter 3

Table 1. 2-Way ANOVA for the effects of juvenile and adult treatment group on female receptivity (S. ocreata)………………………………………………………………………… 72

Table 2. 2-Way ANOVA for the effects of juvenile and adult treatment group on female receptivity (S. rovneri)………………………………………………………………………… 73

Table 3. Correlational relationships female receptivity and amount of juvenile experience for each treatment group………………………………………………………………………. 74

Chapter 4

Table 1. Repeated measure ANOVA for the effects of diet on body condition (Pilot study)…………………………………………………………………………………………...109

Table 2. ANCOVAs for effects of diet on body condition by day (Experiment 1)………….....110

Table 3. One-Way ANOVA of RBC on days 1, 3, 5 (Experiment 1)………….……………....111

Table 4. Multivariate repeated measures ANOVA (RBC) and ANCOVAs (MBC, VBC,

DBC) for effect of diet on body condition at day 10…………………………………………...112

Table 5. ANCOVAS for effects of diet on body condition by day (Experiment 2)……………113

Table 6. One-Way ANOVAs of RBC on days 1, 3, 5 (Experiment 2)…………………………114

Table 7. ANCOVAs of MBC, VBC, DBC on days 3, 5, 8, and 10……………………….115-116

Table 8. One-Way ANOVA of RBC for days 1, 3, 5, 8 and 10………………………………..117

LIST OF FIGURES.

Chapter 2

Figure 1. Sample of GC/MS chromatograms………………………………………………….43

Figure 2. Female receptivity to each of the four possible male phenotypes based on type of juvenile experience…………………………………………………………………….45

Figure 3. Mating outcomes when females were paired with each of the four possible male phenotypes by juvenile experience……………………………………………………………47

Chapter 3

Figure 1. Experimental apparatuses of juvenile and adult trials for both species……………76

Figure 2. Experimental Design…...…………………………………………………………...78

Figure 3. Effect of female exposure on total composite receptivity rate in response to heterospecific or conspecific males…………..………………………………………………80

Figure 4. Comparison of aggression between species towards hetero- and con-specific males before and after the barrier was removed……………………………………………...82

Chapter 4

Figure 1. (Experiment 1) Change in body condition over time in response to diet………….119

Figure 2. Survivorship of spiders maintained in the absence of food and water…………….121

Figure 3. (Experiment 2) Change in body condition over time in response to diet………….123

Appendix I

Figure 1. Logistic regressions of mating outcome versus female composite receptivity and composite receptivity as a rate………………………………………………………………145

Figure 2. Comparison of female receptivity rates between species for mated and unmated females………………………………………………………………………………………..147

Chapter 1: Introduction and General Overview

Jenai M. Rutledge

Department of Biological Sciences

University of Cincinnati

614 Rieveschl Hall

Cincinnati, OH USA 45221-0006

Corresponding author – Email: [email protected]; Tel: 513-556-9753, Fax: 513-556-5299

In essence, all scientific research is designed around testing the natural/external validity of hypotheses and related predictions that are generated by scientific theory and mathematical modeling. Virtually all areas of empirical research that test hypotheses rely on the use of assumptions or sets of assumptions to help design, analyze and/or interpret experiments and data.

Assumptions are a necessary part of conducting research and are useful for simplifying complex systems, such as those encountered in ecology, animal behavior, and evolution. As a consequence, predictions about these systems can be made and tested through the interpretation of statistical analyses. However, the validity of conclusions drawn from any empirical study is only as good as the assumptions upon which the research design and interpretations were made; if any of the assumptions are demonstrated to be false or illogical, the accuracy of results and subsequent research based on those results may be jeopardized.

Accepting assumptions about the natural history, physiology, or behavioral capacities of certain types of animals without directly testing their validity may hinder research progress and as a consequence, restrict our ability to understand the processes of evolution and ecology. In some cases, the reason assumptions go untested stems from a relatively dogmatic approach that accepts the findings of previous researchers to be entirely accurate, or assumes processes are the same across all circumstances or biological taxa. However, in much research, it seems likely that factors implicit in certain assumptions are simply overlooked.

The use of artificial experimental manipulations that modify natural phenotypic characteristics or that attempt to mimic natural conditions that are otherwise difficult to measure without bias in nature are common. For example, sexual selection theory suggests that

2 secondary sexual characteristics, i.e., characteristics that do not appear to serve any obvious function in contributing to the overall survival of an animal (e.g., long and colorful tail feathers of male peacocks; foreleg tufts of bristles of spiders; large antlers of various ungulates) evolve as a result of difference in reproductive success between individuals who possess the traits and those who do not (Darwin, 1871). In order to study effects of sexual selection, it is generally common to examine a trait (physical or behavioral) that has no apparent role in survival of an animal (or that has been demonstrated to be detrimental to overall survival in extreme forms) and test the assumption that having the trait provides some kind of reproductive advantage (e.g., improved ability to gain access to or to attract a mate). The use of artificial manipulations can be especially useful when testing hypotheses about how specific phenotypic traits influence mate choice (e.g., Andersson 1982; Petrie & Halliday 1994; Scheffer et al. 1996; Hebets 2003; Pryke

& Andersson 2005). Thus, manipulations of physical phenotypes are often made in an attempt to represent population extremes or heterospecific phenotypes, which can help to parse out subtle differences in response variables along a continuum, e.g. female preference for a larger secondary sexual trait in a male.

Although sham treatments or other types of procedures are implemented as controls to determine whether artificial treatments or conditions have effects beyond those intended, many studies that use artificial manipulations in an effort to mimic natural conditions assume that responses to artificial manipulations would be similar to the naturally occurring conditions/phenotypes. This can be a dangerous assumption, as in some cases, even seemingly benign experimental techniques have been shown to impact the animals in the study in unforeseen ways, e.g., colour of leg-bands in lab-maintained populations of finches affected

3 mating success (Burley 1986; Jennions 1998); home-cage colouration of lab mice was found to affect body mass, food consumption and various behavioral parameters (Sherwin & Glen 2003); lighting conditions for lab-maintained populations of starlings affected mate preferences (Evans et al. 2006), which may inadvertently bias the results of the study. When such effects are detected, important information about pre-existing biases and sensitivities in various animals to certain stimuli has been revealed (e.g., preference for red and pink over blue and green in finches: Burley 1986; critical flicker fusion frequency of the eyes of European starlings and its effects on mate choice: Evans et al. 2006). However, it is more common that data are not inspected for these types of effects. Ultimately, the failure to investigate whether an artificial experimental manipulation has unintended effects may result in studies that make conclusions that are misleading or erroneous (see reviews by: Major and Kendal 1996; Moore and Robinson

2004). Because an effort is generally made to keep manipulations relatively the same for all animals across treatments, it is commonly assumed that if the laboratory environment has an effect, that the effect will be similar for all animals. While this is generally a statistically-sound assumption, studies in which unintended, undetected responses to various controlled factors like housing environment, may produce results and conclusions that have diminished external validity (i.e., results that do not reflect responses to experimental variables that would be observed in natural populations).

Although the focus of my doctoral research is broad and incorporates a wide range of evolutionary, ecological and behavioral questions (some of which are relatively unrelated to each other) there is one common theme that unites the bulk of my dissertation research. Each project I have conducted has sought to examine a common assumption or established practice in animal

4 behavior research, to determine its overall validity. Historically, many aspects of spider physiology and behavior have been assumed to mirror that of insect and other arthropods animals. Although in some cases, there are similarities, distinct differences have been demonstrated. Anatomically, one clear difference between spiders and insects is the reliance of spiders on hydrostatic pressure for locomotion (Foelix 1996). In insects hemolymph pressure within the body remains relatively low and does not generally aid in locomotion (Borror et al.

1989). In constrast, spiders rely almost exclusively on hydrostatic pressure of the hemolymph in the body for leg extension during locomotion (Foelix 1996). In addition, clear differences in metabolic and respiration rates exist between spiders and other poikilotherm arthropods, including insects (e.g., Anderson & Prestwich 1982). Both of these unique aspects of spider biology influence the manner in which spiders can interact with their environment and other organisms. In light of findings I report in the fourth chapter regarding the effects of dehydration and starvation on body condition, recognizing these differences between insect and spider anatomy and physiology becomes increasingly important. Although in insects as well as spiders, certain aspects of metabolic responses to starvation and/or changes in hemolymph volume as a result of dehydration may be similar, the ultimate effects of these changes are likely to quite different between these groups of organisms.

For the purposes of my dissertation research, I chose to focus on two wolf spider species,

Schizocosa ocreata (Hentz) and S. rovneri Dondale & Uetz. These species are common in the leaf litter of deciduous forests throughout the eastern United States. These species are sympatric but do not interbreed due to behavioral barriers; however, forced interspecies copulation yields viable offspring (Stratton & Uetz, 1981, 1983). Female S. ocreata and S. rovneri are

5 morphologically indistinguishable, but male S. rovneri lack the secondary sexual characteristics of male S. ocreata (foreleg pigmentation and tufts) and differ in their courtship display. Females of both species distinguish between conspecific and heterospecific males on the basis of male courtship signals (Stratton & Uetz, 1981; Orr & Uetz unpubl.). However, males court in response to female silk (and pheromones) of either species equally (Roberts & Uetz, 2004). Thus, reproductive isolation is presumably maintained via female preference. Although these species are rapidly becoming model organisms for the study of multimodal communication and sexual selection, much is still unknown about their physiology, development, and behavioral sensitivities to environmental variables.

In the first two chapters of my dissertation I conducted research to investigate the extent to which mate preferences in S. ocreata and S. rovneri, are flexible and how experience can influence these preferences. Recent work in the areas of animal learning and the effects of social experience on arthropod behavior have demonstrated that behavioral flexibility in response to environmental conditions is more common than previously recognized (e.g., Wagner et al. 2001;

Hebets 2003) . These findings contradict historical assumptions that invertebrate animal behaviors represent inflexible, genetically pre-determined responses to stimuli. The third chapter focuses on the novel discovery that spiders (Schizocosa ocreata) may be able to modulate their apparent body condition (calculated by traditional body condition measures) when faced with short-term food and/or water deprivation. This suggests that current measures of body condition may not provide an accurate estimate of current body condition. The final element of my dissertation research, presented as an appendix here, identifies a novel, and more powerful assay for predicting mate outcome and assessing female mate choice in wolf spiders, based on female

6 receptive behaviors in behavioral trials that have variable durations (e.g., multiple data sets from different studies, studies that involve mating trials, etc.). The following is a brief outline of the dissertation research presented in the following chapters.

Chapter 2: “Exposure to multisensory cues affects female mate preferences in wolf spiders” (In Press: Rutledge, J.M., et al., Exposure to multiple sensory cues as a juvenile affects adult female mate preferences in wolf spiders, Animal Behaviour (2010), doi:10.1016/j.anbehav.2010.05.027)

This chapter examines unintended consequences of using artificial coloring methods

(e.g., nail polish) to modify phenotypes of research subjects, specifically adult male wolf spiders.

Artificial manipulations of physical phenotypes are used throughout behavioral research often in an attempt to test the effects of novel traits or to create phenotypes that mimic phenotypic extremes found in a natural population. However, data are rarely inspected to determine whether an artificial manipulation affected research subjects in an unintended manner. Results show that adult female behavior is indeed influenced by prior exposure to males while juvenile, but not in the manner shown in earlier studies (e.g., Hebets 2003; Hebets & Vink 2007). Females exposed as juveniles to males that had been visually modified with nail polish (forelegs were painted black) were significantly less likely to mate with males possessing the visual phenotype to which they had be previously exposed. Although, females exposed as juveniles to males that had minimal visual modifications but were chemically different from unmodified spiders (because clear nail polish was added to their forelegs) were slightly less likely to mate with males possessing familiar male phenotypes, most astoundingly 100% of these females that were paired with males that had been visually modified (black painted forelegs) mated. Ultimately, this study suggests that using nail polish as a phenotype modifier may have unanticipated effects on results on the behavior of spiders, calling into question its value as a phenotype modifier for future behavioral studies.

Chapter 3: “Juvenile exposure to adult heterospecific and conspecific males has limited

effect on adult female mate preferences in closely related sympatric wolf spider species”

This chapter examines the influence of juvenile experience on adult mate choice in two closely-related sibling species (S. ocreata and S. rovneri) that are reproductively isolated by behavioral mechanisms alone. Studies of other congeneric species within the same clade suggest that experience as juveniles leads to changes mate preference as adults (Hebets & Vink 2007).

My results show that in contrast to these recent studies, mate preference is largely fixed and species-specific, and only minimally affected by prior exposure.

Chapter 4: “Testing the limits of traditional measures of body condition for spiders”

Body condition indices based on morphology are traditionally calculated for spiders as well as other animals using a residual index from a regression of weight or abdomen volume

(which vary with feeding rate) against cephalothorax width (which is invariant over time within each instar). Studies of food deprivation and satiation have suggested these indices reflect feeding experience (e.g., Uetz et al. 2002). However, for spiders, water consumption is critical because arachnids use hydrostatic pressure for locomotion and other behaviors. In a short-term diet manipulation study, spiders given only food (without access to water) or only water (without access to food) did not significantly differ from each other in terms of mass or body condition for up to five days, suggesting spiders may regulate weight and apparent condition. Further analysis may reveal additional or alternative trends.

Appendix I: “Behavioral assays for predicting mating success in two species of wolf spiders: rates of general behaviors, male signals, and female receptivity”

In studies of mate preference, the ultimate measure of female choice is mating. However, in many studies (e.g., experimental cue isolation, video and seismic playback), a “stand-in” value that is highly predictive of mating probability is needed. This novel measure improves upon the

“composite receptivity” standard that is used throughout wolf spider sexual selection research.

References.

Burley, N. 1986. Comparison of the band-color preferences of 2 species of estrildid finches.

Animal Behaviour, 34: 1732-1741.

Borror, D.J., C.A. Triplehorn, & N.F. Johnson. 1989. Chapter 3: The anatomy, physiology, and

development of insects. Pp. 24-73, In: An Introduction to the Study of Insects, 6th Ed.

Thomson Learning, Australia.

Darwin, C.R. 1871. The Descent of Man, and Selection in Relation to Sex. John Murray, London.

Evans, J.E., Cuthill, I.C. & Bennett, A.T.D. 2006. The effect of flicker from fluorescent lights on

mate choice in captive birds. Animal Behaviour, 72: 393-400.

Hebets, E.A. 2003. Sub-adult experience influences adult mate choice in an arthropod: exposed

female wolf spiders prefer males of a familiar phenotype. Proceedings of the National

Academy of Science 100: 13390-13395.

Hebets, E.A. & Vink, C.J. 2007. Experience leads to preference: experienced females prefer

brush-legged males in a population of syntopic wolf spiders. Behavioral Ecology, 18:

1010-1020.

Jennions, M.D. 1998. The effect of leg band symmetry on female-male association in zebra

finches. Animal Behaviour, 55: 61-67.

Major, R.E. & Kendal, C.E. 1996. The contribution of artificial nest experiments to

understanding avian reproductive success: a review of methods and conclusions. Ibis,

138: 298-307.

Moore, R.P. & Robinson, W.D. 2004. Aritificial bird nests, external validity, and bias in

ecological field studies. Ecology, 85: 1562-1567.

Roberts, J.A. and G.W. Uetz. 2004. Chemical signaling in a wolf spider: a test of ethospecies

discrimination. Journal of Chemical Ecology 30: 1271-1284.

Sherwin, C.M. & Glen, E.F. 2003. Cage colour preferences and effects of home cage colour on

anxiety in laboratory mice. Animal Behaviour, 66: 1085-1092.

Stratton, G.E. and G.W. Uetz. 1981. Acoustic communication and reproductive isolation in two

species of wolf spiders. Science, New Series 214: 575-577.

Stratton, G.E. and G.W. Uetz. 1983. Communication via substratum-coupled stridulation and

reproductive isolation in wolf spiders (Araneae: Lycosidae). Animal Behaviour 31: 164-

172.

Uetz, G.W., R. Papke, and Beril Kilinc. 2002. Influence of feeding regime on body size, body

condition and a male secondary sexual character in Schizocosa ocreata wolf spiders

(Araneae, Lycosidae): condition-dependence in a visual signaling trait. Journal of

Arachnology 30: 461-469.

Wagner, W.E., M.R. Smeds, and D.D. Wiegmann. 2001. Experience affects female responses to

male song in the variable field cricket Gyrllus lineaticeps (Orthoptera, Gryllidae).

Ethology 107: 769-776.

Chapter 2: Exposure to multiple sensory cues as a juvenile affects adult female mate

preferences in wolf spiders

(In Press: Rutledge, J.M., et al., Exposure to multiple sensory cues as a juvenile affects adult

female mate preferences in wolf spiders, Animal Behaviour (2010),

doi:10.1016/j.anbehav.2010.05.027)

Jenai M. Rutledge, Amber Miller & George W. Uetz

Department of Biological Sciences

University of Cincinnati

614 Rieveschl Hall

Cincinnati, OH USA 45221-0006

Corresponding author – [email protected]; Tel: 513-556-9753, Fax: 513-556-5299

Abstract.

Experience is known to influence female mating in a variety of vertebrate animals; however, the effects of experience on mate choice have been less well-studied in invertebrates.

In a series of recent studies conducted on spiders, it was shown that females develop preferences for novel, artificially modified male phenotypes when exposed to them as juveniles. However, because Schizocosa wolf spiders are known to respond to multiple sensory cues in mate choice, concerns have been raised about the use of nail polish to modify male visual phenotypes. Here we attempted to repeat the earlier experiment, but address effects of chemical vs. visual learning separately, using the wolf spider species Schizocosa rovneri. Results indicate that exposure to novel visual/chemical male phenotypes influenced adult female mate preference for visual and to some extent chemical phenotypes, but in this case, females avoided familiar male phenotypes and preferred those to which they had not been exposed. Our results suggest that female mate preferences may be based on more factors than previously recognized, and that experience may reinforce behavioural isolation of species by increasing avoidance of mating with visually or chemically distinct male phenotypes.

Keywords: mate choice, plasticity, subadult experience, experimental design, multimodal communication

Introduction

Much research has shown that male secondary sexual traits (physical or behavioural,e.g., courtship) reflect female mate preferences and are thus reinforced by sexual selection

(Andersson 1994; Wiens 2001; Shuster & Wade 2003). However, the extent to which female preferences direct the evolution of male traits may ultimately depend on the level of plasticity in female mate preferences (Wagner 1998; Saetre 2000). Age, predation risk, hunger, and other environmental factors have been shown to influence female preferences (Wagner 1998; Kodric-

Brown & Nicoletto 2001; Coleman et al. 2004; Norton & Uetz 2005; Uetz & Norton 2007;

Hebets et al. 2008; Fox & Moya-Laraño 2009). In addition, social experience during both adult and juvenile stages has been shown to affect female mate choice in some vertebrates through mate-choice copying (White & Galef 2000; Witte & Ryan 2002; Walling et al. 2008), “previous male effects”/sequential mate choice (Bakker & Milinski 1991; Milinski & Bakker 1992; Hovi &

Ratti 1994; Collins 1995; Kavaliers et al. 2003), sexual imprinting (Owen et al. 1999; Verzijden

& ten Cate 2007) or basic familiarity with potential mates/mate phenotypes (Zajitschek et al.

2006; Tudor & Morris 2009). Although much less abundant, there is also growing evidence that experience can influence female mate preferences in invertebrate animals as well (Reid &

Stamps 1997; Wagner et al. 2001; Hebets 2003; Johnson 2005; Hebets & Vink 2007; Ödeen &

Moray 2008; Wilder & Rypstra 2008), challenging previous assumptions that invertebrate mate preferences are largely predefined by an animal’s genetic make-up.

In many species, asynchrony of male-female maturation can result in encounters between mature individuals of one sex and immatures of the other. As a consequence, juveniles of the sex that matures later are likely to gain experience with potential mates prior to sexual maturity,

13 which may influence their future mate preferences (Hebets 2003; Hebets & Vink 2007; Wilder &

Rypstra 2008). In a series of studies conducted on wolf spiders of the genus Schizocosa (Hebets

2003; Hebets & Vink 2007), females were shown to develop preferences for artificial male phenotypes to which they were exposed as juveniles. In one study (Hebets 2003), adult female S. uetzi preferred familiar males with modified phenotypes (i.e., brown or black forelegs, which are used in a visual courtship display) to which they had been exposed as juveniles, and behaved aggressively toward novel phenotype males. This was one of the first studies to provide evidence of plasticity in female mate preferences in response to juvenile experience for an invertebrate, and thus challenged the historically common assumption that invertebrate behaviour is inflexible and genetically controlled (Hebets 2003). However, the external validity of this study is in question because of methods used to create the male phenotypes. Hebets (2003; Hebets et al.

2006) modified male visual phenotypes by painting forelegs with nail polish – a substance known to contain and release a number of volatile compounds – which raises the question whether observed effects could be a learning response based on chemical cues alone, or a combination of chemical and visual cues (results were interpreted as if only visual cues were provided to the female spiders).

Many wolf spiders, including several species of the genus Schizocosa, use multimodal communication, and have been shown to respond to both individual modes and combinations of sensory cues (Scheffer et al. 1996; Hebets & Uetz 1999; Uetz 2000; Rypstra et al. 2009). It is well-known that chemical cues (pheromones) contained within female spider silk induce male courtship, and are used by males of this and many other spider species to locate potential mates

(Rovner 1977; Uetz & Denterlein 1979; Stratton & Uetz 1981; Foelix 1996; Roberts & Uetz

2004; Gaskett 2005; Rypstra et al. 2009). However, little is known about the extent to which female spiders attend to chemical cues (contact or olfactory) within their environment (but see

Persons & Uetz 1996; Punzo & Kukoyi 1997), and how they might use chemical cues to assess mates. In this study, we address this question by repeating an earlier experiment (Hebets 2003) with a different Schizocosa species. We sought to not only examine the possible effects of subadult exposure to artificial visual male phenotypes, but also to determine how exposure to different types of cues (visual versus chemical cues provided by nail polish) may impact adult female mate preferences.

Methods

Study Species.

We chose the wolf spider, Schizocosa rovneri Uetz and Dondale, a member of a well- studied group of spiders (the S. ocreata clade within the genus Schizocosa) for our study rather than S. uetzi (as used by Hebets 2003) because of their unique natural history, ease of availability and phylogenetic position (Stratton 2005). Most S. rovneri are similar in size to other members of the genus Schizocosa (males 6.48-8.07 mm; females 6.01-7.95 mm), but males lack prominent decorations of the forelegs seen in a closely related, sympatric sibling species, S. ocreata, whose male secondary sexual characteristics (leg pigmentation and tufts of bristles) are well-studied

(Dondale & Redner 1978; Uetz & Dondale 1979; Uetz 2000; Stratton 2005). Male S. rovneri also have primarily unimodal courtship (seismic signals) rather than the multimodal courtship

(visual and seismic signals) of S. ocreata and S. uetzi. Even though S. rovneri courtship is primarily unimodal, females of this species are sensitive to visual cues when assessing a potential mate, as female S. rovneri showed a preference for males with foreleg tufts when only

15 visual courtship cues were presented in video playback studies (McClintock & Uetz 1996).

Female S. rovneri are morphologically indistinguishable from female S. ocreata and produce similar chemical signals, and males of both species court in response to female silk (and pheromones) of either species equally (Roberts & Uetz 2004). On the other hand, males are morphologically distinct upon maturation, and females of these species distinguish between conspecifics and heterospecifics on the basis of male courtship displays and male traits (Uetz &

Denterlein 1979; Stratton & Uetz 1981, 1983; Uetz 2000). When raised in isolation, both female

S. ocreata and S. rovneri, as well as other members of this clade (including S. uetzi), exhibit mate recognition based on species-specific male traits (secondary sexual characteristics and courtship displays) (Stratton & Uetz 1981, 1983, 1986; Hebets & Uetz 2000; Uetz 2000; Hebets et al.

2006). Thus, reproductive isolation is presumably maintained via female preference.

In the field, it is highly probable that females are exposed to male courtship multiple times prior to maturity of conspecifics and heterospecifics alike, as males mature before females and occur in high densities. Where S. rovneri and S. ocreata co-occur, S. ocreata tends to mature slightly ahead of S. rovneri. As females reach maturity over the 6-8 week breeding season, males are engaged in near constant courtship activity, stimulated by pheromones in silk "draglines" laid by adult females as they move throughout the leaf litter habitat. As a consequence, it is possible that juvenile exposure might play a role in development of mating preferences, as suggested by

Hebets (2003). Additionally, because female S. rovneri may encounter more heterospecifics than conspecifics as juveniles, it is possible that early experience as juveniles with adult males may help these females avoid heterospecific males as adults. On the other hand, interspecies S. ocreata/S. rovneri hybrids are occasionally found, indicating that exposure might lead to

16 interbreeding, as observed by Hebets & Vink (2007). Given the potential for exposure to multiple males as juveniles, and the role played by chemical, visual and seismic communication among species in the S. ocreata clade (Hebets & Uetz 1999, 2000; Uetz 2000), S. rovneri would seem to be an excellent choice to repeat Hebets’ (2003) study.

Collection and Housing.

Immature male and female Schizocosa rovneri were collected in spring 2008 during daylight hours from Conrad Park, adjacent to the Ohio River floodplain in Boone County, KY

(39° 7'44.22"N; 84°43'5.50"W), and returned to the laboratory. All specimens for this research were collected with the permission of the Boone County Parks and Recreation Service (Permit not required). Spiders were housed individually in opaque round plastic containers (360mL deli- dish; 10cm diameter) and maintained under identical, controlled conditions (temperature 23-

25°C and stable relative humidity of 65-75%) and a 13/11h light/dark cycle to simulate late spring/summer light conditions. Spiders had continuous access to water via a cotton wick suspended in water held in a separate container beneath the housing container and were fed 2-3 ten-day old crickets twice a week. Females were also fed one cricket the day before a trial in addition to the regularly scheduled feedings (if on different days) in an effort to minimize hunger-related female aggression towards males.

Experimental Methods.

To investigate the effects of juvenile experience with novel visual or chemical adult male phenotypes on adult female mate preferences, a two-stage experiment was conducted. In the first stage of the experiment, sub-adult females were exposed multiply to courting adult conspecific

17 males whose forelegs had been painted with either black or clear nail polish. In the second stage, females were allowed to mature and were then measured for receptivity and/or willingness to mate with males possessing familiar or unfamiliar visual or chemical phenotypes. All trials

(juvenile and adult) took place in round, clear plastic containers (15.4 cm diam., 6.5 cm ht) lined with white filter paper.

Male Phenotype Manipulation

Two unique male phenotypes were created by painting the male tibia and patella with either black or clear nail polish (CoverGirl© Boundless ColorTM “Midnight Magic-610”; and

CoverGirl © Boundless ColorTM “Blown Glass-440”, respectively). Although every effort was made to mirror the methods utilized by Hebets (2003), the line of nail polishes used for that study had been discontinued at the time of our study; consequently we used nail polishes made by the same manufacturer and brand (i.e., Cover Girl©), but not the same type. Black was chosen as a phenotype because it mimics the pigmented forelegs of males of the closely related sympatric sibling species, S. ocreata, which has a range and microhabitat that overlaps that of S. rovneri and is thus similar to male phenotypes S. rovneri females are likely to encounter in the wild (McClintock & Uetz 1996). Clear nail polish was chosen as a means of modifying males’

“chemical” phenotypes without dramatically modifying their visual phenotypes. For adult mating trials, two additional male phenotypes - containing the chemical phenotypes of each treatment but without the visual cues - were created by painting a small black or clear dot out of view on the center on the dorsal surface of the cephalothorax of previously unmodified adult males.

For painting, males were anesthetized using CO2. Males were anesthetized directly prior to painting (for no longer than a minute prior). It generally took spiders between 1-2 minutes to respond to the CO2. As soon as males were immobile, they were removed from their home containers for painting. After painting, males were returned to their home containers and allowed to recover in without lids in a fume hood in part to ensure fast drying of the nail polish and to dissipate the initial fumes from the nail polish as it dried. Painting took no more than 30 seconds to one minute per male, and males revived and generally began grooming within 3 minutes of painting. With the exception of two males, the spiders did not groom off the nail polish applied to the forelegs or the cephalothorax. In an effort to standardize the amount of nail polish used on each male, five males of each phenotype were randomly selected and weighed before and after painting. No significant differences in weight were detected between the four male phenotypes, so it was assumed that amount of nail polish was roughly equal across phenotype treatments. All males were allowed to recover for 24-48 hours before they were used in either experience or mating trials,and observed prior to trials to ensure that painting did not affect mobility.

Subsequent analyses of male behaviour showed no differences in rate of male courtship (ln- transformed bouts/min) as a result of phenotype modification (One-way ANOVA: F3, 149

=0.0532, P=0.9837).

Juvenile Female Experience Trials

Penultimate instar (one molt prior to maturity) females were randomly assigned to one of two juvenile exposure groups or to the control group (unexposed females raised in isolation to maturity; N=78). Females in exposure groups were paired with a courting adult male conspecific with either black (N=40) or clear (N=38) forelegs (depending on the group to which they had

19 been assigned) in a round, clear plastic arena lined with filter paper and were allowed to interact for 30 minutes every other day until they reached maturity. In an effort to control for individual male effects on female mate preferences and to mimic natural conditions where females are likely to encounter many different males, females were never paired with the same male twice.

Opaque barriers were placed around each arena to minimize external visual distractions.

Containers were cleaned thoroughly with 70% EtOH after each trial and allowed to air dry to eliminate any residual chemical cues. Although generally males do not mount females without the female indicating a certain degree of receptivity, on a few rare occasions males did attempt to mount juvenile females during the 30 minute exposure trial. Although no successful mounts were directly observed, any successful mounts were not likely to result in intromission as the openings to female genitalia remain sealed until maturation. Of the 78 females that gained experience as juveniles with adult courting males having either black or clear painted forelegs, 72 gained experience with two or more (up to nine) different males prior to maturation.

Adult male S. rovneri display courtship in response to pheromones contained within adult female silk (in the presence or absence of a visual female stimulus) but do not court juvenile female silk (Uetz & Denterlein 1979; Stratton & Uetz 1981, 1983, 1986; Roberts & Uetz 2005).

Therefore, to elicit male courtship during juvenile exposure trials, adult females were placed in the trial arenas overnight (~12 hours) prior to the trial and were allowed to deposit silk (and the pheromones therein) onto the filter paper linings. Males were used multiply for experience trials, but no female was paired with the same male twice. In addition, to ensure that males did not become fatigued from multiple usages, males were used in a maximum of one trial per day.

Adult Mating Trials

As adults, females were assigned at random to be paired with a male possessing one of four phenotypes: black forelegs (N=41); clear forelegs (N=36); black “hidden” dot on cephalothorax (N=37); clear “hidden” dot on cephalothorax (N=42). Females were tested as adults with an adult male only once. Because in related species of wolf spiders, female receptivity peaks 2-3 weeks post-maturity (Uetz 2000; Norton & Uetz 2005), adult trials began

7-21 days following each female’s final molt. To induce male courtship, females were placed into their trial arenas the night before the trial to allow them to deposit silk (and pheromones) onto the filter paper lining. In the morning, females were returned to their housing containers for at least 2 hours to allow them to rehydrate. Males were placed in the trial arena 1-2 minutes before the female to allow the male to begin courtship prior to the introduction of the female.

Trials lasted for 10 minutes or until copulation occurred. All trials were videotaped for later analysis of male and female behaviour.

Female S. rovneri exhibit stereotypic behaviours in response to male courtship, which indicate receptivity and/or willingness to mate (Uetz & Denterlein 1979; Scheffer et al. 1996;

Delaney et al. 2007) including: slow pivot (90to 180 slow turn(s)), tandem leg extend (forward extension of front two pairs (LI, LII and RI, RII) of legs while tilting body towards the substratum with abdomen slightly lifted), and settle (lowering of the cephalothorax to the substratum while keeping abdomen slightly lifted). Previous studies have examined the relationship between receptive displays and mating success in S. rovneri, and found that 90% of females who displayed more than one of the receptive behaviours described above mated, while very few mated without performing any receptive behaviours (Scheffer et al. 1996; Delaney M.S.

Thesis 1997). When unreceptive, females lunge at and may cannibalize the male. The number of receptive displays and aggressive behaviours exhibited towards the male performed by each female were counted from video using Noldus Observer 5.0TM. Female receptive displays were summed and a composite receptivity score (sum of receptive displays minus lunges) was calculated (as in Uetz & Roberts 2002; Uetz & Norton 2007). Because trials in which mating occurred were shorter than trials in which mating did not occur, rates of receptive displays and composite receptivity were adjusted (composite receptivity score/trial duration) and used in the final analyses (referred to as “composite receptivity rates” from here on). To improve normality, composite receptivity rates were log-transformed for analysis. Log-composite receptivity rate was a reliable predictor of mating outcome (Logistic Regression: X2=47.709, p<0.0001). In addition, latency to first contact, mating outcome (yes/no), and latency to mate (when applicable) were measured. To evaluate and account for the potential effect of male behaviour on female response, the proportion of the trial that males spent courting as well as number of bouts of courtship were also recorded. Trials in which males courted for less than 90 seconds of the total ten-minute trial time and in which mating did not occur were excluded from statistical analyses

(N=3 trials). All statistical analyses were performed using JMP IN, ver. 5.1.2. statistical software.

A chemical analysis of these two nail polish compounds (clear and black) along with another brown-coloured nail polish (CoverGirl© Boundless ColorTM - “Perfect Penny-540”) was conducted at the University of Cincinnati Mass Spectrophotometry facility to ascertain if chemical differences could be discerned between aromatic compounds released from the nail polishes after drying. Because the previous study by Hebets (2003) used two different pigmented

22 nail polishes (brown and black), analysis of a third nail polish (brown) in addition to the black and clear nail polishes used in our study was conducted to verify that any differences observed between the nail polishes were not strictly the result of the presence/absence of pigment. These nail polishes were analysed using two methods both of which are headspace methods intended to capture the volatiles coming off of the nail polish (as would be happening during the trials) using a DB5-MS UI column (Agilent Technologies: length 30m; internal diameter 0.250; film

0.25μm). In the first analysis, the nail polishes were painted onto separate glass sample tubes and were allowed to dry for 24-hours. The samples were then purged with nitrogen prior to sealing and were sampled with a 100m Polydimethylsiloxane SPME fiber. In the second analysis, the nail polishes were painted onto glass sample tubes and were allowed to dry for 18 hours. These samples were dried under air to allow for the potential oxidation of aldehydes and similar chemicals, and were not purged with nitrogen before sampling. These samples were sampled with a 75m CarboxenTM-PDMS SPME fiber. For both analyses, extremely volatile chemicals that would likely be lost within an 18-24 hour drying period were not closely investigated.

Results

When fitted in a two-way ANOVA model, neither juvenile treatment nor adult treatment alone accounted for a significant portion of variation in composite receptivity rate; however, the interaction term between juvenile treatment and adult treatment on composite receptivity rate was significant (Table 1). Likewise, juvenile treatment alone and adult treatment alone did not affect mating outcome, but in combination had a significant effect on mating outcome depending on treatment group (Table 2).

Exposure as juveniles to males with unique visual and chemical phenotypes (created using nail polish) influenced adult female mate preferences for males possessing familiar visual/chemical (painted forelegs) and chemical only phenotypes (cephalothorax dot). Females exposed to males with black forelegs as juveniles were less receptive as adults to males with black forelegs than any other male phenotype (Oneway ANOVA: F3, 40 =8.428, P=0.0002 Fig.

2A). They were also less likely to mate with males with black forelegs than were females

2 exposed to the clear male phenotype or unexposed females (Likelihood Ratio: X 2, 41=13.876,

P=0.001, Fig. 3A). Conversely, females exposed as juveniles to males with clear (painted) forelegs did not show a preference for or discriminate against males with clear forelegs in adult trials relative to other male phenotypes, but were significantly more receptive (Oneway

ANOVA: F3,37=4.544, P=0.009, differences between treatments were derived from multiple comparisons of means via Tukey HSD analysis,Fig. 2B) and most willing to mate with males possessing black forelegs (10/10 trials resulted in mating Fig. 3B).

The effects of juvenile exposure to males with black forelegs appear to be primarily visual, as females exposed to males with black forelegs did not demonstrate a preference for or negative bias against males with the familiar chemical phenotype (black hidden). Females exposed to males with black forelegs as juveniles were were equally receptive to males with the black hidden, clear foreleg and clear hidden phenotypes (Fig. 3a); however, females exposed to the clear male phenotype were significantly less receptive towards males possessing the familiar chemical phenotype (“clear hidden”) than to males with the unfamiliar chemical phenotype

(Oneway ANOVA: Clear Hidden vs. Black Hidden: F1, 19=4.905, P=0.041). In addition, females exposed to males with clear (painted) forelegs as juveniles were significantly less receptive to the

“clear-hidden” (i.e., familiar chemical) male phenotype than females exposed to males with black forelegs (Oneway ANOVA: F2, 42=4.116, P=0.024; comparison of all pairs via Tukey-

Kramer HSD). Unexposed females did not show an innate preference for or aversion to any one of the four adult male phenotypes (Oneway ANOVA: F3,76=0.7529, P=0.554 Fig. 3c) and were

2 equally willing to mate with all male phenotypes ( Likelihood Ratio: X 3, 76=2.294, P=0.5136 Fig.

3).

Overall female aggression (measured as number of lunges by females towards males) was low across all mating trials and did not vary significantly as a result of juvenile or adult treatment

2 (Two-way ANOVA: F11, 153 =1.321, P=0.198) or amount of juvenile exposure (R 77=0.027,

2 P=0.150) . Female aggression did not predict mating outcome (Logistic Regression: X 1, 153

=0.263, P=0.608). Only four trials ended in female cannibalism of the male and each of the four cannibalism events occurred in different treatment groups. Amount of juvenile experience (i.e., number of times a female was paired with a male prior to maturity) did not have a significant

2 effect on mating outcome (Logistic Regression: X 1, 77=0.940, P=0.332) or adult female

2 preference (composite receptivity rate – R 77=.026, P=0.156) when considered across all exposed females.

Chemical analysis of the different nail polish types via Gas Chromatography Mass

Spectrophotometry verified that when analysed 18-24 hours after drying a number of the volatile chemicals released by each of the three nail polishes differed from each other (Table 3; Fig. 1).

Discussion

Our results confirm that juvenile experience can influence adult female mate preferences in wolf spiders, and that effects of experience on female preferences appear to vary depending on what females are exposed to, as was shown in previous studies using a related species (Hebets

2003; Hebets and Vink 2007). However, our findings contrast with the results of these earlier studies, as where effects of experience were detected, females in this study were less receptive overall and less likely to mate with familiar male visual phenotypes (i.e., phenotypes which they had seen as juveniles). Additionally, this study indicates that females may be sensitive to chemical cues associated with potential mates and learn to recognize and discriminate against chemical phenotypes in addition to visual phenotypes to which they have been exposed as juveniles. Therefore, our findings suggest that female mate preferences may be based on more factors than previously recognized.

While no inherent preference for any male phenotype was seen in unexposed adult females, females with experience exhibited distinct patterns of preference, depending on type of exposure. Females exposed to males with black forelegs as juveniles were less receptive toward that phenotype, and less likely to mate with them as adults. Moreover, females exposed to males with forelegs painted with clear nail polish – intended to modify the chemical phenotype rather than visual phenotype – were more receptive to and preferred to mate with (100% of trials) males with “unfamiliar” visual phenotypes (i.e., black forelegs vs. any other male phenotype). These females were also more receptive to males with the unfamiliar chemical phenotype alone (black hidden) than to males with the familiar chemical phenotype. When females were exposed to males with minimal visual modifications – i.e. forelegs painted with clear nail polish – chemical

26 cues may have had a more direct impact on adult female mate preferences. Females exposed to males with clear forelegs as juveniles were less receptive as adults towards males with the familiar chemical phenotype (‘clear hidden’ & ‘clear foreleg’) and were more receptive towards males with the novel chemical phenotype (‘black hidden’ & ‘black foreleg’). In contrast, females exposed to males with black forelegs did not demonstrate a preference/bias towards males with the familiar chemical phenotype alone (‘black hidden’). It would appear that female mate choice in S. rovneri may follow a hierarchy of sensory criteria, since females exposed to males with clear painted forelegs showed a preference for the novel visual and chemical phenotypes (black foreleg and black hidden), but females in the other treatment groups did not.

While these results would seem to conflict with the previous findings of Hebets (2003), there are some possible parallels between these data and data presented in a more recent study by

Hebets and Vink (2007). In what Hebets and Vink (2007) term a ‘syntopic/potentially interbreeding population’ of wolf spiders resembling Schizocosa ocreata and S. rovneri, they found that females exhibited a preference for males with darkly pigmented and tufted forelegs (S. ocreata-like morph) vs. males with unornamented forelegs (S. rovneri-like morph) regardless of the male phenotype females were exposed to as juveniles. Although females in our study exposed as juveniles to males with black painted forelegs discriminated against those same male phenotypes as adults, the opposite was true for females exposed to males with minimal foreleg modification (clear nail polish). Previous studies presenting virgin, unexposed females with video playback of digitally manipulated male phenotypes have provided evidence of a sensory bias for male leg ornamentation in females of some species in the S. ocreata clade (McClintock

& Uetz 1996; Hebets & Uetz 2000). Specifically, video playback studies suggest a bias in female

S. rovneri for the visual characters of male S. ocreata (i.e., black pigmented and tufted forelegs, leg-tapping and waving behaviour - McClintock & Uetz 1996) at least when seismic courtship cues were not provided. Interestingly, unexposed females in this study did not show a preference for males with black painted forelegs, suggesting perhaps that the bias found previously in this species may have more to do with the leg tufts themselves than leg pigmentation. Alternatively, the lack of preference for ornamented males when visual and seismic courtship cues are both present could reflect dominance of seismic signals over visual signals in this species (Uetz et al.

2009).

Contrasting results across treatment groups, i.e. opposite direction effects of exposure on mate choice seen in this study relative to that of Hebets (2003), might suggest that the observed effects arise more from exposure to the substance used to modify the phenotypes rather than the visual phenotypes themselves (intended to resemble naturally-occurring phenotypes). Perhaps the fact that we utilized different types of nail polish than were used in the original study by

Hebets (2003) explains these differences. If the aromatic compounds present in the original nail polishes were mostly attractive, a preference for (rather than aversion to) the nail polishes may have developed as a result. Unfortunately, because relatively little is known about chemical communication between potential mates in spiders, it is difficult to say to what extent these results may translate to mating decisions in the field. Under ecologically valid permutations of the same experimental methods using wolf spiders, i.e. exposing females to closely related heterospecific species, the effects of experience on mate preferences have been mixed (Hebets &

Vink 2007; Hebets 2007; Rutledge et al., in prep). Clearly there is some degree of plasticity in

28 female mate preferences, but it is not clear whether experience with natural chemical stimuli is able to elicit such effects or if responses are only observable when artificial cues are presented.

Our results suggest that juvenile female experience may reinforce behavioural species isolation, by increasing recognition of (and thereby avoidance of mating with) male phenotypes that are visually or chemically distinct from the species norm. The finding that experience can alter female mate preferences may suggest that certain aspects of female mate choice are frequency-dependent with a preference for rarer visual (and possibly chemical) phenotypes. In both S. rovneri and S. ocreata, males tend to mature before females, and it is likely that females encounter adult male conspecifics prior to maturity. However, phenology data collected for S. ocreata and S. rovneri populations in Ohio and Illinois show that male S. ocreata mature slightly before male S. rovneri in the field (Uetz & Denterlein 1979; Delaney M.S. Thesis 1997; G.W.

Uetz, unpubl.). Therefore, in areas where the two species overlap, the relative density of adult male conspecifics (S. rovneri) would remain fairly low early in the season while females are still juveniles, and exposure to male S. ocreata seems more likely. Discriminating against the adult male phenotype most abundantly encountered as a juvenile, or developing a preference for male phenotypes not previously encountered, may help females avoid making the costly mistake of mating with a heterospecific. This would be adaptive under cue-restricted conditions, such as those found in the complex leaf litter in which these two species often co-occur. Selection pressures acting to prevent crosses between the two species are presumably high, as females tend to mate only once (Norton & Uetz 2005), and hybrid mating between S. ocreata and S. rovneri yields viable but behaviourally sterile offspring (i.e., hybrid males are rejected by females of either species and hybrid females; and hybrid females do not mate with hybrids or males of

29 either species - Stratton & Uetz 1981, 1986). This may also explain why experience leads to the development of a negative bias for familiar phenotypes in S. rovneri but a positive preference in species studied previously (e.g., Hebets 2003; Hebets & Vink 2007).

Because scientists often work with animals whose natural history is not fully understood, it can be difficult to predict whether certain sensory stimuli will have effects on research subjects beyond those intended. Assumptions that artificial manipulations of natural conditions will yield data that reflect actual, biologically relevant effects are often untested and in many cases may be flawed. A classic example of this was highlighted by Burley et al. (1982) who observed that male zebra finches with red or pink leg bands (used to identify individuals in a captive population) were more likely to find a mate than males with blue or green leg bands. Male spiders use chemical cues to distinguish between mated vs. unmated females (Rypstra et al., 2009).

Furthermore, within the genus Schizocosa there is evidence that males are also sensitive to airborne olfactory cues produced by females (Tietjen, 1978; Tietjen & Rovner, 1982; Searcy et al., 1999; Schonewolf et al., 2006). Females possess the same putative chemosensory structures

(e.g., tarsal organs, ‘taste hairs’) as do males (Tietjen & Rovner, 1982), and in at least one species of wolf spider (Pardosa milvina), females were shown to exhibit predator avoidance behaviour in response to chemical cues present in the silk of a potential heterospecific spider predator (Persons et al. 2001; Persons & Rypstra 2001). Thus, it is probable that female wolf spiders, like males, are capable of detecting both olfactory and contact chemical cues and may use them in their decisions about potential mates. The importance of the earlier study by Hebets

(2003) lies in the demonstration of previously undetected/under-studied flexibility in invertebrate mate choice. However, results from our study suggest that experience may lead females to

30 develop preferences for novel phenotypes or discriminate against familiar phenotypes, depending on the modality of cues available at the time of exposure. As a consequence, our results would urge caution in using artificial manipulation of phenotypes with aromatic compounds like nail polish or paint in mate choice studies of invertebrates.

Acknowledgements

This work represents a portion of a thesis submitted by JMR in partial fulfillment of the requirements for the Ph. D. degree from the Department of Biological Sciences at the University of Cincinnati. This research was supported by grant IBN 0239164 from the National Science

Foundation (to GWU), the American Arachnological Society (JMR), the University of

Cincinnati Research Council (JMR), and the Wiemen/Wendel/Benedict Student Research Fund

(JMR). Thanks to Justin Allen, Jessica Diersing, Jessica Moore, Shira Gordon, Brian Moskalik,

Chris Hartzel, and Kristin Plott for helping to collect and maintain the spiders in the laboratory and for helping to paint spiders and set-up trials. Also, thanks to Larry Sallans, Ph.D., from the

UC GC Mass Spectrophotometry Facility for running the analyses of the nail polishes and for helping us with the interpretation of the results. And thanks to Eric Maurer and two anonymous reviewers for comments on the manuscript.

References

Andersson, M. 1994. Sexual Selection. Princeton: Princeton University Press.

Bakker, T. C. M. & Milinski, M. 1991. Sequential female choice and the previous mate effect in

sticklebacks. Behavioral Ecology and Sociobiology, 29, 205-210.

Burley, N. 1986. Comparison of the band-color preferences of 2 species of estrildid finches.

Animal Behaviour, 34, 1732-1741.

Burley, N. T., Kratzenberg, G. & Radman, P. 1982. Influence of colour banding on the

conspecific preferences of zebra finches. Animal Behaviour, 30, 444-455.

Coleman, S. W., Patricelli, G. L. & Borgia, G. 2004. Variable female preferences drive complex

male displays. Nature, 428, 742-745.

Collins, S. A. 1995. The effect of recent experience on female choice in zebra finches. Animal

Behaviour, 49, 579-486.

Delaney, K. J. 1997. Communication in the context of courtship and aggression in two species of

wolf spider (Araneae: Lycosidae). M. S. Thesis, University of Cincinnati.

Delaney, K. J., Roberts, J. A. & Uetz, G. W. 2007. Male signaling behavior and sexual selection

in a wolf spider (Araneae: Lycosidae): a test for dual functions. Behavioral Ecology and

Sociobiology, 62, 67-75.

Dondale, C. D. & Redner, J. H. 1978. Revision of the neartic wolf spider genus Schizocosa

(Araneida: Lycosidae). Canadian Entomologist 110, 143-181.

Foelix, R. F. 1996. Biology of spiders, 2nd Ed. New York: Oxford University Press.

Fox, C. W. & Moya-Laraño, J. 2009. Diet affects female mating behaviour in a seed-feeding

beetle. Physiological Entomology, 34, 370-378.

Gaskett, A. C. 2005. Spider sex pheromones: emission, reception, structures, and functions.

Biological Reviews, 82, 27-48.

Hebets, E. A. 2003. Subadult experience influences adult mate choice in an arthropod: exposed

female wolf spiders prefer males of a familiar phenotype. Proceedings of the National

Academy of Sciences, USA, 100, 13390-13395.

Hebets, E. A. 2007. Subadult female experience does not influence species recognition in the

wolf spider Schizocosa uetzi Stratton 1997. Journal of Arachnology, 35, 1-10.

Hebets, E. A. & Uetz, G. W. 1999. Female responses to isolated signals from multimodal male

courtship displays in the wolf spiders genus Schizocosa (Araneae: Lycosidae). Animal

Behaviour, 57, 865-872.

Hebets, E. A. & Uetz, G. W. 2000. Leg ornamentation and the efficacy of courtship display in

four species of wolf spider (Araneae: Lycosidae). Behavioral Ecology and Sociobiology,

47, 280-286.

Hebets, E. A. & Vink, C. J. 2007. Experience leads to preference: experienced females prefer

brush-legged males in a population of syntopic wolf spiders. Behavioral Ecology, 18,

1010-1020.

Hebets, E. A., Cuasay, K. & Rivlin, P. K. 2006. The role of visual ornamentation in female

choice of a multimodal male courtship display. Ethology, 112, 1062-1070.

Hebets, E. A., Wesson, J. & Shamble, P. S. 2008. Diet influences mate choice selectivity in adult

female wolf spiders. Animal Behaviour, 76, 355-363.

Hovi, M. & Ratti, O. 1994. Mate sampling and assessment procedures in female pied flycatcher

(Ficedula hypoleuca). Ethology, 96, 127-137.

Johnson, C. 2005. Cohabitation of juvenile females with mature males promotes sexual

cannibalism in fishing spiders. Behavioral Ecology, 16, 269-273.

Kavaliers, M., Colwell, D. D., Braun, W. J. & Choleris, E. 2003. Brief exposure to the odour of a

parasitized male alters the subsequent mate odour responses of female mice. Animal

Behaviour, 65, 59-68.

Kodric-Brown, A. & Nicoletto, P. F. 2001. Age and experience affect female mate choice in the

guppy (Poecilia reticulate). American Naturalist, 157, 316-323.

McClintock, W. J. & Uetz, G. W. 1996. Female choice and pre-existing bias: visual cues during

courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour, 52,

167-181.

Milinski, M. & Bakker, T. C. 1992. Costs influence sequential mate choice in sticklebacks,

Gasterosteus aculeatus. Proceedings of the Royal Society of London, Series B, 250, 229-

233.

Norton, S. & Uetz, G. W. 2005. Mating frequency in Schizocosa ocreata (Hentz) wolf spiders:

evidence for a mating system with female monogamy and male polygamy. Journal of

Arachnology, 33, 16-24.

Ödeen, A. & Moray, C. M. 2008. Drosophila melanogaster virgins are more likely to mate with

strangers than familiar flies. Naturwissenschaften, 95, 253-256.

Owen, I. P. F., Rowe, C. & Thomas, A. L. R. 1999. Sexual selection, speciation and imprinting:

separating the sheep from the goats. Trends in Ecology and Evolution, 14, 131-132.

Persons, M. H. & 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.

Persons, M. H. & Uetz, G. W. 1996. Wolf spiders vary patch residence time in the presence of

chemical cues from prey (Araneae, Lycosidae). Journal of Arachnology, 24, 76-79.

Persons, M. H., Walker, S. E., Rypstra, A. L. & Marshall, S. D. 2001. Wolf spider predator

avoidance tactics and survival in the presence of diet-associated predator cues (Araneae:

Lycosidae). Animal Behavior, 61, 43-51.

Punzo, F. & Kukoyi, O. 1997. The effects of prey chemical cues on patch residence time in the

wolf spider Trochosa parthenus (Chamberlin) (Lycosidae) and the lynx spider Oxyopes

salticus Hentz (Oxyopidae). Bulletin of the British Arachnology Society, 10, 323-326.

Reid, M. L. & Stamps, J. A. 1997. Female mate choice tactics in a resource-based mating

system: field test of alternative models. The American Naturalist, 150, 98-121.

Roberts, J. A. & Uetz, G. W. 2005. Information content of female chemical signals in the wolf

spider, Schizocosa ocreata: male discrimination of reproductive state and receptivity.

Animal Behaviour, 70, 217-223.

Roberts, J. A. & Uetz, G. W. 2004. Chemical signaling in a wolf spider: a test of ethospecies

discrimination. Journal of Chemical Ecology, 30, 1271-1284.

Rovner, J. 1977. Spider behavior. In: International Encyclopedia of Psychiatry, Psychology,

Psychoanalysis, and Neurology. pp. 419-423. New York: Aesculapius Publishers, Inc.

Rypstra, A. L., Schlosser, A. M., Sutton, P. L., & Persons, M. H. 2009. Multimodal signaling:

the relative importance of chemical and visual cues from females to the behaviour of

male wolf spiders (Lycosidae). Animal Behaviour, 77, 937-947.

Saetre, G. P. 2000. Sexual signals and speciation. In: Animal Signals: Signaling and Signal

Design in Animal Communication (Ed. by Espmark, Y., Amundsen, T. & Rosenquist,

G.), pp. 237-258. Trondheim: Tapir Academic Press.

Scheffer, S. J., Uetz, G. W. & Stratton, G. E. 1996. Sexual selection, male morphology, and the

efficacy of courtship signaling in two wolf spiders (Araneae: Lycosidae). Behavioral

Ecology and Sociobiology, 38, 17-23.

Schonewolf, K. W., Bell, R., Rypstra, A. L. & Persons, M. H. 2006. Field evidence of an

airborne enemy-avoidance kairomone in wolf spiders. Journal of Chemical Ecology, 32,

1565-1576.

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.

Shuster, S. M. & Wade, M. J. 2003. Mating Systems and Strategies. Princeton: Princeton

University Press.

Stratton, G. E. 2005. Evolution of ornamentation and courtship behavior in Schizocosa: insights

from a phylogeny based on morphology (Araneae: Lycosidae). The Journal of

Arachnology, 33, 347-376.

Stratton, G. E. & Uetz, G. W. 1981. Acoustic communication and reproductive isolation in two

species of wolf spiders. Science, 214, 575-577.

Stratton, G. E. & Uetz, G. W. 1983. Communication via substratum-coupled stridulation and

reproductive isolation in wolf spiders (Araneae: Lycosidae). Animal Behaviour, 31, 164-

172.

Stratton, G. E. & Uetz, G. W. 1986. The inheritance of courtship behavior and its role as a

reproductive isolating mechanism in two species of Schizocosa wolf spiders

(Araneae:Lycosidae). Evolution, 40, 129-141.

Tietjen, W. J. 1978. Tests for olfactory communication in four species of wolf spiders (Araneae,

Lycosidae). Journal of Arachnology, 6, 197-206.

Tietjen, W. J. & Rovner, J. S. 1982. Chemical communication in lycosids and other spiders. In:

Spider Communication: Mechanisms and Ecological Significance (Ed. by P. N. Witt & J.

S. Rovner), pp. 249–279. Princeton: Princeton University Press.

Tudor, M. S. & Morris, M. R. 2009. Experience plays a role in female preferences for symmetry

in the sword tail fish Xiphophorus malinche. Ethology, 115, 812-822.

Uetz, G. W. 2000. Signals and multi-modal signaling in spider communication. In: Animal

Signals: Signaling and Signal Design in Animal Communication. (Ed. by Y. Espmark, T.

Amundsen, & G. Rosenquist), pp. 243-261. Trondheim: Tapir Academic Press.

Uetz, G. W. & Denterlein, G. 1979. Courtship behavior, habitat and reproductive isolation in

Schizocosa rovneri Uetz & Dondale (Araneae: Lycosidae). Journal of Arachnology, 7,

121-128.

Uetz, G. W. & Dondale C. D. 1979. A new wolf spider in the genus Schizocosa (Araneae:

Lycosidae) from Illinois. Journal of Arachnology 7, 86-88.

Uetz, G. W. & Norton, S. 2007. Preference for male traits in female wolf spiders varies with the

choice of available males, female age and reproductive state. Behavioral Ecology and

Sociobiology, 61, 631-641.

Uetz, G. W. & Roberts, J. A. 2002. Multisensory cues and multimodal communication in

spiders: insights from video/audio playback studies. Brain, Behavior and Evolution, 59,

222-230.

Uetz, G. W., Roberts, J. A. & Taylor, P. W. 2009. Multimodal communication and mate choice

in wolf spiders: female response to multimodal versus unimodal signals. Animal

Behaviour, 78, 299-305.

Verzijden, M. N. & ten Cate, C. 2007. Early learning influences species assortative mating

preferences in Lake Victoria cichlid fish. Biology Letters, 3, 134-136.

Wagner, W. E. 1998. Measuring female mating preferences. Animal Behaviour, 55, 1029-1042.

Wagner, W. E., Smeds, M. R. & Wiegmann, D. D. 2001. Experience affects female responses to

male song in the variable field cricket Gyrllus lineaticeps (Orthoptera, Gryllidae).

Ethology, 107, 769-776

Walling, C. A., Royle, N. J., Lindström, J., & Metcalfe, N. B. 2008. Experience-induced

preference for short-sworded males in the green swordtail, Xiphophorus helleri. Animal

Behaviour, 76, 271-276.

White, D. J. & Galef, B. G. Jr. 2000. ‘Culture’ in quail: social influences on mate choices of

female Coturnix japonica. Animal Behaviour, 59, 975-979.

Wiens, J. J. 2001. Widespread loss of sexually selected traits: how the peacock lost its spots.

Trends in Ecology and Evolution, 16, 517-523.

Wilder, S & Rypstra, A. L. 2008. Prior encounters with the opposite sex affect male and female mating behavior in a wolf spider (Araneae, Lycosidae). Behavioral Ecology and Sociobiology,

62, 1813–1820.

Witte, K. & Ryan, M. J. 2002. Mate choice copying in the sailfin molly, Poecilia latipinna, in

the wild. Animal Behaviour, 63, 943-949.

Zajitschek, S. R. K., Evans, J. P. & Brooks, R. 2006. Independent effects of familiarity and

mating preferences for ornamental traits on mating decisions in guppies. Behavioral

Ecology, 17, 911-916.

Table 1. Results of 2-Way ANOVA for the effects of juvenile and adult treatment group on adult female composite receptivity rate toward courting adult males

df SS F p

Overall Model 11 7.765 3.308 0.0005*

Juvenile Treatment 2 0.883 2.069 0.1302

Adult Treatment 3 0.861 1.345 0.2624

Juvenile Treatment x Adult Treat 6 6.160 4.812 0.0002*

Table 2. Nominal Logistic Fit for effects of juvenile and adult treatment group on mate outcome

(Y/N)

df X2 P

Overall Model 11 20.318 0.0412*

Juvenile Treatment 2 2.421 0.2981

Adult Treatment 3 5.605 0.1325

Juvenile Treatment x 6 14.426 0.0252* Adult Treat

Table 3. Mass spectrometry results for three different colours of nail polish (black, brown, and clear). Strength of detected concentrations present in each nail polish relative to each other based on visual inspection of chromatograms (see Fig. 1 for example) are indicated by, +, ++, +++, or –

(barely present, present, strongly present, absent – respectively). R.T. stands for retention time and reflects the relative order in which each chemical was detected.

R.T. Component Black Brown Clear (minutes) 4.8 6-hydroxy-hexan-2-one ++ ++ -

5.0 Chloro-trifluromethylbenzene - ++ -

5.6 Xylene + + +++

6.9 2,6-dimethyl-heptan-2-one or isomer ++ - -

7.2 4,6-dimethyl-heptan-2-one or isomer ++ - -

7.6 Unspecified methacrylate ester ++ ++ -

8.0 Dichlorobenzene + + +++

8.6 Methyl-benzaldehyde + + +++

9.8 2-methyl benzenemethanol + + +++

10.2 Ethyl benzoate ++ + +++

11.2 2-methylbenzoic acid or isomer ++ + +++

12.5 Propyl benzoate ++ + +++

13.1 Butyl benzoate ++ + +++

Figure 1. Sample of the GC/MS chromatogram results from nail polish analysis. Numbers reported directly above each peak refers the retention time (RT, as reported in Table 3 – indicating at what point in the analysis the chemical was detected). Total Ion Current (TIC), reported in the upper right corner of each chromatogram, indicates scale in Daltons for which % saturation on the y-axis are calculated. Numbers and arrows indicate chemicals found to differ between the nail polish colors as reported in Table 3, including: (1) 6-hydroxy-hexan-2-one; (2)

Xylene; (3) 4,6-dimethyl-heptan-4-one or isomer; (4) 2,6-dimethyl-heptan-2-one; (5)

Methacrylate ester; (6) Dichlorobenzene; (7) Methylbenzaldehyde.

Clear

Black

Brown

Figure 2. Adult female receptivity to each of four possible male phenotypes (“black foreleg,”

“black hidden,” “clear foreleg,” “clear hidden”) based on type of juvenile experience. Error bars represent ± S.E. Letters on graph indicate significant differences based on Tukey-Kramer HSD test. (A) Females exposed to males with black-painted forelegs as juveniles; (B) females exposed to males with clear-painted forelegs as juveniles; (C) unexposed females.

Figure 3. Mating outcomes when paired with each of the four possible male phenotypes

(BLKFORE: “black foreleg,” BLKHID: “black hidden,” CLRFORE: “clear foreleg,” CLRHID:

“clear hidden”) based on juvenile experience. (A) Females exposed to males with black-painted forelegs as juveniles; (B) females exposed to males with clear-painted forelegs as juveniles; (C) unexposed females.

(A)JuvenileTreatment=Black N=10 1.0 N=10 N=12 0.8 N=10

0.6 Yes 0.4 Proportion No 0.2

0.0 BlackForeleg BlackHidden ClearForeleg ClearHidden AdultMalePhenotype

(B)JuvenileTreatment=Clear N=10

1.0 N=10 0.8 N=8 0.6 N=10 Yes 0.4 Proportion No 0.2

0.0 BlackForeleg BlackHidden ClearForeleg ClearHidden AdultMalePhenotype

(C)JuvenileTreatment=None N=18 1.0 N=22 N=18 N=20 0.8

0.6 Yes 0.4 Proportion No 0.2

0.0 BlackForeleg BlackHidden ClearForeleg ClearHidden AdultMalePhenotype

Chapter 3: Effects of Juvenile Experience on Adult Female Mating Preferences in Two

Closely Related Sympatric Wolf Spider Species

(to be submitted to Ethology)

Jenai M. Rutledge and George W. Uetz

Department of Biological Sciences

University of Cincinnati

614 Rieveschl Hall

Cincinnati, OH USA 45221-0006

Corresponding author – [email protected]; Tel: 513-556-9753, Fax: 513-556-5299

Abstract

Social experience is known to influence female mate preference in vertebrate animals, but such effects have not been well-studied for invertebrates. Earlier studies (Hebets 2003, 2007; Hebets

& Vink 2007) have documented previously unseen flexibility in female mate choice in the wolf spider genus Schizocosa as result of juvenile female experience with courting adult males. Here we investigate whether juvenile exposure to male courtship influences adult female mate recognition in the wolf spider Schizocosa ocreata and its sympatric sibling species S. rovneri.

Because these species overlap in range, contact between them is probable and interspecies hybrids are occasionally found in nature. Juvenile females were exposed multiply to conspecific or heterospecific male courtship. Upon maturing, exposed females were paired with an adult male of the same or different species to which they have been previously exposed, and were observed to determine receptivity and willingness to copulate. Although overall, results suggest that juvenile experience plays only a minor role (if any) in development of female mate recognition, some context-specific effects of experience were observed. In S. ocreata, the overall amount of juvenile experience (rather than type) influenced adult female receptivity and aggression toward heterospecific males. In S. rovneri, both type and amount of juvenile exposure had effects on female receptivity towards conspecific males.

Keywords: mate choice, social experience, behavioral plasticity, spider

Introduction

Recent interest in the mechanisms that lead to speciation has increasingly revealed that experience and learning can have surprisingly important roles in maintaining reproductive barriers between closely related species, especially in sympatric species in variable environments where morphological and ecological divergence is minimal (Magurran & Ramnarine 2004;

Verzijden & ten Cate 2007; Pfennig 2007; Kujtan & Dukas 2009). Selection pressure to maintain species barriers are usually considered to be intense, as mating between species often has costly fitness consequences (sterile/inviable offspring, reduced reproductive output, etc) (White 1978;

Otte & Endler 1989). In many animals, evolution of species recognition mechanisms appears to be driven by female preferences for male traits or courtship behaviors (Endler & Houde 1995;

Wagner 1998), reinforced by benefits of selectivity as well as costs of recognition errors

(Pfennig 1998). However, the extent to which female preferences influence the direction of evolution of male traits (and ultimately speciation) may depend on the level of plasticity in female choice behavior (Wagner, 1998; Saetre, 2000; Kodric-Brown & Nicoletto, 2001;

Coleman et al., 2004; Uetz & Norton, 2007). In particular, social experience during juvenile development, e.g., sexual imprinting, can influence mate preferences, and result in recognition errors at adulthood (Owen et al., 1999; Hebets 2003; Hebets & Vink 2007).

Although evidence that social experience can affect female mate preferences comes primarily from studies of vertebrate animals, invertebrate behaviors – including mate preference

- can be influenced by experience as well (e.g., Jackson & Wilcox, 1993; Reid & Stamps 1997;

Punzo, 2000, 2004; Wagner et al. 2001; Hoefler, 2002; Johnson 2005). Hebets (2003) found that adult females of the wolf spider Schizocosa uetzi Stratton mated more often with males

50 possessing phenotypes to which they were exposed as juveniles. Likewise, Hebets and Vink

(2007) found that juvenile experience influences adult female mate preference in a potentially interbreeding population of two wolf spider species. While these studies suggest that invertebrate mating preferences may be less genetically ‘hard-wired,’ than previously assumed

(Parri et al. 1997; Wagner et al. 2001; Hebets 2003), recent studies with related wolf spider species have demonstrated that certain aspects of mate recognition at the species level remain inflexible and are not influenced by experience (Hebets 2007).

Although speciation is most often attributed to geographic isolation, the occurrence of behavioral reproductive isolating mechanisms in closely-related sympatric species suggests that behavioral barriers are sufficient to restrict gene flow and result in speciation (Doherty &

Gerhardt 1984; Stratton & Uetz, 1981, 1983, 1986). In this study, we investigated how juvenile experience influences adult female mate recognition, using two well-studied sympatric wolf spider species within the genus Schizocosa (Araneae: Lycosidae). Female S. ocreata and S. rovneri are morphologically indistinguishable, however male S. rovneri lack the tufts of bristles on their forelegs that are characteristic of mature male S. ocreata. Courtship displays also differ dramatically between the two species. S. ocreata male courtship is multi-modal with visual (leg tapping and leg waving) and seismic components (substrate-borne vibration and/or stridulation); whereas, S. rovneri male courtship is primarily unimodal and is made up of patterned seismic vibrations/stridulation only. Females of both species distinguish between conspecific and heterospecific males on the basis of male courtship displays and male traits (Uetz & Denterlein

1979; Stratton & Uetz 1981, 1983; Uetz 2000). Males, however, court in response to female silk

(and pheromones) of either species equally even in the absence of a female spider (Roberts &

Uetz, 2004). Thus, reproductive isolation is presumably maintained via female mate choice preferences. Nonetheless, on rare occasions hybrids have been collected from the field, suggesting that a breakdown in behavioral species barriers does occur, perhaps due to constraints on sensory modes (e.g., limited vibration transmission or restricted visual line-of-sight) in complex litter environments (Scheffer et al., 1996; Uetz, 2000). The existence of hybrids in nature may also imply that female preference/mate recognition is not entirely genetically-based, and that exposure to males early in the mating season might influence female mate choice. Here we test the hypothesis that sub-adult exposure to male courtship of heterospecific vs. conspecific males influences adult mate recognition in female S. ocreata and S. rovneri.

Methods

Study Species

S. ocreata and S. rovneri are common ground-dwelling wolf spiders that occur in the leaf litter of deciduous forests in the eastern United States. These species are sympatric sibling species (Stratton 2005) but do not interbreed due to behavioral barriers; however, forced interspecies copulation yields viable hybrid offspring. However, these hybrids are behaviorally sterile; i.e., courtship behavior of male hybrids is a mix of the two speces courtship displays and is not preferred by females of either species, and female hybrids do not demonstrate receptivity to males of either species or to hybrid males (Stratton & Uetz, 1981, 1983, 1986). When raised in isolation, both female S. ocreata and S. rovneri exhibit mate recognition based on species- specific male traits (secondary sexual characteristics and courtship displays). Nonetheless, it is probable that in the field, females are exposed to male courtship (of both species, where species co-occur) multiple times prior to maturity. In both species, males begin to mature before females,

52 and occur in high densities. Because during the breeding season mature female ‘dragline’ silk, laid down by females as they move about, is quite common throughout the environment males are engaged in near constant courtship activity in response to chemical cues associated with the silk.

General Methods

Data for this study were collected in two stages. In 2005, we investigated how experience with conspecifics and heterospecifics affected female S. ocreata mate choice. S. ocreata were collected during spring as sub-adults from the Cincinnati Nature Center Rowe Woods (Clermont

Co., OH) where S. rovneri do not occur. In spring 2009, the study was repeated to examine the same set of questions for female S. rovneri. S. rovneri were reared in the laboratory from egg sacs produced by females collected as adults from the Ohio River flood plain at Sand Run Creek

(Boone Co., KY) during spring 2008. All spiders were housed individually in opaque plastic containers (10-cm diam. deli-dishes) and kept under a 13/11 h light/dark cycle at approximately

25º C, and constant relative humidity (between 65-75%). Spiders were fed one to two 10-day old crickets (approximately 0.6cm, Acheta domesticus) twice a week. To control for effects of hunger, female spiders were also fed one cricket the day before a trial in addition to regularly scheduled feedings (if on different days).

To examine how sub-adult male courtship influences adult female mate recognition and/or mate preferences, we conducted two-stage experiments in which females first gained experience with male courtship during their pre-adult, or penultimate life stage (one molt prior to maturity), and were then measured as adults for receptivity to male courtship. Females remain in

53 the penultimate stage for approximately two weeks during which time females were exposed multiple times to courting adult heterospecific or conspecific males. Once mature, we recorded female behavioral responses to an adult male of the same or different species to which they had been exposed as juveniles. As a control condition, a group of females were raised to maturity in isolation and tested at adulthood without prior exposure to males.

Juvenile Exposure

Upon reaching their penultimate instar (one molt prior to maturity), juvenile female S. ocreata (N=87) and S. rovneri (N=78) were randomly assigned to one of three exposure treatment groups: 1) conspecific male (S. ocreata females N=27; S. rovneri females N=23) 2) heterospecific male (S. ocreata females N=35; S. rovneri females N=26); and 3) control – no exposure (S. ocreata female N=25; S. rovneri female N=29). Although an effort to create equal sample sizes across treatment groups was initially made, both populations experienced parasitism and/or mortality leading to decreased and uneven sample sizes.

For S. ocreata female exposure trials, juvenile females were individually placed into an arena consisting of a transparent, plastic, open-bottom box (9.5 x 9.5 x 9.8 cm LxWxH) adjacent to an adult courting conspecific/heterospecific male (depending on treatment group) (Figure 1A) allowing females to gain experience with both substrate-borne seismic courtship cues (through the shared posterboard substrate) as well as visual cues. For S. rovneri exposure trials, juvenile females were placed in open-bottom clear plastic (acetate) cylinders (diameter: 6.4 cm, height:

5.7 cm), surrounded by a larger plastic cylinder (diameter: 15.2 cm, height: 7 cm) in which the male was placed (Figure 1B). The cylindrical apparatuses were used for the later experiments

54 because they kept the spiders better contained (spiders occasionally climbed the corners of the rectangular containers) and allowed males to circle females while courting, increasing total exposure. Males and females were kept physically isolated from one another during this phase of the experiment to control for variation in male chemical cues (cuticular or silk-borne). Juvenile females were exposed to male courtship for 30 minutes every other day until they matured, as in

Hebets (2003) experiment with S. uetzi (Figure 2).

Males were used multiply for exposure trials, but no female was paired twice with the same male. To ensure that multiple usages did not affect male courtship vigor, males were used for one exposure trial per day. To induce male courtship during the trial periods, silk (and pheromones therein) of a mature conspecific female (at least 10 days post-maturity; Uetz &

Norton 2007) were deposited on the male portion of the trial substrate overnight (~12 hours).

Males of both species court in response to the presence of adult female silk (and the pheromones contained therein), even in the absence of the female visual stimulus (Stratton & Uetz 1986).

Adult Mate Recognition Trials

In the second stage of the experiment, females from all three treatment groups were assigned at random to one of two adult treatment groups: 1) conspecific male (S.ocreata females

N=41; S. rovneri females N=40); 2) heterospecific male (S. ocreata females N=46; S. rovneri females N=38). Testing began 7-14 days following each female’s final molt (Norton & Uetz

2005; Uetz 2000). Based on their assigned treatment group, females were randomly paired with either an adult heterospecific or conspecific male. Female S. ocreata were placed in a rectangular test arena (19.5cm x 12.5cm - Figure 1B) and female S. rovneri were placed in the same

55 cylindrical apparatus used for juvenile exposure (Figure 3). During the first five minutes of every trial the male and female were separated by a clear barrier so that female receptivity could be measured without tactile and/or chemical stimuli from the male. At the end of five minutes the clear barrier was removed and interactions between the male and female were observed for an additional five minutes to determine whether the pairing would result in copulation. Trials were videotaped for later analysis of female behavior.

Females of both species perform identical stereotypic behaviors that indicate receptivity and/or willingness to copulate (Uetz & Denterlein 1979; Scheffer et al. 1996; Delaney MS

Thesis 1997; Uetz & Norton 2007) including a slow pivot (90-180 degree slow turn(s)), tandem leg extend (the extension of both pairs of legs I and II together anteriorly while lowering cephalothorax towards substrate and raising abdomen slightly), and settle (the lowering of cephalothorax to the substrate while keeping the abdomen slightly lifted). As females of both species behave aggressively towards males if unreceptive or if a mate is unsuitable, all cases of cannibalism and female aggression (lunging) towards the male were also recorded. A composite receptivity score (sum of receptive displays minus lunges—as in Uetz & Roberts 2002; Uetz &

Norton 2007) was also calculated for each female. Because total trial length varied depending on whether mating occurred or not, a composite receptivity rate was calculated (Composite receptivity / trial length in seconds) and used for analyses (as in Rutledge et al., in press).

Statistical Analyses

Data for the two species were analyzed separately. All composite receptivity rate data deviated significantly from a normal distribution. To improve normality, the data were log-

56 transformed in an effort to meet the assumptions of parametric statistics. Trials in which mating did not occur and males courted for less than 10% of the total trial time (1 minute) were excluded from analysis. In all, one trial from the S.ocreata data set and two trials from the S. rovneri data set were excluded for this reason. To test whether juvenile and/or adult treatments had an effect on female mate preferences, female receptivity data were analyzed using a 2-way ANOVA with composite receptivity rate (referred to as “female receptivity” from here on) as the dependent variable and juvenile treatment group and adult treatment group as independent variables.

Female receptive and aggressive behaviors that occurred prior to the removal of the clear barrier were analyzed separately from those that occurred following the removal of the barrier. In previous related studies, amount of juvenile experience was found to also influence female mate preference and/or willingness to mate with a male possessing a novel phenotype, in addition to type of experience. Correlational statistical tests were used to test whether amount of exposure as juveniles had an effect on adult female receptivity and female aggression towards heterospecific versus conspecific males. To determine effects of juvenile and adult treatment groups on mating outcome, separate Chi-square analyses were conducted. . All statistical analyses were performed using the statistical software program JMP 8 (SAS Institutes, USA).

Results

Adult treatment (heterospecific/conspecific) for both species most strongly explained variance in both female receptivity (Table 1 and 2) and the presence/absence of copulation (S.

2 2 ocreata:X1=68.4, p<0.0001; S. rovneri: X 1=4.39; p=0.036). In a pooled analysis of exposed (in which type of juvenile treatment was not included) and unexposed females of both species were significantly more receptive to conspecific males than to heterospecific males (t-test: S. ocreata:

57 t=-11.156, p<0.0001; S. rovneri: t=-6.232, p<0.0001; Figure 3). Receptivity towards heterospecific males did not differ between exposed and unexposed females in either species (t- test: S. rovneri: t=-0.8407, p=0.4062; S. ocreata: t=1.7086, p=0.0946). With two notable exceptions, mating did not occur between heterospecific individuals, nor did experience (type or amount) seem to influence mating success in conspecific pairings. Heterospecific mating occurred one time in each species. In both cases, the females were exposed to heterospecific male courtship as juveniles five or six times prior to maturation. Because female S. ocreata and

S. rovneri are morphologically identical, to ensure that both matings were true hybridization events (and not the result of experimenter or collection error), the females were allowed to lay egg sacs and the resulting offspring were reared to adulthood. Upon maturation, male offspring were examined morphologically and behaviorally for hybrid indicator traits. From a previous study involving forced hybridization between these two species (Stratton & Uetz 1986), it is known that male hybrids generally have reduced leg tufts (in both size and area of leg covered) relative to those found on S. ocreata males, and exhibit courtship behavior that is a mixture of the courtship displays of both species (See Stratton & Uetz 1986 for a full description). In both cases, hybridization was verified.

Type of juvenile experience did not affect mating outcome for either species (S. ocreata:

Likelihood Ratio: X2=1.956, p=0.3760; S. rovneri: Likelihood Ratio: X2=0.850, p=0.6539).

However, in S. rovneri (but not S. ocreata), type of juvenile experience (heterospecific, conspecific, or none) explained a significant portion of the variation in adult female receptivity rates before the barrier between the male and female was removed (Table 1), but not after. A multiple comparisons analysis revealed that during the first five minutes of trials (while the male

58 and female remained physically separated from each other) female S. rovneri that were exposed as juveniles to heterospecifics were significantly less receptive to conspecifics than unexposed females (ANOVA: F=4.512, df=2, p=0.0178; means were compared via a Tukey HSD analysis).

Receptivity towards conspecifics did not differ significantly between females exposed to conspecifics and females exposed to heterospecifics.

Adult female aggression towards adult male conspecifics vs. heterospecifics differed between species (Figure 4). S. ocreata females were significantly more aggressive towards heterospecific males both before and after the barrier between the spiders was removed (t-tests: t=2.1508, p=0.0343 and t=4.5867, p<0.0001, respectively). However, aggression by adult female

S. rovneri towards conspecific and heterospecific males was not significantly different before or after the barrier between the spiders was removed (t-tests: t=-0.1190, p=0.9056 and t=0.5020, p=0.6171, respectively). Additionally, adult female S. ocreata were significantly more aggressive towards heterospecific males after the barrier was removed than were female S. rovneri (t-test: t=-1.8254, p=0.0105, N=84; Figure 4).

Amount of juvenile exposure to adult male courtship affected adult receptivity of females in both species, but in different ways, and only before the barrier was removed (not after). In S. ocreata, females who received the most juvenile exposure to male courtship (of either species) showed increased receptivity to heterospecific males prior to the removal of the barrier (R2 =

0.0973, p = 0.0348). In S. rovneri, females with the highest number of juvenile exposures with male courtship (of either species) were significantly less receptive than females with fewer juvenile exposures (R2=0.2553, p=0.0010). These effects were independent of treatment group,

59 as amount of juvenile exposure did not have an effect on adult female receptivity towards conspecific or heterospecific males in either species (before or after barrier removal) with respect to treatment group (Table 3). In addition, there was a significant correlation between the number of juvenile exposures in combination with type of exposure on aggression towards males by female S. ocreata (but not female S. rovneri). Female S. ocreata that received higher numbers of exposures to conspecific male courtship as juveniles were less aggressive towards conspecific males after the barrier was removed than females who had received fewer exposures (R2=0.0551,

N=15, p=0.0136). Exposure to heterospecific male courtship did not have a detectable effect on aggression in S. ocreata.

Discussion

Taken together, results of this study suggest that experience plays a minor role at best in species recognition for these behaviorally isolated sibling species, as the type of juvenile experience did not influence adult female mate choice. Our results differ with those of another nearly identical study conducted recently by Hebets & Vink (2007) using a population from

Mississippi which they suggest is a mixed (freely-interbreeding) syntopic population of S. ocreata and S. rovneri. In their experiment, juvenile females from the mixed population were repeatedly exposed to adult male courtship of either a tufted (S. ocreata) or non-tufted (S. rovneri) male, as in the study described here. In contrast to our finding, Hebets & Vink (2007) report that juvenile experience with male courtship of either male type led to preference for tufted males (S. ocreata) by females. However, the basis of the claim that the Mississippi population used in the study represents a mixed, interbreeding population of S. ocreata and S. rovneri has not been fully substantiated (molecularly or behaviorally). Although interbreeding

60 between S. ocreata and S. rovneri does occur in nature (i.e., hybrids have occasionally been collected from field sites in Ohio), and can be forced in the lab, inheritance of male traits in these two species results in males exhibiting intermediate morphology (reduced male tufts) and a mixture of parental species courtship behavior (Stratton & Uetz 1986). It would therefore be necessary to demonstrate that both male morphs in the Mississippi population originate from a single egg sac from the same mother, or that hybrids are common and experience higher rates of mating success than previously observed (Stratton & Uetz 1986). If the Mississippi population is truly polymorphic and freely interbreeding, differences observed in the effects of experience on female mate choice would suggest geographic behavioral divergence across a latitudinal gradient. Unfortunately, it is unclear whether observed effects of experience result specifically from exposure to male courtship rather than undetected mating along species lines by two cryptic female species. Consequently, it is difficult to compare results of both studies in an effort to assess how and why juvenile exposure to male courtship can have such different impacts on female preferences.

In the study presented here, S. rovneri females exposed to heterospecific male courtship as juveniles were initially (prior to barrier removal) less receptive towards conspecific males than unexposed females. This finding could indicate that juvenile experience with male courtship leads females of this species to be more cautious or even choosier about their mate decisions, which may ultimately serve to reinforce species isolation. Because this effect was observed only prior to the removal of the separation barrier, this may also provide evidence that physical interaction between the male and the female is important for species identification in S.rovneri.

These results support other recent findings involving the effects of juvenile experience with

61 different artificial conspecific male phenotypes on adult female mate preferences for those phenotypes in S. rovneri (Rutledge et al. 2010). Rutledge et al. (2010) found that females of this species were less receptive as adults to males possessing phenotypes to which they had been exposed to as juveniles, and found evidence that chemical cues within the environment during juvenile experience and mate choice may ultimately influence mating decisions. Although the same effects of experience type on female receptivity were not observed in S. ocreata, females who were exposed multiply to conspecific male courtship as juveniles showed decreased aggression towards conspecific males vs. females with less or no exposure. In addition, aggression toward heterospecific males by S. ocreata females was higher overall regardless of exposure than aggression towards heterospecifics by female S. rovneri (both before and after the females and males were allowed to physically interact). One interpretation of these finding might be that female S. ocreata have a stronger innate recognition templates for species identification than S. rovneri, which is reinforced rather than modified by experience. It is also plausible that physical interaction plays an important role in female mating decisions and agonistic behavior, as one difference between the methods utilized here and those employed by Hebets (2003) is the manner in which females were exposed to males. In the study described here, females were kept physically isolated from males during exposure and thus were not exposed to male tactile or chemical cues prior to maturity; whereas in Hebets’ study, juvenile females had the opportunity to physically interact with adult males during exposure.

Visual and seismic signals are known to be important in species recognition of S. ocreata and S. rovneri (Hebets & Uetz 1999; Uetz 2000; Uetz & Roberts 2002), thus it might be maladaptive for females to exhibit plasticity in preference for male visual and/or seismic

62 courtship. The ability of female S. ocreata and S. rovneri to correctly identify potential

(conspecific) mates has important fitness consequences, as the reproductive costs of choosing to mate with a heterospecific male are high (Stratton & Uetz 1986). Previous studies have hypothesized that the foreleg tufts and multimodal courtship of S. ocreata males are adaptations that help males overcome the visual and seismic barriers for communication that result from the complex leaf litter environment in which they live (Stratton & Uetz 1986; Scheffer et al. 1996;

Uetz et al. 2009). The most recent revised phylogeny of the genus Schizocosa (Stratton 2005;

Hebets & Vink 2007) continues to support the hypothesis that S. rovneri and S. ocreata are sister species. However in contrast to previous phylogenies, recent studies suggest that S. rovneri males have secondarily lost leg tufts (considered to be a secondary sexual characteristic) and multimodal courtship (Stratton 2005). Nonetheless, both female S. ocreata and S. rovneri exhibit preferences for male foreleg ornamentation (a possible pre-existing bias in S. rovneri)

(McClintock & Uetz 1996; Scheffer et al. 1996), suggesting that factors apart from sexual selection have driven the loss of secondary male traits in S. rovneri.

Although these species are sympatric throughout a large portion of their ranges, populations of S. rovneri tend to be most dense in leaf litter along flood plains. Though the litter does not differ dramatically in terms of the types of leaves found in these habitats, generally the floodplain leaf litter has less vertical structure than forest floor leaf litter, and is cemented together by mud, creating a contiguous substrate across the floor of the habitat (Scheffer et al.

1996). These differences have significant impacts on the transmission and reception of seismic and visual signals (Scheffer et al.1996) and are likely to have played a role in the evolution of male morphology and behavior of both species. The enhanced visual courtship displays produced

63 by the addition of leg tufts, as seen in S. ocreata may improve a male’s chances of communicating effectively in the complex leaf litter and thus reduce his risk of predation by females. However, in the floodplain leaf litter, males are likely to perform the bulk of their courtship on the upper surface of the leaf litter, and having such high contrast visual characteristics may ultimately make them overly conspicuous to predators. If true, then natural selection would favor males with reduced or absent visual traits, not because females lack a preference for them, but because they are the males that would survive to mate. Nonetheless, reduction in male courtship signals may ultimately hinder a female’s ability to assess species identity of a potential mate. As a consequence, this might have led females to become more hesitant about accepting and/or attacking males, since costs of mistaken identity (accepting a heterospecific male as a mate) are severe, but costs of not accepting a male may limit reproductive options.

In some cases it seems that juvenile experience may help to inform female expectations about mate availability at maturity which may lead to shifts in female choosiness and receptivity.

In a different study conducted by Johnson (2005), in which juvenile female Dolmedes triton

(fishing spiders) were raised in the presence or absence of adult male chemical, visual and seismic cues, experience had a significant impact on female mating decisions and rates of cannibalism, as virgin females, exposed to male cues prior to maturation, were more likely to cannibalize males than naïve females. In this study, repeated juvenile exposure to heterospecific males led S. rovneri females to be less receptive to conspecific males than unexposed females. It is possible that this exposure to heterospecific males led to modified expectations about the

64 frequency distribution of conspecific mates, leading females to be more cautious (and ultimately less receptive) about accepting males as mates.

From the growing body of literature on the effects of experience on female mate choice in spiders as well as other invertebrate animals, it is clear that plasticity in female mate choice is higher than previously expected. However, experience can have unpredictable effects that appear to be largely species-specific, and often depend on the type of experience. In S. uetzi for example, it was shown that exposure to artificially modified males as juveniles led to female preference for the familiar artificial male phenotype to which they had been previously exposed

(Hebets 2003). However, exposure to male courtship from a closely related species (S. stridulans) did not result in any shifts in female preferences for heterospecific males (Hebets

2007). In S. rovneri, females that were exposed to artificial male phenotypes were shown to discriminate against familiar visual and chemical conspecific male phenotypes (Rutledge et al. in press). However, as reported here, experience with heterospecific males had limited effects on female mate choice in S. rovneri. Ultimately, future research concerning plasticity of female mate preferences should consider how chemical and/or tactile cues might influence female mate choice, which may allow more insight about the role of social factors and sexual selection in species divergence.

Acknowledgements

Foundation (to GWU), the American Arachnological Society (JMR), the University of

Cincinnati Research Council (JMR), and the Wiemen/Wendel/Benedict Student Research Fund

(JMR). Thanks to J. Andrew Roberts, Dave Clark, Julianna Johns, Kerri Wrinn, Jeremy Gibson,

Melita Skelton, Shanquala Pruitt, Adam Stein, and Adam Olsen for helping to collect and maintain the spiders in the laboratory and for helping to set-up trials. Also, thanks to Elke

Buschbeck, Ken Petren, Ann Rypstra and Eric Maurer for providing feedback and advice on this work and for comments on the manuscript.

References

Coleman, S.W., G.L. Patricelli and G. Borgia. 2004. Variable female preferences drive complex

male displays. Nature 428: 742-745.

Delaney, K.J. 1997. Communication in the context of courtship and aggression in two species of

wolf spider (Araneae: Lycosidae). M. S. Thesis, University of Cincinnati.

Doherty, J.A. and H.C. Gerhardt. 1984. Evolutionary and neurobiological implications of

selective phonotaxis in the spring peeper Hyla crucifer. Animal Behaviour 32: 875-881.

Endler, J.A, and A.E. Houde. 1995. Geographic Variation in Female Preferences for Male Traits

in Poecilia reticulate. Evolution 49:456-468.

Hebets, E.A. 2003. Sub-adult experience influences adult mate choice in an arthropod: exposed

female wolf spiders prefer males of a familiar phenotype. Proceedings of the National

Academy of Science 100: 13390-13395.

Hebets, E.A. 2007. Subadult female experience does not influence species recognition in the

wolf spider Schizocosa uetzi Stratton 1997. Journal of Arachnology, 35, 1-10.

Hebets, E.A. and G.W. Uetz. 1999. Female responses to isolated signals from multimodal male

courtship displays in the wolf spiders genus Schizocosa (Araneae: Lycosidae). Animal

Behaviour 57: 865-872.

Hebets, E.A. & Vink, C.J. 2007. Experience leads to preference: experienced females prefer

brush-legged males in a population of syntopic wolf spiders. Behavioral Ecology, 18,

1010-1020.

Hoefler, C.D. 2002. Is contest experience a trump card? The interaction of residency status,

experience, and body size on fighting success in Misumenoides formosipes (Araneae:

Thomisidae). Journal of Insect Behavior 15: 779-790.

Jackson, R.R., and R.S. Wilcox. 1993. Spider flexibly chooses aggressive mimicry signals for

different prey by trial and error. Behaviour 127: 21-36.

Johnson, C. 2005. Cohabitation of juvenile females with mature males promotes sexual

cannibalism in fishing spiders. Behavioral Ecology 16: 269-273.

Kodric-Brown, A. and P.F Nicoletto. 2001. Age and experience affect female choice in the

guppy (Poecilia reticulate). American Naturalist 157: 316-323.

Kujtan, L. and R. Dukas. 2009. Learning magnifies individual variation in heterospecific mating

propensity. Animal Behaviour 78: 549-554.

Magurran, A.E. and I.W. Ramnarine. 2004. Learned mate recognition and reproductive isolation

in guppies. Animal Behaviour 67: 1077-1082.

McClintock, W.J. & Uetz, G.W. 1996. Female choice and pre-existing bias: visual cues during

courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour, 52,

167-181.

Norton, S. and G.W. Uetz. 2005. Mating frequency in Schizocosa ocreata (Hentz) wolf spiders:

evidence for a mating system with female monogamy and male polygamy. Journal of

Arachnology 33:16-24.

Otte, D. and J.A. Endler. 1989. Speciation and Its Consequences. Sinauer Associates, Inc.,

Sunderland

Owen, I.P.F., C. Rowe, and A.L.R. Thomas. 1999. Sexual selection, speciation and imprinting:

separating the sheep from the goats. Trends in Ecology and Evolution 14: 131-132.

Parri, S., R.V. Alatalo, J. Kotiaho, and J. Mappes. 1997. Female choice for male drumming in

the wolf spider Hygrolycosa rubrofasciata. Animal Behaviour 53: 305-312.

Pfennig, K. 1998. The evolution of mate choice and the potential for conflict between species

and mate quality recognition. Proceedings of the Royal Society of London B 265: 1743-

1748.

Pfennig, K. 2007. Facultative mate choice drives adaptive hybridization. Science 318:965-967.

Punzo, F. 2000. An experimental analysis of maze learning in the wolf spider, Trochosa

parthenus (Araneae: Lycosidae). Florida Scientist 63: 155-159.

Punzo, F. 2004. The capacity for spatial learning in spiders: a review. Bulletin of the British

Arachnology Society 13: 65-72.

Reid, M.L. and J.A. Stamps. 1997. Female mate choice tactics in a resource-based mating

system: field test of alternative models. The American Naturalist 150: 98-121.

Roberts, J.A. and G.W. Uetz. 2004. Chemical signaling in a wolf spider: a test of ethospecies

discrimination. Journal of Chemical Ecology 30: 1271-1284.

Rutledge, J.R., A. Miller, and G.W. Uetz. 2010. Exposure to multiple sensory cues as a juvenile

affects adult female mate preferences in wolf spiders. Animal Behaviour 80: 419-426.

Saetre, G.P. 2000. Sexual signals and speciation. Pgs. 237-258 In: Espmark, Y., T. Amundsen,

and G. Rosenquist (Eds.). Animal Signals: Signaling and Signal Design in Animal

Communication. Tapir Academic Press, Trondheim, Norway.

Scheffer, S.J., G.W. Uetz, and G.E. Stratton. 1996. Sexual selection, male morphology, and the

efficacy of courtship signaling in two wolf spiders (Araneae: Lycosidae). Behavioral

Ecology and Sociobiology 38: 17-23.

Stratton, G. E. 2005. Evolution of ornamentation and courtship behavior in Schizocosa: insights

from a phylogeny based on morphology (Araneae, Lycosidae). The Journal of

Arachnology 33: 347-376.

Stratton, G.E. and G.W. Uetz. 1981. Acoustic communication and reproductive isolation in two

species of wolf spiders. Science, New Series 214: 575-577.

Stratton, G.E. and G.W. Uetz. 1983. Communication via substratum-coupled stridulation and

reproductive isolation in wolf spiders (Araneae: Lycosidae). Animal Behaviour 31: 164-

172.

Stratton, G.E. and G.W. Uetz. 1986. The inheritance of courtship behavior and its role as a

reproductive isolating mechanism in two species of Schizocosa wolf spiders

(Araneae:Lycosidae). Evolution 40:129-141.

Uetz, G.W. 2000. Signals and multi-modal signaling in spider communication. Pgs. 243-261 In:

Espmark, Y., T. Amundsen, and G. Rosenquist (Eds.). Animal Signals: Signaling and

Signal Design in Animal Communication. Tapir Academic Press, Trondheim, Norway.

Uetz, G.W. and G. Denterlein. 1979. Courtship behavior, habitat and reproductive isolation in

Schizocosa rovneri Uetz & Dondale (Araneae: Lycosidae). Journal of Arachnology 7:

121-128.

Uetz, G.W., and Roberts, J. A. 2002. Multi-sensory cues and multi-modal communication in

spiders: insights from video/audio playback studies. Brain Behaviour & Evolution 59:

222-230.

Uetz, G.W. and S. Norton. 2007. Preference in male traits in female wolf spiders varies with the

choice of available males, female age, and reproductive state. Behavioral Ecology and

Sociobiology 61: 631-641.

Uetz, G.W., Roberts, J.A., and P.W. Taylor. 2009 Multimodal communication and mate choice

in wolf spiders: Female responses to multimodal vs. unimodal male signals in two sibling

wolf spider species. Animal Behaviour 278:299-305.

Verzijden, M.N. and C. ten Cate. 2007. Early learning influences species assortative mating

preferences in LakeVictoria cichlid fish. Biology Letters 3: 134-136.

Wagner, W.E. Jr. 1998. Measuring female mating preferences. Animal Behaviour 55: 1029-

1042.

Wagner, W.E., M.R. Smeds, and D.D. Wiegmann. 2001. Experience affects female responses to

male song in the variable field cricket Gyrllus lineaticeps (Orthoptera, Gryllidae).

Ethology 107: 769-776.

White, M.J.D. 1978. Modes of Speciation. W.H. Freeman and Company, San Francisco.

Table 1. Results of a Two-Way ANOVA on the effects of juvenile and adult treatment on female composite receptivity rates in S. ocreata before and after removal of the transparent barrier between the male and the female. Asterisks indicate significant effects (P<0.05).

Sum of Factor df F-Ratio P Squares Model 5 0.2144 14.2081 <0.0001* Juvenile 2 0.0010 0.1656 0.8477 Before Treatment Barrier Adult 1 0.2116 70.1150 <0.0001* Removal Treatment Juv. Treat. x 2 0.0005 0.0893 0.9146 Adult Treat. Model 5 0.1120 27.5989 <0.0001* Juvenile 2 0.0174 2.1448 0.1238 After Treatment Barrier Adult 1 0.5123 126.2252 <0.0001* Removal Treatment Juv. Treat. x 2 0.0173 2.1371 0.1247 Adult Treat.

Table 2. Results of a Two-Way ANOVA on the effects of juvenile and adult treatment on female composite receptivity rates in S. rovneri, before and after removal of the transparent barrier between the male and the female. Asterisks indicate significant effects (P<0.05).

Sum of Factor df F-Ratio P Squares Model 5 0.0435 9.4417 <0.0001* Juvenile 2 0.0083 4.5251 0.0142* Before Treatment Barrier Adult 1 0..0265 28.7442 <0.0001* Removal Treatment Juv. Treat. x 2 0.0054 2.9332 0.0598 Adult Treat. Model 5 0.0038 7.7845 <0.0001* Juvenile 2 0.0005 0.5506 0.5791 After Treatment Barrier Adult 1 0.0178 36.5962 <0.0001* Removal Treatment Juv. Treat. x 2 0.0001 0.1384 0.8710 Adult Treat.

Table 3. Relationships between amount of juvenile exposure and female receptivity rates by treatment group for S. ocreata and S. rovneri.

Before After Female Treatment N R2 P R2 P Species (Juv./Adult) S. ocreata/ 12 0.0023 0.8829 0.0002 0.9632 S.ocreata S. ocreata/ 15 0.0075 0.7596 0.2044 0.0907 S. rovneri S. ocreata S. rovneri/ 17 0.0296 0.5091 0.0613 0.3381 S.ocreata S. rovneri/ 17 0.1651 0.1056 0.0543 0.3680 S. rovneri S. rovneri/ 11 0.1198 0.2971 0.2763 0.0968 S. rovneri S. rovneri/ 12 0.0003 0.9554 0.0010 0.9228 S.ocreata S. rovneri S. ocreata/ 13 0.1480 0.1943 0.1220 0.2421 S. rovneri S. ocreata/ 13 0.0098 0.7472 0.0002 0.9684 S.ocreata

Figure 1. Experimental apparatuses (not to scale). (A) Juvenile and adult exposure apparatuses used for S. ocreata trials. (B) Apparatus used for S. rovneri trials. The same apparatus was used for juvenile exposure and adult trials..

(A)

Juvenile Apparatus Adult Apparatus

Transparent Barrier containers removed after 5 minutes

Male

Shared substratum Male Female (posterboard)

(B) Barrier removed after 5 minutes Transparent Shared substratum containers (posterboard)

Male Female

Figure 2. Experimental design. As juveniles and adults females were exposed to either adult courting conspecific or heterospecific males. Females in the control group were raised to maturity in isolation (no experience).

Figure 3. Effect of female exposure on total (entire trial) composite receptivity rate in response to heterospecific or conspecific males. Error bars indicate standard error about the mean.

Figure 4. Comparison of aggression between species towards heterospecific and conspecific males (S.o. = S. ocreata, S.r. = S. rovneri) before and after the barrier between the male and female was removed. Error bars show standard error about the mean

Chapter 4: Testing the limits of body condition indices: a comparison of approaches to

estimating condition during short-term starvation in wolf spiders

(to be submitted to Behavioral Ecology)

Jenai M. Rutledge, Jessica Diersing, Brian Moskalik & George W. Uetz

Department of Biological Sciences

University of Cincinnati

614 Rieveschl Hall

Cincinnati, OH USA 45221-0006

Corresponding author – [email protected]; Tel: 513-556-9753, Fax: 513-556-5299

Abstract

An animal’s body condition may affect its survival, mating success, and thereby fitness. Debate over which non-invasive body condition index (BCI) most accurately and reliably predicts the energetic state and/or physiological health of an animal is ongoing. Here we test the power of four different BCI’s to detect differences in recent feeding and hydration histories of spiders.

Juvenile (mixed sexes) and adult male Schizocosa ocreata (Hentz) were maintained in the lab for

5 days (juveniles) or 10 days (adults) on one of four food/water regimes including, 1) food and water (F+W+), 2) food only (F+W-), 3) water only (F-W+), 4) no food or water (F-W-).

Although by day 5, body condition of both juvenile and adult male spiders in F+W+ and F-W- treatment groups clearly differed regardless of the BCI used, none of the indices detected a difference between spiders in the F+W- and F-W+ treatments until 10 days under deprivation conditions. This suggests that spiders may compensate for short-term starvation by increasing water intake, allowing them to maintain weight and/or internal hydrostatic pressure (and apparent body condition) even after multiple days of food deprivation. These findings raise questions about the use of BCIs to make inferences about current energetic stores in spiders, but also reveal previously unrecognized aspects of ecological physiology for spiders. Further study is needed to determine which indices are best able to separate effects of feeding history from the effects of hydration.

Introduction

Body condition is a term most commonly used to describe the nutritional state/ health of an animal or energetic stores contained within an animal’s body. These characteristics can contribute to an animal’s overall survival, fitness, and/or success in other aspects of its daily life, and as such, metrics that estimate or measure body condition directly are used for a variety of purposes in research and in many different animals. Specifically, body condition (i.e., energetic stores) has been shown to relate to have numerous important behavioral correlates including, mating success (Danielson-François et al. 2002; Scheuber et al. 2003) and mating decisions/tactics (Eraly et a. 2009; Lomborg & Toft 2009; Wilgers et al. 2009; Barry et al. 2010;

Moskalik et al. in prep), fecundity (Danielson-François et al. 2002; Barry et al. 2010), survival

(Fox & Moya-Loraño 2009).

Although measuring fat stores and body content directly is the most accurate method for measuring nutritional health and/or quantifying energetic reserves, such methods are necessarily invasive and often destructive. Because it is frequently undesirable to use invasive/destructive methods to estimate an animal’s energy stores, methods that rely on morphometric measurements of condition-dependent traits scaled for body size are widely used to estimate body condition. There are multiple methods for estimating body condition from morphometric indices and the debate over which method or index provides the most accurate and useable results is ongoing (e.g., Jakob et al. 1996; Kotiaho 1999; Green 2001; Garcia-Berthou 2001;

Moya-Loraño et al. 2008). One of the biggest challenges of using morphological condition indices to infer information about an animal’s energetic reserves is that it is rarely known how well the condition-indicating traits used for the index correlate with actual fat stores; and where

85 this relationship has been investigated, often only weak correlations (if any) between morphological traits and fat content are observed (Virgl & Messier 1993; Krebs & Singleton

1993). One possible explanation for a poor-relationship between estimates of fat stores from

BCIs and actual fat stores is that most indices are not designed to detect differences in types of nutrients stored (e.g., fat vs. lean muscle mass vs. water content, Schulte-Hostedde et al. 2001;

Moya-Loraño et al. 2008). Nutrient storage within an animal’s body can be diverse and body size as well as body mass reflects not only fat storage, but also water content, lean muscle tissue, etc., the relative ratios of which will have differing implications on actual energetic reserves. This concept has been addressed recently by Moya-Loraño et al. (2008), who recommends the usage of a density body condition index. By using a density-based index, differences in ratios of fat, protein, lean muscle mass, etc. should be detected, as each has distinctly different densities.

Moya-Loraño et al. (2008) found their density body condition index to best reflect simulated changes in lipid body contents and manipulations of diet in living spiders; however one seemingly minor oversight of this study as well as many others, is that modifying diet regimes may impact the amount of water in the food and/or water intake of an animal. Because hydration is also likely to affect body density, it may be equally important to investigate separately how hydration affects body condition.

Although much research has investigated the effects of low-quality or low-quantity diets on body condition and subsequent effects on various parameters of fitness (e.g., Scheuber et al.

2003; Moya-Loraño et al. 2003; Hebets et al. 2008; Lomborg & Toft 2009) little is known about how well traditional body condition indices separate out the effects of starvation from dehydration. Hydration is critical to an animal’s survival, and is one of the most significant

86 factors in determining how well an animal can perform important behaviors, including foraging and courtship/mating behaviors. In many cases, studies that attempt to manipulate nutritional diet quality or quantity in an effort to understand the effects of starvation on body condition and behavior may also inadvertently vary the relative water content of the diets. Although this is probably not a critical issue in studies where constant access to water is provided, in others, it may be. For example, many studies involve arthropod predators that require little water beyond that contained in the prey provided, e.g. Pholcids, Brown Recluses. In these cases, differences in nutritional quality of diets as well as their relative water content may be important, but understudied. Dehydration can cause similar declines in mass and body volumes as one might expect to occur as a result of starvation, but both have very different implications for survival, success, and fitness.

The studied effects of dehydration are broad, but some of the most critical responses to dehydration in animals that directly relate to potential body condition and energy stores include reduction or increase in metabolism (Ito 1964; Anderson 1970, 1974; Bjerke & Zachariassen

1997; Sibul et al. 2006), and increased concentrations of osmotically active solutes in the body fluids as a consequence of water loss which can affect various cardiovascular functions (Bjerke

& Zachariassen 1997). In carabid beetles, dehydration can result in a drop in hemolymph volume which increases the concentrations of solutes in various parts of the body (Bjerke & Zachariassen

1997). To cope with the resulting increases in solute concentration, these insects must either cope with the increase in some manner, or remove the solutes - a task that may be metabolically demanding.

Here we attempt to address the question of how well body condition indices are able to separate out the effects of starvation on body condition from the effects of dehydration in a study of wolf spiders (Lycosidae). In an effort to evaluate which morphological index (if any) best detects changes in short-term feeding and hydration histories, body condition was estimated using four different body condition indices. A common method for estimating body condition uses the residuals from an ordinary least squares (OLS) linear regression of a static body size indicator (e.g., cephalothorax width) against a condition-dependent body size trait (e.g., abdomen width). This method has received heavy criticism over the last decade for a variety of reason including concerns about how well underlying statistical assumptions of linear regression of the method and assumptions about the relationship between the traits from which body condition is being estimated can be met using morphological data (Kotiaho 1999; Green 2001; Garcia-

Berthou 2001; Darlington & Smulders 2001). Yet another common criticism is that the estimates of body condition provided by this index may not be strongly correlated with actual size of energy reserves (e.g., Krebs & Singleton 1993; Rolff & Joop 2002). Nonetheless, this index continues to be one of the most popular indices used for behavioral and ecological studies involving body condition. Furthermore, some have argued that the residual OLS regression index

(referred to as RBC from here on) yields easily interpreted body condition values that can be compared across taxa (Jakob et al. 1996) and others have gone as far as to indicate that statistical concerns regarding this metric have been overstated (Schulte-Hostedde 2005).

Juvenile (mixed sex) and adult male Schizocosa ocreata (Hentz) wolf spiders were maintained on one of four different diets in which access to food and/or water was manipulated for 5 or 10 days, respectively. Daily measurements were taken of the spiders to examine how

88 different condition-dependent traits vary under dehydration and/or starvation conditions. S. ocreata spiders are ideals organisms for examining this question because they must cope with periods of food deprivation and drought in the field which are associated with living in temperate environments. In addition, most if not all of the ‘key assumptions’ for using the OLS residual index for estimating body condition, as indentified by Green (2001), are met in data collected on

S. ocreata.

In addition to the residual body condition index (RBC), we also used three of the four condition estimation indices described by Moya-Loraño et al. (2008) including the mass body condition index (MBC), volume body condition index (VBC), as well as the size-corrected density body condition index (DBC, referred to as ‘SDBC’ in Moya-Loraño et al. 2008). Here we compare these four body indices using data collected from the two diet manipulation and discuss how well each index detected differences between starved and/or dehydrated treatments.

Materials and Methods

Experiment 1: Juveniles

Subjects and Housing.

Immature Schizocosa ocreata spiders were collected from the Cincinnati Nature Center,

Rowe Woods, and brought to the laboratory, where they were kept under controlled conditions.

In the laboratory, spiders were housed in opaque, plastic deli containers (360ml, round) where they had access to water via a dental wick that was immersed in water in a separate container below the housing container. Spiders were kept on a 13:11h, light:dark cycle and were maintained in a constant humidity environment. The spiders were fed 2-3 pinhead (approx. 0.5

89 cm in length) or ten-day old (approx. 0.6 cm in length) crickets twice a week depending on the size of the spider. Growth of each spider was tracked with daily molt checks.

Experimental Methods.

The inter-molt interval for juvenile spiders varies widely and can range from 2-12 weeks

(unpublished data). Generally as spiders age, length of inter-molt interval decreases. To avoid having animals molt during the study (which would result in an increase in structural body size indicators – cephalothorax width), only juvenile spiders that had molted within 3-4 days of the start of the experiment were used. If a spider molted during the study, they were removed and were not included in the final data. Because spiders were field-caught, it was not possible to know their precise ages, but an effort to control for initial spider size was made; all spiders used were approximately the same mass (ranging from 0.011g to 0.034g) and size (cephalothorax widths ranged from 1.4mm to 1.9mm). Sixty spiders (15 per treatment) were randomly assigned to one of four feeding/watering regimes including: “food and water” (F+W+), “food, no water”(F+W-), “water, no food”(F-W+), and “no food or water” (F-W-).

One day prior to the start of the experiment, all spiders were fed one pinhead cricket to assure that all were at relatively the same level of satiation. On the first day of the study, all spiders were placed in fresh deli-dish containers (as described above); however, for the treatment groups without water, the reservoir container beneath the housing container was left empty.

Spiders were weighed using an electronic balance (Fisher Scientific, XE Series, Model No.

100A) and were digitally photographed daily for five days after the starting day. Digital photographs of the dorsal surface of all spiders were taken using a Pixera Professional digital

90 microscope camera system (Pixera Corporation San Jose, CA, U.S.A.) at 12X magnification, so that measurements of structural body size indicators (BSI, i.e., cephalothorax width) and condition-dependent body indicators (CDBSI, i.e., abdomen width and abdomen length) could be taken. Digital measurements of BSI and CDBSI traits were taken on each spider using the ruler function in UTHSCSA ImageTool© ver: 3.00 calibrated for a 12X magnification. To reduce variance due to measurement error, body traits were measured three times each and the average of these measurements were used for analysis. Spiders in the F+ treatment groups were fed one pinhead cricket every other day. Spiders in fed treatments were weighed both before and after they were fed. On feeding days, all photos were taken 30-60 minutes after feeding.

Three spiders in the F-W- treatment were discovered to be carrying parasites mid-way through the study and were consequently removed from the study and were not included in the data for analysis. Another spider from the same treatment group had to be excluded owing to limb loss between days.

Calculating Body Condition and Statistics

Body condition estimates were made using four different body condition indices. The first index used was a residual ordinarly least squares (OLS) linear regression index (RBC) in which body condition was estimated from the residuals of a linear regression of mass against cephalothorax width. To meet the assumptions of regression, both mass and cephalothorax width were ln-transformed (Jakob et al. 1996; Marshall et al. 1999).

In addition to traditional RBC, three indices defined most recently by Moya-Laraño

(2008) were used, including the “mass body condition index” (MBC); the “abdomen-volume body condition index” (VBC) and the “statistically-corrected density body condition index”

(SDBC). The MBC and VBC indices are ratio indices in which body condition is calculated as a condition-dependent measure (mass or abdomen volume), divided by a structural body size measure (e.g., cephalothorax width; Moya-Loraño 2008). VBC was calculated as cube-root abdomen volume / cephalothorax width. Abdomen volume was used because in spiders fat stores (i.e., energy stores) are thought to be most concentrated in the abdomen. For the VBC index abdomen volume was calculated as described by Moya-Loraño (2008) and Jakob et al.

(1996) using the equation for the volume of an ellipse: 4/3 lwd, where l = abdomen length/2, w=abdomen width/2, and d=abdomen width/2. Abdomen width was used as a proxy for abdomen depth because abdomen depth was not measured directly (Moya-Loraño 2008). For the

DBC index, body condition was estimated as the cube-root of mass over the cube-root of abdomen volume, corrected for body size (divided by cephalothorax width, Moya-Loraño et al.

2008). We investigated the effects of treatment on VBC, MBC, and DBC using ANCOVA as described in Moya-Loraño et al. (2008 – see description of SDBC for analysis details regarding what we term here as DBC).

To test the null hypothesis that treatment and day had no effect on spider body condition, a repeated-measures ANOVA was performed for each BCI to investigate differences in body condition between diet treatments over time. We used body condition (RBC, MBC, VBC, or

DBC – calculated as ratios scaled by cephalothorax width) as the response variable, time as the within-subjects factor, and diet treatment as the between-subjects factor. Because the

92 assumption of elliptical sphericity, required for using univariate analyses, was not met for most of the analyses, only the results from multivariate Wilkes’ Lambda tests are reported. In addition, multiple one-way ANOVAs (for RBC) and ANCOVAs for (MBC, VBC, and DBC) were performed for days 0, 3, and 5 in experiment 1 and days 0, 3, 5, 8, and 10 in experiment 2 to allow for comparisons between the sensitivity of each BCI to be made. These days were chosen arbitrarily. When significant differences were observed, a post-hoc Tukey HSD comparison of means was performed to determine which treatment groups were different from each other. In addition, to correct for multiple sampling of the data, Bonferroni corrected p- values (p = /n where n was equivalent to the number of independent tests used and was set at

0.05) were used when interpreting results. All analyses were conducted using JMP 8.0.2 (SAS

Institute Inc., Cary, NC, U.S.A.).

Results

Multivariate repeated measures ANOVAs for each index reveal that diet, independent of time, influenced body condition when estimated using the MBC, VBC and RBC indices, but not with the DBC index (Table 1). Time, independent of diet also had a significant effect on body condition when estimated with the MBC, VBC, and DBC indices, but not with the

RBC index. In general, over the course of the study, body condition of spiders decreased in the

F-W- treatment increase very slightly in the F+W+ treatment and remained relatively steady in both the F+W- and F-W+ treatments. Finally there was a significant interaction between diet and time: on average body condition decreased across all treatments over time until day 5 (Table 1).

Results of separate comparisons of means between treatments via ANCOVAs (for MBC,

VBC, and DBC) and one-way ANOVAs (for RBC) on days 1, 3, and 5 (with Bonferroni corrected (/3) and only in cases where p-values less than 0.016). allowed rejection of the null hypothesis that treatment had no effect on body condition for some cases, but not all (Tables 2 &

3). Ultimately, differences detected between treatment groups were similar using the RBC,

MBC, and VBC indices. On day one, no differences in RBC, MBC, VBC or DBC were detected between treatment groups. By day three, the RBC, MBC, and VBC of spiders in the F+W+ and

F-W- differed significantly from each other, as did the RBC, MBC, and VBC of spiders in the

F+W+ and F+W- treatments (Fig. 1). Similar differences between treatments were observed on day five in RBC, MBC, and VBC. Abdomen volume and body mass were significantly correlated across days (average r2=0.6184, p<0.0001), which may explain some of the similarities between condition indices. In contrast, no difference in DBC between treatment groups was detected. Interestingly, none of the body condition indices detected a significant difference between the F+W- and F-W+ treatments even after 5 days (Fig. 1).

Experiment 2: Adult males

Based on the findings of the first pilot study which showed that body condition

(regardless of how it was estimated) of spiders that were starved but had access to water (F-W+) did not differ from that of spiders that were fed but were denied access to water (F+W-), a second, more controlled study was conducted. This effort replicated the first study and further investigated the relationship between the F+W- and F-W+ treatment groups. In this second experiment, adult male spiders were maintained on their assigned food/water regimes for 10 days

(rather than 5) in an effort to determine whether and at what point the body condition of spiders reared in the single deprivation groups (i.e., F-W+ or F+W-) diverged.

Subjects and Housing.

Adult female S. ocreata carrying egg sacs or babies were captured from leaf litter at the

Cincinnati Nature Center Rowe Woods (Clermont Co., OH) during the fall of 2009 and brought back to the lab. Wolf spiders (Araneae, Lycosidae) attach their egg sacs to their spinnerets and carry them around until the spiderlings emerge. Upon hatching, spiderlings crawl onto their mother’s abdomen, where they remain for 7 to 14 days post-emergence before dispersing into the environment. Within the egg sac, spiderlings molt twice prior to emerging from the egg sac, thus the spiderlings found on female abdomens were considered to be 3rd instar spiders (Foelix,

1996). In the lab, spiderlings were allowed to disperse naturally from their mother’s abdomen at which point they were collected and placed into individual semi-opaque, screw-cap, 120 ml specimen cups (6 cm-diam. x 7 cm-height) filled with a mixture of soil and peat moss (1:1 ratio) seeded with Collembola (sp. unknown) to provide a constant food source. A small piece of

Russet potato and a small amount of yeast (Fleishmann’s active dried yeast) were added to each container to sustain the Collembola cultures (which were checked weekly for productivity and supplemented as necessary). In addition to collembola, three to five wingless fruitflies (apterous

Drosophila melanogaster) were added to each container once a week as a source of diet enrichment. So that the development of each spiderling could be tracked, containers were checked daily for molts and a record was kept of each time a spider molted. Containers were misted once a week with water to maintain humidity/access to water and were kept capped to prevent evaporation (and escape). Upon reaching the 5th or 6th instar, spiderlings were moved to

95 white, opaque round deli-dish containers (360mL deli-dish; 10cm diameter) where they were given constant access to water via a cotton dental wick which was suspended in a reservoir of water in a deeper deli-dish below the housing container. Spiders in deli-dish containers were fed two to three pinhead or ten-day old crickets (depending on spider size) twice a week. All spiders in the lab were kept under a 13/11 h light/dark cycle at approximately 25º C, and constant relative humidity.

Experimental Methods.

To test the effects of short-term diet manipulations on body condition a group of adult males (N=44) were randomly assigned to one of the four diet treatments (as described in the experiment 1). We chose to focus on adult males and not adult females because there is evidence that male and female lycosids respond to dietary stress differently (Moya-Laraño et al. 2008), yet little is known about the underlying causes for these differences. In addition, variation in condition-dependent body size indicators (e.g., abdomen width and length) in females is likely to correlate with egg development as well as diet. Because it is often assumed that egg development/investment reflects long-term feeding history (over an individual’s lifetime) and because the relationship between recent feeding/hydration history and egg development is not well understood it would be more difficult to draw sound conclusions about the effects of short- term starvation and dehydration on body condition in females.

Twenty-four hours prior to the start of the study all spiders were fed two pinhead crickets

(8th instar spiders) or two ten-day old crickets (adult males) and were housed in their original containers with constant access to water (via the cotton wick as described above). At the start of

96 the study (referred to as Day 0 in the results), each spider was weighed and digitally photographed from above (as previously described). All spiders were then moved into clean housing containers with identical dimensions to those of the general population (as described above). Methods for maintaining spiders in each diet treatment were identical to those described in experiment 1 with one notable exception; in an effort to maintain similar levels of humidity in housing containers across treatments, spiders in ‘no water’ treatments were housed in containers that had a piece of mesh screen that covered the hole through which the cotton wick would protrude and the reservoir containers were filled with 0.53g of a non-dessicating water absorbing material called Watersorb (http://www.watersorb.com/) mixed with 50mL of water. Water alone was not used in reservoirs for the ‘no water’ treatments to eliminate the possibility of spiders gaining access to water due to the splashing of water onto the screen partition as a result of handling of the containers. The reservoirs of the ‘water’ group were filled with 75mL of water into which a cotton wick was suspended. To verify that the relative humidity within containers of both treatments remained similar across the entire experiment, measurements of temperature and relative humidity were taken inside the containers (without spiders) daily for 5 days. The relative humidity in containers with a wick inserted into a reservoir filled with water was on average 20-

25% higher than the relative humidity of dry containers (containers with a wick but no water, as used in experiment 1), but only 10-14% higher than containers with mesh across the wick opening and watersorb/water combo in the reservoir below the housing container.

Spiders in all treatments were weighed and digitally photographed (to allow for morphometric measurements to be taken at a later date) daily for 10 days after the starting day.

In the absence of food and water, spiders began to senesce after six days. If a spider in any of the

97 treatment groups appeared to be nearing death (determined by a general inability to extend their legs fully or move in a normal manner), that spider was removed from the study. At the end of the tenth day (or upon removal from the study), all spiders in deprivation treatment groups were returned to normal laboratory rearing conditions with access to water and were fed a single ten- day old cricket. Twenty-four hours after returning to normal rearing conditions, spiders were weighed and photographed one final time to determine how the body condition of the spiders in each treatment changed in response to renewed access to food, water, or both. Body conditions were estimated using the four body condition indices described above.

Results

Because the sample size in the F-W- treatment group decreased dramatically by day 8

(due to severe dehydration and/or mortality, Fig. 2), analysis of the data for this experiment was broken into two parts. Comparisons between all four treatment groups were made using the data collected between days 0-8 as described for the previous experiment. To analyze the data across all ten-days of the study, data from the F-W- treatment were excluded and the analyses were re- run, this time including data collected from day 0-10. On average, between the first day of the experiment and day 10, S. ocreata in the F+W- treatment lost significantly less weight (mean change in weight+ SE: -4.82+0.83 mg) than spiders in the F-W+ treatment (mean change in weight+ SE: -9.46+ 0.83 mg; tdf=22=-3.8581, p=0.0009). Body condition of spiders in the F-W+ treatment also remained significantly lower than those in the F+W- and F+W+ (VBC ANCOVA:

F5,27=8.5612, p<0.0001; MBC ANCOVA: F5,27=41.3375, p<0.0001; RBC One-Way ANOVA:

F2,30=10.6260, p=0.0003) treatment after one feeding. In contrast, upon receiving access to water

98 for 24-hours body conditions of spiders in the F+W- did not differ significantly from that of spiders in the F+W+ treatment group (Means comparison via Tukey-Kramer HSD). Twenty-four hours after one feeding of one cricket and access to water, spiders in F+W- and F-W- treatments returned to their starting weights. In contrast, spiders on the F-W+ diet remained on average +

SE 5.23 + 0.98mg lower than their starting weights, which was significantly less than the change in weight of spiders in the F+W+ and F+W- treatments (F2,30=16.097; p<0.0001). Body condition of spiders in all but the F+W+ treatments also decreased slightly over the duration of the study (Fig. 3). Spiders in the F-W+ treatment group generally experienced a greater negative change in body condition between day 0 and day 10 than spiders in the F+W- treatment and

F+W+. While difference in total change in body condition from day 0 to day 10 was significantly different between F+W+ and the single deprivation treatments (F+W-, F-W+) using any of the four body condition indices, between F-W+ and F+W- spiders (DBC: F2,30=18.241, p<0.0001;

MBC: F2,30=9.293, p=0.0007; RBC: F2,30=10.769, p=0.0003; VBC: F2,30=25.169, p<0.0001), total change in body condition was significant only when body condition was calculated with

VBC or DBC indices. Body condition increased over the study period in spiders in the F+W+ treatment (indicated by post-hoc Tukey HSD comparison of means).

A repeated measures ANOVA conducted for each body condition index with all but the

F-W- treatment groups between days 0-10 shows that the different BCIs vary in sensitivity

(Table 4). Because the assumption of elliptical sphericity, required for using univariate analyses, was not met in most cases, only the results from multivariate Wilkes’ Lambda tests are reported.

As in experiment 1, both diet alone and day alone had significant effects on MBC, VBC, and

RBC. DBC was not significantly affected by diet as in all previous analyses, but was influenced

99 by day. As in each of the previous analyses, there was a significant interaction effect between diet and day on all four estimates of body condition, indicating that the pattern of differences between mean body condition for each diet treatment changed over time.

In addition to the repeated measures ANOVAs, separate ANCOVAs (for MBC, VBC, and DBC) and one-way ANOVAs (for RBC) were performed for each index for days 0, 3, 5 using data collected from all four treatment groups between days 0 and days 5 (Tables 5 & 6, respectively) and days 0, 3, 5, 8, and 10 for data collected from the F+W+, F+W- and F-W+ groups between days 0-10 (Tables 7 & 8). These days were chosen arbitrarily to determine how the body condition of spiders would differ at different points in time within and between diet treatments. To correct for this multiple comparisons method, results were Bonferroni corrected using the equation (/n) where n was equivalent to the number of independent tests used (4) and

was set at 0.05. From this correction, only results from the ANOVAs or ANCOVAs of data analyzed at each day that yielded p-values of 0.0125 or less were considered to be significant. At the start of the experiment average body condition (MBC, VBC, RBC, and DBC) did not differ between groups; however, by day 3 RBC, MBC and VBC of spiders in the F+W+ and F-W- treatments were significantly different from each other. On day 8, a significant difference between the body condition of spiders in the F-W- and that of spiders in all other treatments

(when estimated with the RBC and VBC indices) emerged (Table 6). Estimated body condition of F+W+ spiders was significantly higher than that of spiders in the other treatments by day 3 using the VBC index, and by day 8 using the MBC and RBC indices. Although by day 5 the body condition of F-W+ spiders was also significantly different from that of spiders in the F+W+ treatment according to all but the DBC index, only on day 10 did the body conditions of spiders

100 in the single deprivation diet treatments (F+W- and F-W+) differ significantly (except in DBC, which showed no significant differences between these treatments).

Discussion

Our results show that, in S. ocreata, relative changes in morphological condition- dependent body size traits traditionally used in calculations of body condition (e.g., mass, abdomen volume, etc.) are similar in magnitude in response to short-term dehydration or starvation. Apparent body condition between spiders maintained on food only (F+W-) and water only (F-W+) diets remained similar for up to nine days, regardless of the BCI used to estimate body condition. Mass and abdomen volume of spiders in the F+W- and F-W+ treatments declined initially at approximately the same rate for the first 3 days of the experiment; however on day four a leveling off in these two variables was observed for spiders in the F+W- treatment, while spiders in the F-W+ experienced continued declines. Nonetheless, the rate at which both mass and abdomen dimensions decreased between the food only and water only treatments was slow, and only after 10 days on their respective diets did abdomen volume differ significantly between spiders in these groups (Table 6). From these results, it appears that starved and dehydrated spiders are compensating for the deprivations they experience and are seemingly able to resist the effects of these extreme conditions for up to 10 days before clear differences emerge.

However, it is not possible to determine here whether the lack of a perceived difference by the estimations of body condition is due to the inadequacy of current BCIs to detect differences between starvation and dehydration or reflects actual physiological states of the animals (i.e., energetic stores remain similar in short-term starved and short-term dehydrated spiders for a little over a week).

101

The persistence of a decreased body condition in spiders in the F-W+ treatment relative to spiders in the F+W- and F+W+ treatments suggests that physiological differences in energetic/fat stores likely exist well before external morphological differences appear. These finding also clearly show confirm that a decrease in body condition as a result of dehydration is recovered more quickly than when the decrease results from starvation. This indirectly supports the claim that morphological BCIs are able to detect real differences in body nutrient content.

As many have already demonstrated (e.g., Jakob et al. 1996; Moya-Loraño et al. 2008), we found that different body condition indices can produce different results, which can affect interpretations of data in important ways. Although all but the DBC indices indicated major differences in body condition between the negative and positive control groups (F+W+ and F-W-

, respectively) by day 3 (except DBC); in both experiments, the body condition of spiders maintained on single deprivation diets (F+W- and F-W+) did not statistically differ after 5 days regardless of the BCI used to estimate body condition. Indeed, body condition of spiders maintained on single deprivation diets remained similar to each other throughout the extended

10-day study (Exp. 2) and only on the final day of experiment 2 (day 10) did estimated body condition of spiders in these groups diverge significantly (RBC, VBC, and MBC, but not DBC).

Based on slight differences in the initial detection of differences in body condition between the

F+W+ and F-W- and between the F+W+ and single deprivation diet treatments (F+W- and F-

W+), the VBC index appears to be the most sensitive to immediate effects of starvation and/or dehydration in this study. Nonetheless, relative differences in estimated body condition between treatment groups were similar using MBC, VBC, and RBC. The DBC index was the least sensitive to short-term changes in diet and hydration; however, variance about the mean

102 estimates of body condition was generally high within treatments and across days which resulted in very low power (Fig. 2). Consequently, the probability of Type II error was high when using

DBC to estimate body condition. In contrast to the results reported by Moya-Loraño et al. 2008,

DBC (referred to as ‘SDBC’ in Moya-Loraño et al. 2008) had the lowest power to detect differences for the relatively small sample sizes used here, whereas the power to detect an effect of treatment on body condition was high using VBC and MBC from very early on in the experiment (Fig. 2). Given the conflicting results of the current study and those described in

Moya-Loraño et al. (2008), it is difficult to draw a conclusion about which index is superior. It is worth mentioning here that differences seen between this study and that of Moya-Laraño et al.

(2008) may be a result of species differences. Lycosa tarentula is much larger than S. ocreata, lives 18+ months, and is adapted to an arid environment (Moya-Loraño et al. 2008). On the other hand, S. ocreata is much smaller, lives less than a year, and is adapted to a mesic environment (with variable rainfall and RH% in the spring breeding season).

Although our results indicated that MBC, RBC and VBC show similar patterns in how diet affects body condition, relying on a combination of condition indices may prove to provide the best estimation of actual energetic reserves as suggested by Moya-Loraño et al. (2008). More studies that use larger sample sizes and longer sampling periods as well as those that compare direct measurements of fat reserves/body nutrient contents with various estimations of body condition are needed (especially for invertebrate animals).

The life span of S. ocreata is approximately one year, and because these spiders occur in temperate habitats, resource abundance is likely to fluctuate dramatically with season. In

103 addition, wolf spiders are ground-dwelling “sit-and-wait” predators and as such they are likely to experience variation in prey capture rates due to various factors that affect the distribution and abundance of size-appropriate prey items in the environment. It is then perhaps not surprising that these spiders should be able to cope with short-term starvation and or periods of dehydration, as they appear to be able to do. Similar resistance to dehydration and starvation have been observed in a number of other spider species. Resistance to dessication in spiders is owed in part to structures that help reduce water loss during excretion of metabolic wastes (e.g., coxal glands and malpighian tubules, Foelix 1996) and, in some spiders (e.g., those in which tracheae are found in addition to book lungs), mechanisms that help limit evaporation along respiratory surfaces (e.g., spiracle opening and closing, Davies and Edney 1952). In addition, many spider species show remarkable tolerances to long periods of starvation (e.g., black widows have been known to live for up to 200 days without food, Foelix 1996) which they are thought to accomplish through various adaptations (e.g., low resting metabolic rates and/or oxygen consumption, Anderson 1970; decreased metabolic rates in response to starvation: Itô

1964; Anderson 1970, 1974; Canals et al. 2007). Because spiders rely on a hydrostatic skeleton for movement, and the fluid within the hydrostatic skeleton (i.e., hemolymph) is the same as the fluid that functions in respiration (Foelix 1996), the effects of dehydration on both hydrostatic pressure and respiration may have more immediate consequences on behavioral performance and reproductive success than starvation. Nonetheless, little is known about how hydration affects morphological condition-dependent traits, and few studies consider how short-term or long-term hydration history (especially in field studies) relates to relationships between estimated body condition and behavioral and/or reproductive correlates. In the future, more attention should be given to understanding: 1) how dehydration, in combination with food deprviation or alone,

104 influences body condition and its correlates; as well as 2) how hydration history impacts discrepancies between estimated energetic stores and actual fat reserves.

105

References

Anderson, J. 1970. Metabolic rates of spiders. Comparative Biochemistry & Physiology 33: 51-

72.

Anderson, J. 1974. Responses to starvation in the spiders Lycosa lenta Hentz and Filistata

hibernalis (Hentz). Ecology 55: 576-585.

Barry, K.L., G.I. Holwell and M.E. Herberstein. 2010. Multimodal mate assessment by male

praying mantids in a sexually cannibalistic mating system. Animal Behavior 79: 1165-

1172.

Bjerke, R. & K.E. Zachariassen. 1997. Effects of dehydration on water content, metabolism, and

body fluid solutes of a Carabid beetle from dry savanna in East Africa. Comparative

Biochemistry and Physiology 118A: 779-787.

Canals, M., M.J. Salazar, C. Durán, D. Figueroa and C. Veloso. 2007. Respiratory refinements in

the mygalomorph spider Grammostola rosea Walckenaer 1837 (Araneae,

Theraphosidae). The Journal of Arachnology 35: 481-486.

Danielson-François, A., C.A. Fetterer and P.D. Smallwood. 2002. Body condition and mate

choice in Tetragnatha elongata (Araneae, Tetragnathidae). The Journal of Arachnology

30: 20-30.

Darlington, R.B. and T.W. Smulders. 2001. Problems with residual analysis. Animal Behaviour

62: 599-602.

Davies, M.E. and E.B. Edney. 1952. The evaporation of water from spiders. Journal of

Experimental Biology 29, 571–582.

106

Eraly, D., F. Hendrickx and L. Lens. 2009. Condition-dependent mate choice and its implications

for population differentiation in the wolf spider Pirata piraticus. Behavioral Ecology 20:

856-863.

Foelix, R. F. 1996. Biology of spiders, 2nd Ed. New York: Oxford University Press.

Fox, C.W. & J. Moya-Laraño. 2009. Diet affects female mating behaviour in a seed-feeding

beetle. Physiological Entomology 34: 370-378.

García-Berthou, E. 2001. On the misuse of residuals in ecology: testing regression residuals vs.

the analysis of covariance. Journal of Animal Ecology 70: 708-711.

Green, A.J. 2001. Mass/length residuals: Measures of body condition or generators of spurious

results? Ecology 82: 1473-1483.

Hebets, E.A., Wesson, J., and Shamble, P. 2008. Diet influences mate choice selectivity in adult

female wolf spiders. Animal Behaviour 76: 355-363.

Itô, Y. 1964. Preliminary studies on the respiratory energy loss of a spider, Lycosa

pseudoannulata. Researches on Population Ecology 6: 13-21.

Jakob, E.M., S.D. Marshall, and G.W. Uetz. 1996. Estimating fitness: a comparison of body

condition indices. OIKOS 77: 61-67.

Kotiaho, J.S. 1999. Estimating fitness: comparison of body condition indices revisited. Oikos 87:

399-400.

Krebs, C.J. & G.R. Singleton. 1993. Indices of condition for small mammals. Australian Journal

of Zoology 41: 317-323.

Lomborg, J.P. & S. Toft. 2009. Nutritional enrichment increases courtship intensity and

improves mating success in male spiders. Behavioral Ecology 20: 700-708.

107

Marshall, S.D., E.M. Jakob, & G.W. Uetz. 1999. Re-estimating fitness: can scaling issues

confound condition indices? Oikos 87: 401-402.

Moya-Laraño, J., R. Macías-Ordóñez, W.U. Blanckenhorn and C. Fernández-Montraveta. 2008.

Analysing body condition: mass, volume or density? Journal of Animal Ecology 77:

1099-1108.

Rolff, J. & G. Joop. 2002. Estimating condition: pitfalls of using weight as a fitness correlate.

Evolutionary Ecology Research 4: 931-935.

Scheuber, H., A. Jacot and M.W.G. Brinkhof. 2003. Condition dependence of a multicomponent

sexual signal in the field cricket Gryllus campestris. Animal Behavior 65: 721-727.

Schulte-Hostedde, A.I., J.S. Millar and G.J. Hickling. 2001. Evaluating body condition in small

mammals. Canadian Journal of Zoology 79: 1021.

Schulte-Hostedde, A.I., B. Zinner, J.S. Millar and G.J. Hickling. 2005. Restitution of mass-size

residuals: validating body condition indices. Ecology 86: 155-163.

Sibul, I., A. Kuusik, A. Luik and K. Voolma. 2006. Influence of environmental conditions on the

breathing rhythms of the pine weevil Hylobius abietis (Coleoptera: Curculionidae).

Agronomy Research 4: 63-77.

Virgl, J.A. & F. Messier. 1993. Evaluation of body size and body condition indices in muskrats.

The Journal of Wildlife Management. 57: 854-860.

Wilgers, D.J., A.C. Nicholas, D.H. Reed, G.E. Stratton and E.A. Hebets. 2009. Condition-

dependent alternative mating tactics in a sexually cannibalistic wolf spider. Behavioral

Ecology 20: 891-900.

108

Table 1. Pilot study data: Results of repeated measures ANOVAs for examining effects of diet manipulations on body condition using four different body condition indices. Approx. BCI Source df F p MBC Diet 3 14.6402 <0.0001 Day 5 79.0441 <0.0001 Day*Diet 15 6.4742 <0.0001 VBC Diet 3 13.5067 <0.0001 Day 5 89.1296 <0.0001 Day*Diet 15 7.1276 <0.0001 DBC Diet 3 2.1414 0.1064 Day 5 69.2109 <0.0001 Day*Diet 15 6.0089 <0.0001 RBC Diet 3 12.8443 <0.0001 Day 5 0.2534 0.936 Day*Diet 15 9.5058 <0.0001

109

Table 2. Exp.1: Effects of treatment on estimated body condition (MBC, VBC, DBC), calculated via ANCOVA for all treatments on days 1, 3 and 5. * indicates statistical significance (p<0.016) BCI Day Source df SS F p MBC 1 Treat. 3,3 0.0005 2.51 0.0701 CW 1,1 0.01371 208.11 <.0001* Treat * CW 3,3 0.00018 0.9116 0.4426 3 Treat. 3,3 0.00232 12.039 <.0001* CW 1,1 0.01357 211.64 <.0001* Treat * CW 3,3 0.00015 0.7561 0.5244 5 Treat. 3,3 0.00551 24.567 <.0001* CW 1,1 0.01204 160.99 <.0001* Treat * CW 3,3 2.1E-05 0.0932 0.9635 VBC 1 Treat. 3,3 0.07639 2.6626 0.0588 CW 1,1 0.25614 26.784 <.0001* Treat * CW 3,3 0.0149 0.5194 0.671 3 Treat. 3,3 0.26455 10.412 <.0001* CW 1,1 0.20821 24.585 <.0001* Treat * CW 3,3 0.01204 0.4737 0.7021 5 Treat. 3,3 0.59141 20.784 <.0001* CW 1,1 0.09607 10.128 0.0026 Treat * CW 3,3 0.00106 0.0371 0.9903 DBC 1 Treat 3,3 1.5E-05 0.3041 0.8222 CW 1,1 0.00269 162.85 <.0001* Treat*CW 3,3 6.1E-05 1.2219 0.3146 Ab Volume1/3 1,1 0.00151 91.552 <.0001* Treat* Ab Volume1/3 3,3 3.2E-05 0.6362 0.5962 CW* Ab Volume1/3 1,1 2.5E-05 1.4862 0.2301 Treat*CW* Ab Volume1/3 3,3 1.2E-05 0.2388 0.8687 3 Treat 3,3 6.1E-05 1.388 0.2604 CW 1,1 0.00222 152.18 <.0001* Treat*CW 3,3 6.7E-06 0.1521 0.9277 Ab Volume1/3 1,1 0.00131 89.925 <.0001* Treat* Ab Volume1/3 3,3 1.3E-05 0.2999 0.8253 CW* Ab Volume1/3 1,1 3E-08 0.0021 0.9638 Treat*CW* Ab Volume1/3 3,3 1.2E-05 0.274 0.8438 5 Treat 3,3 0.0002 2.2129 0.1015 CW 1,1 0.00235 76.914 <.0001* Treat*CW 3,3 0.00023 2.4717 0.0756 Ab Volume1/3 1,1 0.00153 50.227 <.0001* Treat* Ab Volume1/3 3,3 6.5E-05 0.7101 0.5517 CW* Ab Volume1/3 1,1 1.8E-06 0.0582 0.8106 Treat*CW* Ab Volume1/3 3,3 0.00013 1.4543 0.2415

110

Table 3. Exp. 1: Results from one-way ANOVAs for days 1, 3 and 5 to investigate the effect of treatment on RBC after specific durations.

Day df MS F p 1 3,49 0.00001 2.5782 0.0642 3 3,50 0.00004 13.1183 <0.0001 5 3,50 0.0001 21.1946 <0.0001

111

Table 4. Results from the multivariate repeated measures ANOVA (RBC) and ANCOVAs (MBC, VBC, DBC) to test for effects of starvation/dehydration on body condition, calculated four ways, over 10-days. Results reflect data from F+W+, F+W-, and F-W+.

BCI Source df F p MBC Diet 2,30 9.6372 0.0006 Day 10,21 5.8095 0.0003 Day*Diet 20,42 2.2776 0.0123 VBC Diet 2,30 17.0784* <0.0001 Day 10,21 10.5085*<0.0001 Day*Diet 20,42 3.6076 0.0002 DBC Diet 2,30 3.1050* 0.0595 Day 10,21 9.1763 <0.0001 Day*Diet 20,42 3.0905 0.001

RBC Diet 1,30 9.1108* 0.0008 Day 10,21 3.7622* 0.0051 Day*Diet 20,42 2.0365 0.0261

112

Table 5. Exp.2: Effects of treatment on estimated body condition (MBC, VBC, DBC), calculated via ANCOVA for all treatments on days 1, 3 and 5. * indicates statistical significance (p<0.016) BCI Day Source df SS F p MBC 0 Treat 3,3 0.0004 1.0086 0.4005 CW 1,1 0.0262193.7265 <0.0001* Treat*CW 3,3 0.0003 0.6297 0.6007 3 Treat 3,3 0.0038 9.9041 <0.0001* CW 1,1 0.0279219.1567 <0.0001* Treat*CW 3,3 0.0004 1.1623 0.338 5 Treat 3,3 0.0050 12.6338 <0.0001* CW 1,1 0.0260198.4627 <0.0001* Treat*CW 3,3 0.0002 0.5505 0.6512 VBC 0 Treat 3,3 0.1493 1.5599 0.2165 CW 1,1 0.7988 25.033 <0.0001* Treat*CW 3,3 0.0741 0.7739 0.5164 3 Treat 3,3 1.7590 16.9051 <0.0001* CW 1,1 1.077531.0668 <0.0001* Treat*CW 3,3 0.0615 0.5915 0.6247 5 Treat 3,3 1.8504 17.0251 <0.0001* CW 1,1 1.0518 0.4608 0.7115 Treat*CW 3,3 0.0501 0.4608 0.7115 DBC 0 Treat 3,3 0.0002 0.8751 0.4662 CW 1,1 0.004776.4449 <0.0001* Treat*CW 3,3 0.0001 0.6489 0.5905 Ab Volume1/3 1,1 0.002642.2897 <0.0001* Treat* Ab Volume1/3 3,3 0.0001 0.3371 0.7986 CW* Ab Volume1/3 1,1 0.0000 0.6522 0.4264 Treat*CW* Ab Volume1/3 3,3 0.0000 0.2281 0.876 3 Treat 3,3 0.0001 0.7785 0.5163 CW 1,1 0.003161.6559 <.0001* Treat*CW 3,3 0.0001 0.366 0.7781 Ab Volume1/3 1,1 0.002549.6027 <.0001* Treat* Ab Volume1/3 3,3 0.0001 0.8086 0.5002 CW* Ab Volume1/3 1,1 0.0000 0.0956 0.7596 Treat*CW* Ab Volume1/3 3,3 0.0000 0.2866 0.8346 5 Treat 3,3 0.0002 1.332 0.2846 CW 1,1 0.002452.1274 <.0001* Treat*CW 3,3 0.0001 0.8637 0.4719 Ab Volume1/3 1,1 0.00279960.7961 <.0001* Treat* Ab Volume1/3 3,3 0.000112 0.8135 0.4976 CW* Ab Volume1/3 1,1 2.21E-06 0.0481 0.8281 Treat*CW* Ab Volume1/3 3,3 7.91E-05 0.5723 0.6381

113

Table 6. Exp. 2: Results from one-way ANOVAs for days 1, 3 and 5 to investigate the effect of treatment on RBC after specific durations. Day df SS F p 0 3,39 0.00003 1.0145 0.3967 3 3,39 0.000252 8.7298 0.0001* 5 3,39 0.000293 13.1472 <0.0001*

114

Table 7. Results from ANCOVAs (MBC, VBC, DBC) on days 3, 5, 8 and 10 to investigate the effect of treatment on body condition in adult males. F-W- data excluded. *indicates significant difference (Bonferroni correct p=0.0125) BCI Day Source df SS F p MBC 0 Treat 2,2 0.0004 1.3413 0.2784 CW 1,1 0.0196141.2299 <.0001* Treat*CW 2,2 0.0002 0.8627 0.4333 3 Treat 2,2 0.00248.4281 0.0014* CW 1,1 0.0204144.5698 <.0001* Treat*CW 2,2 0.0003751.3321 0.2807 5 Treat 2,2 0.0024078.1679 0.0017* CW 1,1 0.0197133.4097 <.0001* Treat*CW 2,2 0.0002 0.6981 0.5063 8 Treat 2,2 0.003904 15.6475 <.0001* CW 1,1 0.0200160.3189 <.0001* Treat*CW 2,2 0.0001 0.4242 0.6586 10 Treat 2,2 0.0051 21.8171 <.0001* CW 1,1 0.0195166.9811 <.0001* Treat*CW 2,2 0.0001 0.3982 0.6754 VBC 0 Treat 2,2 0.1386 1.9892 0.1564 CW 1,1 0.510814.6587 0.0007* Treat*CW 2,2 0.0528 0.7583 0.4782 3 Treat 2,2 1.000713.0754 0.0001* CW 1,1 0.816721.3442 <.0001* Treat*CW 2,2 0.0611 0.7986 0.4603 5 Treat 2,2 0.92181611.2285 0.0003* CW 1,1 0.75270218.337 0.0002* Treat*CW 2,2 0.0445920.5432 0.5871 8 Treat 2,2 1.4115 21.8414 <.0001* CW 1,1 0.652420.1922 0.0001* Treat*CW 2,2 0.0264 0.4083 0.6688 10 Treat 2,2 1.8277 35.0297 <.0001* CW 1,1 0.569121.8168 <.0001* Treat*CW 2,2 0.0465 0.8921 0.4215 DBC 0 Treat 2,2 4.47E-05 0.3419 0.7143 CW 1,1 0.00394660.3938 <.0001* Treat*CW 2,2 0.0001010.7694 0.4759 Ab Volume1/3 1,1 0.00193829.6572 <.0001* Treat* Ab Volume1/3 2,2 4.56E-05 0.3493 0.7092 CW* Ab Volume1/3 1,1 6.1E-05 0.9329 0.3451 Treat*CW* Ab Volume1/3 2,2 1.23E-050.0941 0.9106 3 Treat 2,2 5.22E-050.4391 0.6504 CW 1,1 0.00211735.6512 <.0001*

115

BCI Day Source df SS F p DBC(cont) 3 Treat*CW 2,2 5.52E-05 0.4647 0.6346 Ab Volume1/3 1,1 0.00207935.0035 <.0001* Treat* Ab Volume1/3 2,2 0.000114 0.9571 0.4001 CW* Ab Volume1/3 1,1 1.18E-08 0.0002 0.9889 Treat*CW* Ab Volume1/3 2,2 2.48E-05 0.209 0.8131 5 Treat 2,2 0.0001731.6818 0.2101 CW 1,1 0.00155130.2045 <.0001* Treat*CW 2,2 0.0001081.0544 0.3661 Ab Volume1/3 1,1 0.00266551.8984 <.0001* Treat* Ab Volume1/3 2,2 0.000113 1.1032 0.3503 CW* Ab Volume1/3 1,1 3.54E-06 0.0689 0.7956 Treat*CW* Ab Volume1/3 2,2 7.9E-050.7692 0.476 8 Treat 2,2 5.76E-050.5503 0.5849 CW 1,1 0.00187935.8864 <.0001* Treat*CW 2,2 9.8E-050.9361 0.4079 Ab Volume1/3 1,1 0.00164131.3272 <.0001* Treat* Ab Volume1/3 2,2 6.76E-05 0.6452 0.5346 CW* Ab Volume1/3 1,1 2.29E-05 0.4371 0.5157 Treat*CW* Ab Volume1/3 2,2 1.58E-050.1509 0.8609 10 Treat 2,2 0.0001171.1296 0.342 CW 1,1 0.00165331.8863 <.0001* Treat*CW 2,2 6.85E-050.6609 0.5268 Ab Volume1/3 1,1 0.00094918.3101 0.0003* Treat* Ab Volume1/3 2,2 0.000138 1.3294 0.286 CW* Ab Volume1/3 1,1 9.5E-07 0.0184 0.8934 Treat*CW* Ab Volume1/3 2,2 4.02E-050.3881 0.6831

116

Table 8. Exp. 2: Results from one-way ANOVAs for days 1, 3, 5, 8 and 10 to investigate the effect of treatment on RBC after specific durations. Exclused F-W- treatment. *indicates significant effect (Bonferroni corrected p=0.0125). Day df MS F p 0 2,30 0.00005 1.4545 0.273 3 2,30 0.00016 4.6019 0.0181 5 2,30 0.0002 7.6238 0.0021* 8 2,30 0.00029 13.7093<0.0001* 10 2,30 0.00045 20.1516<0.0001*

117

Figure 1. Experiment 1: Response in body condition to feeding/hydration treatments over time .

Estimated mean Adjusted means shown for MBC, VBC, and DBC are least square means controlled for body size (cephalothorax width).

118

119

Figure 2. Survivorship (%) of spiders maintained in the absence of food and water (F-W-) over

10 days.

120

121

Figure 3. Experiment 2: Response in body condition to feeding/hydration treatments over 10- days. Estimated mean adjusted means shown for MBC, VBC, and DBC are least square means controlled for body size (cephalothorax width). ‘After’ category on the x-axis represents measurements taken of body condition 24-hours after spiders were removed from their feeding treatments and were fed and given access to water.

122

123

Chapter 5: General Conclusions

Jenai M. Rutledge

Department of Biological Sciences

University of Cincinnati

614 Rieveschl Hall

Cincinnati, OH USA 45221-0006

Corresponding author – [email protected]; Tel: 513-556-9753, Fax: 513-556-5299

124

Within the research presented here, I examined and tested a number of assumptions both about an emerging animal model (wolf spiders) and some commonly used experimental methods/techniques. From this research, it is clear that behavioral flexibility in female mate preferences in wolf spiders is higher than previously recognized; however, the manner in which female preferences are affected by environmental/social factors is unpredictable (Chapter 1).

Because the effects of external cues on mating behavior is not yet well understood, precautions designed to limit the extent to which environmental factors and/or social experience vary between research subjects (unless it is the express purpose of the project to examine the effects of these factors) should be taken. Furthermore, results from research presented in chapter 2 regarding the use of nail polish for manipulating animal phenotypes highlights the importance of acknowledging where lapses in our understanding an animal’s sensory systems exist rather than overlooking them. Finally, results from research presented in the third chapter regarding the use of various body condition indices to estimate energetic reserves in spiders, suggests that body condition indices are not well-adapted for distinguishing between differences in morphological traits caused by starvation and those caused by dehydration.

Historically, most invertebrates (including spiders) have been considered as behaviorally inflexible; i.e., behaviors are innate and under strong genetic control. However, evidence to the contrary continues to mount (e.g., field crickets, Gyrllus lineaticeps, Wagner et al. 2001; fruit flies, Drosophila melanogaster, Ödeen & Moray 2008; fishing spiders, Dolomedes tenebrosous

Johnson 2005). For many invertebrates, the assumption of behavioral inflexibility has been one of convenience, as relatively little is known about their physiology, behavior and/or ecology.

Even for invertebrates that serve as model organisms in molecular, cellular, and/or genetic

125 research (e.g., Drosophila, Caenorhabditis elegans) and for which entire genomes have been sequenced, knowledge of their natural history remains surprisingly limited. Within the well- studied wolf spider genus Schizocosa a number of interesting discoveries regarding plasticity in female mate preferences for both naturally occurring (e.g., Hebets & Vink 2007) and novel/artificial traits (Hebets 2003) have been made. The implications of these findings are interesting on many levels, but especially when considering the evolution of male ornamentation within this genus (common in many of these species, Stratton 2005), because of its importance in testing sexual selection theory. Results from my study about the effects of juvenile exposure to artificially-manipulated male phenotypes on female mate preferences (Rutledge et al. In press), supports previous findings that adult female mate preferences in these wolf spiders can be influenced by experience, and that the specific type of experience appears to be important

(Hebets 2003, Hebets & Vink 2007). However, in contrast to the results of past studies, I found that experience did not lead to increased female mate preference for males or male traits to which the females had been previously exposed as juveniles; instead, females were less likely to accept males possessing familiar phenotypes as mates.

Despite growing knowledge about how visual, seismic, and even tactile cues influence female mating behavior and mate choice in Schizocosa spiders (Hebets & Uetz 1999;

McClintock & Uetz 1996; Scheffer et al. 1996; Uetz & Roberts 2002; Hebets et al. 2006), our understanding of how they use chemical cues in their environment for prey capture, predator detection and mate decisions is lacking (but see Persons et al. 2001; Persons & Rypstra 2001).

Nonetheless, it is clear from the study on the effects of juvenile female experience with artificially-modified male phenotypes on female mate preferences (Chapter 1), that chemical

126 cues may be important to female mate choice. Females may be able to associate chemical cues with potential mates and/or recognize and discriminate against chemical phenotypes that they encountered as juveniles – which may be indicative of heterospecifics.

In addition to examining the effects of juvenile experience on adult female mate preferences, it was also shown how an artificial method for modifying male phenotypes, i.e., nail polish, can have unintended and uncontrolled effects when used in behavioral studies. Nail polish was chosen as the targeted substance for analysis because it was used in prior studies to modify male phenotypes in an effort to examine how subadult female experience with different male phenotypes might influence subsequent adult mate preferences (Hebets 2003; Hebets et al.

2006). In these previous studies, results were interpreted as if only visual modifications were made to male phenotypes. As a phenotype modifier, nail polish has a number of advantages, e.g., quick drying formulas are readily available, it is highly viscous which makes it easier to apply evenly to even small surfaces and once dry, it is difficult to remove mechanically (i.e., via grooming). However, nail polish also contains highly volatile compounds, many of which remain and continue to volatilize into the air around the surface upon which it is applied for at least 24- hours. For the purposes of investigating if and how different colors of nail polish vary in their chemical properties, a chemical analysis of three different colors of nail polish (black, brown, and clear) was conducted using a GC Mass Spetrophotometer. From the GC Mass-spectrometer analysis, it was shown that different colors of nail polish have different aromatic chemical properties. As a result, using different nail polish colors may not only modify the visual phenotype of the animal it is applied to, but also its chemical phenotype. Unfortunately, many of the readily available tools that could be used to manipulate male phenotypes are also likely to

127 contain a chemical element (e.g., Sharpie® markers, other types of paint, art dye, etc.), the effects of which are ultimately difficult to detect and/or study. As a consequence, these results demonstrate the importance of choosing and using artificial materials that may contain aromatic compounds to mimic nature when studying spiders, or invertebrates in general.

Although effects of juvenile experience with different conspecific male phenotypes on adult female mate preferences was observed in S. rovneri (Rutledge et al. In press) and S. uetzi

(Hebets 2003; Hebets et al. 2006), exposure of juvenile females to adult heterospecific male courtship in the wolf spiders S. ocreata and S. rovneri had limited effects on adult female behavior. These results do not match the results of a previously published study on a population of wolf spiders in Mississippi that resemble S. ocreata and S. rovneri, in which experience with either tufted (S. ocreata) or untufted (S. rovneri) males led to increased adult female mate preference for the male ‘morph’ to which they were exposed as juveniles. In contrast, the data presented here show that where effects of experience were observed, preference for familiar males was not clearly established, as female S. ocreata that gained experience as juveniles with either heterospecific or conspecific males were generally more receptive to heterospecific males than females that had received no juvenile exposure. In S. rovneri female experience seemed to cause a decrease in the performance of female receptive behavior overall, regardless of experience type or the species of male with which they were paired as adults.

From these results, it appears that shifts in adult female preference due to juvenile experience with heterospecific or conspecific adult male courtship are limited at best, and are not likely to play any significant role in maintenance/break-down of isolating behavioral species

128 barriers observed in these closely related species. Nor are such effects likely to explain the occasional existence of hybrids in the field. Based on the findings presented in Chapter 2 and 3, it can be concluded that female mate preferences may be affected by environmental and social cues. In order for male secondary sexual ornamentation (i.e., decorations that play little to no role in improving survival of the animal possessing them, but that may increase that individual’s mating success) to evolve, according to the theory of sexual selection, a directional preference for the trait must exist in a high proportion of the female population (Andersson 1994). Plasticity in female mate preferences may ultimately act to reduce directional selection on male traits resulting from female preferences, which could lead to disruptive or stabilizing selection for such trait within a population. From this, variation in male phenotypes within a population, as is seen in the ornamentation of male S. uetzi (Hebets 2003), may evolve.

Body condition, which generally refers to an animal’s lipid storage or primary energetic reserves, is known to be another important factor in both female and male mate preferences

(Danielson-François et al. 2002; Scheuber et al. 2003; Eraly et a. 2009; Lomborg & Toft 2009;

Wilgers et al. 2009; Barry et al. 2010). Methods for estimating body condition are diverse, however usually when invasive sampling/quantification of actual body contents of an animal is not feasible, body condition is commonly estimated from an index that uses a condition- dependent body size indicator (CDBSI, i.e., a trait whose size fluctuates with energy stores). To separate the effects that structural body size may have on variation in condition-dependent traits, from variation of those traits due to fluctuations in energetic reserves, a non-condition dependent trait that correlates with structural body size (e.g., bone length, carapace width) is incorporated into the index, either by directly dividing the size of the condition-dependent trait by the

129 structural body size indicator (SBSI, i.e., ratio body condition index), using residuals from an ordinary least squares regression the CDBSI against SBSI to estimate condition, or by putting the

SBSI into an ANCOVA model as a covariate. Although the short-comings of body condition indices that estimate condition from morphological measurements are well-documented and include concerns about practical and statistical validity of certain metrics (e.g., Jakob et al. 1996,

Kotiaho 1999; Marshall et al. 1999; Green 2001; Darlington & Smulders 2001), as well as the lack of a strong correlation between CDBSI and actual fat stores (Virgl & Messier 1993; Krebs

& Singleton 1993; Moya-Loraño et al. 2008) few have directly examined how dehydration influences estimations of body condition using morphological indices. Under short (5-day) and moderate (10-day) periods during which S. ocreata spiders were maintained with water but no food, or with food but no water, body condition (as estimated from size-adjusted body mass (i.e.,

SBSI = cephalothorax width, MBC), size-adjusted abdomen volume (VBC), size-adjusted abdomen density (DBC) and residuals generated from an OLS regression of body mass (CDBSI) against cephalothorax width (SBSI)), remained similar and differences in body condition as a result of dehydration versus starvation were not observable until day 9 or 10, depending on the body condition index used. Although, tolerance to periods of starvation and dehydration has been well-documented in spiders (Davies & Edney 1952; Anderson 1970, 1974), it was surprising to find that the effects on the external apparent body condition in wolf spiders of starvation and dehydration are similar. From this we can conclude that spiders are to some extent, able to compensate for short-term water and food deprivations, e.g., under conditions where water is not accessible spiders may be able to resist desiccation by taking advantage of water contained within their prey, and likewise, under conditions where food is not available spiders may

130 increase their uptake of water, possibly to maintain locomotory function in even as energetic stores decrease.

As science progresses at a more and more rapid pace, it can be easy to overlook the simple assumptions upon which we base our research. Either as an artifact of historic prejudice/perception about the animals we study, or simple confounds inherent within the tools and/or experimental designs we employ, the validity of many minor (and major) assumptions often goes untested. Although serious concerns can arise about conclusions drawn from assumptions that are later found to be false, it is much more common that by not examining further the fundamental assumptions within our own research, we unnecessarily limit interpretations of data and consequently our understanding of our research systems. Results of the research in this dissertation contribute to the increasing body of work that demonstrates the need to consider more closely how assumptions in our research affect and/or bias our conclusions.

131

References

Anderson, J. 1970. Metabolic rates of spiders. Comparative Biochemistry & Physiology 33: 51-

72.

Anderson, J. 1974. Responses to starvation in the spiders Lycosa lenta Hentz and Filistata

hibernalis (Hentz). Ecology 55: 576-585.

Andersson, M. 1994. Sexual Selection. Princeton University Press: Princeton

Barry, K.L., G.I. Holwell and M.E. Herberstein. 2010. Multimodal mate assessment by male

praying mantids in a sexually cannibalistic mating system. Animal Behavior 79: 1165-

1172.

Danielson-François, A., C.A. Fetterer and P.D. Smallwood. 2002. Body condition and mate

choice in Tetragnatha elongata (Araneae, Tetragnathidae). The Journal of Arachnology

30: 20-30.

Darlington, R.B. and T.W. Smulders. 2001. Problems with residual analysis. Animal Behaviour

62: 599-602.

Davies, M.E. and E.B. Edney. 1952. The evaporation of water from spiders. Journal of

Experimental Biology 29, 571–582.

Eraly, D., F. Hendrickx and L. Lens. 2009. Condition-dependent mate choice and its implications

for population differentiation in the wolf spider Pirata piraticus. Behavioral Ecology 20:

856-863.

Green, A.J. 2001. Mass/length residuals: Measures of body condition or generators of spurious

results? Ecology 82: 1473-1483.

132

Hebets, E.A. 2003. Subadult experience influences adult mate choice in an arthropod: exposed

female wolf spiders prefer males of a familiar phenotype. Proceedings of the National

Academy of Sciences, USA 100: 13390-13395.

Hebets, E.A. and G.W. Uetz. 1999. Female responses to isolated signals from multimodal male

courtship displays in the wolf spiders genus Schizocosa (Araneae: Lycosidae). Animal

Behaviour 57: 865-872.

Hebets, E.A. and C.J. Vink. 2007. Experience leads to preference: experienced females prefer

brush-legged males in a population of syntopic wolf spiders. Behavioral Ecology 18:

1010-1020.

Hebets, E.A., Cuasay, K. & Rivlin, P.K. 2006. The role of visual ornamentation in female choice

of a multimodal male courtship display. Ethology 112: 1062-1070.

Jakob, E.M., S.D. Marshall, and G.W. Uetz. 1996. Estimating fitness: a comparison of body

condition indices. OIKOS 77: 61-67.

Johnson, C. 2005. Cohabitation of juvenile females with mature males promotes sexual

cannibalism in fishing spiders. Behavioral Ecology 16: 269-273.

Kotiaho, J.S. 1999. Estimating fitness: comparison of body condition indices revisited. Oikos 87:

399-400.

Krebs, C.J. & G.R. Singleton. 1993. Indices of condition for small mammals. Australian Journal

of Zoology 41: 317-323.

Lomborg, J.P. & S. Toft. 2009. Nutritional enrichment increases courtship intensity and

improves mating success in male spiders. Behavioral Ecology 20: 700-708.

Marshall, S.D., E.M. Jakob, & G.W. Uetz. 1999. Re-estimating fitness: can scaling issues

confound condition indices? Oikos 87: 401-402.

133

McClintock, W.J. & Uetz, G.W. 1996. Female choice and pre-existing bias: visual cues during

courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour 52:

167-181.

Ödeen, A. and C.M. Moray. 2008. Drosophila melanogaster virgins are more likely to mate with

strangers than familiar flies. Naturwissenschaften 95: 253-256.

Persons, M.H. & 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.

Persons, M.H., Walker, S.E., Rypstra, A.L. & Marshall, S.D. 2001. Wolf spider predator

avoidance tactics and survival in the presence of diet-associated predator cues (Araneae:

Lycosidae). Animal Behavior 61: 43-51.

Scheffer, S.J., G.W. Uetz, and G.E. Stratton. 1996. Sexual selection, male morphology, and the

efficacy of courtship signaling in two wolf spiders (Araneae: Lycosidae). Behavioral

Ecology and Sociobiology 38: 17-23.

Scheuber, H., A. Jacot and M.W.G. Brinkhof. 2003. Condition dependence of a multicomponent

sexual signal in the field cricket Gryllus campestris. Animal Behavior 65: 721-727.

Stratton, G. E. 2005. Evolution of ornamentation and courtship behavior in Schizocosa: insights

from a phylogeny based on morphology (Araneae: Lycosidae). The Journal of

Arachnology 33: 347-376.

Uetz, G.W., and Roberts, J. A. 2002. Multi-sensory cues and multi-modal communication in

spiders: insights from video/audio playback studies. Brain Behaviour & Evolution 59:

222-230.

134

Virgl, J.A. & F. Messier. 1993. Evaluation of body size and body condition indices in muskrats.

The Journal of Wildlife Management. 57: 854-860.

Wagner, W.E., M.R. Smeds, and D.D. Wiegmann. 2001. Experience affects female responses to

male song in the variable field cricket Gyrllus lineaticeps (Orthoptera, Gryllidae).

Ethology 107: 769-776.

Wilgers, D.J., A.C. Nicholas, D.H. Reed, G.E. Stratton and E.A. Hebets. 2009. Condition-

dependent alternative mating tactics in a sexually cannibalistic wolf spider. Behavioral

Ecology 20: 891-900.

135

Appendix I: Behavioral assays for predicting mating success in two species of wolf spiders:

rates of general behaviors, male signals, and female receptivity

Jenai Rutledge

136

In studies of mate preference, the ultimate measure of female choice is mating. However, in many studies (e.g., experimental cue isolation, video and seismic playback, or in systems where mating is infrequent or difficult to observe), indirect measurements that are highly predictive of the probability of mating can be useful. Such indirect measures may include histological evaluations of reproductive structures (e.g., egg development, Contreras-Garduño et al. 2007); hormone assays (e.g., testosterone levels, Vitousek et al. 2010) or simple quantification of behaviors or behavioral sequences that are known to precede most successful copulations. In many studies for which invasive assays of female receptivity are undesirable, or when allowing males and females to physically interact may result in unwanted reductions in sample sizes (e.g., via sexual cannibalism), the latter method is often best. Methods that rely on the quantification of stereotypical behaviors/sequences of behaviors are common across a variety of taxa including fish (e.g., guppies Poecilia reticulata, Houde 1988; swordtails Xiphophorus nigrensis, Cummings & Mollaghan 2006), birds (e.g., Japanese quail Coturnix japonica, White

2004), insects (e.g., parasitoid wasp Spalangia endius, King & Dickenson 2008) and spiders

(e.g., wolf spiders Lycosa rabida: Rovner 1968; Schizocosa ocreata & S. rovneri: Uetz &

Denterlein 1979, Stratton & Uetz 1981).

In spiders, where female cannibalism of males is common, using a behavioral proxy to assess female sexual receptivity rather than copulation can be particularly valuable, especially where males need to be used multiply (e.g., repeated-measures designs, or limited sample sizes – as can be common when using field-caught animals). However, even in live mating trials, having additional, “back-up” metrics to assess female receptivity/mate choice can be useful – the fact that mating does not occur during a given period of time does not always mean that the female

137 was not receptive. In some cases, males are hesitant to mate, or for unknown reasons or seem to have difficulty locating the female (personal observation). Indeed, male “personalities” or behavioral syndromes may affect mating outcome in many ways – bolder males may attempt to force copulation even when the female is unwilling (Johns et al. 2009), or more timid males may be put off by an overly eager female. Although behavioral syndromes in S.ocreata and S. rovneri have not been directly investigated, differences in male aggression between Schizocosa species has been observed (Hebets & Vink 2007) and there is evidence of “personality” effects in territorial interactions of the cellar spider, Pholcus phalangioides (Araneae, Pholcidae; Rypstra et al. 2010). As a consequence, measuring the extent to which a female was receptive to a male during a trial can allow for a more thorough interpretation of data regarding female mate choice.

In general, a good measure of receptivity is one that is strongly correlated with mating, is easily observed (and easily defined), and reproducible. Receptive behaviors are often species specific, and it is rare that overlap between species in receptive displays occurs; however, in closely related sibling species, or even in cryptic species complexes, female receptive behaviors may be similar or identical, as is the case with females in the wolf spider sibling species S. ocreata and S. rovneri (Uetz & Denterlein 1979; Stratton & Uetz 1981). Female S. ocreata and

S. rovneri exhibit stereotypic behaviors in response to male courtship, which typically precede copulation and indicate receptivity and/or willingness to mate (Uetz & Denterlein 1979; Stratton

& Uetz 1981) including: slow pivot (90to 180 slow turn(s)), tandem leg extend (forward extension of front two pairs (LI, LII and RI, RII) of legs while tilting body towards the substratum with abdomen slightly lifted), and settle (lowering of the cephalothorax to the substratum while keeping abdomen slightly lifted). When unreceptive, females lunge at and/or

138 may cannibalize the male. Research on these species commonly relies on a “composite receptivity score” which is a sum of the total number of receptive displays performed by a female during a trial minus aggressive behaviors exhibited towards the male. Although this composite score usually correlates well with mating outcome, it can be misleading when mating is permitted, because the length of those trials will be shorter than trials in which mating does not occur. As a result, when using a composite receptivity score, females who mate early on in a trial may appear to have been less receptive or even ‘coercively mated’ (relative to females who did not mate but were receptive), simply because the time in which the females displayed receptivity was different. Using a composite receptivity score also makes it difficult to generate a description of ‘normal’ female receptiveness for each species for comparison of reported results from multiple studies, as differences in trial durations or conditions may influence the total number of behaviors performed. Such comparisons may ultimately help to further elucidate behavioral differences between females of these two species (which currently are indistinguishable in morphology and behavior).

One possible solution is to calculate female receptivity as a rate (number of receptive displays / trial duration). Using the data set from chapter 3, in which S. ocreata and S. rovneri females were paired with heterospecific or conspecific males after exposure as juveniles to adult male courtship, I demonstrate how composite receptivity scores compare to rate scores in their predictive abilities of mating outcome and show how receptivity rates may be informative for detecting species differences. Because mating occurred in all but one of the trials conducted with

S. ocreata females in which the females were paired with conspecific males, it was not possible to examine differences in female receptivity to conspecific males alone; consequently, the results

139 shown here reflect female responses to both heterospecific and conspecific males. Composite receptivity and receptivity rates were calculated from the all behaviors observed during the entire duration of the trial (ranging from 5-10 minutes, depending on if/when mating occurred relative to the removal of the clear plastic barrier separating the male from the female, see methods in chapter 3).

In general, both composite receptivity and composite receptivity rates are strongly related to mating outcome in both S. ocreata and S. rovneri (Figure 1). And, in both species, females that ended up mating had higher composite receptivity scores (mean + SE, S. ocreata: 24.86 +

2.26; S. rovneri: 13.43 + 1.49) and display rates (mean + SE, S. ocreata: 3.94 + 0.39 displays per minute; S. rovneri: 1.99 + 0.26 displays per minute) than those who did not. Although the data for both species, in this data set, were collected under the same conditions and maximum trial duration (and are thus comparable regardless of whether rate is calculated for each species or not), I have used rate here to show how comparisons between species can be made. When average receptivity rates are compared between species (Figure 2), of the females who mated, female S. ocreata displayed more behaviors per unit time than female S. rovneri (t-test: t=-4.75, df=72, p<0.0001), whereas no difference is detected in female receptivity between species among females who did not mate (t=-0.281, df=168, p=0.7791). This observation that S. ocreata female responses to male courtship appear to be higher on average than that of S. rovneri females, mirrors differences seen in the intensities (and complexity) of male courtship behavior, i.e., S. ocreata males have much more vigorous and conspicuous courtship displays than S. rovneri males. One current hypothesis regarding the evolutionary divergence of these two species is that as populations of the ancestral species invaded floodplain leaf litter habitats,

140 predation pressures were more intense in these new habitats. As a result of decreased habitat complexity, conspicuous secondary sexual characteristics (i.e., foreleg pigment and tufts of bristles) and a conspicuous courtship display may have been selected against, ultimately resulting in the secondary loss of visual ornamentation and courtship displays in male S. rovneri

(Stratton 2005). The fact that female S. rovneri also appear to perform fewer receptive displays than female S. ocreata may lend further support to the hypothesis that conspicuous behavior

(e.g., receptive behavior, complex visual male courtship) is selected against in the floodplain habitats that S. rovneri can be found. Clearly further study of species differences is needed to determine whether the behavioral differences reported here represent actual species-level differences. One way to further investigate whether female receptivity differs between species and/or whether any differences can be linked to environmental factors, could be to examine if/how female responses to male courtship changes in the presence of predator cues and/or whether one species has a stronger response to predator cues than the other. Ultimately, the novel measure proposed here improves upon the “composite receptivity” standard that is used throughout wolf spider sexual selection research, and if implemented, could help better understand differences that exist between closely related species at the level of the female (rather than the male alone).

141

References

Cummings, M. & D. Mollaghan. 2006. Repeatability and consistency of female preference

behaviours in a northern swordtail, Xiphophorus nigrensis. Animal Behaviour 72: 217-

224.

Hebets, E.A. & C.J. Vink. 2007. Experience leads to preference: experienced females prefer

brush-legged males in a population of syntopic wolf spiders. Behavioral Ecology 18:

1010-1020.

Houde, A.E. 1988. The effects of female choice and male-male competition on the mating

success of male guppies. Animal Behaviour 36: 888-896.

Johns, J.L., Roberts, J.A., Clark, D.L., & G. W. Uetz. 2009. Love bites: male fang use during

coercive mating in wolf spiders is potentially costly to females. Behavioral Ecology and

Sociobiology 64:13-18.

Rovner, J.S. 1968. An analysis of display in the lycosid spider Lycosa rabida Walckenaer.

Animal Behaviour 16: 358-369.

Rypstra, A.L., S.A. Nagy, & C.D. Hoefler. 2010. Personality influences the effects of recent

feeding history on web defense by males and females of the cellar spider, Pholcus

phalangioides (Araneae, Pholcidae). Presentation at 34th American Arachnological

Society Meeting, Greenville, North Carolina USA.

Stratton, G. E. 2005. Evolution of ornamentation and courtship behavior in Schizocosa: insights

from a phylogeny based on morphology (Araneae, Lycosidae). The Journal of

Arachnology 33: 347-376.

Stratton, G.E. & G.W. Uetz. 1981. Acoustic communication and reproductive isolation in two

species of wolf spiders (Araneae:Lycosidae). Science 214:575-577.

142

Uetz, G.W. & G. Denterlein. 1979. Courtship behavior, habitat and reproductive isolation in

Schizocosa rovneri Uetz & Dondale (Araneae: Lycosidae). Journal of Arachnology 7:

121-128.

Vitousek, M.N., M.A. Mitchell, L.M. Romero, J. Awerman, & M. Wikelski. 2010. To breed or

not to breed: physiological correlates of reproductive status in a facultatively biennial

iguanid. Hormones and Behavior 57: 140-146.

White, D.J. 2004. Influences of social learning on mate-choice decisions. Learning & Behavior

32: 105-113.

143

Figure 1. Logistic regressions of female composite receptivity scores or female composite

receptivity rates (displays/minute) against mating outcome in response to courtship of

conspecific or heterospecific males. (A) S. ocreata: i. composite receptivity –

2 2 2 X 1=45.941, p<0.0001, (R =0.385); ii. composite receptivity rate - X 1=60.505,

2 2 p<0.0001, (R =0.515). (B) S. rovneri: i. composite receptivity – X 1=41.621, p<0.0001,

2 2 2 (R =0.322); ii. composite receptivity rate - X 1=56.425, p<0.0001, (R =0.436) .

144

(A) S. ocreata

Mated? (y/n) Mated? (y/n)

(B) S. rovneri Mated? (y/n) Mated? (y/n)

145

Figure 2. Comparison of female receptivity rates (displays/minute) between species for mated

and unmated females.

146

147