Vibratory Communication in the Black Widow Spider, Latrodectus Hesperus (Araneae: Theridiidae)

Home , Allocosa brasiliensis, Latrodectus pallidus

Vibratory Communication in the Black Widow Spider, Latrodectus hesperus (Araneae: Theridiidae)

Senthurran Sivalinghem

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Ecology and Evolutionary Biology University of Toronto

Vibratory Communication in the Black Widow Spider, Latrodectus hesperus (Araneae: Theridiidae)

Senthurran Sivalinghem

Doctor of Philosophy

Department of Ecology and Evolutionary Biology University of Toronto

2020 Abstract

Several studies have described vibration producing behaviours across many web-building spiders, and vibratory communication is thought to play an integral role during male-female interactions. Despite the presumed ubiquity of vibratory communication in this group of spiders, very little is known about the characteristics and functions of the signals involved, how signals are produced and transmitted through webs, or how vibrations are perceived. In this thesis, I used the western black widow spider, Latrodectus hesperus, as my focal organism, to investigate the details of vibratory communication from sender to the receiver. My results show that male L. hesperus courtship vibration signals comprise three distinct components (abdominal tremulation, bounce and web plucks), each produced using different signal production mechanism. Larger males produced bounce and web pluck signals with high power, which suggests that these signals may carry information about male traits. I found that during the early phase of courtship, males produced these different signal components haphazardly, with little temporal organization among the individual components (unstructured signaling). However, during the later phase of courtship, as males approach females, males intermittently organized signal components into a stereotyped temporal sequence (structured signaling). I tested the importance of these composite multicomponent signals, and found that males that displayed these structured signals more often ii were more likely to successfully copulate and copulate sooner. Vibrations arriving at the females are transmitted through the legs. My results show that the more distal joints of the legs (i.e. tarsus-metatarsus joint) move more to higher frequencies, and more proximal joints are tuned to lower frequencies. This suggests that female leg-joints play an important role in segregating vibration frequencies, and this may have important implications for prey/mate discrimination, as well as vibration source localization. These results show that L. hesperus males employ multiple signal components during courtship, and provide novel insights into emergent signal complexities in a web-building spider, previously thought to be ‘simple’ signalers. Additionally, the spider body mechanics can play a key role in influence vibration perception.

iii En Kudumbathirkaaha

In Memory Of Thaveswaran Kandavanam (Thavam Sithappa)

iv Acknowledgments

This thesis would not have been possible without the support and input from numerous people. First, and foremost, I am forever grateful to my mentor Andrew Mason for introducing me to research in invertebrate sounds and vibrations, and in particular vibratory communication in spiders. It felt like kismet. I thank Andrew for being a supportive advisor and an inspiring mentor to me during my scientific and personal development. His endless support and guidance gave me the confidence I needed to explore my own research interests independently and complete research projects. It is not possible for me to put in words how much I value Andrew’s mentorship, and what it has meant to me. Andrew was always available to hear my doubts and questions and provide input and feedback to help me find clarity, both in life and research. It was an honour and a privilege to be his student.

I am also thankful to my thesis committee members, Maydianne Andrade and Ken Welch, for their insights, guidance, and perspectives on my research projects, and for giving me feedback, comments and suggestions on my manuscripts. I thank Maydianne for allowing me to use her lab resources, when needed, to keep my spider populations alive and to conduct mating experiments. Maydianne has also provided me numerous opportunities to communicate my science to elementary school children and to the public, to help me grow as a science educator and communicator. As a dazed and confused freshman many years ago, not knowing what I wanted to do, meeting Maydianne was one of the most significant moments in my life. I am always inspired by her intelligence, integrity, and passion and enthusiasm for science.

My PhD experience was greatly enhanced by all my colleagues and peers in the Mason lab. I thank my current lab-mates Terrence Chang and Andrew Masson (AJ) for the many interesting and thoughtful conversations, which were much needed from time to time. Working with Natasha Mhatre was truly an amazing learning experience, which helped me grow as a researcher. Natasha was always generous with her time and provided great support in training me on the scanning laser vibrometer and explaining many concepts related to biomechanics. I am also grateful for the support and genuine sense of bonhomie I received from Norman Lee, Dean Koucoulas, and Jenn van Eindhoven during my early years. My research would also not be possible without the tireless help of many undergraduate research assistants and work-study students who fed my spiders and helped with experiments. v

Outside of the Mason lab, I was also fortunate enough to have received support and friendship from colleagues in the Andrade lab, as well as the department of biological sciences at UTSC: Emily MacLeod, Luciana Baruffaldi, Sheena Fry, Allan Edelsparre, Anders Vesterberg, Charmaine Condy, Monica Mowery, Malcolm Rosenthal, Catherine Scott, and Nishant Singh. I’m especially grateful for Emily, Luciana, and Sheena, who were always willing to spare me black widow males and females. My research would not have been possible without the initial population of black widows I inherited from them. I would also like to thank Damian Elias at UC Berkley for providing comments and suggestions on my manuscripts. Damian’s research on jumping spiders and his writing style has been a source of influence and inspiration to me, and has guided me on many aspects of the research conducted in my dissertation.

I am very blessed to have continuous, never-ending love and support from my family and friends. They never stopped believing in me, and have always encouraged me to keep moving forward. I owe them a debt of gratitude for always being there for me. I would like to thank my father and mother, Sivalinghem and Selvakumary, who have always given me everything they can. I would not be where I am today without their love and effort. I thank Priya and Mayurran for their love and admiration, and for giving me strength and courage when needed. My family has grown since I started my PhD, and the love and encouragement I have received from my wife Vinusha, and my in-laws have given me additional motivation to keep moving forward. Vinusha has selflessly supported me throughout my writing process and has helped me stay focused and ambitious. I hope I keep making you all proud.

Lastly, I would like to acknowledge the financial support provided by graduate fellowships from the University of Toronto, Natural Science and Engineering Research Council (NSERC) of Canada, and the Ontario Graduate Scholarship; as well as support provided to my advisor Andrew Mason: NSERC Discovery Grant.

Table of Contents

Contents Abstract ...... ii Acknowledgments...... v Table of Contents ...... vii List of Tables ...... ix List of Figures ...... x Chapter 1 Vibratory Communication in Spiders: An Integrative Approach to Understanding Signal Complexity ...... 1

1.1 Introduction ...... 1 1.1.1 Vibratory Communication in Spiders ...... 3 1.1.2 Vibratory Communication in Web-Building Spiders ...... 10 1.2 Western Black Widow Spider (Latrodectus hesperus) ...... 12 1.2.1 General Natural History and Mating Behaviour ...... 12 1.2.2 Vibration Signals of L. hesperus ...... 22 1.3 Thesis Outline ...... 24 1.4 References ...... 26 Chapter 2 Vibratory Communication in a Black Widow Spider (Latrodectus hesperus): Signal Structure and Signaling Mechanisms...... 40

2.1 Abstract ...... 40 2.2 Introduction ...... 41 2.3 Materials and Methods ...... 45 2.3.1 Behavioural Traits and Signal Characterization ...... 45 2.3.2 Vibration Signal Production Mechanism ...... 47 2.3.3 Importance of Male Abdomen for Signal Production ...... 48 2.3.3 Vibration Signal Tranmission ...... 49 2.3.3 Statistical Analysis ...... 50 2.4 Results ...... 51 2.4.1 Vibratory Behaviours and Signals of L. hesperus ...... 51 2.4.2 Signal Production Mechanism ...... 62 2.4.2 Signal Transmission ...... 64 2.5 Discussion...... 71 2.6 Acknowledgement ...... 78 2.7 References ...... 79 2.8 Supplementary Materials ...... 87 Chapter 3 Function of Structured Signaling in the Black Widow Spider, Latrodectus hesperus ...... 91

3.1 Abstract ...... 91

vii

3.2 Introduction ...... 92 3.3 Materials and Methods ...... 95 3.3.1 Study Animal ...... 95 3.3.2 Behavioural Experiments ...... 96 3.3.3 Recording Procedures ...... 98 3.3.4 Structured Signal Analysis and Mating Success ...... 99 3.3.5 Statistical Analysis ...... 100 3.4 Results ...... 100 3.4.1 Experiment 1 ...... 100 3.4.2 Experiment 2 ...... 104 3.5 Discussion...... 104 3.6 Acknowledgement ...... 110 3.7 References ...... 111 Chapter 4 Posture Controls Mechanical Tuning in the Black Widow Spider Mechanosensory System ...... 121

4.1 Abstract ...... 121 4.2 Introduction ...... 122 4.3 Materials and Methods ...... 124 4.3.1 Body Dynamics: Vibrometry ...... 124 4.3.2 Body Dynamics: Modelling ...... 125 4.3.3 Body Dynamics: Modelling ...... 127 4.4 Results ...... 128 4.4.1 Frequency Segregation...... 128 4.4.2 Modelling Whole-spider Vibrational Mechanics...... 129 4.4.3 Full Body Vibrational Mode ...... 132 4.4.4 Body Size Effects ...... 133 4.4.5 Posture Effects ...... 136 4.5 Discussion ...... 140 4.5.1 Simplified Body and Web Mechanics ...... 140 4.5.2 Multifunctional Sensors ...... 141 4.5.3 Sensory Complexity and Intentionality ...... 142 4.6 Authors’ Contribution Statement ...... 143 4.7 Acknowledgements ...... 143 4.8 References ...... 144 4.9 Supplementary Materials ...... 148

Chapter 5 General Discussion ...... 160

5.1 Research Implications and Future Directions ...... 160 5.1.1 Males Produce Multiple Vibration Signals ...... 160 5.1.2 Males Use Different Signal Production Mechanisms ...... 165 5.1.3 Male Signals are Informative and Important for Male Mating Success ....166 5.1.4 Female Body Mechanics Affect Vibration Tuning of Leg Joints ...... 167 5.2 Conclusion ...... 172 5.3 References ...... 173

viii

List of Tables

Table 1.1: Summaries of studies that have characterized vibration signals of web-building spiders...... 13

Table 1.2: Non-exhaustive list of literatures describing courtship vibratory behaviours in different Latrodectus species...... 23

Table 2.1: Summary of Latrodectus hesperus male signal characteristics during courtship and copulation...... 54

Table 2.2: Discriminant function analysis of vibration signal components produced by male Latrodectus hesperus...... 58

Table 2.3: Results from Friedman test and subsequent Wilcoxon Pair-wise comparisons...... 59

Table 2.4: Results from Wilcoxon Pair-wise comparisons...... 60

Table 3.1: Summary, normality test, and skewness of Latrodectus hesperus male weight, proximal-phase duration, and male proximal-phase structured signaling rates...... 101

List of Figures

Figure 1.1: Images of western black widow (Latrodectus hesperus) spider...... 20

Figure 2.1: Web-borne vibratory signals of male Latrodectus hesperus...... 53

Figure 2.2: Linear discriminant function plot of the signaling space for each vibratory signal component of L. hesperus male...... 57

Figure 2.3: Male L. hesperus structured signaling during male-female interactions...... 61

Figure 2.4: Abdominal tremulation signal production mechanism...... 65

Figure 2.5: Male bounce signal production mechanism...... 66

Figure 2.6: Male web pluck signal production mechanism...... 67

Figure 2.7: Comparing spectral differences between a male abdominal tremulation signal velocities and dorso-ventral movement velocities...... 68

Figure 2.8: Comparing spectral differences between a male bounce signal velocities and dorso- ventral movement velocities...... 69

Figure 2.9: Comparing spectral differences between a male web pluck signal velocities and dorso-ventral movement velocities...... 70

Figure S2.1: Custom-made mating arena used for conducting mating trials...... 87

Figure S2.2: Mean maximum velocity of male body segment movements during abdominal tremulation and bounce signal production...... 88

Figure S2.3: Effects of immobilizing male abdomen on abdominal tremulation and bounce signals...... 89

Figure S2.4: Dorso-ventral movement velocities of multiple legs during web pluck signal production...... 90 x

Figure 3.1: Male (L. hesperus) structured vibration signaling during courtship...... 97

Figure 3.2: Variations in L. hesperus male structured signaling rates...... 102

Figure 3.3: Predictors of male mating success and structured vibratory signaling rates...... 103

Figure 4.1: Vibration transmission through spider compared to the model...... 130

Figure 4.2: Vibration spectra of measured and modelled spiders, showing frequency segregation in vibration transmission through body segments...... 131

Figure 4.3: Full body vibration modes predicted by model and measured in real spiders...... 134

Figure 4.4: Abdominal size and density have little effect on the spatial segregation of vibration frequencies...... 135

Figure 4.5: Posture affects joint tuning...... 138

Figure S4.1: A schematic of the experimental setup...... 148

Figure S4.2: Signal quality during vibrometry...... 149

Figure S4.3: The relationship between angular velocity and displacement...... 150

Figure S4.4: Multibody dynamics...... 151

Figure S4.5: Model stiffness sensitivity analyses...... 153

Figure S4.6: Body size sensitivity analyses...... 154

Figure S4.7: Body size variation has little effect on frequency segregation in the spider body.155

Figure S4.8: Posture has a strong effect on frequency segregation in the spider body...... 156

Figure S4.9: We tested model predictions by measuring six females in both the neutral and crouched posture...... 157

Figure S4.10: Male L. hesperus courtship and prey vibration power density spectra...... 158

Figure 5.1: Oscillogram and mean power spectrum of female abdominal twitch signals...... 164 xi

Figure 5.2: Metatarsal Lyriform organ (HS10) of the western black widow spider (Latrodectus hesperus)...... 169

Figure 5.3: Structural details of the HS10 organ of Latrodectus hesperus...... 170

Figure 5.4: Transmission electron micrograph of Latrodectus hesperus HS10 organ slits...... 171

xii

Chapter 1 Vibratory Communication in Spiders: An Integrative Approach to Understanding Signal Complexity 1.1 Introduction

Communication is a complex process involving the exchange of information between a ‘sender’, who emits a signal containing information, and a ‘receiver’, whose sensory system(s) receives and processes this information (Bradbury and Vehrencamp, 2011). Animal communication signals vary greatly in complexity – some consisting of simple, repeated signals with high redundancy in a single sensory modality (e.g. species-specific pulsed calling songs of several orthopteran insects (Drosopoulos and Claridge, 2005)); while others comprise elaborate sequences of varied components across multiple modalities (some examples includes: the complex waggle dance sequences of forager honeybees (De Marco and Menzel, 2005; Kirchner et al., 1988); coordinated visual and seismic courtship signals of jumping spiders (Elias et al., 2012, 2003); colour and movement displays of male guppies (Kodric-Brown and Nicoletto, 2001); the movement and chemical signals in sagebrush lizards (Thompson et al., 2008); multiple visual ornaments in northern cardinals (Jawor et al., 2004; Jawor and Breitwisch, 2004); and the visual, movement, and acoustic displays of bower birds (Borgia and Coleman, 2000; Coleman et al., 2004; Robson et al., 2005)). The diversity and multitude of signals organisms use for communication has generated a plethora of research focused on understanding the underlying mechanisms, function, and evolution of multiple signals and signal complexity (Candolin, 2003; Hebets and Papaj, 2005; Higham and Hebets, 2013; Johnstone, 1996; Ruxton and Schaefer, 2011).

In general, “complex signals” are classified as any signal having multiple distinct components organized into composite temporal sequences (Candolin, 2003; Hebets and Papaj, 2005; Partan, 2013; Partan and Marler, 2005). Individual components of these composite signals can stimulate a single sensory modality (‘multicomponent signals’), and/or stimulate different sensory modalities (‘multimodal signals’) (Hölldobler, 1999; Rowe and Halpin, 2013; Ruxton and Schaefer, 2011; Scheuber et al., 2004; Smith and Evans, 2013). A classic example of multicomponent signals is seen in túngara frogs, Physalaemus pustulosus, where male acoustic signals comprise two components – a frequency-modulated “whine”, followed by several shorter

1 2 broadband “chucks” (Ryan, 1980; Ryan and Rand, 1990). Multimodal signals are seen in the Bornean ranid frog, Staurois guttatus, where males emit short, frequency-modulated acoustic calls, which are accompanied by “foot-flagging” (visual display), where the male extends and waves one of his hind-legs (Grafe and Wanger, 2007). The evolution of complex signals can be driven by several factors including female sensitivity and preference for signal complexity (e.g. runaway selection) and environmental constraints on signal efficacy (e.g. sensory drive) (Cummings and Endler, 2018; Endler, 1992; Endler and Basolo, 1998; Endler and Houde, 1995; Ryan et al., 1993, 1990).

Several non-mutually exclusive hypotheses have been proposed to account for the evolution of complex signals. These hypotheses can be divided into two groups: 1) Content-based hypotheses focus on selection pressures acting on and driving the information content of signals (the ‘strategic design’ of signals) (Hebets and Papaj, 2005); and 2) Efficacy-based hypotheses focus on forces acting on effective transmission, reception, and processing of signals (the ‘tactical design’ of signals) (Hebets and Papaj, 2005). Under content-based hypotheses, multiple signal components can convey similar information (‘backup message’ hypothesis), or convey different information (‘multiple message’ hypothesis) (Doucet and Montgomerie, 2003; Hebets and Papaj, 2005; Thompson et al., 2008). For example, in fiddler crabs, Uca mjoebergi, males repeatedly produce claw-waving displays, as well as vibrational drumming signals (Mowles et al., 2017). The rate of claw-waving and vibrational signaling are positively correlated, suggesting that both signals convey similar information about male stamina (redundant signals). However, vibrational signals can also provide additional information about male borrows to females (multiple messages) (Mowles et al., 2017). Additionally, different signal components can be intended for different receivers (‘multiple receiver’ hypothesis). For example, male treefrogs Eleutherodactylus coqui produce a two-note ‘co-qui’ advertisement call. The ‘co’ signal is primarily intended for rival males, whereas the ‘qui’ signal functions for female attraction (Narins and Capranica, 1978).

Efficacy-based hypotheses propose that variations in signaling environment characteristics and/or receiver sensory systems can drive the evolution of multiple signal components (e.g. ‘sensory drive’ hypothesis; ‘sensory constraints’ hypothesis) (Candolin, 2003; Fleishman, 1988; Guilford and Dawkins, 1993, 1991; Miller and Bee, 2012; Preininger et al., 2013; Rowe, 2013, 1999). For example, Tobias, Aben, Brumfield (2010) demonstrated that differences in acoustic

3 transmission properties between Amazonian forest types was correlated with predictable variation in acoustic signals of 17 bird species. Complex signals can also evolve if signal components interact to enhance and influence the detectability, discriminability, and preference for one or more signals (‘receiver psychology’ hypothesis) (Guilford and Dawkins, 1991; Miller and Bee, 2012; Rowe, 1999; Stange et al., 2017). For example, Rowe (2002) demonstrated that the presence of an acoustic signal increased the speed of visual discrimination learning in domestic chicks (Gallus gallus domesticus). Studying complex communication signals can provide greater insights into the evolution of languages, and how neural processes integrate complex sensory information. This in turn, requires an integrative approach that examines the different aspects of communication in focal organisms; from senders to signaling environments and to receivers. The overarching aim of my Ph.D. thesis is to use an integrative approach to examine signal complexity in a web-building spider (Araneae).

1.1.1 Vibratory Communication in Spiders

Spiders are exceptional organisms for investigating communication (Herberstein et al., 2014; Uetz and Stratton, 1983; Uhl and Elias, 2011). They are found across a wide range of social and ecological environments, and have a rich diversity in life history traits and reproductive tactics (Andrade, 2019; Herberstein and Wignall, 2011; Simkovic and Andrade, 2019). Correspondingly, their communication systems are diverse, incorporating a multitude of signals across different sensory modalities and ranging in degree of signal complexity both within and among genera (Herberstein et al., 2014; Uetz et al., 2016; Uhl and Elias, 2011). For example, in the wolf spider genus Schizocosa, some species communicate using relatively simple vibration- only signals with one element, where other species can have many signals components (multicomponent signals), and can integrate these vibration signals with visual displays (multimodal signaling) (Hebets et al., 2013). In the context of male-female interactions, many spiders have been shown to communicate using chemical signals (Baruffaldi and Andrade, 2015; Uetz and Roberts, 2002; Uhl and Elias, 2011), visual displays and movements (Elias et al., 2006c; Uetz and Roberts, 2002), air and substrate-borne vibrations (Elias et al., 2003; Herberstein et al., 2014; Kronestedt, 1996; Uetz et al., 2016), and/or tactile signals (Uhl and Elias, 2011). Among the different sensory modalities, however, spiders are particularly adept at perceiving substrate-borne vibrations (Barth, 2002, 1997).

For most spider species, vibrations play a vital role during navigation, predator/prey detection and localization, and intra- and inter-specific communication (Barth, 2002; Barth and Schmitt, 1991; Baurecht and Barth, 1993, 1992; Cross et al., 2007; Elias et al., 2014, 2010a, 2010b, 2005; Elias and Mason, 2014, 2011; Gibson and Uetz, 2008; Hebets et al., 2013; Sivalinghem et al., 2010; Vibert et al., 2014). Spiders perceive vibration stimuli primarily using slit sensilla embedded within the exoskeleton (Barth, 2012a, 2012b, 2004). These slit sensilla are essentially elongated holes in the exoskeleton that can detect minute (order of nanometers) mechanical strains in the cuticle (Barth, 2012b, 2004). Slit sensilla can appear either singly, in loose aggregates, or as distinct structures called Lyriform organs (Barth, 2004). Lyriform organs, in particular, have been the focus of most research for their role in sensing substrate-borne vibrations (Barth and Geethabali, 1982; Hergenröder and Barth, 1983; Hößl et al., 2007, 2009). Lyriform organs are comprised of an array of closely arranged parallel slits distributed near leg- joints (Barth and Bohnenberger, 1978; Barth and Pickelmann, 1975). Substrate-borne vibrations are first transmitted along the spider’s legs. The role of the body (i.e. the legs) in influencing vibration transmission and in turn perception is less understood, but thought to play a key role in perception and localization (Aicher et al., 1983). Nevertheless, the vibrations transmitted along the legs cause leg-joints to bend, which deforms the exoskeleton and results in the compression and expansion of individual slits of the Lyriform organs (Barth and Geethabali, 1982; Hößl et al., 2009, 2007, 2006; Schaber et al., 2012). The compression of individual slits elicits neural responses (Gingl et al., 2006; Molina et al., 2009; Seyfarth and French, 1994; Walcott and Kloot, 1959; Walcott, 1969). Since spiders are highly effective predators, it is likely that these sensory organs evolved for prey detection and localization (Hergenröder and Barth, 1983; Masters, 1984a, 1984b). Consequently, substrate-borne vibratory communication may have evolved as a way for mates to distinguish themselves from prey, and this mode of communication is widespread among spiders (Herberstein et al., 2014; Vibert et al., 2014; Wignall and Herberstein, 2013a).

Spiders produce substrate-borne vibratory signals using three main mechanisms: (1) tremulation; (2) percussion; and (3) stridulation (Elias and Mason, 2011). Tremulation signals are widespread among spiders, and are produced by oscillating specific body parts (Dierkes and Barth, 1995; Elias et al., 2003, 2006d; Rovner and Barth, 1981). Vibrations from these oscillations are then transmitted through the leg(s) onto the signaling substrate (Dierkes and Barth, 1995). To date,

5 the most common tremulation signal observed in spiders is the dorso-ventral oscillation of male abdomen during courtship (Dierkes and Barth, 1995; Elias et al., 2003, 2006d). Since tremulation signals are directly linked to the vibrations of specific body parts, there can be a close association between the characteristics of the signals and the physical attributes of the signaler (Elias and Mason, 2011; Morris and De Luca, 1998). In general, tremulation signals in spiders are low- frequency narrowband vibrations that are best suited for communicating on substrates that pass low frequencies with high fidelity (e.g. plant stems and leaves, and webs) (Elias et al., 2010a, 2003; Elias and Mason, 2014, 2011; Rosenthal et al., 2019).

Percussive signals are generated by either striking different body structures together or against the signaling substrate (Elias et al., 2003; Kotiaho et al., 1996; Kronestedt, 1996; Parri et al., 1997; Schüch and Barth, 1985). When a body part strikes the substrate, kinetic energy is transferred onto the substrate eliciting the natural frequency response of that substrate (Elias et al., 2004, 2006d, 2010a; Elias and Mason, 2011; Hebets et al., 2008). These transient impulse signals usually have a broad frequency range at the source point, and as the signals are transmitted along the substrate, their spectral profile is shaped by the frequency-filtering and transmission characteristics of the signaling medium (Elias and Mason, 2011). Percussive signals are well suited for transmission across a variety of substrates, and the intensity and timing of these signals can provide useful information (Elias et al., 2006d; Hebets et al., 2008; Schüch and Barth, 1985). Percussive signaling has been described in several spider species, where males strike the substrate with their abdomen, pedipalp, and/or first legs (Elias et al., 2012, 2010a, 2003; Hebets et al., 2008; Kronestedt, 1996; Schüch and Barth, 1985).

Lastly, stridulatory signals are produced by scraping two body structures together (Dutto et al., 2011; Elias et al., 2003, 2006d; Gwinner-Hanke, 1970; Lee et al., 1986; Stratton and Uetz, 1983). This signaling mechanism involves a specialized structure called a stridulatory organ that functions as a frequency-multiplier (Elias et al., 2003; Stratton and Uetz, 1983). Stridulatory organs comprise two elements: 1) the ‘pars stridens’ or “file”, which are sclerotized array of ridges; and 2) the ‘plectrum’ or “scrapper”, which can be sclerotized peg(s), teeth, ridge(s), or setae (Elias et al., 2003; Stratton and Uetz, 1983). The pars stridens on one body structure is scraped by a plectrum on another body structure, and the resulting vibrations are then transmitted either through the substrate or air (Gwinner-Hanke, 1970; Lee et al., 1986; Peretti et al., 2006; Uhl and Elias, 2011). Stridulatory organs have been described in several spider species, and their

6 location on spider bodies can differ among species (Elias et al., 2003, 2006d; Jocqué, 2005; Stratton and Uetz, 1983). In one observation in Zodariidae spiders, Jocque (2005) described the presence of six different stridulatory organs on different body structures, across three species in the Mallinella genus. To date, however, the signals generated from stridulatory mechanisms have only been examined for a few species (Elias et al., 2003, 2006d). Whether many of the purported ‘stridulatory organs’ described in the literature to date are actually involved in signal production remains unknown. Nevertheless, stridulatory signals, like tremulatory signals, are also best suited for communication on specific substrates with favourable frequency-filtering and transmission properties (Elias and Mason, 2011). Spiders can use any combination of signaling mechanisms mentioned above to produce a range of complex vibratory signals (Elias et al., 2003, 2006d, 2012).

Current literature on substrate-borne vibratory communication in spiders shows that this mode of communication is not only ubiquitous, but remarkably diverse in the degree of signal complexity (Elias et al., 2012; Herberstein et al., 2014; Uetz et al., 2016; Uhl and Elias, 2011). In some spiders, vibratory signals are relatively simple comprising a single repetitive element (Elias et al., 2010b; Hebets et al., 2013; Kotiaho et al., 1996); whereas in other species, signals can consist of many distinct components produced by different signaling mechanisms (Elias et al., 2006d, 2012; Hebets et al., 2013). For example, in the jumping spider, Phidippus clarus, male courtship vibrations are comprised of a single abdominal tremulation component, which males produce repeatedly throughout courtship (Elias et al., 2010b; Sivalinghem et al., 2010). Interestingly, these males also produce a different higher-frequency tremulation signal during male-male aggressive interactions (Elias et al., 2008), and yet a third longer tremulation signal during mate- guarding (evidence for ‘multiple receiver’ hypothesis; Elias et al., 2014). Although P. clarus males have three distinct signals in their repertoire, males use these signals exclusively in different behavioural contexts, and within each context, males display simple repetitive signals. In contrast to Phidippus males, many species in the genus Habronattus produce composite vibratory songs that comprise several distinct components (Elias et al., 2012). For example, vibratory signals of male Habronattus coecatus can have up to 20 different signal elements organized into distinct functional groupings (‘motifs’); and combined with different visual displays (Elias et al., 2012). Within the Habronattus genus, there are also large variations in the number of signal components (in turn signal complexity) among the different species (Elias et

7 al., 2012). Similarly, as mentioned previously, variations in number of signal components is also seen in Schizocosa wolf spiders, where some species (e.g. S. rovneri) only produce signals with a single component, while others (e.g. S. stridulans, S. ocreata, and S. floridana) can have several components (Elias et al., 2006d, 2010a; Hebets et al., 2013; Rosenthal et al., 2019).

In addition to the number of individual signal components involved, vibratory signals of many spiders can also vary in their level of temporal organization (‘signal structure’) (Elias et al., 2012, 2006d, 2006a, 2006b; Hebets et al., 2013). For example, in wolf spiders, males of some species display signal components rather “loosely” with little temporal organization between components, whereas in other species, males display signal elements in a tightly coordinated stereotyped sequences (Hebets et al., 2013). Even within species, males can vary how they combine and display signal components during courtship (Elias et al., 2006d). For example, male Schizocosa stridulans transition from displaying multiple signal components simultaneously (‘parallel signaling’), during early stages of courtship, to displaying signal components sequentially (‘serial signaling’) as they approach closer to females (Elias et al., 2006d). Perhaps the most exquisite demonstration of complex non-random signal structuring/organization is seen in the jumping spider, Habronattus coecatus, where males temporally organize different signal components into stereotyped sequences or motifs (Elias et al., 2012). During the course of courtship interactions, as males progressively get closer to females, they gradually vary their signal motifs, by adding and removing different signal components; building to a crescendo, where signal intensities and rates increase (Elias et al., 2012). The variable non-random temporal sequencing of signal elements adds another level of signal complexity, and is thought to play a functional role during courtship (Elias et al., 2012).

Vibratory signals of many spiders are both necessary and sufficient for mating success (Elias et al., 2005; Gibson and Uetz, 2008; Girard et al., 2015; Hebets et al., 2013; Rundus et al., 2011; Shamble et al., 2009; Sivalinghem et al., 2010; Wilgers and Hebets, 2012). Studies looking at the individual signal components have demonstrated vibratory signals’ functional role in species identity (Schmitt et al., 1994, 1993), mate recognition (Vibert et al., 2014), conveying information about mate quality (Gibson and Uetz, 2008; Kotiaho et al., 1996; Parri et al., 1997; Rivero et al., 2000; Sivalinghem et al., 2010; Wignall et al., 2014), inhibiting or suppressing female predatory/aggressive behaviour (Vibert et al., 2014; Wignall and Herberstein, 2013a), and stimulating female arousal (Maklakov et al., 2003). In a seminal set of studies on the European

8 wolf spider Hygrolycosa rubrofasciata (Lycosidae), male abdominal percussive drumming (i.e. drumming rates) were shown to function as honest indicators of male condition and quality, and females show preference for males with increased drumming rates (Ahtiainen et al., 2005; Alatalo et al., 1998; Kotiaho et al., 1996; Lindström et al., 2006; Mappes et al., 1996; Parri et al., 2002, 1997). Male abdominal drumming was associated with higher energetic output (Kotiaho et al., 1996), and drumming rates were associated with male condition and age (Kotiaho et al., 1999; Lindström et al., 2006; Rivero et al., 2000) and higher immunocompetency (Ahtiainen et al., 2006, 2005, 2004).

In contrast to studies on individual signal components, fewer studies have examined the function and evolution of multicomponent vibratory signaling and the importance of the varying degrees of signal organizations seen in spiders (Elias et al., 2012; Girard et al., 2015; Hebets, 2008; Hebets et al., 2013). Understanding the significance and evolution of vibration complexity in spiders, however, has been the focus of some contemporary studies (Elias et al., 2012; Elias and Mason, 2011; Herberstein et al., 2014; Uetz et al., 2016). The signaling micro-environment that many cursorial spider species inhabit (i.e. jumping spiders [Araneae: Salticidae] and wolf spiders [Araneae: Lycosidae]) can be highly heterogeneous and complex. Sensory drive theory predicts that the transmission properties of these variable signaling substrates can impose strong selection on vibratory signals and signaling behaviours, and drive the evolution of multiple signals/vibration complexity (Elias and Mason, 2014; Rosenthal et al., 2019). Accordingly, spiders on heterogeneous microhabitats can either specialize to communicate on a particular substrate, or they can adopt a more generalist signaling strategy to enable them to communicate on multiple substrate types. Spiders that specialize on particular substrate(s) are hypothesized to have corresponding signals that match the transmission properties of the substrate(s) (Elias and Mason, 2014). For example, studies comparing vibration transmission properties of various natural substrate types, have shown that vibration signals are best transmitted – with least amount of signal attenuation – when passing through leaf litter (Elias et al., 2010a, 2004). In turn, these same studies also show that females are more likely to mate with males courting on leaf litter compared to other substrates; suggesting that the signals are specialized for communicating on leaf litter (Elias et al., 2010a, 2004). In a more recent study, Rosenthal et al. (2019) demonstrate that the wolf spider Schizocosa floridana, are exclusively found, and have high mating rates, on oak litter, which also transmit courtship vibrations with least attenuation. In

9 this species, a novel male abdominal “chirp” component was especially transmitted with high efficacy through oak litter, and Rosenthal et al. (2019) suggest that substrate-specialization, especially on substrates with high signal transmission fidelity, can also relax environmental constraints leading to evolution of novel signal components. On the other hand, substrate generalists can evolve multiple signal components to match the different transmission characteristics of different substrates (Elias and Mason, 2014). Additionally, substrate generalists can also produce more versatile signals that are effective on multiple substrates (e.g. many percussive signals) (Elias and Mason, 2014, 2011).

In addition to environmental constraints due to microhabitat heterogeneity, vibration complexity in spiders is also driven by female preference of signal complexity (Elias et al., 2006a). For example, in a study examining two sky island populations of Habronattus pugillis with different levels of signal complexity (simple vs. complex), females were shown to prefer the complex signals of foreign males that comprised more components than local males (Elias et al., 2006a, 2006b). Furthermore, in jumping spiders and wolf spider, increase in vibration signal complexity is associated with increase in visual signal complexity, further suggesting the influence of sexual selection (Elias et al., 2006b; Hebets et al., 2013). There may be other factors driving vibration complexity in spiders – e.g. female sensory constraints and/or receiver psychology – but further research is needed (but see: Hebets, 2005; Hebets and Papaj, 2005; VanderSal and Hebets, 2007).

Although there is an ever-increasing appreciation and research on vibratory communication in spiders, our understanding of this mode of communication is still very limited (Elias and Mason, 2014; Herberstein et al., 2014; Uetz et al., 2016). Additionally, much of our understanding about vibratory communication in spiders, to date, comes from extensive studies on a few species of cursorial hunting spiders (i.e. wolf spiders [Araneae: Lycosidae], jumping spiders [Araneae: Salticidae], and the American wandering spider [Araneae: Ctenidae]), while other groups of spiders have received very little to no attention (Herberstein et al., 2014). This leaves a significant knowledge gap in our understanding of vibrational ecology and communication in other groups of spiders.

1.1.2 Vibratory Communication in Web-Building Spiders

One such group that has been largely overlooked for vibratory communication studies are web- building spiders (superfamily: Araneoidea) (Herberstein et al., 2014). Web-building spiders are ecologically important “sit-and-wait” predators of many invertebrates and some vertebrates, and are integral constituents of many terrestrial ecosystems (Blackledge et al., 2011; Griswold et al., 1998; Herberstein and Wignall, 2011). In most species, females are relatively sessile and construct elaborate web structures that are functionally optimized for ensnaring and capturing prey, and as a medium for sending and receiving vibratory information (Blackledge et al., 2011; Frohlich and Buskirk, 1982; Klärner and Barth, 1982; Landolfa and Barth, 1996; Masters, 1984a, 1984b; Zevenbergen et al., 2008). Most web-dwelling spiders have poor visual sense, and are highly dependent on acute vibration sense for prey capture and communication (Barth and Geethabali, 1982; Herberstein and Wignall, 2011; Masters, 1984a; Wignall and Herberstein, 2013a). Web-dwelling spiders are substrate specialists as male-female communication occurs on the webs that females build. Given that females have much greater control over the construction of webs, examining vibratory communication in web-building spiders can provide valuable insights into the interplay between signals and signaling substrates, and how these interactions affect signal function, and drive signal evolution and complexity (Elias and Mason, 2014, 2011).

Spider webs can impose significant constraints on vibratory communication and signal evolution (Elias and Mason, 2014, 2011). The physical properties of webs and silks, including tension, stiffness, and composition and architecture, can affect vibration transmission (Frohlich and Buskirk, 1982; Landolfa and Barth, 1996; Masters and Markl, 1981; Mortimer et al., 2019, 2016, 2014; Naftilan, 1999; Vibert et al., 2016). However, many studies have also shown that webs and silks are exceptionally effective substrates for vibratory communication as signals across wide frequency spectra can be transmitted with high efficacy, and webs are robust enough to withstand significant damage and changes (Mortimer, 2019; Mortimer et al., 2014; Tso et al., 2005; Vibert et al., 2016). The robustness and high signal transmission fidelity of spider webs can also relax constraints on vibration signals and drive evolution of novel signals and signal diversity (see Rosenthal et al., 2019).

Across many species of web-building spiders, males have been observed to engage in a variety of vibratory behaviours during courtship – including “abdominal wagging”, “body-rocking”, “leg

11 tapping”, “leg-flexing”, “leg waving”, web shaking”, “push-ups”, “palpal boxing”, “body- jerking”, “web-flexing”, “web-bouncing”, and “plucking” – and some males perform repeated bouts of different vibratory behaviours (Aisenberg, 2009; Barrantes, 2008; Bartos, 1997; Calbacho-Rosa et al., 2013; Coyle and O’Shields, 1990; Dutto et al., 2011; Forster, 1992; Maklakov et al., 2003; Ross and Smith, 1979; Singer et al., 2005, 2000; Suter and Renkes, 1984; Vibert et al., 2014; Wignall and Herberstein, 2013b). For example, in the kleptoparasitic spider, Argyrodes antipodiana, Whitehouse and Jackson (1994) described 32 different types of male vibratory behaviours during courtship interactions. In the funnel-web spider, Agelenopsis aperta, males engage in stereotyped repeated bouts of abdominal waggles, followed by web-flexions during male-female interactions (Singer et al., 2000). It is evident from the literature that within many species of web-dwelling spiders, males can generate many different types of web-borne vibrations, and communication in web-building spider can comprise multiple signal components (Herberstein et al., 2014; Lubin, 1986; Singer et al., 2000; Suter and Renkes, 1984; Uhl and Elias, 2011). Descriptions of vibratory signaling in web-dwelling spiders also highlight the noticeable lack of temporal patterning or organization among the different signals, and web- dwelling spiders are generally classified as ‘simple’ signalers compared to cursorial spiders (Herberstein et al., 2014; Vibert et al., 2014). To date, however, vibratory behaviours of web- dwelling spiders have only been described for a few species, and most studies only provide qualitative (and sometimes vague) descriptions of either the conspicuous movements of body parts, or the overall vibration producing behaviours (Aisenberg, 2009; Barrantes, 2008; Coyle and O’Shields, 1990; Forster, 1992; Maklakov et al., 2003; Ross and Smith, 1979; Suter and Renkes, 1984).

Very few studies have measured web-borne vibration signals directly from webs and quantified signal characteristics (Table 1.1). Recently, a set of papers by Anne Wignall, Marie Herberstein and colleagues show that in the orb-web spider (genus: Argiope), males produce different vibration signals that vary in spectral and temporal characteristics and are distinct from prey generated vibrations (Wignall and Herberstein, 2013b). Male signals can reduce female predatory response and may convey information about male traits and affect female choice (O’Hanlon et al., 2017; Wignall et al., 2014; Wignall and Herberstein, 2013a). These studies show that web-borne vibration signals can vary in characteristics and function within a genus, and emphasize the desperate need for further quantitative studies to examine variation in

12 vibration signal characteristics and function, mechanisms underlying signal production, and signal transmission, perception, and processing. In the following chapters of my Ph.D. thesis, I examine details of vibratory communication in the western black widow spider, Latrodectus hesperus (Theridiidae). Black widow spiders are excellent organisms for studying communication as they are feasibly reared in the laboratory and much is known about their mating behaviours and life history traits (Andrade, 2003, 1996; Andrade and MacLeod, 2015; MacLeod, 2013; Scott et al., 2012, 2019).

1.2 Western Black Widow Spider (Latrodectus hesperus) 1.2.1 General Natural History and Mating Behaviour

Western black widow spiders, Latrodectus hesperus, are found throughout western North America, with populations ranging from British Columbia (Canada) to California (U.S.A), and into Texas (U.S.A.) and parts of Mexico (Garb et al., 2004). Morphologically, L. hesperus are similar to other species within the Latrodectus genus. Both males and females are identified by the red hourglass pattern on the ventral side of their abdomen (opisthosoma). Female L. hesperus are black with no other noticeable patterning on their body (Figure 1.1). Males can vary from light brown to black with two to three small white spots along the edge on the dorsal side of their abdomen. Additionally, males have a single stripe along the centre of the dorsal side of the abdomen, which can range from white to reddish-orange in colour (Figure 1.1). As with all species in the Latrodectus genus, there is a female-biased sexual size dimorphism in L. hesperus: females range in length from 8 - 15 mm and weight from 120 - 400 mg; males range in length from 3 - 6.5 mm and weight from 8 - 18 mg (Kaston, 1970). Female L. hesperus are relatively sedentary, and builds dense cob-webs comprised of a complex irregular-network of silk treads that radiate from a thickly woven refuge. The radiating network of densely interconnected silk- threads forms a sheet-like layer that take the shape of surrounding vegetation. Adult females may inhabit a single web site throughout adulthood, which they actively maintain by adding silk to increase density and size (Kasumovic and Andrade, 2004; MacLeod and Andrade, 2014).

Table 1.1: Summaries of studies that have characterized vibration signals of web-building spiders.

Signals Signal Signal Animal References Notes Measured Characteristics Function Cob-web spiders Achaearanea Male -Signal -Signal function (Wignall and -Peak frequency and duration of male sp. stridulatory characteristics not not tested, but Taylor, 2011) signals did not differ from prey or vibrations provided female predator (assassin bug) aggressive responded to mimicry vibrations male signals by -Male vibrations were distinct in temporal adopting structure and amplitude (low amplitude) copulatory position

Latrodectus Male -Duty Cycle = 68% -Vibration (De Luca et al., -Only overall vibrations were recorded; production 2015) study did not distinguish individual hasselti courtship -Root Mean Square requires four components vibrations (RMS) Amplitude = times more 10.69 mm/s energetic output than resting -Males with intermediate mass had the highest duty cycle -Overall vibration effort may be important for female choice

13 14

Signals Signal Signal Animal References Notes Measured Characteristics Function Latrodectus Male -Peak Frequency = -Abdominal (Vibert et al., -Only signals at the onset of courtship hesperus abdominal 43.38 Hz tremulation 2014) were recorded and analyzed signals may vibration -Duration = 6.45 sec -Signals during the onset of courtship function for lacked noticeable sequencing/structuring -RMS Amplitude = mate 0.462 mm/s discrimination or -Max Amplitude = suppress female 1.848 mm/s predatory attacks

Male web -Peak Frequency = cutting 36.60 Hz -Duration = 1.00 sec -RMS Amplitude = 3.73 mm/s -Max Amplitude = 17.91 mm/s

Male -Peak Frequency = walking 55.38 Hz -Duration = 8.06 sec -RMS Amplitude = 1.53 mm/s -Max Amplitude = 14.44 mm/s

Signals Signal Signal Animal References Notes Measured Characteristics Function * Steatoda Male -Frequency Range = -Male signal (Gwinner-Hanke, -Male stridulation produced a "metallic bipunctata stridulatory 700 - 1400 Hz important for 1970) click-sound" mating success vibrations -Peak Frequency = (Lee et al., 1986) -Only acoustic component of the signals 869 Hz were recorded using a microphone -Duration = 0.053 -Web-borne vibrations were not recorded sec -Inter-pulse Interval = 0.291 sec * Steatoda Male -Frequency Range = (Lee et al., 1986) -Only acoustic component of the signals borealis stridulatory 750 - 1100 Hz were recorded using a microphone vibrations -Peak Frequency = -Web-borne vibrations were not recorded 942 Hz -Duration = 0.068 sec -Inter-pulse Interval = 0.282 sec

* Physocylus Female -Female produced -Males relax (Peretti et al., -Authors recorded airborne component of globosus copulation bursts of pulse like female genitalic 2006) vibration signals stridulation signals (spectral squeezing, characteristics not during described) copulation, in response to -Burst Duration = 1.7 female signals; sec potential -Pulses per Burst = reproductive 8.2 ± 3.3 benefits

Signals Signal Signal Animal References Notes Measured Characteristics Function Funnel-web spiders Agelenopsis Male -Low-Frequency -Successful (Singer et al., -Males began signaling with abdominal aperta abdominal Peak = 2.48 Hz males had higher 2000) wagging, followed by web flex signal, and abdominal then rest; this pattern continued for 90 min wagging -High-Frequency wagging Peak = 15.81 Hz -Sequence of successful and unsuccessful frequency and males involved repetitive series of signals elevated high- Male web -Low-Frequency frequency peaks -Males that made greater transitions to other behaviours (e.g. grooming) were flex Peak = 1.36 Hz -No difference unsuccessful -High-Frequency web flex signal Peak = 15.11 Hz between successful and unsuccessful males

Orb-web spiders Argiope Male -Peak Frequency = -Males with (Wignall and -Signals can be complex, but not properly keyserlingi shuddering 30 Hz higher Herberstein, characterized shuddering rates 2013b) -Duration = 0.42 sec -Males begin courtship with shuddering were accepted (Wignall and and add abdominal wagging signals sooner Herberstein, 2013a) -In later stages, during mating dance Male -Peak Frequency = -Male body phase, males add web plucks and bounce abdominal 116 Hz condition was (O’Hanlon et al., elements to shuddering and abdominal wagging not correlated 2017) -Duration = 0.32 sec wagging signals with any parameter of Male plucks -Plucks Per Second = signal elements 2.13

Signals Signal Signal Animal References Notes Measured Characteristics Function Male -Total Bounces = -Shuddering bounce 63.12 delayed female predatory behaviour Argiope Male -Signal -Male (Wignall et al., -Male shuddering is highly repeatable radon shuddering characteristics not shuddering was 2014) within male and variable between males provided positively -Shudder signal in A. keyserlingi not correlated with related to male condition male body condition

Sheet-web spiders Frontinella Male -Peak Frequency = -Vibrations (Suter and -Transition between vibrations signals is pyramitela abdomen 19.5 Hz functions Renkes, 1984) ordered, but signal complexity not includes: described flexion -Duration = 17.4 sec Species recognition; Male fast -Peak Frequency = suppress female abdomen 23.9 Hz aggression; and flexion -Duration = 13.8 sec stimulate females to mate Male push -Signal down characteristics not provided Female fast -Peak Frequency = abdomen 17 Hz flexion -Duration = 5.4 sec

Signals Signal Signal Animal References Notes Measured Characteristics Function Female -Peak Frequency = dorsad 18.4 Hz flexion -Duration = 2.3 sec

Female web -Peak Frequency = plucks 20.8 Hz -Duration = 1.0 sec

Tegenaria Male palp -Peak Frequency = -Function of (Vibert et al., -Only signals at the onset of courtship agrestis drumming 32.83 Hz signal not 2014) were recorded and analyzed known -Duration = 7.25 sec -Signal structure not described -RMS Amplitude = 0.214 mm/s -Max Amplitude = 1.357 mm/s

Male web -Peak Frequency = tapping 40.64 Hz -Duration = 9.62 sec -RMS Amplitude = 2.185 mm/s -Max Amplitude = 20.368 mm/s

Signals Signal Signal Animal References Notes Measured Characteristics Function Male jerk -Peak Frequency = 50.88 Hz -Duration = 1.58 sec -RMS Amplitude = 3.518 mm/s -Max Amplitude = 19.129 mm/s

* Indicates studies that have characterized only airborne (sound) components of web-borne vibrations.

Figure 1.1: Images of western black widow (Latrodectus hesperus) spider. (A) Female and (B) male during (C) courtship and (D) copulation.

20 21

Western black widows have two active periods; a late summer/fall period, and a spring/summer period (Scott et al., 2019; Sheena Fry, Personal Communication). The proportion of virgin females is higher in the fall when females start to mature. During this period, there is a high mating frequency. Adult females (both mated and virgin), and developing juveniles, overwinter until the following spring, where mated females lay eggs and juveniles continue development. During this period, mating frequency is low as the proportion of virgin females is relatively low compared to mature males (Scott et al., 2019; Sheena Fry, Personal Communication). These two active periods results in different competitive contexts for males. In the fall period, males are under scramble competition to develop faster and find females early (see: Kasumovic and Andrade, 2006; Scott et al., 2019). In the spring period, males are under direct competition, as males may compete with multiple rivals on female webs (Kasumovic and Andrade, 2009; MacLeod, 2013; Scott et al., 2019). Both males and females can mate multiple times during the breeding season (MacLeod, 2013). Although L. hesperus males are capable of polygyny, mating opportunities may be limited due to high mortality rates during mate searching (Scott et al., 2019) and extensive courtship effort (Stoltz et al., 2009; Stoltz and Andrade, 2010). Unlike some other species within the Latrodectus genus, e.g. the Australian redback spider (L. hasselti), where males engage in ritualized self-sacrifice behaviours to be consumed by females after copulation (Andrade, 1996), L. hesperus females that are well-fed rarely consumed males during copulation (Johnson et al., 2011).

In contrast to relatively sessile females, adult males are nomadic and initiate risky mate- searching behaviour at sexual maturity (Baruffaldi and Andrade, 2015; Kasumovic and Andrade, 2004; MacLeod and Andrade, 2014; Scott et al., 2019). Males rely on airborne pheromones carried from female webs to detect, discriminate, and localize females (Baruffaldi and Andrade, 2015; MacLeod and Andrade, 2014; Scott et al., 2018). Upon arriving on female webs, males immediately initiate courtship behaviour, which includes web-borne vibration signaling, tactile signaling, web modification, and wrapping females with silk (Ross and Smith, 1979; Scott et al., 2012; Vibert et al., 2014) (Figure 1.1). Courtship in western black widows can last up to 3 hours on average, and can be divided into three phases (Scott et al., 2012). During the early (distal) phase, males stay away from females and engage in exploratory and web-modification behaviours, where they cut silk and bundle up different silk threads, and add their own silk threads onto the web (Scott et al., 2012). In the later (proximal) phase, males approach females

22 and make repeated attempts to mount the female abdomen and engage is various tactile behaviours before copulation (the third phase) (Scott et al., 2012) (Figure 1.1). All males engage in web-borne vibratory signaling through all three phases of male-female interactions (Chapter 2).

1.2.2 Vibration Signals of L. hesperus

Studies have shown that many Latrodectus males display a multitude of vibratory behaviours during courtship (Table 1.2). For example, Forster (1992) described several vibratory courtship behaviours in the Australian redback spiders (L. hasselti) including “body bouncing”, “web- plucking”, “leg waving”, “palpal boxing”, and “abdominal vibrations”. Similar vibratory behaviours were also noted in L. geometricus, L. pallidus, L. mactans, and L. revivensis (Table 1.2). In L. hesperus, Ross and Smith (1979) previously provided qualitative descriptions of several vibratory behaviours of males and females, including “body shakes”, “abdominal vibrations”, “web shaking”, “twitches” and “push-ups”. However, whether these different behaviours generate different signal types, and the signaling mechanisms involved are not yet known. Additionally, given the lack of consistent and standardized nomenclature in the literature, it is difficult to compare signaling behaviours and repertoires among species.

Recently, Vibert et al. (2014) recorded courting male vibrations and characterized male abdominal vibration signals as they entered female webs. Male abdominal vibration signals are low-amplitude and low frequency (<50 Hz) vibrations that are distinct from prey generated vibrations (Vibert et al., 2014). Females may use male abdominal vibration signals to discriminate mates from prey (Vibert et al., 2014). Aside from this study, very little is currently known about vibratory communication in widow spiders.

Table 1.2: Non-exhaustive list of literatures describing courtship vibratory behaviours in different Latrodectus species.

Vibratory Courtship Species Citations Behaviours -Abdominal vibration -(McCrone and Levi, 1964) -Abdominal jerking L. bishopi -Web jerking -Palpal boxing

-(Segoli et al., 2008) L. geometricus -Abdominal vibrations -Abdominal vibration -(Forster, 1992) -Body bouncing -(Andrade, 1996) -(Stoltz et al., 2009, 2008) -Web-plucking -(De Luca et al., 2015) L. hasselti -Leg waving -Palpal boxing/knocking -Push-ups

-Abdominal twitching/vibration -(D’Amour et al., 1936) -Body jerking/bouncing -(Kaston, 1970) -(Ross and Smith, 1979) -Web twanging -(Scott et al., 2012) L. hesperus -Palpal boxing/tapping -(Vibert et al., 2014) -Push-ups -Female abdominal twitching

-"Intense jerking" -(Schmidt, 1991) L. lugubris -Plucking -Abdominal tremulation -(Breene and Sweet, 1985) L. mactans -"Female reciprocating -(Herms et al., 1935) movements" -Abdominal vibration -(Harari et al., 2009) L. pallidus -Jerking -Abdominal vibrations -(Anava and Lubin, 1993) -"Jerking and bouncing -(Berendonck, 2003) L. revivensis movement" -Web strumming -Abdominal vibrations -(Neumann and Schneider, 2011) L. tredecimguttatus -Twitching movements

Thesis Goals

The specific goals of my research were to 1) record and quantitatively characterize web-borne vibration signals during male-female interactions and examine signal variation during courtship, 2) examine signal production mechanisms and transmission, 3) experimentally test the function of vibratory signaling, and 4) examine vibration transmission across female body and mechanical tuning of leg joints. The first goal quantitatively addresses the question of whether different vibratory behaviours correspond to different or similar signals and lays the foundation for examining potential complexities in signals. The second goal addresses the relationship between male movements and signal characteristics and provides insights on the types of vibration signals involved during communication. The third goal addresses the importance and functional significance of vibratory signals. The last goal addresses how mechanics of a spider’s body affect vibration perception. I combine behavioural experiments, laser Doppler vibrometry recording and acoustic signal analysis, and high-speed video analysis to provide, to my knowledge, one of the first detailed examinations of vibratory communication in a web-building spider. My research provides novel insights into emergent signal complexities in a web-building spider, as well as the influence of spider body mechanics for possible vibration frequency discrimination in spiders.

1.3 Thesis Outline

Through a series of three manuscripts, I will address my research goals mentioned above. The three studies and their respective objectives are outline below.

Study 1: Vibratory Communication in a Black Widow Spider (Latrodectus hesperus): Signal Structure and Signaling Mechanism

Research Objectives: In this study, I characterize the spectral and temporal properties of different male vibrations signal during male-female interactions and examine a) variations in signal characteristics and signaling behaviour during different phases of male-female interactions and b) correlations between signal characteristics and male traits. Additionally, I examine the movements of different male body structures during signal production to determine signaling mechanisms. I further experimentally test the importance of male abdomen on signal production.

24 25

Lastly, examine transfer functions between male movements and resulting signals to determine the relationship between movement and signal characteristics.

Study 2: Function of Structured Signaling in the Black Widow Spider, Latrodectus hesperus

Research Objectives: This study is a continuation of Study 1. Here I test the hypothesis that male structured signals are important for mating success. I specifically tested whether males structured signaling rates predicted aspects of male traits and mating success. In a separate experiment, I also tested the effects of attenuating male abdominal tremulation signals on mating success.

Study 3: Posture Controls Mechanical Tuning in the Black Widow Spider Mechanosensory System

Research Objectives: This is a collaborative study examining transfer functions of vibrations transmitted along female leg segments and body to determine the frequency filtering properties of female legs. I further examine the full body vibration modes of females to determine the relative motion of different body segments to different vibration frequencies. These measurements were combined with multibody dynamic modelling to simulate vibration transmission along female legs, and determine effects of female mass, and leg posture on mechanical tuning of leg joints. The predictions from these models were then corroborated with vibration recordings with live females.

1.4 References

Ahtiainen, J.J., Alatalo, R.V., Kortet, R., Rantala, M.J., 2006. Immune function, dominance and mating success in drumming male wolf spiders Hygrolycosa rubrofasciata. Behavioral Ecology and Sociobiology 60, 826.

Ahtiainen, J.J., Alatalo, R.V., Kortet, R., Rantala, M.J., 2005. A trade-off between sexual signalling and immune function in a natural population of the drumming wolf spider Hygrolycosa rubrofasciata. Journal of Evolutionary Biology 18, 985–991.

Ahtiainen, J.J., Alatalo, R.V., Kortet, R., Rantala, M.J., 2004. Sexual advertisement and immune function in an arachnid species (Lycosidae). Behavioral Ecology 15, 602–606.

Aicher, B., Markl, H., Masters, W.M., Kirschenlohr, H.L., 1983. Vibration transmission through the walking legs of the fiddler crab, Uca pugilator (Brachyura, Ocypodidae) as measured by laser Doppler vibrometry. Journal of Comparative Physiology A 150, 483– 491.

Aisenberg, A., 2009. Male performance and body size affect female re-mating occurrence in the orb-web spider Leucauge mariana (Araneae, Tetragnathidae). Ethology 115, 1127–1136.

Alatalo, R.V., Kotiaho, J., Mappes, J., Parri, S., 1998. Mate choice for offspring performance: major benefits or minor costs? Proceedings of the Royal Society of London. Series B: Biological Sciences 265, 2297–2301.

Anava, A., Lubin, Y., 1993. Presence of gender cues in the web of a widow spider, Latrodectus revivensis, and a description of courtship behaviour. Bulletin of the British Arachnological Society 9, 119–122.

Andrade, M.C., 2019. Sexual selection and social context: Web-building spiders as emerging models for adaptive plasticity. Advances in the Study of Behavior, 51, 177–250.

Andrade, M.C.B., 2003. Risky mate search and male self-sacrifice in redback spiders. Behavioral Ecology 14, 531–538.

Andrade, M.C.B., 1996. Sexual Selection for Male Sacrifice in the Australian Redback Spider. Science 271, 70–72.

Andrade, M.C.B., MacLeod, E.C., 2015. Potential for CFC in black widows (Genus Latrodectus): Mechanisms and social context, in: Peretti, A.V., Aisenberg, A. (Eds.), Cryptic Female Choice in Arthropods: Patterns, Mechanisms and Prospects. Springer International Publishing, Cham, pp. 27–53.

Barrantes, G., 2008. Courtship behavior and copulation in Tengella radiata (Araneae, Tengellidae). Journal of Arachnology 36, 606–608.

Barth, F.G., 2012a. Learning from animal sensors: The clever “design” of spider mechanoreceptors, in: Bioinspiration, Biomimetics, and Bioreplication 2012. Presented at

26 27

the Bioinspiration, Biomimetics, and Bioreplication 2012, International Society for Optics and Photonics, p. 833904.

Barth, F.G., 2012b. Spider strain detection, in: Barth, F.G., Humphrey, J.A.C., Srinivasan, M.V. (Eds.), Frontiers in Sensing: From Biology to Engineering. Springer, Vienna, pp. 251– 273. https://doi.org/10.1007/978-3-211-99749-9_17

Barth, F.G., 2004. Spider mechanoreceptors. Current Opinion in Neurobiology 14, 415–422.

Barth, F.G., 2002. Spider senses – technical perfection and biology1. Zoology 105, 271–285.

Barth, F.G., 1997. Vibratory communication in spiders: Adaptation and compromise at many levels, in: Lehrer, M. (Ed.), Orientation and Communication in Arthropods, EXS. Birkhäuser, Basel, pp. 247–272.

Barth, F.G., Bohnenberger, J., 1978. Lyriform slit sense organ: Thresholds and stimulus amplitude ranges in a multi-unit mechanoreceptor. Journal of Comparative Physiology A 125, 37–43.

Barth, F.G., Geethabali, 1982. Spider vibration receptors: Threshold curves of individual slits in the metatarsal lyriform organ. Journal of Comparative Physiology A 148, 175–185.

Barth, F.G., Pickelmann, P., 1975. Lyriform slit sense organs. Journal of Comparative Physiology A 103, 39–54.

Barth, F.G., Schmitt, A., 1991. Species recognition and species isolation in wandering spiders (Cupiennius spp.; Ctenidae). Behavioral Ecology and Sociobiology 29, 333–339.

Bartos, M., 1997. Quantitative analyses of male courtship behaviour in Pholcus phalangioides (Fuesslin, 1775) (Araneae, Pholcidae), in: Proceedings of the 17th European Colloquium of Arachnology, Edinburgh. pp. 171–176.

Baruffaldi, L., Andrade, M.C.B., 2015. Contact pheromones mediate male preference in black widow spiders: avoidance of hungry sexual cannibals? Animal Behaviour 102, 25–32.

Baurecht, D., Barth, F.G., 1993. Vibratory communication in spiders. Journal of Comparative Physiology A 173, 309–319.

Baurecht, D., Barth, F.G., 1992. Vibratory communication in spiders. Journal of Comparative Physiology A 171, 231–243.

Berendonck, B., 2003. Reproductive strategies in Latrodectus revivensis (Araneae: Theridiidae): functional morphology and sexual cannibalism (Ph.D. Thesis). University of Dusseldorf, Germany.

Blackledge, T.A., Kuntner, M., Agnarsson, I., 2011. The form and function of spider orb webs, in: Advances in Insect Physiology. Elsevier, pp. 175–262.

Borgia, G., Coleman, S.W., 2000. Co-option of male courtship signals from aggressive display in bowerbirds. Proceedings of the Royal Society of London. Series B: Biological Sciences 267, 1735–1740.

Bradbury, J.W., Vehrencamp, S.L., 2011. Principles of animal communication, 2nd edn Sunderland. MA: Sinauer Associates.

Breene, R.G., Sweet, M.H., 1985. Evidence of insemination of multiple females by the male black widow spider, Latrodectus Mactans (Araneae, Theridiidae). The Journal of Arachnology 13, 331–335.

Calbacho-Rosa, L., Galicia-Mendoza, I., Dutto, M.S., Córdoba-Aguilar, A., Peretti, A.V., 2013. Copulatory behavior in a pholcid spider: males use specialized genitalic movements for sperm removal and copulatory courtship. Naturwissenschaften 100, 407– 416.

Candolin, U., 2003. The use of multiple cues in mate choice. Biological Reviews 78, 575–595.

Coleman, S.W., Patricelli, G.L., Borgia, G., 2004. Variable female preferences drive complex male displays. Nature 428, 742–745.

Coyle, F.A., O’Shields, T.C., 1990. Courtship and mating behavior of Thelechoris Karschi (Araneae, Dipluridae), an African funnel-web spider. The Journal of Arachnology 18, 281–296.

Cross, F.R., Jackson, R.R., Pollard, S.D., Walker, M.W., 2007. Cross-modality effects during male-male interactions of jumping spiders. Behavioural Processes 75, 290–296.

Cummings, M.E., Endler, J.A., 2018. 25 Years of sensory drive: the evidence and its watery bias. Current Zoology 64, 471–484.

D’Amour, F.E., Becker, F.E., van Riper, W., 1936. The black widow spider. The Quarterly Review of Biology 11, 123–160.

De Luca, P.A., Stoltz, J.A., Andrade, M.C.B., Mason, A.C., 2015. Metabolic efficiency in courtship favors males with intermediate mass in the Australian redback spider, Latrodectus hasselti. Journal of Insect Physiology 72, 35–42.

De Marco, R., Menzel, R., 2005. Encoding spatial information in the waggle dance. Journal of Experimental Biology 208, 3885–3894.

Dierkes, S., Barth, F.G., 1995. Mechanism of signal production in the vibratory communication of the wandering spider Cupiennius getazi (Arachnida, Araneae). Journal of Comparative Physiology A 176, 31–44.

Doucet, S.M., Montgomerie, R., 2003. Multiple sexual ornaments in satin bowerbirds: ultraviolet plumage and bowers signal different aspects of male quality. Behavioral Ecology 14, 503–509.

Drosopoulos, S., Claridge, M.F., 2005. Insect sounds and communication: Physiology, behaviour, ecology, and evolution. CRC Press.

Dutto, M.S., Calbacho-Rosa, L., Peretti, A.V., 2011. Signalling and sexual conflict: Female spiders use stridulation to inform males of sexual receptivity. Ethology 117, 1040–1049.

Elias, D.O., Hebets, E.A., Hoy, R.R., 2006a. Female preference for complex/novel signals in a spider. Behavioral Ecology 17, 765–771.

Elias, D.O., Hebets, E.A., Hoy, R.R., Maddison, W.P., Mason, A.C., 2006b. Regional seismic song differences in sky island populations of the jumping spider Habronattus pugillis Griswold (Araneae, Salticidae). The Journal of Arachnology 34, 545–556.

Elias, D.O., Hebets, E.A., Hoy, R.R., Mason, A.C., 2005. Seismic signals are crucial for male mating success in a visual specialist jumping spider (Araneae: Salticidae). Animal Behaviour 69, 931–938.

Elias, D.O., Kasumovic, M.M., Punzalan, D., Andrade, M.C.B., Mason, A.C., 2008. Assessment during aggressive contests between male jumping spiders. Animal Behaviour 76, 901–910.

Elias, D.O., Land, B.R., Mason, A.C., Hoy, R.R., 2006c. Measuring and quantifying dynamic visual signals in jumping spiders. Journal of Comparative Physiology A 192, 785–797.

Elias, D.O., Lee, N., Hebets, E.A., Mason, A.C., 2006d. Seismic signal production in a wolf spider: parallel versus serial multi-component signals. Journal of Experimental Biology 209, 1074–1084.

Elias, D.O., Maddison, W.P., Peckmezian, C., Girard, M.B., Mason, A.C., 2012. Orchestrating the score: complex multimodal courtship in the Habronattus coecatus group of Habronattus jumping spiders (Araneae: Salticidae). Biological Journal of the Linnean Society 105, 522–547.

Elias, D.O., Mason, A.C., 2014. The role of wave and substrate heterogeneity in vibratory communication: Practical issues in studying the effect of vibratory environments in communication, in: Cocroft, R.B., Gogala, M., Hill, P.S.M., Wessel, A. (Eds.), Studying Vibrational Communication, Animal Signals and Communication. Springer, Berlin, Heidelberg, pp. 215–247.

Elias, D.O., Mason, A.C., 2011. Signaling in variable environments: substrate-borne signaling mechanisms and communication behavior in spiders. The Use of Vibrations in Communication: properties, mechanisms and function across taxa. Kerala: Research Signpost.

Elias, D.O., Mason, A.C., Hebets, E.A., 2010a. A signal-substrate match in the substrate-borne component of a multimodal courtship display. Current Zoology 56, 370–378.

Elias, D.O., Mason, A.C., Hoy, R.R., 2004. The effect of substrate on the efficacy of seismic courtship signal transmission in the jumping spider Habronattus dossenus (Araneae: Salticidae). Journal of Experimental Biology 207, 4105–4110.

Elias, D.O., Mason, A.C., Maddison, W.P., Hoy, R.R., 2003. Seismic signals in a courting male jumping spider (Araneae: Salticidae). Journal of Experimental Biology 206, 4029– 4039.

Elias, D.O., Sivalinghem, S., Mason, A.C., Andrade, M.C.B., Kasumovic, M.M., 2014. Mate- guarding courtship behaviour: tactics in a changing world. Animal Behaviour 97, 25–33.

Elias, D.O., Sivalinghem, S., Mason, A.C., Andrade, M.C.B., Kasumovic, M.M., 2010b. Vibratory communication in the jumping spider Phidippus clarus: Substrate-borne courtship signals are important for male mating success. Ethology 116, 990–998.

Endler, J.A., 1992. Signals, signal conditions, and the direction of evolution. The American Naturalist 139, S125–S153.

Endler, J.A., Basolo, A.L., 1998. Sensory ecology, receiver biases and sexual selection. Trends in Ecology & Evolution 13, 415–420.

Endler, J.A., Houde, A.E., 1995. Geographic variation in female preferences for male traits in Poecilia Reticulata. Evolution 49, 456–468.

Fleishman, L.J., 1988. Sensory influences on physical design of a visual display. Animal Behaviour 36, 1420–1424.

Forster, L.M., 1992. The stereotyped behavior of sexual cannibalism in Latrodectus hasselti Thorell (Araneae, Theridiidae), the Australian Redback Spider. Australian Journal of Zoology 40, 1–11.

Frohlich, C., Buskirk, R.E., 1982. Transmission and attenuation of vibration in orb spider webs. Journal of Theoretical Biology 95, 13–36.

Garb, J.E., González, A., Gillespie, R.G., 2004. The black widow spider genus Latrodectus (Araneae: Theridiidae): phylogeny, biogeography, and invasion history. Molecular phylogenetics and evolution 31, 1127–1142.

Gibson, J.S., Uetz, G.W., 2008. Seismic communication and mate choice in wolf spiders: components of male seismic signals and mating success. Animal Behaviour 75, 1253– 1262.

Gingl, E., Burger, A.-M., Barth, F.G., 2006. Intracellular recording from a spider vibration receptor. Journal of Comparative Physiology A 192, 551–558.

Girard, M.B., Elias, D.O., Kasumovic, M.M., 2015. Female preference for multi-modal courtship: multiple signals are important for male mating success in peacock spiders. Proceedings of the Royal Society B: Biological Sciences 282, 20152222.

Grafe, T.U., Wanger, T.C., 2007. Multimodal signaling in male and female foot-flagging frogs Staurois guttatus (Ranidae): An alerting function of calling. Ethology 113, 772–781.

Griswold, C.E., Coddington, J.A., Hormiga, G., Scharff, N., 1998. Phylogeny of the orb-web building spiders (Araneae, Orbiculariae: Deinopoidea, Araneoidea). Zoological Journal of the Linnean Society 123, 1–99.

Guilford, T., Dawkins, M.S., 1993. Receiver psychology and the design of animal signals. Trends in Neurosciences 16, 430–436.

Guilford, T., Dawkins, M.S., 1991. Receiver psychology and the evolution of animal signals. Animal Behaviour 42, 1–14.

Gwinner-Hanke, H., 1970. Zum Verhalten zweier stridulierender Spinnen Steatoda bipunctata Linné und Teutana grossa Koch (Theridiidae, Araneae), unter besonderer Berücksichtigung des Fortpflanzungsverhaltens. Zeitschrift für Tierpsychologie 27, 649– 678.

Harari, A.R., Ziv, M., Lubin, Y., 2009. Conflict or co-operation in the courtship display of the white widow spider, Latrodectus pallidus. Journal of Arachnology 37, 254–260.

Hebets, E.A., 2008. Seismic signal dominance in the multimodal courtship display of the wolf spider Schizocosa stridulans Stratton 1991. Behavioral Ecology 19, 1250–1257.

Hebets, E.A., 2005. Attention-altering signal interactions in the multimodal courtship display of the wolf spider Schizocosa uetzi. Behavioral Ecology 16, 75–82.

Hebets, E.A., Elias, D.O., Mason, A.C., Miller, G.L., Stratton, G.E., 2008. Substrate- dependent signalling success in the wolf spider, Schizocosa retrorsa. Animal Behaviour 75, 605–615.

Hebets, E.A., Papaj, D.R., 2005. Complex signal function: developing a framework of testable hypotheses. Behavioral Ecology and Sociobiology 57, 197–214.

Hebets, E.A., Vink, C.J., Sullivan-Beckers, L., Rosenthal, M.F., 2013. The dominance of seismic signaling and selection for signal complexity in Schizocosa multimodal courtship displays. Behavioral Ecology and Sociobiology 67, 1483–1498.

Herberstein, M.E., Wignall, A., 2011. Introduction: Spider biology. Spider Behaviour: Flexibility and Versatility 1–30.

Herberstein, M.E., Wignall, A.E., Hebets, E.A., Schneider, J.M., 2014. Dangerous mating systems: Signal complexity, signal content and neural capacity in spiders. Neuroscience & Biobehavioral Reviews, Beyond Sexual Selection: the evolution of sex differences from brain to behavior 46, 509–518.

Hergenröder, R., Barth, F.G., 1983. The release of attack and escape behavior by vibratory stimuli in a wandering spider (Cupiennim salei Keys.). Journal of Comparative Physiology A 152, 347–359.

Herms, W.B., Bailey, S.F., McIvor, B., 1935. The Black Widow Spider. Bulletin of the California Agricultural Experiment Station.

Higham, J.P., Hebets, E.A., 2013. An introduction to multimodal communication. Behavioral Ecology and Sociobiology 67, 1381–1388.

Hölldobler, B., 1999. Multimodal signals in ant communication. Journal of Comparative Physiology A 184, 129–141.

Hößl, B., Böhm, H.J., Rammerstorfer, F.G., Barth, F.G., 2007. Finite element modeling of arachnid slit sensilla—I. The mechanical significance of different slit arrays. Journal of Comparative Physiology A 193, 445–459.

Hößl, B., Böhm, H.J., Rammerstorfer, F.G., Müllan, R., Barth, F.G., 2006. Studying the deformation of arachnid slit sensilla by a fracture mechanical approach. Journal of Biomechanics 39, 1761–1768.

Hößl, B., Böhm, H.J., Schaber, C.F., Rammerstorfer, F.G., Barth, F.G., 2009. Finite element modeling of arachnid slit sensilla: II. Actual lyriform organs and the face deformations of the individual slits. Journal of Comparative Physiology A 195, 881–894.

Jawor, J.M., Breitwisch, R., 2004. Multiple ornaments in male northern cardinals, Cardinalis cardinalis, as indicators of condition. Ethology 110, 113–126.

Jawor, J.M., Gray, N., Beall, S.M., Breitwisch, R., 2004. Multiple ornaments correlate with aspects of condition and behaviour in female northern cardinals, Cardinalis cardinalis. Animal Behaviour 67, 875–882.

Jocqué, R., 2005. Six stridulating organs on one spider (Araneae, Zodariidae): Is this the limit? Journal of Arachnology 33, 597–603.

Johnson, J.C., Trubl, P., Blackmore, V., Miles, L., 2011. Male black widows court well-fed females more than starved females: silken cues indicate sexual cannibalism risk. Animal Behaviour 82, 383–390.

Johnstone, R.A., 1996. Multiple displays in animal communication: ‘backup signals’ and ‘multiple messages.’ Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 351, 329–338.

Kaston, B.J., 1970. Comparative biology of American black widow spiders. Transactions of the San Diego Society of Natural History.

Kasumovic, M.M., Andrade, M.C.B., 2009. A change in competitive context reverses sexual selection on male size. Journal of Evolutionary Biology 22, 324–333.

Kasumovic, M.M., Andrade, M.C.B., 2006. Male development tracks rapidly shifting sexual versus natural selection pressures. Current Biology 16, R242–R243.

Kasumovic, M.M., Andrade, M.C.B., 2004. Discrimination of airborne pheromones by mate- searching male western black widow spiders (Latrodectus hesperus): species- and population-specific responses. Canadian Journal of Zoology. 82, 1027–1034.

Kirchner, W.H., Lindauer, M., Michelsen, A., 1988. Honeybee dance communication: acoustical indication of direction in round dances. Naturwissenschaften 75, 629–630.

Klärner, D., Barth, F.G., 1982. Vibratory signals and prey capture in orb-weaving spiders (Zygiella x-notata, Nephila clavipes; Araneidae). Journal of Comparative Physiology A 148, 445–455.

Kodric-Brown, A., Nicoletto, P.F., 2001. Female choice in the guppy (Poecilia reticulata): the interaction between male color and display. Behavioral Ecology and Sociobiology 50, 346–351.

Kotiaho, J., Alatalo, R.V., Mappes, J., Parri, S., 1996. Sexual selection in a wolf spider: Male drumming activity, body size, and viability. Evolution 50, 1977–1981.

Kotiaho, J.S., Alatalo, R.V., Mappes, J., Parri, S., 1999. Sexual signalling and viability in a wolf spider (Hygrolycosa rubrofasciata): measurements under laboratory and field conditions. Behavioral Ecology and Sociobiology 46, 123–128.

Kronestedt, T., 1996. Vibratory communication in the wolf spider Hygrolycosa rubrofasciata (Araneae, Lycosidae). Revue Suisse de Zoologie 4, 341–354.

Landolfa, M.A., Barth, F.G., 1996. Vibrations in the orb web of the spider Nephila clavipes: cues for discrimination and orientation. Journal of Comparative Physiology A 179, 493– 508.

Lee, R. C. P., Nyffeler, M., Krelina, E., Pennycook, B.W., 1986. Acoustic communication in two spider species of the genus Steatoda (Araneae, Theridiidae). Mitteilungen der Schweizerischen Entomologischen Gesellschaft 59, 337–348.

Lindström, L., Ahtiainen, J.J., Mappes, J., Kotiaho, J.S., Lyytinen, A., Alatalo, R.V., 2006. Negatively condition dependent predation cost of a positively condition dependent sexual signalling. Journal of Evolutionary Biology 19, 649–656.

Lubin, Y.D., 1986. Courtship and Alternative Mating Tactics in a Social Spider. The Journal of Arachnology 14, 239–257.

MacLeod, E., 2013. New insights into the evolutionary maintenance of male mate choice behaviour using the western black widow spider, Latrodectus hesperus (Thesis).

MacLeod, E.C., Andrade, M.C.B., 2014. Strong, convergent male mate choice along two preference axes in field populations of black widow spiders. Animal Behaviour 89, 163– 169.

Maklakov, A.A., Bilde, T., Lubin, Y., 2003. Vibratory courtship in a web-building spider: signalling quality or stimulating the female? Animal Behaviour 66, 623–630.

Mappes, J., Alatalo, R.V., Kotiaho, J., Parri, S., 1996. Viability costs of condition-dependent sexual male display in a drumming wolf spider. Proceedings of the Royal Society of London. Series B: Biological Sciences 263, 785–789.

Masters, W.M., 1984a. Vibrations in the orbwebs of Nuctenea sclopetaria (Araneidae): II. Prey and wind signals and the spider’s response threshold. Behavioral Ecology and Sociobiology 15, 217–223.

Masters, W.M., 1984b. Vibrations in the orbwebs of Nuctenea sclopetaria (Araneidae). Behavioral Ecology and Sociobiology 15, 207–215.

Masters, W.M., Markl, H., 1981. Vibration signal transmission in spider orb webs. Science 213, 363–365.

McCrone, J.D., Levi, H.W., 1964. North American widow spiders of the Latrodectus curacaviensis Group (Araneae, Theridiidae). Psyche 71, 12–27.

Miller, C.T., Bee, M.A., 2012. Receiver psychology turns 20: is it time for a broader approach? Animal Behaviour 83, 331–343.

Molina, J., Schaber, C.F., Barth, F.G., 2009. In search of differences between the two types of sensory cells innervating spider slit sensilla (Cupiennius salei Keys.). Journal of Comparative Physiology A 195, 1031. https://doi.org/10.1007/s00359-009-0477-9

Morris, G., De Luca, P., 1998. Courtship communication in meadow katydids: Female preference for large male vibrations. Behaviour 135, 777–794. https://doi.org/10.1163/156853998792640422

Mortimer, B., 2019. A spider’s vibration landscape: Adaptations to promote vibrational information transfer in orb webs. Integrative and Comparative Biology 59, 1636–1645.

Mortimer, B., Gordon, S.D., Holland, C., Siviour, C.R., Vollrath, F., Windmill, J.F.C., 2014. The speed of sound in silk: Linking material performance to biological function. Advanced Materials 26, 5179–5183.

Mortimer, B., Soler, A., Siviour, C.R., Zaera, R., Vollrath, F., 2016. Tuning the instrument: sonic properties in the spider’s web. Journal of The Royal Society Interface 13, 20160341.

Mortimer, B., Soler, A., Wilkins, L., Vollrath, F., 2019. Decoding the locational information in the orb web vibrations of Araneus diadematus and Zygiella x-notata. Journal of The Royal Society Interface 16, 20190201.

Mowles, S.L., Jennions, M., Backwell, P.R., 2017. Multimodal communication in courting fiddler crabs reveals male performance capacities. Royal Society Open Science 4, 161093.

Naftilan, S.A., 1999. Transmission of vibrations in funnel and sheet spider webs. International Journal of Biological Macromolecules 24, 289–293.

Narins, P.M., Capranica, R.R., 1978. Communicative significance of the two-note call of the treefrog Eleutherodactylus coqui. Journal of Comparative Physiology A 127, 1–9.

Neumann, R., Schneider, J.M., 2011. Frequent failure of male monopolization strategies as a cost of female choice in the black widow spider Latrodectus tredecimguttatus. Ethology 117, 1057–1066.

O’Hanlon, J.C., Wignall, A.E., Herberstein, M.E., 2017. Short and fast vs long and slow: Age changes courtship in male orb-web spiders (Argiope keyserlingi). The Science of Nature 105, 3.

Parri, S., Alatalo, R.V., Kotiaho, J., Mappes, J., 1997. Female choice for male drumming in the wolf spider Hygrolycosa rubrofasciata. Animal Behaviour 53, 305–312.

Parri, S., Alatalo, R.V., Kotiaho, J.S., Mappes, J., Rivero, A., 2002. Sexual selection in the wolf spider Hygrolycosa rubrofasciata: Female preference for drum duration and pulse rate. Behavioral Ecology 13, 615–621.

Partan, S.R., 2013. Ten unanswered questions in multimodal communication. Behavioral Ecology and Sociobiology 67, 1523–1539.

Partan, S.R., Marler, P., 2005. Issues in the Classification of Multimodal Communication Signals. The American Naturalist 166, 231–245.

Peretti, A., Eberhard, W.G., Briceño, R.D., 2006. Copulatory dialogue: Female spiders sing during copulation to influence male genitalic movements. Animal Behaviour 72, 413– 421.

Preininger, D., Boeckle, M., Freudmann, A., Starnberger, I., Sztatecsny, M., Hödl, W., 2013. Multimodal signaling in the small torrent frog (Micrixalus saxicola) in a complex acoustic environment. Behavioral Ecology and Sociobiology 67, 1449–1456.

Rivero, A., Alatalo, R.V., Kotiaho, J.S., Mappes, J., Parri, S., 2000. Acoustic signalling in a wolf spider: can signal characteristics predict male quality? Animal Behaviour 60, 187– 194.

Robson, T.E., Goldizen, A.W., Green, D.J., 2005. The multiple signals assessed by female satin bowerbirds: could they be used to narrow down females’ choices of mates? Biology Letters 1, 264–267.

Rosenthal, M.F., Hebets, E.A., Kessler, B., McGinley, R., Elias, D.O., 2019. The effects of microhabitat specialization on mating communication in a wolf spider. Behavioral Ecology 30, 1398–1405.

Ross, K., Smith, R.L., 1979. Aspects of the courtship behavior of the black widow spider, Latrodectus hesperus (Araneae: Theridiidae), with evidence for the existence of a contact sex pheromone. The Journal of Arachnology 7, 69–77.

Rovner, J.S., Barth, F.G., 1981. Vibratory communication through living plants by a tropical wandering spider. Science 214, 464–466.

Rowe, C., 2013. Receiver psychology: a receiver’s perspective. Animal Behaviour 85, 517–523.

Rowe, C., 2002. Sound improves visual discrimination learning in avian predators. Proceedings of the Royal Society of London. Series B: Biological Sciences 269, 1353–1357.

Rowe, C., 1999. Receiver psychology and the evolution of multicomponent signals. Animal Behaviour 58, 921–931.

Rowe, C., Halpin, C., 2013. Why are warning displays multimodal? Behavioral Ecology and Sociobiology 67, 1425–1439.

Rundus, A.S., Sullivan-Beckers, L., Wilgers, D.J., Hebets, E.A., 2011. Females are choosier in the dark: Environment dependent reliance on courtship components and its impact on fitness. Evolution 65, 268–282.

Ruxton, G.D., Schaefer, H.M., 2011. Resolving current disagreements and ambiguities in the terminology of animal communication. Journal of Evolutionary Biology 24, 2574–2585.

Ryan, M.J., 1980. Female mate choice in a Neotropical frog. Science 209, 523–525.

Ryan, M.J., Fox, J.H., Wilczynski, W., Rand, A.S., 1990. Sexual selection for sensory exploitation in the frog Physalaemus pustulosus. Nature 343, 66–67.

Ryan, M.J., Rand, A.S., 1990. The sensory basis of sexual selection for complex calls in the túngara frog, Physalaemus Pustulosus (sexual selection for sensory exploitation). Evolution 44, 305–314.

Ryan, M.J., Rand, A.S., Butlin, R.K., Guilford, T., Krebs, J.R., 1993. Sexual selection and signal evolution: the ghost of biases past. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 340, 187–195.

Schaber, C.F., Gorb, S.N., Barth, F.G., 2012. Force transformation in spider strain sensors: white light interferometry. Journal of The Royal Society Interface 9, 1254–1264.

Scheuber, H., Jacot, A., Brinkhof, M.W.G., 2004. Female preference for multiple condition- dependent components of a sexually selected signal. Proceedings of the Royal Society of London. Series B: Biological Sciences 271, 2453–2457.

Schmidt, G., 1991. Further crossing experiments in Latrodectus species (Araneida: Theridiidae). Bulletin Sociéte Neuchâteloise des Sciences Naturelles 116, 215–222.

Schmitt, A., Friedel, T., Barth, F.G., 1993. Importance of pause between spider courtship vibrations and general problems using synthetic stimuli in behavioural studies. Journal of Comparative Physiology A 172, 707–714.

Schmitt, A., Schuster, M., Barth, F.G., 1994. Vibratory communication in a wandering spider, Cupiennius getazi: Female and male preferences for features of the conspecific male’s releaser. Animal Behaviour 48, 1155–1171.

Schüch, W., Barth, F.G., 1985. Temporal patterns in the vibratory courtship signals of the wandering spider Cupiennius salei Keys. Behavioral Ecology and Sociobiology 16, 263– 271.

Scott, C., Vibert, S., Gries, G., 2012. Evidence that web reduction by western black widow males functions in sexual communication. The Canadian Entomologist 144, 672–678.

Scott, C.E., Anderson, A.G., Andrade, M.C.B., 2018. A review of the mechanisms and functional roles of male silk use in spider courtship and mating. Journal of Arachnology 46, 173–206.

Scott, C.E., McCann, S., Andrade, M.C.B., 2019. Male black widows parasitize mate- searching effort of rivals to find females faster. Proceedings of the Royal Society B: Biological Sciences 286, 20191470.

Segoli, M., Arieli, R., Sierwald, P., Harari, A.R., Lubin, Y., 2008. Sexual Cannibalism in the Brown Widow Spider (Latrodectus geometricus). Ethology 114, 279–286.

Seyfarth, E.A., French, A.S., 1994. Intracellular characterization of identified sensory cells in a new spider mechanoreceptor preparation. Journal of Neurophysiology 71, 1422–1427.

Shamble, P.S., Wilgers, D.J., Swoboda, K.A., Hebets, E.A., 2009. Courtship effort is a better predictor of mating success than ornamentation for male wolf spiders. Behavioural Ecology 20, 1242–1251.

Simkovic, V., Andrade, M.C.B., 2019. Seasonal variation in sexual behavior and web aggregation in a little-known long-jawed spider (Tetragnatha straminea) (Araneae: Tetragnathidae). Journal of Arachnology 47, 28–36.

Singer, F., Riechert, S., Becker, E., 2005. Male induction of female quiescence/catalepsis during courtship in the spider, Agelenopsis aperta. Behaviour 142, 57–70.

Singer, F., Riechert, S., Xu, H., Morris, A., Becker, E., Hale, J., Noureddine, M., 2000. Analysis of courtship success in the funnel-web spider Agelenopsis aperta. Behaviour 137, 93–117.

Sivalinghem, S., Kasumovic, M.M., Mason, A.C., Andrade, M.C.B., Elias, D.O., 2010. Vibratory communication in the jumping spider Phidippus clarus: polyandry, male courtship signals, and mating success. Behavioural Ecology 21, 1308–1314.

Smith, C.L., Evans, C.S., 2013. A new heuristic for capturing the complexity of multimodal signals. Behavioral Ecology and Sociobiology 67, 1389–1398.

Stange, N., Page, R.A., Ryan, M.J., Taylor, R.C., 2017. Interactions between complex multisensory signal components result in unexpected mate choice responses. Animal Behaviour 134, 239–247.

Stoltz, J.A., Andrade, M.C.B., 2010. Female’s courtship threshold allows intruding males to mate with reduced effort. Proceedings of the Royal Society B: Biological Sciences 277, 585–592.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2009. Male courtship effort determines female response to competing rivals in redback spiders. Animal Behaviour 77, 79–85.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2008. Females reward courtship by competing males in a cannibalistic spider. Behavioral Ecology and Sociobiology 62, 689–697.

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.

Suter, R.B., Renkes, G., 1984. The courtship of Frontinella Pyramitela (Araneae, Linyphiidae): patterns, vibrations and functions. The Journal of Arachnology 12, 37–54.

Thompson, J.T., Bissell, A.N., Martins, E.P., 2008. Inhibitory interactions between multimodal behavioural responses may influence the evolution of complex signals. Animal Behaviour 76, 113–121.

Tobias, J.A., Aben, J., Brumfield, R.T., Derryberry, E.P., Halfwerk, W., Slabbekoorn, H., Seddon, N., 2010. Song divergence by sensory drive in Amazonian birds. Evolution 64, 2820–2839.

Tso, I.-M., Wu, H.-C., Hwang, I.-R., 2005. Giant wood spider Nephila pilipes alters silk protein in response to prey variation. Journal of Experimental Biology 208, 1053–1061.

Uetz, G.W., Clark, D.L., Roberts, J.A., 2016. Multimodal communication in wolf spiders (Lycosidae)—An emerging model for study, in: Advances in the Study of Behavior. Elsevier, pp. 117–159.

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., Stratton, G., 1983. Communication in spiders. Endeavour 7, 13–18.

Uhl, G., Elias, D., 2011. Spider behaviour: Flexibility and versatility. Cambridge University Press, Cambridge Virant-Doberlet M, Cokl C (2004) Vibrational communication in insects. Neotropical Entomology 33, 121134.

VanderSal, N.D., Hebets, E.A., 2007. Cross-modal effects on learning: A seismic stimulus improves color discrimination learning in a jumping spider. Journal of Experimental Biology 210, 3689–3695.

Vibert, S., Scott, C., Gries, G., 2016. Vibration transmission through sheet webs of hobo spiders (Eratigena agrestis) and tangle webs of western black widow spiders (Latrodectus hesperus). Journal of Comparative Physiology A 202, 749–758.

Vibert, S., Scott, C., Gries, G., 2014. A meal or a male: the ‘whispers’ of black widow males do not trigger a predatory response in females. Frontiers in Zoology 11, 4.

Walcott, C., Kloot, W.G. van der, 1959. The physiology of the spider vibration receptor. Journal of Experimental Zoology 141, 191–244.

Walcott, Charles., 1969. A spider’s vibration receptor: Its anatomy and physiology. Integrative and Comparative Biology 9, 133–144.

Whitehouse, M.E.A., Jackson, R.R., 1994. Intraspecific interactions of Argyrodes antipodiana, a kleptoparasitic spider from New Zealand. New Zealand Journal of Zoology 21, 253– 268.

Wignall, A.E., Herberstein, M.E., 2013a. Male courtship vibrations delay predatory behaviour in female spiders. Scientific Reports 3, 1–5.

Wignall, A.E., Herberstein, M.E., 2013b. The Influence of Vibratory Courtship on Female Mating Behaviour in Orb-Web Spiders (Argiope keyserlingi, Karsch 1878). PLoS One 8.

Wignall, A.E., Kemp, D.J., Herberstein, M.E., 2014. Extreme short-term repeatability of male courtship performance in a tropical orb-web spider. Behavioral Ecology 25, 1083–1088.

Wignall, A.E., Taylor, P.W., 2011. Assassin bug uses aggressive mimicry to lure spider prey. Proceedings of the Royal Society B: Biological Sciences 278, 1427–1433.

Wilgers, D.J., Hebets, E.A., 2012. Seismic signaling is crucial for female mate choice in a multimodal signaling wolf spider. Ethology 118, 387–397.

Zevenbergen, J.M., Schneider, N.K., Blackledge, T.A., 2008. Fine dining or fortress? Functional shifts in spider web architecture by the western black widow Latrodectus hesperus. Animal Behaviour 76, 823–829.

Chapter 2 Vibratory Communication in a Black Widow Spider (Latrodectus hesperus): Signal Structure and Signaling Mechanisms

Senthurran Sivalinghem and Andrew C. Mason

2.1 Abstract

Despite the ubiquity of web-borne vibratory communication among many web-dwelling spiders, few empirical studies, to date, have quantified the characteristics and mechanisms of the signals involved. In this study, we used western black widow spiders, Latrodectus hesperus, and examined male signal characteristics, signal production mechanisms, and signal transmission efficacy on webs. It is not clear how vibrations are transmitted through the chaotic cobweb typical of this species and relatives, and previous work argued that L. hesperus courtship vibrations lacked structure. We video-recorded courtship and copulation behaviour, and used laser vibrometry to record and characterize web-borne vibrations. We examined signaling mechanisms using synchronous high-speed video, and vibrometry recordings. Lastly, we examined signal transmission by measuring transfer functions between body movements and the resulting vibration signal. We found that males displayed three distinct signal types (abdominal tremulation, bounce, and web plucks), each of which were generated by a different signal production mechanism. Our results also show that male bounces and web plucks may convey information about male size. Contrary to earlier work, we show that during the later phases of male-female interactions, males intermittently organize individual signal types into stereotyped sequential displays (‘structured signaling’). Moreover, despite the cobweb structure, transfer function analyses showed that female webs transmit male signals with high efficacy. These results suggest that vibratory communication in L. hesperus, previously classified as simple signalers, can comprise emergent forms of signal complexity, and lay the foundation for future studies on this mode of communication in other web-dwelling species.

2.2 Introduction

Spiders are useful organisms for studying complex communication systems. In particular, vibratory communication in spiders has received considerable focus over the years (Baurecht and Barth, 1992; Elias and Mason, 2014, 2011; Hebets, 2008; Hebets et al., 2013; Herberstein et al., 2014; Kotiaho et al., 1996; Uetz et al., 2009; Uhl and Elias, 2011). Most of the vibratory communication studies to date focus on cursorial hunting spiders such as jumping spiders (Salticidae), wolf spiders (Lycosidae), and wandering spiders (Ctenidae). Extensive research in cursorial spiders show that signals in different species within and among genera can vary in complexity (Elias et al., 2012, 2006b; Hebets et al., 2013; Herberstein et al., 2014), function (Barth and Schmitt, 1991; Elias et al., 2006a, 2005, 2004; Gibson and Uetz, 2008; Hebets, 2005; Hebets et al., 2008; Parri et al., 1997; Rivero et al., 2000; Rosenthal et al., 2019; Sivalinghem et al., 2010; Uetz et al., 2013; Wilgers and Hebets, 2012), and production mechanisms (Elias et al., 2006c, 2003). For example, in many jumping spider species (i.e. genera Habronattus, Phidippus, and Maratus) males use complex ‘multi-modal’ displays combining vibratory signals with elaborate visual signals and movements (Elias et al., 2012, 2003; Girard et al., 2011; Sivalinghem et al., 2010). These multi-modal signals convey information about mates (Sivalinghem et al., 2010), affect mating success (Elias et al., 2005, 2006a, 2010b; Girard et al., 2015; Sivalinghem et al., 2010), play a role during mate guarding and territorial disputes (Elias et al., 2014, 2010a, 2008; Kasumovic et al., 2010, 2009), and are optimized for effective transmission on variable signaling substrates (Elias et al., 2014, 2004; Elias and Mason, 2011). The vibratory component of the multi-modal display itself can be highly complex, where in some species (e.g. Habronattus coecatus) male vibratory signals comprise several components (‘multi- components signals’) that are temporally organized into distinct functional groupings (Elias et al., 2012).

In contrast to knowledge about vibratory communication in cursorial spiders, much less is known about the details of vibratory communication in web-dwelling spiders (Herberstein et al., 2014). Web-dwelling spiders are generally regarded as relatively “simple” with few signal elements, and are generally under-utilized in communication research (Herberstein et al., 2014). Web- dwelling spiders are “sit-and-wait” predators and key constituents of terrestrial ecosystems (Herberstein and Wignall, 2011). In many species, females construct elaborate web structures

41 42 that function to ensnare prey and as substrates for receiving vibratory information (Blackledge et al., 2011; Blackledge and Zevenbergen, 2007; Eberhard, 1990; Landolfa and Barth, 1996). Most species have poor vision and are highly dependent on web-borne vibrations for short-range communication (Barth and Bohnenberger, 1978; Barth and Geethabali, 1982; Herberstein and Wignall, 2011). Despite being considered simple communicators, across many species, courting males engage in a multitude of vibratory behaviours (e.g. “abdominal wagging”, “body rocking”, “leg tapping”, “leg-flexing”, “body jerking”, “web-flexing”, “bouncing”, and “web plucks”) (Aisenberg, 2009; Barrantes, 2008; Forster, 1992; Ross and Smith, 1979; Singer et al., 2000). For example, Whitehouse and Jackson (1994) observed 32 different types of male courtship vibratory behaviours in the kleptoparasitic spider, Argyrodes antipodiana. Despite this, empirical studies on vibratory communication in web-dwelling spiders are scarce and we know very little about the characteristics of the signals involved, and their functions and mechanisms (but see: Herberstein et al., 2014; Vibert et al., 2014; Wignall and Herberstein, 2013a, 2013b).

To date, vibratory behaviours of web-dwelling spiders have only been described for a few species. The large majority of these studies only provide qualitative descriptions of the signals; mostly describing the conspicuous movements of body parts, or the general vibration behaviours of males (Aisenberg, 2009; Barrantes, 2008; Forster, 1992; Ross and Smith, 1979; Singer et al., 2000). Only recently have a few studies recorded and characterized web-borne vibratory signals in some orb-web and cobweb spiders (Vibert et al. 2014; Wignall et al., 2014; Wignall and Herberstein, 2013a, 2013b). Male A. keyserlingi use vibratory signals that are different from prey generated vibrations; at least one of these signals (male “shudders”) reduce female predatory behaviours, and convey information about male traits to influence female mate choice (Wignall et al., 2014; Wignall and Herberstein, 2013a, 2013b). Nevertheless, research in vibratory communication in web-building spiders is still in its early years, as there still remain significant gaps in our understanding of the properties of web-borne vibratory signals, their function, and how signals are generated and transmitted on female webs (Herberstein et al., 2014). Web- dwelling spiders essentially build the substrate on which they signal (“substrate specialists”), and have greater control over the signaling environment as well as the signals (Elias and Mason, 2014, 2011). Thus, they are ideal for investigating signal-substrate/environment interactions, and how these interactions drive signal function and evolution. In this study, we use the western

43 black widow spider (Latrodectus hesperus) to investigate details of vibratory signaling in a web- dwelling spider.

Widow spiders (Theridiidae: Latrodectus) are notorious for their neurotoxic venom, and are exceptional model organisms for studying ecological and social factors underlying sexual selection, and the evolution of life history traits and reproductive tactics (Andrade, 2019, 2003, 1996; Andrade and MacLeod, 2015; Baruffaldi et al., 2019; Elias et al., 2011; Kasumovic and Andrade, 2009, 2006, 2004). As a result, they are ideal organisms for studying communication. Western black widows, L. hesperus, are found across western North America, ranging from Alberta and British Columbia (Canada) to California (U.S.A.), Texas (U.S.A.), and into Mexico (Garb et al., 2004). Females on average are twice as large (e.g., leg length), and can be 15 to 22 times heavier than males (Kaston, 1970). Females build elaborate dense cob-webs comprised of a complex irregular-network of silk treads that functions to ensnare prey and as a medium for perceiving vibrations (Zevenbergen et al., 2008). Females are relatively sessile, and may inhabit a single web site throughout adulthood (MacLeod, 2013). After maturing, adult L. hesperus males initiate mate searching behaviour, and use airborne semiochemical signals from female webs to localize mates (Baruffaldi and Andrade, 2015; MacLeod and Andrade, 2014; Ross and Smith, 1979; Scott et al., 2019). Upon arriving on female webs, males immediately initiate courtship behaviours, lasting up to three hours; these behaviours involve producing web-borne vibrations, modifying female web structure, wrapping females with silk, and various tactile behaviours (MacLeod, 2013).

Previously, Ross and Smith (1979) and Forster (1992) provided qualitative descriptions of several courtship behaviours of L. hesperus and L. hasselti (Australian redback spiders) males that are likely to produce vibrations including behaviours characterized as “body shakes”, “abdominal vibrations”, “web shaking”, “push-ups”, “body bouncing”, “web-plucking”, “leg waving”, and “palpal boxing”. Similar behaviours have also been referred to in L. geometricus (Segoli et al., 2008), L. pallidus (Harari et al., 2009), L. mactans (Breene and Sweet, 1985), and L. revivensis (Anava and Lubin, 1993) without quantification. However, it is unclear whether these different courtship behaviours result in similar or distinct vibration signals, and the mechanisms involved are difficult to discern from simple observations with the naked eye. Additionally, the lack of consistent and standardized nomenclature in the literature makes it difficult to compare signaling repertoires of different species of web-dwelling spiders. For

44 example, the ‘body bounce’ behaviour described in L. hesperus shares resemblance to ‘web bounce’ behaviour in L. hasselti (Forster, 1992), ‘web-flexion’ behaviour in the funnel-web spider Agelenopsis aperta (Singer and Riechert, 1995), and ‘8-leg flexion’ behaviour described in Frontinella pyramitela (Suter and Renkes, 1984).

Recently, Vibert et al. (2014) characterized vibrations of courting male L. hesperus using laser vibrometry. Specifically, they focused on male abdominal tremulation signals during the early phase of courtship as males entered female webs. They showed that abdominal tremulation signals are low-amplitude vibrations, distinct from prey vibrations, and vibrations generated during walking, web cutting, and silk bundling behaviours (Vibert et al., 2014). However, male vibrations were recorded only in the early phase of courtship, in the absence of females on the web, which can affect the signals recorded, and male signaling behaviour. Currently, the characteristics of other signal types, previously observed (see Ross and Smith, 1979), and how signals vary across time within male-female interactions, and potential information content in male vibratory signals remains unknown.

The purpose of this study is to empirically characterize the different types of web-borne vibratory signals, determine signal production mechanisms, and examine the efficacy of signal transmission on female webs of western black widow spiders, Latrodectus hesperus. We first use synchronous video and laser vibrometry recordings to characterize different vibration signals during courtship and copulation, and examine variations between signal types and across time during male-female interactions. Additionally, we also examine potential information content in signals. We then use synchronized high-speed video and laser recordings to examine the movements of body parts during different signal production. We further examine the source of signal generation be comparing relative velocities of body-part movements. In addition, we experimentally test the importance of the abdomen movement for signal production. Lastly, we measure transfer functions of vibration signals relative to body movements to examine vibration transmission on female webs. By examining the characteristics, production mechanisms, and transmission properties of vibratory signals in L. hesperus, and web-dwelling spiders in general, we gain insights into the complexities and function of vibratory signals in web-dwelling spider. Examining the details of communication signals and mechanisms are crucial for generating testable hypotheses on the function and evolution of signals, and can provide insights into possible processing demands on receiver sensory systems.

2.3 Materials and Methods 2.3.1 Behavioural Traits and Signal Characterization

Study Animal

Western black widow (Latrodectus hesperus) males and females used in experiments were descendants of females collected from Hastings Natural History Reserves (Carmel Valley, California, USA), and reared in temperature controlled rooms at the University of Toronto Scarborough. Spiders were kept at 22-24°C, and 14:10 light cycle. Adult males were fed fruit flies, Drosophila melanogaster, twice a week; and adult females were fed one cricket, Acheta domesticus, once a week. Both males and females were given water twice a week. For all adult males and females, at least one week elapsed after final molting (i.e., date of sexual maturity) before their use in experiments.

Experimental Set-up

Custom-made mating arenas consisted of a plastic Tupperware (length x width x height cm3 = 25 x 17 x 7), with four metal bolts (height = 11.43 cm) attached at each corner. The Tupperware was filled with soil to emulate a pseudo-natural environment (Figure S2.1). On one of the metal bolts we attached a square clear plastic lid (5 cm2) on top with hot glue. Adult virgin females were placed on the underside of the lid and allowed at least one week to build webs; most females used these lids to construct their refuge.

We conducted mating trials by placing adult L. hesperus males on female webs; each female was paired with only one male. All males were allowed to court freely, and were given up to six hours to copulate; after which, they were categorized as ‘mated’ or ‘not mated’. We only considered males that engaged in courtship behaviours. Trials where males remained motionless after two hours, or tried to leave the arena, were excluded from our data. Both males and females from those trials were not used afterwards to eliminate any experience effects. All experiments were conducted in an anechoic chamber (Eckel audiometric rooms G-Series xht-batten) on a vibration isolation table (Newport 18235 MT). Mating arenas were illuminated with two LED lights (Ledgo CN-B308).

Vibration Signal Recording Procedure

We continuously recorded web-borne vibrations throughout the entire trial. We recorded vibration signals near the females’ refuge, as this allowed us to approximate the transmitted signals available to females. Vibration signals were recorded using a laser Doppler vibrometer (LDV) (Polytec PDV 100; 20 mm/s peak measurement range; 0.5 Hz - 22.0 kHz frequency range; 24 bits). To enhance reflectivity of the LDV, three pieces of reflective tape (~1mm2) were carefully placed near the entrance of females’ refuge - specifically on silk threads attached to female’s legs. The reflective tapes also served as our measurement points for the LDV. Since the LDV can only detect surface movements parallel to the laser beam, we were limited to only recording the transverse wave component of the vibration signals. All vibration signals were recorded and stored onto an audio data recorder (Sound Devices 722; 16 bits; 48.0 kHz audio sampling rate) for later analysis. Vibration signals from the LDV were concurrently synchronized and recorded onto the audio track of a Sony HD video camera recorder used to video record mating trials (HANDYCAM® NEX-VG10; Sony e-mount lens SEL18200; 44.1 kHz audio sampling rate). We video recorded all male-female interactions, including courtship and copulation.

Vibration Signal Analysis

We recorded complete courtship behaviour and vibratory signals from 16 males, and copulatory vibrations from 12 of the 16 males. We examined three vibration signal types previously described for male L. hesperus (abdominal tremulation, bounce, and web plucks) (Ross and Smith, 1979). For each male, we selected five samples of each vibration types during courtship and five samples during copulation for our analysis. We specifically selected portions where males produced signals roughly at the same distance from our recording point, as this allowed for reliable comparisons among signal types without introducing variation due to web structure. For all recordings, we first measured the peak amplitude of background noise and used this as the threshold amplitude. Only vibration signals with amplitudes above that threshold were analyzed, as any signals below would be indistinguishable from background noise. For each signal, we measured five parameters: (a) signal duration; (b) peak frequency; (c) peak power; (d) peak amplitude; and (e) Root Mean Square (RMS) amplitude (which provide an index of average signal magnitude/strength). All vibration signal analyses were done using Raven Bioacoustics Research Program (Pro version 1.4, Cornell Laboratory of Ornithology, Ithaca, NY, USA). In

Raven, Power Spectral Density (PSD) was computed using a rectangular window Fast Fourier Transform (FFT) with 50% overlap, and 3000 sample Hop size. We used a window sample size set at 6000 samples, with a 3 dB filter bandwidth of 7.08 Hz. For a given signal selection, we used the frequency with the highest PSD value to obtain ‘peak frequency’ and ‘peak power’ of that signal. Amplitude is the magnitude of the maximum relative deviation of the signal waveform. We measured ‘peak amplitude’ as the greater of the absolute value of the maximum and minimum relative deviation of the signal waveform.

Morphometrics

All males were weighed using an electronic balance (Ohaus Explorer; accurate to ±0.01 mg) before mating trials, and preserved in 70% ethanol after trials. We then photographed the dorsum of male cephalothorax using a Nikon Digital Sight camera (Nikon DS-Fi1) mounted on a Nikon SMZ800 dissecting microscope. Photographs were captured using Act-1 (Nikon Corp. 2000). We measured cephalothorax width using Image J (all magnifications on the microscope were pre-calibrated with a ruler).

2.3.2 Vibration Signal Production Mechanism

Synchronized High-speed Video and Vibration Recording

We captured high-speed video recording of male L. hesperus vibratory signaling using a digital high-speed video camera filming at 500 frames/sec (Lightning® RDT/16 camera, High Speed Imaging Inc.; Navitar Zoom 7000 lens; resolution 1280x1024). Simultaneously, vibration signals from the LDV were digitized (National Instruments BNC-2110, 40 samples/frame, 20 kHz sample rate), and synchronized with high-speed video, using a video/data synchronization software (MiDASDA v4.0, Xcitex Inc.).

High-speed Video Analysis

To infer body segments involved in the production of different vibration signals, the movement of different male body segments during vibration signal production was tracked frame-by-frame from high-speed video recordings and converted to velocity. We compared these to corresponding oscillograms of synchronized vibration and video recordings. We focused specifically on the movement of male abdomen tips, cephalothorax, and the femur-patella joint of either the right or left leg attached to the web and visible in our recordings (depending on the

48 orientation of male spider). For all measurements, we considered the long axis of the male body as our x-axis, and measured movements perpendicular to this (y-axis). Males are naturally oriented ventral-side up on female webs. Therefore, displacement in the positive direction indicates upward motion (moving ventrally) towards the horizontal web-sheet, and displacement in the negative direction indicates downward motion (moving dorsally) away from the web- sheet. For the femur-patella joints we also measured movements in the x-axis (lateral), such that, movements in the positive direction indicate leg extension, and movements in the negative direction indicate leg flexion. All measurements were relative to the initial starting position of the body segment, and we used known measurements of either male femur or patella-tibia length as our calibration reference. For each signal type (abdominal tremulation [N = 8 males]); bounce [N = 9 males]; and web plucks [N = 1 male]), we recorded between two to eight bouts per male. Frame-by-frame analyses were done using Tracker (version 4.82; Open Source Physics).

Additionally, we also measured and compared the maximum velocities of each body part during production of abdominal tremulation and bounce signals to infer the main source of the signals. Since velocity values can be positive or negative, depending on the direction, we used the absolute values to obtain maximum velocities.

2.3.3 Importance of Male Abdomen for Signal Production

Experimental Manipulation

In a separate experiment, we examined the importance of male abdomen on signal production. Since male abdominal tremulation and bounce signals both involve conspicuous abdomen movements, we wanted to test whether the abdomen is crucial for producing these signals. Specifically, we performed experiments where we prevented abdominal movements by waxing together male abdomen and cephalothorax, and we tested the effects of immobilizing male abdomen on male signals. This treatment was reversible, so males were recorded with and without their abdomen immobilized. To immobilize male abdomen, we first anesthetized males using CO2 (Praxair Inc.). Then, under a dissecting microscope (Nikon SMZ800), we carefully waxed together their cephalothorax and abdomen with hair-removal wax (placed dorsally). This restricted abdominal movements.

We tested the same males in both the experimental (immobilized) and control conditions (N = 11). We randomly assigned each male to either the experimental-first (N = 6) or control-first (N = 5) condition. Males in the experimental-first group had their abdomen immobilized and allowed to court females, while males in the control-first group were first anesthetized, but were not waxed, and then allowed to court females. All males were allowed to court for up to one hour. After the first courtship trial, males were removed from female webs and prepared for testing the second condition. Males in the experimental-first group had their wax carefully removed, while males in the control-first group had their abdomen and cephalothorax waxed together. All males were given 24 hrs between trials, and were given water. Each male was paired with the same female and web.

Recording Procedures and Signal Analysis

We recorded vibration signals continuously throughout a one hour courtship period (see recording procedures described above). To ensure consistency between conditions, we recorded vibration signals from the same spot on the web. We placed a piece of reflective tape (~1mm2) on web-silks near the female refuge, and all vibration signals were recorded from this spot during experimental and control trials.

For each male, in each condition, we randomly selected five samples of the different vibration signals produced near the recording point; close to the female. Since only abdominal tremulation and bounce signals corresponded to movements of the abdomen in previous observations, we only considered these signal types. We analyzed peak frequency and peak power of signals in each condition.

2.3.3 Vibration Signal Transmission

Transfer Function Analysis

We compared power-spectrum differences between the web-borne vibration signals (recorded with the LDV) and the vibrations of different body parts (from high-speed video analysis above) to make inferences about the transmission of male signals on female webs. Using the body segment movement velocities from high-speed video analysis, we measured the transfer function of the resulting web vibration (output) relative to velocities of male body parts (input). We first interpolated the body segment movement velocity data to match the data points from the LDV

50 recording. We then calculated a transfer function of output vibration from LDV recordings with respect to that of body segment movement velocities.

We also calculated the coherence to determine how much of the recorded web-borne vibrations are predicted by different body segment vibrations. Values of coherence range between zero and one. A coherence of one would indicate a linear relationship between body and web vibrations. Coherence values less than one, but greater than zero, can indicate that body segment vibration alone does not account for resulting web-vibrations, and that other factors (e.g. noise or mechanical properties of the web) may be involved in influencing the power-spectrum of the web vibrations. Lastly, a coherence of 0 would indicate no relationship between body segment movements and web vibrations. All transfer function and coherence analyses were performed using Matlab (MathWorks® version 2018b).

2.3.3 Statistical Analysis

All data, unless otherwise specified, are presented as mean±SD. To determine whether male vibration signal types were distinct, we first performed a principal components analysis (PCA) on signal duration, peak frequency, peak power, peak amplitude, and RMS amplitude to calculate new uncorrelated variables. We then used a discriminant function analysis (DFA) to estimate how well the new PCA scores predicted signal types.

We then examined differences in measured signal parameters among different signal types. Since various parameters were not normally distributed, we used the repeated-measures non-parametric Friedman test. For significant differences, we then conducted pair-wise comparisons using a Wilcoxon signed-rank test to see which signal types differed. Next we examined differences between courtship and copulation vibration signals. We used a Wilcoxon test to compare differences in signal parameters. Because signal peak power and amplitudes are dependent on distance between the signal source and recording position, and we could not control for distance, we only compared differences in signal duration and peak frequency. However, variation in signaling distance at the time of recording is expected to introduce random variation with respect to body movements, and thus this is was a conservative analysis.

We then examined whether signal parameters of different signal types were predicted by male mass and size. Since male mass and size were highly correlated we first performed a PCA on

51 mass and size to calculate a new PCA scores. Only one PC score was calculated that accounted for 89.7% of the variance in male measurements. PC1 was positively associated with male weight (0.95) and size (0.95) with an eigenvalue of 1.79. We used a general linear regression analysis to examine whether the new PCA scores predicted signal parameters of different signal types.

To infer the body segment involved in signal production, we compared maximum velocities of different body parts during abdominal tremulation and bounce signal production. Since velocities of different body segments were measured from the same set of males, we used a repeated measures Friedman’s test. For significant differences, we subsequently used a Wilcoxon pair- wise comparison. To determine the effects of immobilizing male abdomen on abdominal tremulation and bounce signal characteristics, we used a Wilcoxon test to examine differences in peak frequency and peak power of abdominal tremulation and bounce signals between experimental and control condition.

2.4 Results 2.4.1 Vibratory Behaviours and Signals of L. hesperus

Male-female interaction can be divided into three phases (Figure 2.1A) (Scott et al., 2012). We called phase 1 the ‘distal phase’, defined as the time from when the male was placed on the web to when the male approached and initiated first contact with female (40.21 ± 18.31 min, N = 16). Phase 2, termed ‘proximal phase’, was defined as the time from initial contact with female to copulation (insertion of male pedipalp; 99.16 ± 47.29 min, N = 16). The final phase is ‘copulation’, which was the duration of the first copulation (for males that mated; 6.27 ± 2.38 min, N = 12). During the distal phase, males remained distant from females, only moving around the perimeter of the mating arena, and primarily engaged in exploratory and web-modification behaviours. During this phase, males cut silk strands, bundled up loose strands, and added their own silk threads onto female webs, while intermittently producing vibratory signals. In the proximal phase, males repeatedly approached females, and produced vibration signals.

Males produced web-borne vibrations during all three phases of male-female interactions (Figure 2.1A). We recorded at least three signal types (abdominal tremulation, bounce, and web plucks) each associated with different temporal and spectral characteristics, and stereotyped body

52 movements (Figure 2.1B-D & Table 1). These three signal types were previously described, primarily qualitatively, in other Latrodectus species including L. hesperus (Anava and Lubin, 1993; Forster, 1992; Ross and Smith, 1979; Scott et al., 2012).

Abdominal tremulation signals are conspicuous low-amplitude, relatively narrow-band signals produced throughout courtship and during copulation (Figure 2.1Ci & Di). Abdominal tremulation occurred either simultaneously with web plucks or preceding bounce signals. Bounce signals are high-amplitude, broadband percussive signals (Figure 2.1Cii & Dii). Bounces can occur as a single signal following abdominal tremulation, or preceding web plucks, or can occur simultaneously with web plucks. When produced together with web plucks, male movements on the web appears jittery or ‘jerky’. During copulation, bounce signals always followed abdominal tremulation.

Unlike abdominal tremulation and bounce signals, web plucks were relatively inconspicuous, such that they could be confused with incidental vibrations produced during walking behaviour. We therefore classified web pluck signals as vibrations produced while males remained relatively at the same spot on the web plucking silk threads. These were differentiated from vibrations produced while males were in continuous forward motion. Web pluck signals are relatively high frequency narrow-band signals displayed either in groupings, following bounce signals, or simultaneously with abdominal tremulation or bounce signals (Figure 2.1Ciii & Diii). Compared to previously described walking and web-cutting vibrations (see: Vibert et al., 2014), web plucks were characteristically shorter in duration and higher in frequency (Table 2.1). All three male vibration signals showed high degree of variability both within and between males (Table 2.1).

Figure 2.1: Web-borne vibratory signals of male Latrodectus hesperus. (A) Oscillogram of web- borne vibrations recorded during mating trial. Male-female interactions were divided into three phases (distal, proximal, and copulation), and males produced vibratory signals in all three phases. (B) Expanded view of the oscillogram section marked in A (red box). Males produced three distinct types of vibration signals: abdominal tremulation [Ab (green)], bounce [Bo (red)], and web plucks [Pl (orange)]. Web plucks occurred in groups, sometimes simultaneously with bounce signals. (C) Oscillogram of male (i) abdominal tremulation, (ii) bounce, and (iii) web pluck signals. (D) Mean power spectra (thick black line) of male (i) abdominal tremulation, (ii) bounce, and (iii) web pluck signals. Thin coloured lines show means of individual males (N=16).

Table 2.1: Summary of Latrodectus hesperus male signal characteristics during courtship and copulation.

Male Vibration Signal Types Courtship Copulation Abdominal Abdominal Signal Parameters Bounce Mean Web Plucks Bounce Mean Web Plucks Vibration Mean Vibration Mean (SD) Mean (SD) (SD) Mean (SD) (SD) (SD) Duration (s) 8.83 (3.68) 0.2 (0.06) 0.16 (0.02) 26.19 (14.74) 0.45 (0.24) 0.2 (0.2)

Peak Frequency (Hz) 52.3 (15.2) 68.41 (28.59) 107.16 (51.07) 54.9 (22.38) 46.97 (25.25) 128.91 (88.48)

Peak Power (dB) 87.84 (6.74) 106.19 (4.13) 99.61 (6.45) 78.17 (5.73) 102.87 (4.21) 91.68 (6.95)

Peak Amplitude 0.08 (0.06) 0.41 (0.14) 0.2 (0.12) 0.02 (0.01) 0.29 (0.15) 0.12 (0.09) (mm/s) RMS Amplitude 0.02 (0.02) 0.12 (0.05) 0.06 (0.03) 0.005 (0.003) 0.06 (0.02) 0.03 (0.02) (mm/s) Individuals 16 16 16 11 11 12

Signal Characteristics, Variation, and Information Content

For each signal type we measured five parameters (signal duration, peak frequency, peak power, peak amplitude, and RMS amplitude). Two principle components scores were calculated; PC1 accounted to for 65% and PC2 accounted for 24% of the variance in signal parameters. PC1 was positively associated with RMS amplitude (0.97), peak amplitude (0.96), and peak power (0.96) and negatively associated with signal duration (-0.7) with an eigenvalue of 3.27. PC2 was positively associated with peak frequency (0.96) and negatively associated with signal duration

(-0.54) with an eigenvalue of 1.23. A discriminant function analysis was significant (F2,45 = 13.91, Wilks’ λ = 0.62, P < 0.001; Figure 2.2) and had a correct classification rate of 81.3% (Table 2.2).

We found significant differences in all five measured signal parameters among signal types (Table 2.3). Wilcoxon pair-wise comparisons of signal duration showed that abdominal tremulation signals were significantly longer then bounce and web plucks, and bounce signals were significantly longer than web plucks (Table 2.3). Comparisons of peak frequencies showed that abdominal tremulations exhibited significantly lower peak frequencies than bounce and web pluck signals, and bounce signals had significantly lower frequencies than web pluck signals (Table 2.3). Abdominal tremulation signals had significantly lower peak power, amplitudes, and RMS amplitudes compared to bounce and web plucks (Table 2.3). Bounce signals had significantly higher peak power, amplitudes, and RMS amplitudes compared to web pluck signals (Table 2.3).

We used a Wilcoxon pair-wise test to compare duration and peak frequencies of courtship and copulatory vibrations (Table 2.4). Courtship abdominal tremulation and bounce signals had significantly shorter durations compared to copulation tremulation and bounce signals, respectively (Table 2.4). However, there were no differences in peak frequency between courtship and copulation signals (Table 2.4). We also found no significant differences in signal duration or peak frequency between web pluck signals produced during courtship and copulation (Table 2.4). We examined whether male traits (PC1; associated with male weight and size) predicted signal characteristics using a linear regression analysis. Since male size and weight were highly correlated (r = 0.79; p < 0.001), we used PC1 scores to account for both traits. We found that, for abdominal tremulation signals, none of the measured signal parameters were

55 56 predicted by male traits (P > 0.05). However, we found that PC1 was significantly correlated 2 with bounce signal power (R = 0.29; F1,14 = 5.57; P = 0.033; correlation coefficient (r) = 0.53). Larger males produced bounce signals with higher power (y-intercept = 106.19; B = -2.20; Beta = -0.53; T = -2.36; p = 0.033). Male traits were also significantly correlated with web pluck 2 signal power (R = 0.26; F1,14 = 5.03; p = 0.042; correlation coefficient (r) = 0.51). Larger males produced plucks with higher power (y-intercept = 99.61; B = -3.32; Beta = -0.51; t = -2.24; p = 0.042). Therefore, our results show that both bounce and web pluck signal power can carry information about male weight and size.

Stereotyped Sequencing of Signal Components

Males intermittently produced all three signal types throughout all phases of male-female interactions (Figure 2.3A). Each signal type was produced either individually or simultaneously with other signal types. For example, males sometimes produced abdominal tremulation signals and web plucks simultaneously, or produced web plucks and bounce signals. In the distal phase, the occurrence of different signals lacked any detectable temporal organization (Figure 2.3Bi) and thus we labeled this phase ‘unstructured signaling’ (Figure 2.3Bi). The temporally unstructured nature of male signaling, along with simultaneous production of signal types, results in the apparent jitteriness of male movements during courtship, which was previously described as ‘jerky movements’ (Anava and Lubin, 1993; Kaston, 1970; Ross and Smith, 1979).

However, during the proximal phase, as males approached females, males would intermittently transition between unstructured signal displays to more stereotyped ‘structured signaling’ displays (Figure 2.3Bii). During structured signaling, males displayed individual signal components in a highly stereotyped temporal sequence (Figure 2.3Bii). Structured signal sequences begin with a brief pause where males stop all movements. Males then begin to display abdominal tremulation signal. This abdominal tremulation signal is then followed by an intense bounce signal, which is then followed by a set of web pluck signals (Figure 2.3Bii). We define one structured signal as this stereotyped sequence of three component signals, delivered in order, and preceded and ended by a period without vibration. All males produce this specific sequence of signals intermittently throughout the proximal phase (Figure 2.3Bii). During copulation, male vibrations are highly structured as they only produced structured signals repeatedly (Figure 2.3A).

Figure 2.2: Linear discriminant function plot of the signaling space for each vibratory signal component of L. hesperus male: Abdominal tremulation (blue squares); bounce (red triangles); web plucks (green circles); group centroid (black asterisks; P < 0.001).

Table 2.2: Discriminant function analysis of vibration signal components produced by male Latrodectus hesperus during courtship (actual rows by predicted columns) based on principal component scores incorporating signal duration, peak power, peak amplitude, RMS amplitude, and peak frequency.

Abdominal Web Bounce Vibration Plucks Abdominal Vibration 16 0 0 Bounce 1 11 4 Web Plucks 1 3 12

Table 2.3: Results from Friedman test (2) and subsequent Wilcoxon Pair-wise comparisons (Z) male L. hesperus vibration signal parameters between abdominal tremulation (Ab), bounce (Bo), and web pluck (Pl) signals. Sample size (N), degrees of freedom (df), and significance (P) are indicated.

Friedman Test Signal Parameter Comparison N df 2 P Duration Ab x Bo x Pl 16 2 26.95 < 0.001 Peak Frequency Ab x Bo x Pl 16 2 23.63 < 0.001 Peak Power Ab x Bo x Pl 16 2 28.50 < 0.001 Amplitude Ab x Bo x Pl 16 2 30.13 < 0.001 RMS Amplitude Ab x Bo x Pl 16 2 30.13 < 0.001 Wilcoxon Pair-wise Comparison Signal Parameter Comparison N Z P Ab x Bo 16 -3.52 < 0.001 Duration Ab x Pl 16 -3.52 < 0.001 Bo x Pl 16 -2.74 0.006 Ab x Bo 16 -2.84 0.004 Peak Frequency Ab x Pl 16 -3.52 < 0.001 Bo x Pl 16 -3.47 0.001 Ab x Bo 16 -3.52 < 0.001 Peak Power Ab x Pl 16 -3.52 < 0.001 Bo x Pl 16 -3.21 0.001 Ab x Bo 16 -3.52 < 0.001 Amplitude Ab x Pl 16 -3.52 < 0.001 Bo x Pl 16 -3.41 0.001 Ab x Bo 16 -3.52 < 0.001 RMS Amplitude Ab x Pl 16 -3.52 < 0.001 Bo x Pl 16 -3.46 0.001

Table 2.4: Results from Wilcoxon Pair-wise comparisons (Z) of duration and peak frequencies between male L. hesperus courtship and copulation vibrations. Sample size (N), and significance (P) are indicated.

Signal Signal Type Comparison N Z P Parameter Abdominal Tremulation Courtship x Copulation 11 -2.85 0.004 Duration Bounce Courtship x Copulation 11 -2.67 0.008 Web Pluck Courtship x Copulation 12 -0.31 0.754 Abdominal Tremulation Courtship x Copulation 11 -0.45 0.657 Peak Bounce Courtship x Copulation 11 -0.80 0.424 Frequency Web Pluck Courtship x Copulation 12 -0.31 0.754

Figure 2.3: Male L. hesperus structured signaling during male-female interactions. (A) Oscillogram of web-borne vibrations recorded during a male-female interaction, with red stars indicating occurrence of male structured vibratory signaling. (B) Expanded view of five minute oscillogram sections marked in A during (i) distal phase (black box) and (ii) proximal phase (orange box). All structured signaling sequences (black dots) began with a transient pause followed by a characteristically stereotyped ordering of different signal types: males first displayed abdominal tremulation signals (green line), which was followed by a bounce signal (red star), and then a set of web plucks (orange diamonds). Males intermittently transitioned between random signal displays to stereotyped structured signaling. (C) Male signaling during distal phase lacked any noticeable temporal organization; males rarely displayed structured signaling. However, during proximal phase and copulation, males displayed structured signals significantly more (higher rates) (P < 0.005).

We counted the number of times males displayed structured signals within each interaction phase (Figure 2.3A). We then divided this number by the phase duration to get structured signaling rates for each phase (Figure 2.3C). We compared mean structured signaling rates between the distal (0.15 ± 0.18 signals/min; N = 15), proximal (0.84 ± 0.45 signals/min; N = 18), and copulation (1.84 ± 0.69 signals/min; N = 15) phase. A Kruskal-Wallis test showed that structured 2 signaling rates were significantly different among the three phases ( 2 = 33.83, P < 0.001) (Figure 2.3C). A Mann-Whitney U pair-wise test showed that males produced structured signals at significantly higher rates during in the proximal phase compared to distal phase (U = 16.0, Z = −4.30, P < 0.001), and significantly higher during copulation compared to distal phase (U = 0.0, Z = −4.67, P < 0.001) and proximal phase (U = 34.0, Z = −3.65, P < 0.001) (Figure 2.3C). Taken together, L. hesperus males intermittently organize individual signal types/components into stereotyped temporal sequences, which they display significantly more often in the later phases of male-female interactions.

2.4.2 Signal Production Mechanism

Abdominal Tremulation

Abdominal tremulation signals are generated by rapid dorso-ventral oscillation of the abdomen (Figure 2.4). Oscillograms of abdominal tremulation signals and male abdomen movement velocities show a temporal match between the abdominal movement and the web-vibration (Figure 2.4i-ii). However, since only the legs are in contact with the web, abdominal vibrations are transferred to the web via the individual legs attached. Abdominal vibrations are first transmitted through the cephalothorax, then to the legs, and then onto the web (Figure 2.4i-v). This consequently results in males’ whole body to oscillate on the web.

We compared maximum velocities of different body parts to examine source of signal generation (Figure S2.2). A Friedman test showed significant differences in mean maximum velocities 2 between body segments (N = 8,  3 = 16.2, P = 0.001) (Figure S2.2). Pair-wise Wilcoxon signed ranks post-hoc tests showed that male abdominal movements had higher maximum velocities (25.35 ± 7.95 mm/s) compared to cephalothorax (10.42 ± 2.96 mm/s; Z = -2.52, P = 0.012), femur-patella (F-Pt) joint dorso-ventral (9.53 ± 3.04 mm/s; Z = -2.52, P = 0.012), and F-Pt joint lateral movement (8.47 ± 3.87 mm/s; Z = -2.52, P = 0.012) movements (Figure S2.2). There

62 63 were no differences in the maximum velocities between cephalothorax and F-Pt joint movements (P > 0.05) (Figure S2.2). Our results show that the abdomen is the main source of the tremulation signal.

We further corroborate this with our abdominal immobilization experiments. Immobilizing male abdomen only attenuated abdominal tremulation signal power (Wilcoxon pair-wise test: Z = - 2.94, N = 11, P = 0.003; partial eta2 = 0.78; Cohen’s d = 3.44), but had no effect on signal frequency (Z = -1.07, N = 11, P = 0.286) (Figure S2.3).

Bounce

Bounce signals are high intensity percussive signals generated by males striking the web or female body with high velocities (Figure 2.5). The high velocities of the cephalothorax and abdomen are generated by initial flexion of male legs (Figure 2.5ii). This initial leg flexion propels male cephalothorax and the abdomen ventrally, causing them to strike the web or female abdomen (when males are near females) (Figure 2.5ii-v). After the initial web (or female) strike, males would transiently oscillate and then either start walking or producing web pluck signals. During structured signaling, males transitioned from relatively low velocity abdominal tremulation signals to rapid, high velocity bounce signals.

A Friedman test comparing maximum velocities of body segments during bounce signaling 2 showed no significant differences (N = 9,  3 = 5.40, P = 0.145) between abdomen (176.73 ± 45.92), cephalothorax (145.03 ± 44.45), femur-patella joint dorso-ventral (160.82 ± 43.48), and lateral (187.20 ± 52.31) movements (Figure S2.2). Abdominal immobilization experiments showed that immobilizing male abdomen had no significant effect on bounce signal power (Wilcoxon pair-wise test: Z = -0.889, P = 0.374) or peak frequency (Z = -0.756, N = 11, P = 0.449) of bounce signals (Figure S2.3). Taken together, male bounce signals are not driven by abdominal movements, but rather are generated by leg flexion and subsequent impact of males’ whole body against the web or female abdomen.

We also compared mean maximum velocities between abdominal tremulation and bounce signals across body segments. A Mann-Whitney U comparison showed that production of bounce signals involved significantly higher velocities of male abdomen (U = 0.0, Z = -7.05, P < 0.001), cephalothorax (U = 0.0, Z = -7.05, P < 0.001), F-Pt joint dorso-ventral (U = 0.0, Z = -

7.05, P < 0.001), and F-Pt joint lateral (U = 0.0, Z = -7.05, P < 0.001) movements (Figure S2.2). Bounce signal production involved overall significantly higher velocity body movements (Figure S2.2).

Web Plucks

Web pluck signals are solely produced by males pulling and releasing silk threads (Figure 2.6). Male abdomen and cephalothorax played no role in the production of these signals (Figure 6iv- v). Web pluck signals can be produced by a single leg plucking silk threads or two legs plucking different silk threads simultaneously (Figure S2.4).

2.4.2 Signal Transmission

Abdominal Tremulation

Power spectral density and transfer functions were calculated comparing velocities of individual male body segments (abdomen, cephalothorax, and femur-patella (F-Pt) joint) and abdominal tremulation signal vibrations recorded near the female. Abdominal tremulation signals are highly attenuated as they are transmitted from males to females (Figure 2.7A & B). Although lower frequencies (<40 Hz) were attenuated slightly more than higher frequencies, overall female webs passed all frequencies produced by males during abdominal tremulation signaling approximately equally (Figure 7B). Coherence calculations showed a fairly linear relationship between body segment vibrations and abdominal tremulation signals, suggesting that much of abdominal tremulation signal frequencies can be predicted by body movement vibrations (Figure 2.7C). However, higher frequency components of the signal (>70 Hz) were not predicted by male movement vibrations, suggesting other factors involved in shaping some of the frequency characteristics of abdominal tremulation signals (Figure 2.7B & C).

Bounce

Power spectral and transfer function analysis of bounce signals showed that the low-frequency movements of the bounce generated an attenuated broad-band signal (Figure 2.8A & B). The webs passed all frequencies of the bounce signals equally. Coherence calculations showed no relationship between bounce signal frequencies and male movement frequencies (Figure 2.8C). Bounce signals are produced by relatively low-frequency body movement leading to a percussive impulse, whereas the bounce signals themselves contain broad range of frequencies.

Figure 2.4: Abdominal tremulation signal production mechanism. Left panels show (i) vibration amplitude from synchronized laser vibrometry recording, (ii – iv) dorso-ventral movement velocities of male abdomen, cephalothorax and a femur-patella joint, respectively, and (v) lateral movement velocities of the same femur-patella joint in iv. All left panels i – v have the same time scale. Right panels illustrate (ii – v) the corresponding body segment movement. Abdominal tremulation signals are generated by dorso-ventral oscillations of male abdomen (right panel i [1 and 2]), which causes similar movements in the cephalothorax and femur-patella joints, but at lower amplitudes.

Figure 2.5: Male bounce signal production mechanism. Left panels illustrate (i) the body segments and movement direction used for analysis, and (ii – v) sequence of movements involved in generating bounce signals. Bounce signals are impulse-like percussive signals initiated by contraction of leg joints (left panel ii [movement 1]). This results in the female cephalothorax and abdomen to either strike the web or female abdomen at high velocities (left panel iii [movement 2]). Male cephalothorax and abdomen striking the web produces the bounce signal [right panel i]. Males will transiently oscillate on the web due to the impact (left panel iv [movement 3] and v [Movement 4]). Right panels show (i) vibration amplitude from synchronized laser vibrometry recording, (ii – iv) dorso-ventral movement velocities of male abdomen, cephalothorax and a femur-patella joint, respectively, and (v) lateral movement velocities of the same femur-patella joint in iv. All right panels i – v have the same time scale. Red vertical hashed line indicates the onset of femur-patella joint contraction (right panel ii and iii); which precedes bounce signals (right panel i).

Figure 2.6: Male web pluck signal production mechanism (also see Supplementary Figure S2.4). Left panels show (i) vibration amplitude from synchronized laser vibrometry recording, (ii – iv) dorso-ventral movement velocities of male abdomen, cephalothorax and a femur-patella joint, respectively, and (v) lateral movement velocities of the same femur-patella joint in iv. All left panels i – v have the same time scale. Right panels illustrate (i) the body segments and movement direction used for analysis, and (ii) sequence of movements involved in generating web plucks signals. Plucks are percussive-like signals initiated by males pulling and releasing individual silk threads (right panel ii [movement 1]). Red vertical hashed line indicates the onset of leg contraction (right and left panel ii). Production of individual web pluck signals corresponds to movements of different legs.

Figure 2.7: Comparing spectral differences between a male abdominal tremulation signal velocities and dorso-ventral movement velocities of the (i) abdomen, (ii) cephalothorax, and (iii) femur-patella joint. (A) Shows power spectral densities of male signal vibrations (blue lines) and body vibrations (black likes). Means (thick lines) and individuals (thin lines) are shown. (B) Transfer functions of male abdominal tremulation signals relative to individual body segment vibrations. Means (thick lines) and individuals (thin lines) are shown. All dB are relative to input body segment movement velocities. (C) Coherence of transfer functions calculated in B, between male signals and body vibrations. Means (thick lines) and individuals (thin lines) are shown.

Figure 2.8: Comparing spectral differences between a male bounce signal velocities and dorso- ventral movement velocities of the (i) abdomen, (ii) cephalothorax, and (iii) femur-patella joint. (A) Shows power spectral densities of male signal vibrations (blue lines) and body vibrations (black likes). Means (thick lines) and individuals (thin lines) are shown. (B) Transfer functions of male abdominal tremulation signals relative to individual body segment vibrations. Means (thick lines) and individuals (thin lines) are shown. All dB are relative to input body segment movement velocities. (C) Coherence of transfer functions calculated in B, between male signals and body vibrations. Means (thick lines) and individuals (thin lines) are shown.

Figure 2.9: Comparing spectral differences between a male web pluck signal velocities and dorso-ventral movement velocities of (i) abdomen and (ii) cephalothorax, and both dorso-ventral and lateral movement velocities of femur-patella joint (iii – iv, respectively). (A) Shows power spectral densities of male signal vibrations (blue lines) and body vibrations (black likes). Means (thick lines) and individuals (thin lines) are shown. (B) Transfer functions of male abdominal tremulation signals relative to individual body segment vibrations. Means (thick lines) and individuals (thin lines) are shown. All dB are relative to input body segment movement velocities. (C) Coherence of transfer functions calculated in B, between male signals and body vibrations. Means (thick lines) and individuals (thin lines) are shown.

Web Plucks

Similar to bounce signals, the low-frequency movement during web plucks generated broad-band signals (Figure 2.9A & B). The webs passed all frequencies approximately equally. Coherence calculations showed no relationship between web pluck signal frequencies and male movement frequencies (Figure 2.9C).

2.5 Discussion

In the present study, we characterized web-borne vibratory signals of the western black widow spider, Latrodectus hesperus, and examined signal structure, signal production mechanisms, and signal transmission efficacy. Our results show that, during courtship and copulation, L. hesperus males produced three characteristically distinct types of web-borne vibration signals (abdominal tremulation, bounce, and web pluck signals). These signals may convey information about males, as larger males produced bounce and web pluck signals with higher power. We also show that, during courtship, L. hesperus males transitioned between unstructured signaling (disorganized/haphazard display of signal component types with no noticeable temporal patterning), to intermittent bouts of structured signaling, where males grouped each signal type into a characteristically stereotyped temporal sequence. Males displayed these structured multicomponent signals more frequently during later phases of male-female interactions as males approach females, suggesting that these structured signals may play an important role in male mating success. Results from our high-speed video analyses show that the production of each signal type corresponded to movements of different body segments; and thus, different signaling mechanisms. Our transfer function analysis between male body vibration and corresponding vibratory signals showed a linear relationship between body segment vibration frequencies and abdominal tremulation signals. Female cobwebs passed all frequencies produced by males during abdominal tremulation signaling. However, we found no relationships between male body movement frequencies and bounce or web pluck signal frequencies; which further corroborates that these signals are produced via different mechanisms. Overall, our study demonstrates that: (a) vibratory communication in L. hesperus comprises multiple signals, some of which can carry information about male size; (b) males intermittently organize individual signal components into stereotyped sequence; (c) males generate each signal type using independent production mechanisms; and (d) female webs transmit male signals with high efficacy. To our knowledge,

71 72 our study is the first to show that, in L. hesperus, male web-borne vibratory signals can comprise multiple components. Our study also demonstrates emergent forms of complex signal structure in this species.

Our finding that L. hesperus males produce three different vibration signal types, during courtship and copulation, varying in both spectral and temporal characteristics, provides the first empirical evidence that black widow spiders communicate using multiple signals. Communication using multiple signals is widespread across many cursorial spider groups including: jumping spider (Elias et al., 2012, 2003; Girard et al., 2011), wolf spiders (Hebets et al., 2013; Rosenthal et al., 2019; Rundus et al., 2010), and wandering spiders (Schmitt et al., 1994). In contrast, although many web-dwelling spiders are presumed to generate multiple web- borne vibratory signals during courtship, to date, most studies have only provided qualitative descriptions, generally based on unaided visual observations (Barrantes, 2008; Coyle and O’Shields, 1990; Maklakov et al., 2003; Singer et al., 2000; Suter and Renkes, 1984). These studies infer, rather than measure, the production and transmission of vibratory signals. For example, in the cob-web spider Argyrodes antipodiana, males were observed to display 32 different vibratory behaviours during courtship (Whitehouse and Jackson, 1994). But it is not known whether (or how) these signals differ, nor the functional significance of these signals. Recently, Wignall and Herberstein (2013a) characterized four different courtship vibratory signals of male orb-web spider Argiope keyserlingi (‘shuddering’, ‘abdominal wagging’, ‘plucks’, and ‘bounce’). Although the signal parameters were not correlated with male body condition (Wignall and Herberstein, 2013a), male shudder vibrations were shown to play an important role in inhibiting female aggressive/predatory response (Wignall and Herberstein, 2013b). In black widows (L. hesperus), Vibert et al. (2014) recently characterized the abdominal tremulation signals of males as they entered female webs. The courtship abdominal tremulation signals characterized in our study are comparable to those previously described (Vibert et al., 2014). However, in this study, we further show that male abdominal tremulation signals are distinct from bounce and web-pluck signals, which were not previously characterized. Currently, there is a dearth of research and knowledge on the diversity of web-vibration signals in web- dwelling spiders, and it may not be uncommon that many web-dwelling spiders communicate using multiple distinct signals types.

Our results showing that larger males produce bounce and web plucks with higher intensities demonstrate that these signals can carry information about male size which could have functional implications in inter- or intra-sexual interactions. In widow spiders, larger males have a mating advantage during direct competition (Kasumovic and Andrade, 2009; Stoltz et al., 2009, 2008). For example, in the Australian redback spider, L. hasselti, females have been shown to distinguish and prematurely cannibalize smaller males who minimize courtship investment when paired with larger males (Stoltz et al., 2008). It is possible that females may use vibration signals to discriminate between mates. Interestingly, we found no relationship between male abdominal tremulation signals and male size, suggesting that abdominal tremulation signals may serve a different function or convey different information (e.g. species/mate identity). Previously, Vibert et al. (2014) demonstrated that the low-amplitude abdominal tremulation signals of male L. hesperus inhibited female predatory response during playback experiments. The abdominal tremulation signals of black widow males are similar in characteristics to male shudder signals seen in the orb-web spider Argiope keyserlingi; and male shudder signals were also shown to suppress female predatory response (Herberstein et al., 2014; Wignall and Herberstein, 2013a, 2013b). In contrast, however, shudder rates of male Argiope radon, were shown to be highly repeatable within males and positively correlated with male body condition (Wignall et al., 2014). The evolution of complex multi-component signals can be driven by selection for signal content, where different signal components may convey different information to receivers (‘multiple messages’ hypothesis) or act as back-up signals and convey similar information (‘redundant’ message hypothesis) (Hebets and Papaj, 2005). Our results suggest that, in L. hesperus, male bounce and web pluck signals may convey similar/redundant information about mate quality to females, while abdominal tremulation signals could provide different information about males (see Vibert et al., 2014). However, whether females can distinguish and differentiate between individual signal components is not known. In most spiders, males engage in lengthy and energetically costly courtship, and courtship effort (i.e. signaling rate), rather than properties of individual signal components, has been shown to be important for male mating success (Gibson and Uetz, 2008; Kotiaho et al., 1996; Shamble et al., 2009; Sivalinghem et al., 2010). Given that males in our trials were allowed to move freely during courtship, the distance between males and our point of vibration measurements was not standardized. This limits our interpretations of any potential information content in male signal amplitudes. Therefore, further work is needed to elucidate the information content in male signals.

In addition, we also found male abdominal tremulation and bounce signals vary (i.e. in duration) between courtship and copulation. These copulatory vibrations have not been previously described, and at present, the importance of vibratory signaling during copulation is not known for widow spiders. In L. hesperus, both males and female can mate multiple times, and rates of sexual cannibalism are variable in this species, and may depend on food availability. It is possible that copulatory vibrations may play a role during cryptic female choice; for example, rejection of subsequent rival males may depend on signal quality (Eberhard, 1994; Huber and Eberhard, 1997; Sivalinghem et al., 2010). Future studies should examine female re-mating rates as a function of male vibratory signaling. Copulatory signaling may also function in preventing premature interruption of copulation and/or inhibiting aggression (Garcia Diaz et al., 2015). Females of most web-dwelling spiders are aggressive predators, and in some species, are much larger than males. Males courting on female webs, and during copulation, risk the potential of premature cannibalism. Therefore, vibratory signaling may be crucial during both courtship and during copulation. In L. hesperus, males sequentially cycled through all three signal components during copulation. It is possible that the individual components of these multi-component displays may serve different functions during copulation. The presence of copulatory signaling in other widow species is not currently known, and future studies should examine copulatory vibration signaling in congeners (e.g. L. hasselti), where sexual cannibalism is common.

A key finding in our study is that L. hesperus males organize individual signal components into a stereotyped temporal sequence. These composite multi-component structured signal displays begin with momentary pause, followed by abdominal tremulation, then a bounce signal, and then a set of web plucks. Males produced these structured signals, comprising all signal components, more often during later proximal phase of male-female interaction, as males repeatedly approach females; suggesting that structured signals may be important for male mating success. Previously, L. hesperus male signals were thought to have no noticeable temporal patterning (Vibert et al., 2014); and males were thought to display individual signal elements haphazardly during courtship. Our results demonstrate nascent level of signal complexity in a widow spider, which, to our knowledge, has not been described for any web-dwelling spiders. Non-random sequencing of individual signal components is best-studied in birdsong (Catchpole and Slater, 2008). In contrast, arthropods are considered to have relatively simple signals with fixed patterns – e.g. calling songs of crickets (Desutter-Grandcolas and Robillard, 2011; Robillard and

Desutter-Grandcolas, 2011). There are few empirical examples, among arthropods, of species arranging signal components into meaningful sequences (Elias et al., 2012, 2006c). To date, the best example is seen in a jumping spider (Habronattus coecatus) where males vary signal structures throughout courtship by adding and removing different vibration elements (Elias et al., 2012). These distinct groupings of different signal components are thought to play a functional role during courtship, and male Habronattus coecatus are considered to have some of the most varied and complex signals among arthropods (Elias et al., 2012).

The structured signals of L. hesperus males, described in this study, are less complex compared to many jumping spiders, and only displayed intermittently, but provide the first empirical evidence of signal complexity in a web-dwelling spider. The potential for complex signaling had been alluded to in previous descriptions of male courtship behaviours in some web-dwelling spiders. For example, Singer et al. (Singer et al., 2000) described stereotyped sequences of male vibratory behaviours during courtship in the funnel-web spider, Agelenopsis aperta, where males displayed abdominal vibration followed by web flex behaviour, and then rest. Successful males were less likely to deviate from these stereotyped behavioural transitions (Singer et al., 2000). Likewise, in Tengella radiata, males signals comprise sequences of one or two abdominal vibrations followed by bouts of body rocking behaviour (Barrantes, 2008). It is possible that complex signaling in other web-dwelling spiders are more common than previously appreciated, and given the lack of empirical studies, these may have been previously overlooked. Web- dwelling spiders may provide ideal models for researching evolution of signal complexity.

Our results suggest that the three signal types correspond to different movements of male body segments, and thus three different signal production mechanisms: (i) abdominal tremulation signals are solely produced by the rapid oscillations of the abdomen; (ii) bounce signals are triggered by initial leg contractions, which propels the cephalothorax and abdomen towards the web at high velocities; and (iii) web plucks are produced by the rapid pull-and-release of silk threads. These signals can be further classified as either tremulatory signals (i.e. abdominal tremulation signal) or percussive signals (i.e. bounce and web pluck signals). Tremulation signals, in particular abdominal tremulation signals, are widespread are across many species that communicate using substrate-borne vibrations (Cocroft and Rodríguez, 2005; Virant-Doberlet and Cokl, 2004). These signals are generally narrowband and are best suited for communication on substrates that allow for effective transmission (e.g. plant stem and leaves and spider webs)

(Elias and Mason, 2011). Thus, they are commonly seen in organisms that specialize on particular substrates (Elias and Mason, 2011).

Unlike abdominal tremulation signals, where some of the frequency profile of the signals matches the abdominal oscillation frequencies, both bounce and web pluck signal frequencies likely correspond to the natural resonant frequencies of the web elicited by the impulsive force of male cephalothorax or recoiled silk thread, respectively. Percussive signals are versatile signals, in that, they are suited for communication on a variety of substrates, and the timing and intensity of these signals can convey information about the signaler (Elias et al., 2006c, 2004; Elias and Mason, 2011; Schüch and Barth, 1985). In web-dwelling spiders, especially cobweb spiders, percussive signals can be particularly useful for communicating through the dense network of interconnected silks threads. Further research examining the transmission of different signal types are needed to further elucidate how different vibrations are transmitted through spider webs, and how this relates to signal function.

Western black widow males constantly transitioned between these signaling mechanisms throughout courtship. During structured signaling, males coordinate and perform all three signaling mechanisms sequentially. Lengthy multicomponent signaling in spiders likely requires coordination of intense neuro-muscular processes and/or changes in internal hydrostatic pressure (responsible for leg extension in spiders) (Parry and Brown, 1959; Stewart and Martin, 1974). This can require large energetic costs. For example, Cady et al. (2011) demonstrated that in wolf spiders, species with complex signal displays (Schizocosa ocreata) expend relatively more energy during courtship signaling compared sister species (S. rovneri) that produce simple repetitive signals. In widow spiders (Latrodectus hasselti), the metabolic rate associated with vibratory signaling was shown to be four time more than resting metabolic rate (De Luca et al., 2015). The underlying mechanisms driving these distinct body segment vibrations (e.g. muscle- ligament force; hemolymph/hydrostatic pressure) are not known (Blickhan and Barth, 1985). In the wandering spider, Cupiennius getazi, the dorsal and ventral abdominal movements during tremulation corresponded to two different pedicel muscle contractions (Dierkes and Barth, 1995).

Of special interest is the bounce signal production mechanism, as it is currently unclear how leg flexion causes the subsequent whole-body to strike the web. The initial contraction of male legs

77 appears to act as a release mechanism, releasing stored potential energy into kinetic, and it is possible that the vigor in which males display these signals can convey important information to females and/or rivals (Byers et al., 2010). Spider legs have only flexor muscles (Blickhan and Barth, 1985), and leg-joints are extended by hemolymph hydraulic pressure, generated from the cephalothorax (Parry and Brown, 1959; Stewart and Martin, 1974). Therefore, generating bounce signals may involve coordination of both processes. Further research is needed to examine the finer details of signal production mechanisms in widow spiders.

Our transfer function analysis shows for the first time, to our knowledge, that female cobwebs transmit male signals with high efficacy (also see Vibert et al., 2016). Unlike highly structured orb webs, cobwebs are cross connected with several points of intersecting silk connections, which can be variable among females and contexts (DiRienzo and Aonuma, 2018; DiRienzo and Montiglio, 2016a, 2016b); and it might be assumed that this structure would lead to poor signal transmission. We show that female cobwebs transmit male abdominal tremulation signals with high efficacy. The high coherence values (Figure 2.7) suggest that the oscillation frequencies of male abdomen, cephalothorax and femur-patella joints predicted frequencies of male abdominal tremulation signals; thus there is a linear relationship between what the male is inputting onto the web and the signals arriving at females. However, given that we did not examine how other mechanical and material properties of the cob-webs (e.g. silk tension) influence signal transmission, our interpretations of the transmission efficacy of female cob-webs may be incomplete.

Nevertheless, these results are consistent with previous vibration transmission studies in L. hesperus cob-webs that demonstrated that frequencies <50Hz were much less attenuated (Vibert et al., 2016). Vibration measurements in the present study were limited to only the transverse wave components of male signals, and it is not known whether male signals comprise other wave components (i.e. longitudinal vibrations). Female L. hesperus webs were shown to have similar transmission efficiencies for both transverse and longitudinal vibrations (Vibert et al., 2016). This is different from many orb-webs, where longitudinal vibrations were shown to attenuate much less and can provide directional information (Masters and Markl, 1981; Mortimer, 2019; Mortimer et al., 2019). Web-dwelling spiders are great models for studying the interplay between signals and signaling medium as most web-dwellers construct the medium (webs) through which they transmit and receive vibratory information; and therefore can have greater control over their

78 signaling environment to optimize for effective transmission (Masters and Markl, 1981; Mortimer, 2019; Mortimer et al., 2019, 2016). Currently, there is a dearth of studies examining vibration transmission across various web types (but see Naftilan, 1999; Vibert et al., 2016). It is possible that different web types are optimized for efficiently transmitting different vibration types, which can influence the vibratory signals and signaling behaviours of different species.

In conclusion, our study quantified: 1) web-borne vibratory signals of male western black widow spider; 2) male vibratory signaling mechanisms; and c) the transmission efficacy of female webs. Furthermore, we provide the first demonstration of nascent complex signaling in a web-dwelling spider. It is likely that many other web-dwelling spiders communicate using complex multicomponent signals, and more empirical studies are needed in this group of spiders. Among spiders, communication systems can range from species that use simple repetitive signals to some of the most complex communicators known among arthropods (Herberstein et al., 2014). Studying communication in web-dwelling spiders may provide valuable insights into the evolution of signal complexity, and the interplay between substrate and signals. Currently, the present study is one of only four other empirical works that have directly measured vibratory signals from webs (Singer et al., 2000; Suter and Renkes, 1984; Vibert et al., 2014; Wignall and Herberstein, 2013a), and there is clearly a need for further research.

2.6 Acknowledgement

We thank the Mason lab undergraduate assistants for helping with spider husbandry and conducting mating experiments. We thank Drs. Damian Elias, Kenneth Welch, and Maydianne Andrade for providing comments and suggests on this manuscript. We also thank members of the Andrade and Mason labs for helpful comments and feedback both during and after experiments, analyses, and writing of this manuscript. This research was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) Postgraduate Scholarship, and Ontario Graduate Scholarship to S.S, and NSERC Discovery Grant to A.C.M. The authors declare that they have no conflict of interest.

2.7 References

Aisenberg, A., 2009. Male performance and body size affect female re-mating occurrence in the orb-web spider Leucauge mariana (Araneae, Tetragnathidae). Ethology 115, 1127–1136.

Andrade, M.C., 2019. Sexual selection and social context: Web-building spiders as emerging models for adaptive plasticity. Advances in the Study of Behavior 51, 177–250.

Andrade, M.C.B., 2003. Risky mate search and male self-sacrifice in redback spiders. Behavioral Ecology 14, 531–538.

Andrade, M.C.B., 1996. Sexual selection for male sacrifice in the Australian redback spider. Science 271, 70–72.

Barrantes, G., 2008. Courtship behavior and copulation in Tengella radiata (Araneae, Tengellidae). The Journal of Arachnology 36, 606–608.

Barth, F.G., Bohnenberger, J., 1978. Lyriform slit sense organ: Thresholds and stimulus amplitude ranges in a multi-unit mechanoreceptor. Journal of Comparative Physiology A 125, 37–43.

Barth, F.G., Geethabali, 1982. Spider vibration receptors: Threshold curves of individual slits in the metatarsal lyriform organ. Journal of Comparative Physiology A 148, 175–185.

Barth, F.G., Schmitt, A., 1991. Species recognition and species isolation in wandering spiders (Cupiennius spp.; Ctenidae). Behavioral Ecology and Sociobiology 29, 333–339.

Baruffaldi, L., Andrade, M.C.B., 2015. Contact pheromones mediate male preference in black widow spiders: Avoidance of hungry sexual cannibals? Animal Behaviour 102, 25–32.

Baruffaldi, L., Siddiqui, H., Thambiappah, A., Andrade, M.C.B., 2019. Male responses suggest both evolutionary conservation and rapid change in chemical cues of female widow spiders. Animal Behaviour 157, 61–68.

Baurecht, D., Barth, F.G., 1992. Vibratory communication in spiders. Journal of Comparative Physiology A 171, 231–243.

Blackledge, T.A., Kuntner, M., Agnarsson, I., 2011. The form and function of spider orb webs, in: Advances in Insect Physiology. Elsevier, pp. 175–262.

79 80

Blackledge, T.A., Zevenbergen, J.M., 2007. Condition-dependent spider web architecture in the western black widow, Latrodectus hesperus. Animal Behaviour 73, 855–864.

Blickhan, R., Barth, F.G., 1985. Strains in the exoskeleton of spiders. Journal of Comparative Physiology A 157, 115–147.

Breene, R.G., Sweet, M.H., 1985. Evidence of insemination of multiple females by the male black widow spider, Latrodectus Mactans (Araneae, Theridiidae). The Journal of Arachnology 13, 331–335.

Byers, J., Hebets, E., Podos, J., 2010. Female mate choice based upon male motor performance. Animal Behaviour 79, 771–778.

Cady, A.B., Delaney, K.J., Uetz, G.W., 2011. Contrasting energetic costs of courtship signaling in two wolf spiders having divergent courtship behaviors. The Journal of Arachnology 39, 161–165.

Catchpole, C., Slater, P.J.B., 2008. Bird song: biological themes and variations. 2008. Cambridge [England].

Cocroft, R.B., Rodríguez, R.L., 2005. The behavioral ecology of insect vibrational communication. BioScience 55, 323.

Coyle, F.A., O’Shields, T.C., 1990. Courtship and mating behavior of Thelechoris Karschi (Araneae, Dipluridae), an African funnel web spider. The Journal of Arachnology 18, 281–296.

Desutter-Grandcolas, L., Robillard, T., 2011. Evolution of calling songs as multicomponent signals in crickets (Orthoptera: Grylloidea: Eneopterinae). Behaviour 148, 627–672.

DiRienzo, N., Aonuma, H., 2018. Plasticity in extended phenotype increases offspring defence despite individual variation in web structure and behaviour. Animal Behaviour 138, 9–17.

DiRienzo, N., Montiglio, P.-O., 2016a. The contribution of developmental experience vs. condition to life history, trait variation and individual differences. Journal of Animal Ecology 85, 915–926.

DiRienzo, N., Montiglio, P.-O., 2016b. Linking consistent individual differences in web structure and behavior in black widow spiders. Behavioral Ecology 27, 1424–1431.

Eberhard, W.G., 1994. Evidence for widespread courtship during copulation in 131 species of insects and spiders, and implications for cryptic female choice. Evolution 48, 711–733.

Eberhard, W.G., 1990. Function and phylogeny of spider webs. Annual Review of Ecology and Systematics 21, 341–372.

Elias, D.O., Andrade, M.C.B., Kasumovic, M.M., 2011. Dynamic population structure and the evolution of spider mating systems, in: Casas, J. (Ed.), Advances in Insect Physiology, Spider Physiology and Behaviour. Academic Press, pp. 65–114.

Elias, D.O., Botero, C.A., Andrade, M.C.B., Mason, A.C., Kasumovic, M.M., 2010a. High resource valuation fuels “desperado” fighting tactics in female jumping spiders. Behavioral Ecology 21, 868–875.

Elias, D.O., Hebets, E.A., Hoy, R.R., 2006a. Female preference for complex/novel signals in a spider. Behavioral Ecology 17, 765–771.

Elias, D.O., Kasumovic, M.M., Punzalan, D., Andrade, M.C.B., Mason, A.C., 2008. Assessment during aggressive contests between male jumping spiders. Animal Behaviour 76, 901–910.

Elias, D.O., Lee, N., Hebets, E.A., Mason, A.C., 2006c. Seismic signal production in a wolf spider: Parallel versus serial multi-component signals. Journal of Experimental Biology 209, 1074–1084.

Elias, D.O., Maddison, W.P., Peckmezian, C., Girard, M.B., Mason, A.C., 2012. Orchestrating the score: Complex multimodal courtship in the Habronattus coecatus group of Habronattus jumping spiders (Araneae: Salticidae). Biological Journal of the Linnean Society 105, 522–547.

Elias, D.O., Mason, A.C., Maddison, W.P., Hoy, R.R., 2003. Seismic signals in a courting male jumping spider (Araneae: Salticidae). Journal of Experimental Biology 206, 4029– 4039.

Elias, D.O., Sivalinghem, S., Mason, A.C., Andrade, M.C.B., Kasumovic, M.M., 2014. Mate- guarding courtship behaviour: Tactics in a changing world. Animal Behaviour 97, 25–33.

Forster, L.M., 1992. The stereotyped behavior of sexual cannibalism in Latrodectus hasselti Thorell (Araneae, Theridiidae), the Australian redback spider. Australian Journal of Zoology 40, 1–11.

Garb, J.E., González, A., Gillespie, R.G., 2004. The black widow spider genus Latrodectus (Araneae: Theridiidae): Phylogeny, biogeography, and invasion history. Molecular Phylogenetics and Evolution 31, 1127–1142.

Garcia Diaz, V., Aisenberg, A., Peretti, A.V., 2015. Communication during copulation in the sex-role reversed wolf spider Allocosa brasiliensis: Female shakes for soliciting new ejaculations? Behavioural Processes 116, 62–68.

Gibson, J.S., Uetz, G.W., 2008. Seismic communication and mate choice in wolf spiders: Components of male seismic signals and mating success. Animal Behaviour 75, 1253– 1262.

Girard, M.B., Elias, D.O., Kasumovic, M.M., 2015. Female preference for multi-modal courtship: Multiple signals are important for male mating success in peacock spiders. Proceedings of the Royal Society B: Biological Sciences 282, 20152222.

Girard, M.B., Kasumovic, M.M., Elias, D.O., 2011. Multi-modal courtship in the peacock spider, Maratus volans (O.P.-Cambridge, 1874). PLoS One 6.

Harari, A.R., Ziv, M., Lubin, Y., 2009. Conflict or cooperation in the courtship display of the white widow spider, Latrodectus pallidus. The Journal of Arachnology 37, 254–260.

Hebets, E.A., 2008. Seismic signal dominance in the multimodal courtship display of the wolf spider Schizocosa stridulans Stratton 1991. Behavioral Ecology 19, 1250–1257.

Hebets, E.A., 2005. Attention-altering signal interactions in the multimodal courtship display of the wolf spider Schizocosa uetzi. Behavioral Ecology 16, 75–82.

Hebets, E.A., Elias, D.O., Mason, A.C., Miller, G.L., Stratton, G.E., 2008. Substrate- dependent signalling success in the wolf spider, Schizocosa retrorsa. Animal Behaviour 75, 605–615.

Hebets, E.A., Papaj, D.R., 2005. Complex signal function: Developing a framework of testable hypotheses. Behavioral Ecology and Sociobiology 57, 197–214.

Herberstein, M.E., Wignall, A., 2011. Introduction: Spider biology. Spider Behaviour: Flexibility and Versatility 1–30.

Huber, B.A., Eberhard, W.G., 1997. Courtship, copulation, and genital mechanics in Physocyclus globosus (Araneae, Pholcidae). Canadian Journal of Zoology 75, 905–918.

Kaston, B.J., 1970. Comparative biology of American black widow spiders. Transactions of the San Diego Society of Natural History.

Kasumovic, M.M., Andrade, M.C.B., 2009. A change in competitive context reverses sexual selection on male size. Journal of Evolutionary Biology 22, 324–333.

Kasumovic, M.M., Andrade, M.C.B., 2006. Male development tracks rapidly shifting sexual versus natural selection pressures. Current Biology 16, R242–R243.

Kasumovic, M.M., Andrade, M.C.B., 2004. Discrimination of airborne pheromones by mate- searching male western black widow spiders (Latrodectus hesperus): Species- and population-specific responses. Canadian Journal of Zoology 82, 1027–1034.

Kasumovic, M.M., Elias, D.O., Punzalan, D., Mason, A.C., Andrade, M.C.B., 2009. Experience affects the outcome of agonistic contests without affecting the selective advantage of size. Animal Behaviour 77, 1533–1538.

Kasumovic, M.M., Elias, D.O., Sivalinghem, S., Mason, A.C., Andrade, M.C.B., 2010. Examination of prior contest experience and the retention of winner and loser effects. Behavioral Ecology 21, 404–409.

Kotiaho, J., Alatalo, R.V., Mappes, J., Parri, S., 1996. Sexual selection in a wolf spider: Male drumming activity, body size, and viability. Evolution 50, 1977–1981.

Landolfa, M.A., Barth, F.G., 1996. Vibrations in the orb web of the spider Nephila clavipes: Cues for discrimination and orientation. Journal of Comparative Physiology A 179, 493– 508.

MacLeod, E., 2013. New insights into the evolutionary maintenance of male mate choice behaviour using the western black widow spider, Latrodectus hesperus (Thesis).

MacLeod, E.C., Andrade, M.C.B., 2014. Strong, convergent male mate choice along two preference axes in field populations of black widow spiders. Animal Behaviour 89, 163– 169.

Maklakov, A.A., Bilde, T., Lubin, Y., 2003. Vibratory courtship in a web-building spider: Signalling quality or stimulating the female? Animal Behaviour 66, 623–630.

Masters, W.M., Markl, H., 1981. Vibration signal transmission in spider orb webs. Science 213, 363–365.

Mortimer, B., 2019. A spider’s vibration landscape: Adaptations to promote vibrational information transfer in orb webs. Integrative and Comparative Biology 59, 1636–1645.

Mortimer, B., Soler, A., Siviour, C.R., Zaera, R., Vollrath, F., 2016. Tuning the instrument: Sonic properties in the spider’s web. Journal of The Royal Society Interface 13, 20160341.

Naftilan, S.A., 1999. Transmission of vibrations in funnel and sheet spider webs. International Journal of Biological Macromolecules 24, 289–293.

Parri, S., Alatalo, R.V., Kotiaho, J., Mappes, J., 1997. Female choice for male drumming in the wolf spider Hygrolycosa rubrofasciata. Animal Behaviour 53, 305–312.

Parry, D.A., Brown, R.H.J., 1959. The Hydraulic Mechanism of the Spider Leg. Journal of Experimental Biology 36, 423–433.

Rivero, A., Alatalo, R.V., Kotiaho, J.S., Mappes, J., Parri, S., 2000. Acoustic signalling in a wolf spider: Can signal characteristics predict male quality? Animal Behaviour 60, 187– 194. https://doi.org/10.1006/anbe.2000.1452

Robillard, T., Desutter-Grandcolas, L., 2011. The complex stridulatory behavior of the cricket Eneoptera guyanensis Chopard (Orthoptera: Grylloidea: Eneopterinae). Journal of Insect Physiology 57, 694–703.

Rosenthal, M.F., Hebets, E.A., Kessler, B., McGinley, R., Elias, D.O., 2019. The effects of microhabitat specialization on mating communication in a wolf spider. Behavioral Ecology 30, 1398–1405.

Rundus, A.S., Santer, R.D., Hebets, E.A., 2010. Multimodal courtship efficacy of Schizocosa retrorsa wolf spiders: Implications of an additional signal modality. Behavioral Ecology 21, 701–707.

Schüch, W., Barth, F.G., 1985. Temporal patterns in the vibratory courtship signals of the wandering spider Cupiennius salei Keys. Behavioral Ecology and Sociobiology 16, 263– 271.

Scott, C., Vibert, S., Gries, G., 2012. Evidence that web reduction by western black widow males functions in sexual communication. The Canadian Entomologist 144, 672–678.

Segoli, M., Arieli, R., Sierwald, P., Harari, A.R., Lubin, Y., 2008. Sexual cannibalism in the brown widow spider (Latrodectus geometricus). Ethology 114, 279–286.

Shamble, P.S., Wilgers, D.J., Swoboda, K.A., Hebets, E.A., 2009. Courtship effort is a better predictor of mating success than ornamentation for male wolf spiders. Behavioral Ecology 20, 1242–1251.

Singer, F., Riechert, S., Xu, H., Morris, A., Becker, E., Hale, J., Noureddine, M., 2000. Analysis of courtship success in the funnel-web spider Agelenopsis aperta. Behaviour 137, 93–117.

Singer, F., Riechert, S.E., 1995. Mating system and mating success of the desert spider Agelenopsis aperta. Behavioral Ecology and Sociobiology 36, 313–322.

Sivalinghem, S., Kasumovic, M.M., Mason, A.C., Andrade, M.C.B., Elias, D.O., 2010. Vibratory communication in the jumping spider Phidippus clarus: Polyandry, male courtship signals, and mating success. Behavioral Ecology 21, 1308–1314.

Stewart, D.M., Martin, A.W., 1974. Blood pressure in the tarantula, Dugesiella hentzi. Journal of Comparative Physiology A 88, 141–172.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2009. Male courtship effort determines female response to competing rivals in redback spiders. Animal Behaviour 77, 79–85.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2008. Females reward courtship by competing males in a cannibalistic spider. Behavioral Ecology Sociobiology 62, 689–697.

Suter, R.B., Renkes, G., 1984. The courtship of Frontinella Pyramitela (Araneae, Linyphiidae): Patterns, vibrations and functions. The Journal of Arachnology 12, 37–54.

Uetz, G.W., Roberts, J.A., Clark, D.L., Gibson, J.S., Gordon, S.D., 2013. Multimodal signals increase active space of communication by wolf spiders in a complex litter environment. Behavioral Ecology and Sociobiology 67, 1471–1482.

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.

Vibert, S., Scott, C., Gries, G., 2014. A meal or a male: the ‘whispers’ of black widow males do not trigger a predatory response in females. Frontiers in Zoology 11, 4.

Virant-Doberlet, M., Cokl, A., 2004. Vibrational communication in insects. Neotropical Entomology 33, 121–134.

Whitehouse, M.E.A., Jackson, R.R., 1994. Intraspecific interactions of Argyrodes antipodiana, a kleptoparasitic spider from New Zealand. New Zealand Journal of Zoology 21, 253– 268.

Wignall, A.E., Herberstein, M.E., 2013a. The influence of vibratory courtship on female mating behaviour in orb-web spiders (Argiope keyserlingi, Karsch 1878). PLoS One 8.

Wignall, A.E., Herberstein, M.E., 2013b. Male courtship vibrations delay predatory behaviour in female spiders. Scientific Reports 3, 1–5.

Wignall, A.E., Kemp, D.J., Herberstein, M.E., 2014. Extreme short-term repeatability of male courtship performance in a tropical orb-web spider. Behavioral Ecology 25, 1083–1088.

Wilgers, D.J., Hebets, E.A., 2012. Seismic signaling is crucial for female mate choice in a multimodal signaling wolf spider. Ethology 118, 387–397.

2.8 Supplementary Materials

Figure S2.1: Custom-made mating arena used for conducting mating trials.

Figure S2.2: Mean maximum velocity of male body segment movements during (gray bars) abdominal tremulation and (black bars) bounce signal production. Male velocities of all measured body segments were significantly higher (p > 0.001) during bounce signal production than abdominal tremulation signaling.

Figure S2.3: Effects of immobilizing male abdomen on abdominal tremulation and bounce signals. (A) Difference in peak frequency of abdominal tremulation and bounce signals of males in control and immobilized (experimental) treatments. Immobilizing male abdomen had no effect on peak frequency of abdominal tremulation and bounce signals (Wilcoxon test P > 0.05). (B) Differences in peak intensities of abdominal tremulation and bounce signals of males in control and experimental treatments. Abdominal tremulation signals of males with immobilized abdomen had significantly lower intensities (Wilcoxon test P = 0.003; red asterisks). Immobilizing male abdomen had no effect on bounce signal intensities.

Figure S2.4: Dorso-ventral movement velocities of multiple legs during web pluck signal production. Left panels show (i) vibration amplitude from synchronized laser vibrometry recording of three individual plucks, and the dorso-ventral movement velocities of (ii) right leg 1 [red line], (iii) left leg 1 [blue line], and (iv) right leg 2 [brown line] femur-patella joints. All left panels i – iv have the same time scale. Right panels illustrate (i) the direction of movement used for analysis, and (ii – iv) corresponding leg movements involved in generating web plucks signals. Male web pluck signals can result from either a signal leg plucking a thread or multiple legs plucking different threads.

Chapter 3 Function of Structured Signaling in the Black Widow Spider, Latrodectus hesperus

Senthurran Sivalinghem and Andrew C. Mason

3.1 Abstract

Many animals communicate using complex signals comprised of multiple individual components. Although much research has focused on the function of individual signal components, few studies have examined how organisms organize signal components into distinct sequences (i.e., signal structure), and how this structural complexity relates to signal function. Male western black widow spiders (Latrodectus hesperus) intermittently transition from a “haphazard” production of multiple vibratory signal components to a temporally structured multicomponent signal display (organized sequence) during courtship. In this study, we examined the function of structured signaling during male-female interaction in L. hesperus. We ran mating trials using virgin L. hesperus males and females and used video recordings and laser vibrometry to quantify how often males produced structured signals (structured signaling rates). We examined whether structured signaling rates were predictive of male mating success and male mass. In a separate experimental study, we examined the effects of removing one of the signal components (i.e. abdominal tremulation) on male mating success. Our results showed that males that produced structured signals more frequently (higher structured signaling rates) were more likely to mate successfully and mated sooner. However, larger males had lower structured signaling rates overall, suggesting a trade-off between structured signaling and size/mass. We also showed that impairing the abdominal tremulation component had no effect mating success, but these ‘muted’ males took longer to mate. Our study demonstrates that the temporal arrangement of signal components plays a functional role during male-female interaction in L. hesperus.

3.2 Introduction

Animal communication signals can vary greatly in complexity, with some having simple repetitive signals to others organisms with many distinct individual components stimulating the same (“multicomponent signals”) or different sensory modalities (“multimodal signals”) in receivers (Hebets and Papaj, 2005; Higham and Hebets, 2013; Partan and Marler, 2005). For example, in brown-headed cowbirds, Molothurs ater, male courtship signals are comprised of both auditory and visual displays; and males can have up to eight different acoustic components and five different visual components in their signal (O’Loghlen and Rothstein, 2012, 2010). Understanding the origin and function of complex communication is the focus of current research (Candolin, 2003; Higham and Hebets, 2013, 2013; Miller and Bee, 2012; Partan, 2013; Partan and Marler, 2005; Ronald et al., 2017; Rowe, 1999; Seyfarth and Cheney, 2017). Complexity in signals can be associated with the number of individual components or elements that comprise the signal (i.e., repertoire size), such that, an organism with a larger repertoire of components is considered to have a more complex signal (Catchpole and Slater, 2008; Garamszegi et al., 2002; Hasselquist et al., 1996; Kipper et al., 2006; McComb and Semple, 2005). Much research has focused on the functions of individual components, and multicomponent signals have been shown to allow organisms to convey different information to different receivers (Hebets and Papaj, 2005; Johnstone, 1996; Ossip-Klein et al., 2013; Seyfarth and Cheney, 2017); communicate effectively in varying signaling environments and behavioural contexts (Elias and Mason, 2014, 2011; Rowe, 1999; Stoffer and Uetz, 2017; Taylor et al., 2011; Uetz et al., 2013; Uy and Safran, 2013); and enhance and influence receiver detection, discrimination, and preference (Byers and Kroodsma, 2009; Gridi-Papp et al., 2006; Hebets, 2005; Miller and Bee, 2012; Preininger et al., 2013; Rowe, 1999; Ruxton and Schaefer, 2011; Stange et al., 2017; Taylor and Ryan, 2013; Wilczynski et al., 1999). One aspect of signal complexity that has received much less attention is how individual components are ordered or combined into distinct sequences (i.e., signal structure), and how this structural/temporal complexity relates to signal function (Arnold and Zuberbühler, 2012; Kershenbaum, 2014).

Non-random sequencing of individual signal components (signal structure) is perhaps best studied in birdsong (Alger et al., 2016; Catchpole and Slater, 2008; Chatfield and Lemon, 1970; Große Ruse et al., 2016; Lemon and Chatfield, 1971; Sasahara et al., 2012; Weiss et al., 2014;

92 93

Zann, 1993), but also in a few non-avian taxa including, Chiropterans (Behr and von Helversen, 2004; Bohn et al., 2009; Kanwal et al., 1994), non-human Primates (Clarke et al., 2006; Fischer et al., 2017; Marler and Mitani, 1989; Seyfarth and Cheney, 2017), Anurans (Gridi-Papp et al., 2006; Stange et al., 2017; Wilczynski et al., 1999), and Cetaceans (McCowan et al., 1999; Payne and McVay, 1971; Suzuki et al., 2006). The specific arrangements of signal components have been shown to play a functional role in communication (Clucas et al., 2004; Hailman et al., 1987; Holland et al., 2000; Vallet and Kreutzer, 1995). For example, playback experiments in the Carolina chickadees, Poecile carolinensis, show that receivers pay attention to note ordering in songs, as they are unresponsive to songs with notes arranged in atypical orders; atypical song structures lose their communicative function (Clucas et al., 2004). Moreover, in some non- human primates with limited signal repertoire, combining signal components allows for communicating more information (Crockford and Boesch, 2003; Schamberg et al., 2016; Zuberbühler, 2002). Many of the studies, however, have focused on quantifying the ways in which temporal sequences are non-random. A few have shown that the temporal arrangement of signal components can provide meaningful information about the signaler (Beckers and Ten Cate, 2001; Catchpole and Rowell, 1993; Green, 1975; Hafner et al., 1979; Kipper and Kiefer, 2010; Payne and McVay, 1971; Petrusková et al., 2010; Schamberg et al., 2016), location of food sources (Freeberg and Lucas, 2002; Grüter and Farina, 2009), predator types (Manser, 2001; Manser et al., 2001; Seyfarth et al., 1980; Slobodchikoff et al., 1991; Templeton et al., 2005), and environmental, social, and behavioural changes (Clucas et al., 2004; Crockford and Boesch, 2003; Evans and Marler, 1994; Freeberg et al., 2012; Ouattara et al., 2009; Schneider and Lewis, 2004; Woolley and Doupe, 2008). Outside of these studies, the functional significance of structured signals has been little researched for most taxa. In the present study, we examine an arthropod system where signals comprise multiple components, but were historically not considered to be structured signals.

Bird and non-human primate calls may be highly complex, comprising many syllables and several motifs, and are useful for understanding the evolution of complex communication systems and language. In contrast, signals of most arthropods are considered to be simple, having a few components displayed in a fixed pattern. Despite this, however, some arthropod species have been shown to communicate using highly complex and structurally variable signal displays. Among arthropods, spiders (Araneae) have become ideal organisms for investigating complex

94 signal evolution (Elias et al., 2012; Elias and Mason, 2014, 2011; Herberstein et al., 2014; Uhl and Elias, 2011). Extensive research on several cursorial hunting spiders (i.e. jumping spiders [Araneae: Salticidae]; wolf spiders [Araneae: Lycosidae]; and wandering spiders [Araneae: Ctenidae]) has shown that courtship signals comprise multiple components, and species within and among genera can vary in repertoire size and the degree of temporal organization among individual components (signal structure) (Elias et al., 2012; Hebets et al., 2013; Herberstein et al., 2014; Uhl and Elias, 2011). For example, the jumping spider, Habronattus coecatus, has one of the most highly organized signals described to date (Elias et al., 2012). Male courtship displays involve an intricate combination of several visual and substrate-borne vibratory signal components that are tightly coordinated and displayed as distinct functional groupings (i.e., motifs) (Elias et al., 2012). In addition, males vary signal structure during courtship by gradually adding new components, as well as changing motifs, and signal display rates (Elias et al., 2012). In the wolf spiders, particularly the genus Schizocosa, male signal structures can range from relatively ‘loose’ display of signals to more stereotyped sequences (Hebets et al., 2013).

In contrast to highly organized signals of many cursorial spiders, the courtship signals of web- building spiders have been thought to be relatively simple, with minimal structure (Coyle and O’Shields, 1990; Eberhard and Huber, 1998; Herberstein et al., 2014; Lubin, 1986; Maklakov et al., 2003; Vibert et al., 2014). In a previous study, however, we showed that during the course of an extended courtship, male western black widow spiders (Latrodectus hesperus) intermittently transition from a “haphazard” production of multiple signal components to a temporally structured multicomponent signal display (organized sequence).

Western black widow spiders are web-dwelling spiders found throughout western North America, from southwestern Canada (i.e. Alberta and British Columbia) to California and Texas (U.S.A.) (Kasumovic and Andrade, 2004; MacLeod, 2013). Females are sit-and-wait predators that build dense cob-webs comprised of a complex irregular network of silk threads, which function for prey capture, and as a medium for sending and receiving vibratory information (Blackledge and Zevenbergen, 2007; Vibert et al., 2014). Females are sedentary, while adult males initiate mate-searching behaviour at sexual maturity (Baruffaldi and Andrade, 2015; MacLeod and Andrade, 2014; Scott et al., 2019). Upon arriving on female webs, males initiate courtship behaviour. The early (distal) phase of courtship occurs on the web, with males engaging in exploratory behaviours while cutting and adding silk strands onto the web (Scott et

95 al., 2012). In the later (proximal) phase, males approach females, make initial direct physical contact, and make repeated attempts to mount the female’s abdomen, and eventually copulate (Scott et al., 2012). We showed previously that males produce three distinct vibratory signals during courtship (abdominal tremulation, bounce and web plucks), and that as they move from distal to proximal, and finally copulatory courtship, the signals are more frequently organized in a stereotyped sequence to form a structured signal (Figure 3.1).

In this study, we test the hypothesis that the temporally structured vibratory signals of male L. hesperus are important for male mating success. Using laser Doppler vibrometry to measure male web-borne vibrations during male-female interactions, we first examine variations in structured vibratory signaling rates, and test whether signaling rates are associated with aspects of male mating success and quality (Experiment 1). In a separate study (experiment 2), we test the effects of removing one of the signal components (abdominal tremulation) on male mating success and latencies. An understanding of how signals components are organized and presented to intended receivers, and their functions, is of general interest as it provides insights into understanding signal evolution, underlying aspects of motor control, and the processing demands on receiver sensory systems (Todt and Hultsch, 1998).

3.3 Materials and Methods 3.3.1 Study Animal

Virgin western black widow (Latrodectus hesperus) males and females used in experiments were descendents of gravid females collected from Hastings Natural History Reserves (Carmel Valley, California, USA), and reared in temperature-controlled rooms at the University of Toronto Scarborough. Individual spiders were housed in separate plastic containers (length x width x height cm3 = 8.73 x 8.73 x 11.27; Amac Plastics), and kept at 22-24°C, with 14:10 light:dark cycle. Adult males were fed fruit flies, Drosophila melanogaster, twice a week; and adult females were fed one cricket, Acheta domesticus, once a week. Both males and females were given water twice a week. For all adult males and females, at least one week was given after final molt (into adults), before being used for experiments.

Male L. hesperus continuously produce vibration signals during both the distal and proximal phase of courtship, as well as during copulation (Scott et al., 2012). Vibration signals comprise

96 three distinct components each produced by a different signaling mechanism: (a) abdominal tremulation, (b) bounce, and (c) web plucks (Figure 3.1). Throughout courtship, males will produce (often simultaneously) these different vibration elements haphazardly, lacking any noticeable temporal organization (i.e., unstructured signaling), previously described as ‘jerky movements’(Anava and Lubin, 1993; Kaston, 1970; Ross and Smith, 1979). For example, males typically begin abdominal tremulation upon contact with female web, and continuously produce abdominal vibration throughout courtship; only stopping occasionally to rest or when interrupted by female movements. While producing abdominal tremulation, males can simultaneously produce either bounce or web pluck vibrations. However, during the proximal phase, as males approach females, and during copulation, males intermittently transition between unstructured signaling and ‘structured signaling’ (Figure 3.1). During structured signaling, males produce individual vibration components in a stereotyped sequence (abdominal tremulation, bounce, then web plucks) separated by pauses with no vibrations produced (Figure 3.1). Males produce these structured signals intermittently throughout the proximal phase of courtship and during copulation.

3.3.2 Behavioural Experiments

We conducted two separate mating experiments to test the importance of structured vibratory signaling. In the first experiment, we examined whether structured vibratory signaling rates were associated with male mating success and male mass. We conducted mating trials between adult virgin males and females, and recorded male courtship vibration signals. We compared structured vibratory signaling rates between males that copulated (successful) and those that did not (unsuccessful). Recordings from these trials were also used to characterize male courtship signals (Sivalinghem and Mason in prep). Trials were conducted in mating arenas in which females had built webs as outlined previously (Chapter 2). In brief, metal bolts where placed in the corners of a 25 x 17 x 7 cm plastic container to serve as web supports, a plastic lid (5 cm2) was attached to one bolt to provide a refuge for the female, and a shallow layer of soil provided a substrate. Females were left in the arena for 1 week to construct a web prior to trials. All trials lasted for 6 hours, and males that did not copulate during that time were categorized as ‘unsuccessful’. All trials were conducted in an anechoic chamber (Eckel audiometric rooms G- Series xht-batten).

Figure 3.1: Male (L. hesperus) structured vibration signaling during courtship. (A) Oscillogram (top panel) and spectrogram (bottom panel) of L. hesperus male structured signaling during proximal phase of male-female interaction. Black rectangles indicate bouts of structured signaling. (B) Expanded view of the Oscillogram (top panel) section marked in A (red box), of a structured signal bout, and corresponding spectrogram (bottom panel). Structured signals comprise a stereotyped sequential arrangement of male signal components. Following a momentary pause, males first produced abdominal tremulation signal (green), followed by a bounce signal (maroon), and then a set of web plucks (orange). Males intermittently produced structured signal bouts throughout courtship.

In the second experiment, we tested the effects of attenuating one of the signal components on mating success. We predicted that disruptions to the signal structure would affect mating success and/or mating latencies. Of the three types of vibration signals males produce – abdominal tremulation, bounce, and web plucks – abdominal tremulation signals were the only signal type that was feasible to experimentally manipulate, while still allowing males to behave naturally. This is because bounce and pluck signal components are produced using the legs, and it was therefore not possible to manipulate these without impeding males’ locomotion on the web (Chapter 2). Previously, we showed that abdominal tremulation signals could be attenuated by limiting movement of the abdomen relative to the cephalothorax (Chapter 2). Males were randomly assigned to either the experimental (‘muted’) or control (‘non-muted’) condition. Following procedures previously described (Elias et al., 2005, 2003), males in the muted condition had their cephalothorax and abdomen held together with hair-removal wax, placed dorsally. This restricted abdominal movements. Non-muted males only had a small amount of wax placed on their cephalothorax dorsum; this had no effect on abdominal movement. Mating trials in the second experiment were conducted on custom-made mating arenas which consisted of a wooden base (length x width x height cm3 = 12 x 12 x 1.5), with four metal carriage bolts (height = 4.5 inches) mounted at each corner. All other experimental procedures are same as described above for experiment one.

3.3.3 Recording Procedures

Mating trials were video recorded using a Sony HD video camera (HANDYCAM® NEX-VG10; Sony e-mount lens SEL18200; 44.1 kHz audio sampling rate). Simultaneously, we synchronized and recorded male web-borne vibrations using a laser Doppler vibrometer (LDV; Polytec PDV 100; 20 mm/s peak measurement range; 0.5 Hz - 22.0 kHz frequency range; 24 bits); which were stored onto an audio data recorder (Sound Devices 722; 16 bits; 48.0 kHz audio sampling rate). Vibration signals were recorded from a tiny reflective tape (~1mm2) placed on web-silk at the entrance of females’ refuge. All and recorded vibration signals from the LDV onto the audio track of the digital HD video camera.

98 99

3.3.4 Structured Signal Analysis and Mating Success

In the first experiment, we counted the number of times males produced ‘structured signals’ to calculated structured signaling rates. We compared structured signaling rates between successful and unsuccessful males. Structured signaling constituted a stereotyped, sequentially organized production of individual vibration components (Figure 3.1). The onset of each structured signaling transition followed a momentary pause of all movements. Males would then start by producing abdominal tremulation signals, which lasted on average 11.8 seconds, followed by a bounce signal, and then a set of web plucks signals (Figure 3.1). Structured signaling behaviours were visually conspicuous.

To determine signaling rates, we first divided each male-female interaction into three phases, a) “distal phase”, b) “proximal phase”, and c) “copulation phase”, and measured the duration of each phase (Scott et al., 2012). We considered the distal phase as the time when males started to produce vibration signals on the web to when they approached females and made initial contact. We considered the initial contact as the onset of proximal phase. For males that successfully mated with females, the duration of proximal phase was from the initial contact to when males inserted an intromittent organ (pedipalp) into the female’s genital tract (epigynum), which constituted the onset of copulation. For males that were unsuccessful, the duration of proximal phase was from the initial contact to end of the mating trial. For each male, we counted the number of transitions to structured signal displays in each phase and divided that by the duration of the phase to get structured signaling rates.

Previously, we showed that males produced structured signals more often during the proximal phase of courtship and during copulation; however, not all males produced structured signals during the distal phase. We therefore focused on proximal phase structured signaling in our analyses. We first examined changes in structured signaling rates within the proximal phase. To do this, we normalized proximal-phase time by dividing proximal-phase duration into quarters. We then counted the number of structured signals displayed in each quarter segment to calculate the structured signaling rates of each quarter. Next, we examined whether variations in proximal- phase structured signaling rates between males predicted mating latencies, and were predicted by male mass. All males were weighed using an electronic balance (Ohaus Explorer; accurate to ±0.01 mg) before mating trials.

100

In the second experiment, we examined effects of attenuating abdominal tremulation on male mating success. Specifically, we compared the proportion of successful males between ‘muted’ and ‘non-muted’ groups, as well as their latencies to copulate.

3.3.5 Statistical Analysis

All statistical analyses were performed using SPSS (version 16). All measured variables were normally distributed (Shapiro-Wilk test p > 0.05; Table 3.1). We specify below the particular statistical analyses used.

3.4 Results 3.4.1 Experiment 1

Variation in Structured Signaling Rates and Male Mating Success

We conducted mating trials between 19 pairs of virgin males and females, out of which, 15 (78.9 %) pairs successfully mated. All males, regardless of mating success, produced structured vibratory signals. Among the successful males, most produced structured signals in all three phases of male-female interaction (distal: 12/15 males; proximal: 15/15 males; and copulation: 15/15 males).

We first compared structured signaling rates between proximal-phase quarters. A one-way

ANOVA test showed no significant difference in structured signaling rates among quarters (F(3,

46) = 0.547, p = 0.65; Figure 3.2A). Signaling rates within proximal phase of male-female interaction was relatively consistent. We then compared proximal-phase structured signaling rates between males that successfully copulated with females (N = 15) and unsuccessful males (N = 4). Given that only a small number of males were unsuccessful, we used a non-parametric test. The Mann–Whitney U test showed that successful males produced structured signals at significantly higher rates (Mean ± SD = 0.96 ± 0.40 signals/min) than unsuccessful males (0.22 ± 0.19 signals/min) (U = 1.0, z = −2.9, p = 0.001, partial eta2 = 0.44; Cohen’s d = 1.99) (Figure 3.2B).

Table 3.1: Summary, normality test (Shapiro Wilk, W), and skewness of Latrodectus hesperus male weight, proximal-phase duration, and male proximal-phase structured signaling rates.

Shapiro-Wilk Test for Normality Variables Min Max Mean ± SD Skewness1 W(df) P

Male Weight (mg) 7.4 26.1 14.55 ± 6.02 0.914(15) 0.16 0.47

Proximal-Phase Duration (min) 19.1 174.8 99.16 ± 47.29 0.960(15) 0.69 0.05

Proximal-Phase Structured 0.35 1.71 0.96 ± 0.4 0.955 0.61 0.17 Signalling Rate (signals/min) (15) 1. Skewness < -1 and > 1 indicates skewed data to the left and right, respectively.

101

Figure 3.2: Variations in L. hesperus male structured signaling rates. (A) Difference in male structured signaling rates within the proximal phase. Structured signaling rates were consistent across proximal phase. (B) Proximal structured signaling rate of successful and unsuccessful males. An asterisk indicates a significant different in structured signaling rates (* p < 0.005).

102

Figure 3.3: Predictors of male mating success and structured vibratory signaling rates. (A) Male latency to copulate was positively predicted by proximal-phase structured signaling rate (P = 0.041). (B) Male mass negatively predicted proximal-phase structured signaling rate (P = 0.031).

103

Correlates of Structured Signaling Rates and Mating Latencies

We used a stepwise linear regression and asked whether proximal-phase duration was predicted by male mass and/or proximal-phase structured signaling rates. In the final model (final model: R2 = 0.283; F = 5.142; df = 1; P = 0.041; correlation coefficient (r) = -0.53), only proximal-phase structured signaling rates were significantly correlated with proximal-phase duration (β = -0.532; t = -2.268; P = 0.041) (Figure 3.3A). Males with higher structured signaling rates had shorter latencies to copulate. We also examined whether proximal-phase structured signaling rates were predicted by male mass. Male mass was predictive of structured signaling rates (final regression model: R2 = 0.31; F = 5.839; df = 1; P = 0.031; correlation coefficient (r) = -0.56). Males with larger mass had lower structured signaling rates (β = -0.557; t = -2.416; P = 0.031) (Figure 3.3B). Taken together, males that displayed structured signals at higher rates mated sooner, and structured signaling rates were dependent on male mass.

3.4.2 Experiment 2

Attenuating Abdominal Tremulation and Mating Success

A Chi-Squared test showed that there was no difference in the mating rates of muted (N = 12/21) and non-muted (N = 11/15) males (2 = 1.02, df = 1, p = 0.31). Attenuating abdominal tremulation component did not affect final male mating success. However, when comparing latency to copulate between muted and non-muted males, our results showed a strong non- significant trend (Mann–Whitney U test: U = 34.0, z = −1.97, p = 0.051; partial eta2 = 0.17; Cohen’s d = -0.97). Although our statistical power was low, we show large effects. Muted males, on average, took longer to copulate (150.3 ± 50.4 min) than non-muted males (94.3 ± 64.3 min).

3.5 Discussion

In this study, we tested the function of temporally structured signals in the western black widow spider Latrodectus hesperus. During male-female interactions, male western black widow spiders, L. hesperus, intermittently transition between relatively disorganized production of multiple signal components (unstructured signaling) and organized multi-component signal displays (structured signaling) where males combine individual components in a stereotyped sequence (Figure 3.1). Our results show that structured signaling is linked to male mating success; successful males produce higher structured signaling rates than unsuccessful males, and

104 105 males with higher structured signaling rates mated sooner. We also found that larger and heavier males produced structured signals at lower rates, suggesting possible trade-offs between mass/size and signaling rates. Interestingly, we found that preventing abdominal movements had no effect on male mating success. However, these muted males on average had longer latencies to copulate compared to non-muted males, suggesting that signal power may affect female receptivity. These findings, along with our recent work characterizing vibratory signals of L. hesperus (Chapter 2), reveal that males intermittently vary the structural complexity of their signals during the later stages of courtship, and these composite multicomponent signals (structured signals) play a functional role in their mating success and are linked to aspects of male quality. We show that the signals of this web-dwelling spider (previously thought to be simple and disorganized) show nascent levels of signal complexity. The degree of organization among signal components and the significance of specific component arrangements have been extensively examined in bird songs, as well as a few non-human Primates and Cetaceans calls (Arnold and Zuberbühler, 2012; Chatfield and Lemon, 1970; Clucas et al., 2004; Kershenbaum, 2014; Schamberg et al., 2016; Suzuki et al., 2006; Zuberbühler, 2002). Few studies have examined complex signals in arthropods, and in general, arthropods are considered to be simple signalers. To our knowledge, our study is the first to demonstrate that the structured signals affects mating success in an arthropod (but also see: Elias et al., 2012; Girard et al., 2015).

Our finding that successful male L. hesperus produced structured signals at higher rates compared to unsuccessful males suggests that structured signals are important for male mating success (Figure 3.2B). Signaling rates have been shown to be important for male mating success in other spider species (Elias et al., 2010; Gibson and Uetz, 2008; Kotiaho et al., 1996; Parri et al., 1997; Rundus et al., 2010; Shamble et al., 2009; Sivalinghem et al., 2010). For example, in the jumping spider Phidippus clarus, successful males had higher signaling rates than unsuccessful males (Elias et al., 2010; Sivalinghem et al., 2010). In the wolf spider, Hygrolycosa rubrofasciata, females show preference for males with higher drumming rates (Kotiaho et al., 1996; Parri et al., 1997), and drumming rates convey honest information about male quality (Ahtiainen et al., 2004; Mappes et al., 1996; Rivero et al., 2000), and are associated with energetic, longevity, and predation costs (Ahtiainen et al., 2005; Kotiaho et al., 1999, 1998). Where previous studies focus on fixed composite signals or individual components (e.g. abdominal vibration rates in P. clarus), in this study, we specifically examined how often males

106 transitioned from temporally unstructured signaling to structured multicomponent signaling in the western black widow spider. Structured signaling rates may indicate male energetic output or courtship effort; and previous studies have shown courtship effort to be a better indicator of male condition than individual signal components (Girard et al., 2015; Shamble et al., 2009). In black widow spiders, adult males stop foraging upon their final molt and rely on stored energy reserves to embark on costly mate searching, and then an energetically demanding lengthy courtship process (Anava and Lubin, 1993; Andrade, 2003, 1996; Baruffaldi and Andrade, 2015; De Luca et al., 2015; Harari et al., 2009; Kasumovic and Andrade, 2004; MacLeod and Andrade, 2014; Ross and Smith, 1979; Segev et al., 2003). Our results show that males with higher structured signaling rates mated sooner (Figure 3.3A), suggesting that black widow females may use structured signaling rates to assess male courtship effort, and in turn mate quality. These results are consistent with previous studies in widow spiders demonstrating that male courtship effort is important for male mating success (see: Harari et al., 2009; Stoltz et al., 2009, 2008; Stoltz and Andrade, 2010).

Producing complex temporally structured, multicomponent signal displays may require intense neuro-muscular activity in spiders, and can be energetically more demanding than unstructured signaling. The wolf spider Schizocosa ocreata, which produces complex multimodal signals, were shown to expend relatively more energy during courtship signaling compared with their sister species S. rovneri, that produce simple repetitive vibration only signals (Cady et al., 2011). Similarly, in passerine birds, song complexity is positively associated with increased metabolism across species (Garamszegi et al., 2006). Currently, it is not known whether there are metabolic rate differences between vibratory signaling and resting in L. hesperus; and future studies would need to examine metabolic rate differences between structured and unstructured signaling.

Previously, De Luca et al. (2015) demonstrated that the metabolic rates during vibratory signaling in a congener widow spider, Latrodectus hasselti, is four times that of resting metabolic rate. Additionally, in L. hasselti, larger males were not the most energetically efficient signalers; rather, males with intermediate mass were more efficient signalers (De Luca et al., 2015). Consistent with this previous study, we found that heavier L. hesperus males produced lower structured signaling rates (Figure 3.3B), which indicates that heavier/larger males may be relatively poor signalers, and that there may be a small-male advantage in black widow spiders (also see: Kasumovic and Andrade, 2009). In many species, small males have specific

107 competitive advantages over larger males in regards to physiological efficiencies and early mate searching and copulation success (Blanckenhorn et al., 1995; Blanckenhorn and Viele, 1999; Crompton et al., 2003; Foellmer and Fairbairn, 2005; Kasumovic and Andrade, 2009, 2006; Moya-Laraño et al., 2007; Schneider et al., 2000). Specifically, in species where males compete through scramble competition for access to females, selection should favour smaller males that mature earlier, are more agile, and have lower energy requirements (e.g. foraging), which allows for increased efforts towards mating (Blanckenhorn and Viele, 1999; Crompton et al., 2003; Kasumovic and Andrade, 2009, 2006; Moya-Laraño et al., 2007). For example, in the seed beetles, Stator limbatus, smaller males are able to warm up their bodies to take flight more quickly, and reach females earlier than larger males (Moya-Laraño et al., 2007). In widow spiders, however, males can also compete directly with each other, and in these direct competitions, larger males have a competitive advantage (Kasumovic and Andrade, 2009; Stoltz et al., 2009, 2008). Therefore, smaller males may benefit from signaling harder in order to speed up courtship.

Moreover, males can alter their developmental strategies in response to variation in the competitive environment (scramble vs. direct competition) (Kasumovic and Andrade, 2009, 2006). In the redback spiders, L. hasselti, males speed up or delay their developmental time in response to higher virgin female or male densities, respectively; and males that delay development have larger body size (Kasumovic and Andrade, 2006). Although larger males have a mating advantage during direct competition (Kasumovic and Andrade, 2009; MacLeod, 2013; Stoltz et al., 2009, 2008), our results with L. hesperus males suggests that one of the possible trade-offs for increased size and mass is lower structured signaling rates. In the field, L. hesperus males that mature in the fall season may be under scramble competition due to the high local density of virgin females, and males that mature in the spring may be under direct competition due to fewer virgin females (Scott et al., 2019); Sheena Fry, personal communication). It is likely that these two competitive environments select for different developmental strategies in L. hesperus males, and males that mature in the fall season are likely smaller in size, but better at finding mates than males that develop in the spring season. At present, the importance of vibratory signaling during competitive contexts is not known for widow spiders, and future studies with two males competing on a female web are needed to elucidate the relative importance of male mass/size and signaling rates in these contexts.

108

In this study, we also examine the importance of the abdominal tremulation component of male black widow signals. Interestingly, our results show that attenuating the abdominal tremulation component had no significant effect on male mating success, suggesting that this component is not necessary for mating success. The abdominal tremulation signals of L. hesperus males may have a different functional role, and attenuating this signal component may not necessarily affect male mating success. Previously, Vibert et al. (Vibert et al., 2014) showed that females may use the low-amplitude abdominal tremulation signals to discriminate mates from prey. In our previous study, we showed that abdominal tremulation signal characteristics were not associated with male traits, but the peak power of male bounce and web pluck signals were positively associated with larger males (Chapter 2). These suggest that the abdominal tremulation signals may carry different information than bounce and web plucks. It is possible that, although muted males had their abdominal tremulation vibrations attenuated, these signal components were still detectable to female black widows, and females can still use them to identify mates. Since males display structured signals at close distance to females, these attenuated vibrations may still be perceptible to females. Previous sensory physiology studies have demonstrated the acute vibration sense of many female spiders, including web-dwelling spiders (Barth and Geethabali, 1982; Friedel and Barth, 1995; Walcott, 1963, 1969; Walcott and Kloot, 1959). It is not clear from this study if females assess individual components of the structured multi-component signals, and further research is needed to examine female perception of different signal components.

It is also important to note that in our trials, overall male mating success was high and our sample was relatively small, and this may have made it more difficult to measure an effect of attenuating abdominal tremulation signals. The design of our trials, in which individual virgin females and males were confined on the female’s web, may have facilitated male courtship success. Trial female webs were smaller than many natural webs, and males were never subject to direct competition. Therefore we cannot rule out the possibility that under more naturalistic circumstances, a higher proportion of males may be unsuccessful and success may correlate with signal components and structured signal rates.

Although attenuating abdominal tremulation had no significant effect on male mating success, our results show a strong trend that muted males took longer to copulate. This increased latency to copulate can have significant effects on male mating success in the field, as male courtship

109 efforts can be exploited by rival intruders (Stoltz and Andrade, 2010). In black widow spiders, there is an advantage for mating first and quickly, as field studies have shown that competing males show up quickly on female webs (Scott et al., 2019). This is due to the dominance of first male sperm precedence and the lack of mate discrimination in widow females (MacLeod, 2013; Snow and Andrade, 2005; Stoltz and Andrade, 2010). For example, Stoltz and Andrade (2010) showed that female L. hasselti do not distinguish between courting males, and will allow copulation to any male once her courtship effort threshold (~100 min of courtship) is met. Therefore, males that court longer potentially risk having their courtship effort parasitized by incoming rival males.

The vibratory courtship signals of male western black widow spider show nascent complexity, as males’ transition from haphazard production of individual signal components to more organized stereotyped multicomponent signal displays. Our research demonstrates that these intermittent bouts of temporally structured signal sequences play an important role during courtship, as males that produced these structured signals more frequently are likely to successfully mate and mate sooner. In this study, we also show that heavier males produced structured signals at lower rates, which suggests the possibility of a trade-off between signaling efficiency and mass (see for example De Luca et al., 2015). Our study is the first to demonstrate that the structural complexity of signal components is functional in a web-dwelling spider. Currently, much of our understanding of complex communication signals in spiders comes from extensive research on cursorial spiders (i.e. jumping spiders and wolf spiders; Herberstein et al., 2014; Uhl and Elias, 2011), where some species (e.g. the jumping spider Habronattus coecatus) produce highly organized signals with varying motifs of signals components (Elias et al., 2012); while in other species (e.g. the wolf spider Schizocosa stridulans), males transition from producing signal components simultaneously (‘parallel signaling’) when further away from females, to sequential signal display (‘serial signaling’) (Elias et al., 2006). The latter example is similar to our observations in the western black widow males that transition from haphazard to temporally structured signaling. In S. stridulans, male parallel signaling was hypothesized to function to overwhelm the female’s sense and suppress her predatory behaviour (‘sensory overload’ hypothesis), whereas serial signaling allows females to assess individual signal components (Elias et al., 2006). To date, no study has examined how females perceive and process these complex signals, and what aspects of signal complexities are important for decision making in

110 females. Therefore, we advocate for future studies to also focus on the sensory side of complex communication. In addition, we also recommend for more empirical studies on vibratory communication in web-dwelling spiders, as this group of spiders are greatly underrepresented in the literature (Herberstein et al., 2014). The diversity and function of temporal complexities in communication signals has been little studied outside of some birdsongs and non-human primates. This leaves a significant gap in our understanding of the evolution and function of signal complexity in many other organisms. We demonstrated that in a web-dwelling spider, males actively organize individual signal components into a stereotyped display, and this has important consequences for mating success. There is a clear need for more studies on the signals of many other arthropods.

3.6 Acknowledgement

We thank undergraduate members of the Mason lab who helped with spider maintenance and care. We also thank Drs. Damian Elias, Maydianne, Andrade and Kenneth Welch who gave us valuable feedback, comments and suggestions on this manuscript. This research was funded by the National Sciences and Engineering Research Council of Canada (NSERC) Postgraduate Scholarship, and Ontario Graduate Scholarship to S.S., and NSERC Discovery Grant to A.C.M. The authors declare they have no conflict of interest.

3.7 References

Ahtiainen, J.J., Alatalo, R.V., Kortet, R., Rantala, M.J., 2004. Sexual advertisement and immune function in an arachnid species (Lycosidae). Behavioral Ecology 15, 602–606.

Alger, S.J., Larget, B.R., Riters, L.V., 2016. A novel statistical method for behaviour sequence analysis and its application to birdsong. Animal Behaviour 116, 181–193.

Andrade, M.C.B., 2003. Risky mate search and male self-sacrifice in redback spiders. Behavioral Ecology 14, 531–538.

Andrade, M.C.B., 1996. Sexual selection for male sacrifice in the Australian redback spider. Science 271, 70–72.

Arnold, K., Zuberbühler, K., 2012. Call combinations in monkeys: Compositional or idiomatic expressions? Brain and Language 120, 303–309.

Barth, F.G., Geethabali, 1982. Spider vibration receptors: Threshold curves of individual slits in the metatarsal lyriform organ. Journal of Comparative Physiology A 148, 175–185.

Baruffaldi, L., Andrade, M.C.B., 2015. Contact pheromones mediate male preference in black widow spiders: Avoidance of hungry sexual cannibals? Animal Behaviour 102, 25–32.

Beckers, G.J.L., Ten Cate, C., 2001. Perceptual relevance of species-specific differences in acoustic signal structure in Streptopelia doves. Animal Behaviour 62, 511–518.

Behr, O., von Helversen, O., 2004. Bat serenades - complex courtship songs of the sac-winged bat (Saccopteryx bilineata). Behavioral Ecology and Sociobiology 56, 106–115.

Blackledge, T.A., Zevenbergen, J.M., 2007. Condition-dependent spider web architecture in the western black widow, Latrodectus hesperus. Animal Behaviour 73, 855–864.

Blanckenhorn, W.U., Preziosi, R.F., Fairbairn, D.J., 1995. Time and energy constraints and the evolution of sexual size dimorphism - to eat or to mate? Evolutionary Ecology 9, 369–381.

Blanckenhorn, W.U., Viele, S.N.T., 1999. Foraging in yellow dung flies: Testing for a small- male time budget advantage. Ecological Entomology 24, 1–6.

111 112

Bohn, K.M., Schmidt-French, B., Schwartz, C., Smotherman, M., Pollak, G.D., 2009. Versatility and stereotypy of free-tailed bat songs. PLoS One 4.

Byers, B.E., Kroodsma, D.E., 2009. Female mate choice and songbird song repertoires. Animal Behaviour 77, 13–22.

Cady, A.B., Delaney, K.J., Uetz, G.W., 2011. Contrasting energetic costs of courtship signaling in two wolf spiders having divergent courtship behaviors. The Journal of Arachnology 39, 161–165.

Candolin, U., 2003. The use of multiple cues in mate choice. Biological Reviews 78, 575–595.

Catchpole, C., Slater, P.J.B., 2008. Bird song: Biological themes and variations. 2008. Cambridge [England].

Catchpole, C.K., Rowell, A., 1993. Song sharing and local dialects in a population of the European wren Troglodytes troglodytes. Behaviour 125, 67–78.

Chatfield, C., Lemon, R.E., 1970. Analysing sequences of behavioural events. Journal of Theoretical Biology 29, 427–445.

Clarke, E., Reichard, U.H., Zuberbühler, K., 2006. The syntax and meaning of wild gibbon songs. PLoS ONE 1, e73.

Clucas, B.A., Freeberg, T.M., Lucas, J.R., 2004. Chick-a-dee call syntax, social context, and season affect vocal responses of Carolina chickadees (Poecile carolinensis). Behavioral Ecology and Sociobiology 57, 187–196.

Coyle, F.A., O’Shields, T.C., 1990. Courtship and mating behavior of Thelechoris karschi (Araneae, Dipluridae), an African funnel web spider. The Journal of Arachnology 18, 281–296.

Crockford, C., Boesch, C., 2003. Context-specific calls in wild chimpanzees, Pan troglodytes verus: Analysis of barks. Animal Behaviour 66, 115–125.

Crompton, B., Thomason, J.C., McLachlan, A., 2003. Mating in a viscous universe: The race is to the agile, not to the swift. Proceedings of the Royal Society of London. Series B: Biological Sciences 270, 1991–1995.

Eberhard, W.G., Huber, B.A., 1998. Courtship, copulation, and sperm transfer in Leucauge mariana (Araneae, Tetragnathidae) with implications for higher classification. The Journal of Arachnology 26, 342–368.

113

Elias, D.O., Lee, N., Hebets, E.A., Mason, A.C., 2006. Seismic signal production in a wolf spider: Parallel versus serial multi-component signals. Journal of Experimental Biology 209, 1074–1084.

Elias, D.O., Mason, A.C., 2011. Signaling in variable environments: Substrate-borne signaling mechanisms and communication behavior in spiders. The Use of Vibrations in Communication: properties, mechanisms and function across taxa. Kerala: Research Signpost.

Elias, D.O., Mason, A.C., Maddison, W.P., Hoy, R.R., 2003. Seismic signals in a courting male jumping spider (Araneae: Salticidae). Journal of Experimental Biology 206, 4029– 4039.

Elias, D.O., Sivalinghem, S., Mason, A.C., Andrade, M.C.B., Kasumovic, M.M., 2010. Vibratory communication in the jumping spider Phidippus clarus: Substrate-borne courtship signals are important for male mating success. Ethology 116, 990–998.

Evans, C.S., Marler, P., 1994. Food calling and audience effects in male chickens, Gallus gallus: Their relationships to food availability, courtship and social facilitation. Animal Behaviour 47, 1159–1170.

Fischer, J., Wadewitz, P., Hammerschmidt, K., 2017. Structural variability and communicative complexity in acoustic communication. Animal Behaviour 134, 229–237.

Foellmer, M.W., Fairbairn, D.J., 2005. Selection on male size, leg length and condition during mate search in a sexually highly dimorphic orb-weaving spider. Oecologia 142, 653–662.

Freeberg, T.M., Dunbar, R.I.M., Ord, T.J., 2012. Social complexity as a proximate and ultimate factor in communicative complexity. Philosophical Transactions of the Royal Society B: Biological Sciences 367, 1785–1801.

Freeberg, T.M., Lucas, J.R., 2002. Receivers respond differently to chick-a-dee calls varying in note composition in Carolina chickadees, Poecile carolinensis. Animal Behaviour 63, 837–845.

114

Friedel, T., Barth, F.G., 1995. The response of interneurons in the spider CNS (Cupiennius salei Keyserling) to vibratory courtship signals. Journal of Comparative Physiology A 177, 159–171.

Garamszegi, L.Z., Boulinier, T., Møller, A.P., Török, J., Michl, G., Nichols, J.D., 2002. The estimation of size and change in composition of avian song repertoires. Animal Behavior 63, 623–630.

Garamszegi L.Z., Moreno J., Moller A.P., 2006. Avian song complexity is associated with high field metabolic rate. Evolutionary Ecology Research 8, 75–90.

Gibson, J.S., Uetz, G.W., 2008. Seismic communication and mate choice in wolf spiders: Components of male seismic signals and mating success. Animal Behaviour 75, 1253– 1262.

Green, S., 1975. Dialects in Japanese monkeys: Vocal learning and cultural transmission of locale-specific vocal behavior? Zeitschrift für Tierpsychologie 38, 304–314.

Gridi-Papp, M., Rand, A.S., Ryan, M.J., 2006. Complex call production in the túngara frog. Nature 441, 38–38.

Große Ruse, M., Hasselquist, D., Hansson, B., Tarka, M., Sandsten, M., 2016. Automated analysis of song structure in complex birdsongs. Animal Behaviour 112, 39–51.

Grüter, C., Farina, W.M., 2009. The honeybee waggle dance: Can we follow the steps? Trends in Ecology & Evolution 24, 242–247.

Hafner, G.W., Hamilton, C.L., Steiner, W.W., Thompson, T.J., Winn, H.E., 1979. Signature information in the song of the humpback whale. The Journal of the Acoustical Society of America 66, 1–6.

Hailman, J.P., Ficken, M.S., Ficken, R.W., 1987. Constraints on the structure of combinatorial “Chick-a-dee” Calls. Ethology 75, 62–80.

Harari, A.R., Ziv, M., Lubin, Y., 2009. Conflict or co-operation in the courtship display of the white widow spider, Latrodectus pallidus. The Journal of Arachnology 37, 254–260.

Hasselquist, D., Bensch, S., Schantz, T. von, 1996. Correlation between male song repertoire, extra-pair paternity and offspring survival in the great reed warbler. Nature 381, 229– 232.

Hebets, E.A., 2005. Attention-altering signal interactions in the multimodal courtship display of the wolf spider Schizocosa uetzi. Behavioral Ecology 16, 75–82.

115

Hebets, E.A., Papaj, D.R., 2005. Complex signal function: Developing a framework of testable hypotheses. Behavioral Ecology and Sociobiology 57, 197–214.

Higham, J.P., Hebets, E.A., 2013. An introduction to multimodal communication. Behavioral Ecology and Sociobiology 67, 1381–1388.

Holland, J., Dabelsteen, T., Paris, A.L., 2000. Coding in the song of the wren: Importance of rhythmicity, syntax and element structure. Animal Behaviour 60, 463–470.

Johnstone, R.A., 1996. Multiple displays in animal communication: ‘Backup signals’ and ‘multiple messages.’ Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 351, 329–338.

Kanwal, J.S., Matsumura, S., Ohlemiller, K., Suga, N., 1994. Analysis of acoustic elements and syntax in communication sounds emitted by mustached bats. The Journal of the Acoustical Society of America 96, 1229–1254.

Kaston, B.J., 1970. Comparative biology of American black widow spiders. Transaction of the San Diego Society of Natural History.

Kasumovic, M.M., Andrade, M.C.B., 2009. A change in competitive context reverses sexual selection on male size. Journal of Evolutionary Biology 22, 324–333.

Kasumovic, M.M., Andrade, M.C.B., 2006. Male development tracks rapidly shifting sexual versus natural selection pressures. Current Biology 16, R242–R243.

Kershenbaum, A., 2014. Entropy rate as a measure of animal vocal complexity. Bioacoustics 23, 195–208.

Kipper, S., Kiefer, S., 2010. Age-related changes in birds’ singing styles: On fresh tunes and fading voices?, in: Brockmann, H.J., Roper, T.J., Naguib, M., Wynne-Edwards, K.E., Mitani, J.C., Simmons, L.W. (Eds.), Advances in the Study of Behavior. Academic Press, pp. 77–118.

116

Kipper, S., Mundry, R., Sommer, C., Hultsch, H., Todt, D., 2006. Song repertoire size is correlated with body measures and arrival date in common nightingales, Luscinia megarhynchos. Animal Behaviour 71, 211–217.

Kotiaho, J., Alatalo, R.V., Mappes, J., Parri, S., 1996. Sexual selection in a wolf spider: Male drumming activity, body size, and viability. Evolution 50, 1977–1981.

Kotiaho, J.S., Alatalo, R.V., Mappes, J., Nielsen, M.G., Parri, S., Rivero, A., 1998. Energetic costs of size and sexual signalling in a wolf spider. Proceedings of the Royal Society of London. Series B: Biological Sciences 265, 2203–2209.

Kotiaho, J.S., Alatalo, R.V., Mappes, J., Parri, S., 1999. Sexual signalling and viability in a wolf spider (Hygrolycosa rubrofasciata): Measurements under laboratory and field conditions. Behavioral Ecology and Sociobiology 46, 123–128.

Lemon, R.E., Chatfield, C., 1971. Organization of song in cardinals. Animal Behaviour 19, 1– 17.

Lubin, Y.D., 1986. Courtship and alternative mating tactics in a social spider. The Journal of Arachnology 14, 239–257.

MacLeod, E., 2013. New insights into the evolutionary maintenance of male mate choice behaviour using the western black widow spider, Latrodectus hesperus (Thesis).

MacLeod, E.C., Andrade, M.C.B., 2014. Strong, convergent male mate choice along two preference axes in field populations of black widow spiders. Animal Behaviour 89, 163– 169.

Maklakov, A.A., Bilde, T., Lubin, Y., 2003. Vibratory courtship in a web-building spider: signalling quality or stimulating the female? Animal Behaviour 66, 623–630.

Manser, M.B., 2001. The acoustic structure of suricates’ alarm calls varies with predator type and the level of response urgency. Proceedings of the Royal Society of London. Series B: Biological Sciences 268, 2315–2324.

Manser, M.B., Bell, M.B., Fletcher, L.B., 2001. The information that receivers extract from alarm calls in suricates. Proceedings of the Royal Society of London. Series B: Biological Sciences 268, 2485–2491.

Marler, P., Mitani, J.C., 1989. A phonological analysis of male gibbon singing behavior. Behaviour 109, 20–45.

McComb, K., Semple, S., 2005. Coevolution of vocal communication and sociality in primates. Biology Letters 1, 381–385.

117

McCowan, B., Hanser, S.F., Doyle, L.R., 1999. Quantitative tools for comparing animal communication systems: information theory applied to bottlenose dolphin whistle repertoires. Animal Behaviour 57, 409–419.

Miller, C.T., Bee, M.A., 2012. Receiver psychology turns 20: Is it time for a broader approach? Animal Behaviour 83, 331–343.

Moya-Laraño, J., El-Sayyid, M.E.T., Fox, C.W., 2007. Smaller beetles are better scramble competitors at cooler temperatures. Biology Letters 3, 475–478.

O’Loghlen, A.L., Rothstein, S.I., 2012. When less is best: Female brown-headed cowbirds prefer less intense male displays. PLoS One 7.

O’Loghlen, A.L., Rothstein, S.I., 2010. Multimodal signalling in a songbird: Male audiovisual displays vary significantly by social context in brown-headed cowbirds. Animal Behaviour 79, 1285–1292.

Ossip-Klein, A.G., Fuentes, J.A., Hews, D.K., Martins, E.P., 2013. Information content is more important than sensory system or physical distance in guiding the long-term evolutionary relationships between signaling modalities in Sceloporus lizards. Behavioral Ecology and Sociobiology 67, 1513–1522.

Ouattara, K., Lemasson, A., Zuberbühler, K., 2009. Campbell’s monkeys concatenate vocalizations into context-specific call sequences. Proceedings of the National Academy of Sciences 106, 22026–22031.

Parri, S., Alatalo, R.V., Kotiaho, J., Mappes, J., 1997. Female choice for male drumming in the wolf spider Hygrolycosa rubrofasciata. Animal Behaviour 53, 305–312.

Partan, S.R., 2013. Ten unanswered questions in multimodal communication. Behavioral Ecology and Sociobiology 67, 1523–1539.

Partan, S.R., Marler, P., 2005. Issues in the classification of multimodal communication signals. The American Naturalist 166, 231–245.

Payne, R.S., McVay, S., 1971. Songs of humpback whales. Science 173, 585–597.

Petrusková, T., Osiejuk, T.S., Petrusek, A., 2010. Geographic variation in songs of the tree pipit (Anthus trivialis) at two spatial scales variation. Auk 127, 274–282.

Rivero, A., Alatalo, R.V., Kotiaho, J.S., Mappes, J., Parri, S., 2000. Acoustic signalling in a wolf spider: can signal characteristics predict male quality? Animal Behaviour 60, 187– 194.

118

Ronald, K.L., Zeng, R., White, D.J., Fernández-Juricic, E., Lucas, J.R., 2017. What makes a multimodal signal attractive? A preference function approach. Behavioral Ecology 28, 677–687.

Rowe, C., 1999. Receiver psychology and the evolution of multicomponent signals. Animal behaviour 58, 921–931.

Rundus, A.S., Santer, R.D., Hebets, E.A., 2010. Multimodal courtship efficacy of Schizocosa retrorsa wolf spiders: Implications of an additional signal modality. Behavioral Ecology 21, 701–707.

Ruxton, G.D., Schaefer, H.M., 2011. Resolving current disagreements and ambiguities in the terminology of animal communication. Journal of Evolutionary Biology 24, 2574–2585.

Sasahara, K., Cody, M.L., Cohen, D., Taylor, C.E., 2012. Structural design principles of complex bird songs: A network-based approach. PLoS One 7.

Schamberg, I., Cheney, D.L., Clay, Z., Hohmann, G., Seyfarth, R.M., 2016. Call combinations, vocal exchanges and interparty movement in wild bonobos. Animal Behaviour 122, 109–116.

Schneider, J.M., Herberstein, M.E., Crespigny, F.C.D., Ramamurthy, S., Elgar, M.A., 2000. Sperm competition and small size advantage for males of the golden orb-web spider Nephila edulis. Journal of Evolutionary Biology 13, 939–946.

Schneider, S.S., Lewis, L.A., 2004. The vibration signal, modulatory communication and the organization of labor in honey bees, Apis mellifera. Apidologie 35, 117–131.

Scott, C., Vibert, S., Gries, G., 2012. Evidence that web reduction by western black widow males functions in sexual communication. The Canadian Entomologist 144, 672–678.

Segev, O., Ziv, M., Lubin, Y., 2003. The male mating system in a desert widow spider. The Journal of Arachnology 31, 379–393.

Seyfarth, R.M., Cheney, D.L., 2017. The origin of meaning in animal signals. Animal Behaviour 124, 339–346.

Seyfarth, R.M., Cheney, D.L., Marler, P., 1980. Monkey responses to three different alarm calls: Evidence of predator classification and semantic communication. Science 210, 801–803.

119

Shamble, P.S., Wilgers, D.J., Swoboda, K.A., Hebets, E.A., 2009. Courtship effort is a better predictor of mating success than ornamentation for male wolf spiders. Behavioral Ecology 20, 1242–1251.

Slobodchikoff, C.N., Kiriazis, J., Fischer, C., Creef, E., 1991. Semantic information distinguishing individual predators in the alarm calls of Gunnison’s prairie dogs. Animal Behaviour 42, 713–719.

Snow, L.S., Andrade, M.C., 2005. Multiple sperm storage organs facilitate female control of paternity. Proceedings of the Royal Society B: Biological Sciences 272, 1139–1144.

Stange, N., Page, R.A., Ryan, M.J., Taylor, R.C., 2017. Interactions between complex multisensory signal components result in unexpected mate choice responses. Animal Behaviour 134, 239–247.

Stoffer, B., Uetz, G.W., 2017. The effects of experience with different courtship modalities on unimodal and multimodal preferences in a wolf spider. Animal Behaviour 123, 187–196.

Stoltz, J.A., Andrade, M.C.B., 2010. Female’s courtship threshold allows intruding males to mate with reduced effort. Proceedings of the Royal Society B: Biological Sciences 277, 585–592.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2009. Male courtship effort determines female response to competing rivals in redback spiders. Animal Behaviour 77, 79–85.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2008. Females reward courtship by competing males in a cannibalistic spider. Behavioral Ecology and Sociobiology 62, 689–697.

Suzuki, R., Buck, J.R., Tyack, P.L., 2006. Information entropy of humpback whale songs. The Journal of the Acoustical Society of America 119, 1849–1866.

Taylor, R.C., Klein, B.A., Stein, J., Ryan, M.J., 2011. Multimodal signal variation in space and time: How important is matching a signal with its signaler? Journal of Experimental Biology 214, 815–820.

Taylor, R.C., Ryan, M.J., 2013. Interactions of multisensory components perceptually rescue Túngara frog mating signals. Science 341, 273–274.

Templeton, C.N., Greene, E., Davis, K., 2005. Allometry of alarm calls: Black-capped chickadees encode information about predator size. Science 308, 1934–1937.

Todt, D., Hultsch, H., 1998. How songbirds deal with large amounts of serial information: Retrieval rules suggest a hierarchical song memory. Biological Cybernetics 79, 487–500.

120

Uy, J.A.C., Safran, R.J., 2013. Variation in the temporal and spatial use of signals and its implications for multimodal communication. Behavioral Ecology and Sociobiology 67, 1499–1511.

Vallet, E., Kreutzer, M., 1995. Female canaries are sexually responsive to special song phrases. Animal Behaviour 49, 1603–1610.

Vibert, S., Scott, C., Gries, G., 2014. A meal or a male: the ‘whispers’ of black widow males do not trigger a predatory response in females. Frontiers in Zoology 11, 4.

Walcott, C., 1963. The effect of the web on vibration sensitivity in the spider, Achaearanea tepidariorum (KOCH). Journal of Experimental Biology 40, 595–611.

Walcott, C., Kloot, W.G. van der, 1959. The physiology of the spider vibration receptor. Journal of Experimental Zoology 141, 191–244.

Walcott, Charles., 1969. A spider’s vibration receptor: Its anatomy and physiology. Integrative and Comparative Biology 9, 133–144.

Weiss, M., Hultsch, H., Adam, I., Scharff, C., Kipper, S., 2014. The use of network analysis to study complex animal communication systems: A study on nightingale song. Proceedings of the Royal Society B: Biological Sciences 281, 20140460.

Wilczynski, W., Rand, S.A., Ryan, M.J., 1999. Female preferences for temporal order of call components in the túngara frog: A Bayesian analysis. Animal Behaviour 58, 841–851.

Woolley, S.C., Doupe, A.J., 2008. Social context-induced song variation affects female behavior and gene expression. PLoS Biology 6.

Zann, R., 1993. Structure, sequence and evolution of song elements in wild Australian zebra finches. Auk 110, 702–715.

Zuberbühler, K., 2002. A syntactic rule in forest monkey communication. Animal behaviour 63, 293–299.

121

Chapter 4 Posture Controls Mechanical Tuning in the Black Widow Spider Mechanosensory System

Natasha Mhatre†, Senthurran Sivalinghem†, Andrew C. Mason

† Contributed equally

4.1 Abstract

Spiders use vibrations for sexual communication, capturing prey, and evading predators. Their vibration sensing organs, slit sensilla, resembles cracks in the exoskeleton, and is distributed all over their bodies. Those crucial to sensing web- and substrate vibrations are called lyriform organs, are localized around leg joints, and activated by joint bending. Little is known of how body biomechanics affects the transfer of vibrations to leg joints. Female black widow spiders have a striking body-form; long thin legs support a large pendulous abdomen. Here, we show that in unrestrained female black widows, leg joints are tuned by the body’s mechanics and that adopting different postures enables females to alter both the sensitivity and tuning of these joints. Therefore, we suggest that posture may be used to flexibly focus attention to different components of web vibration, emphasizing the dynamic loop of interactions between brain and body.

4.2 Introduction

Vibration sensing is crucial to spiders, to perceive and localize predators, prey, and mates (Barth, 2001). The sensory organs that enable vibration sensing, i.e. slit sensilla, function by sensing minute strains in the exoskeleton (Barth, 2004). The slit sensilla most important for detecting substrate-borne vibrations are classified as Lyriform Organs - compound organs comprising several closely arranged parallel slits, densely distributed near leg joints (Barth, 2001). Much of what is known about lyriform organs is derived from neurophysiology. Neurophysiological studies in spiders have concentrated on individual legs, joints, and/or receptors and made highly precise measurements by delivering vibrational stimuli to these individual joints (Barth and Geethabali, 1982; Blickhan and Barth, 1985; French et al., 2002; Walcott, 1963). Extremely detailed work over several decades has shown that leg-joint bending is sufficient to produce neurophysiological responses in lyriform organs (Barth, 2001; Schaber et al., 2012; Speck and Barth, 1982; Speck-Hergenröder and Barth, 1987). The exact physical mechanism that converts leg bending into exoskeletal strains to excite neuronal responses is not completely understood. It is believed that strains are transmitted to the exoskeleton during joint bending through a combination of local hydrostatic pressure and direct mechanical ligament attachments (Blickhan and Barth, 1985; Hößl et al., 2009, 2007; Speck and Barth, 1982). Lyriform organs are exquisitely sensitive (Barth and Geethabali, 1982; Blickhan and Barth, 1985; French et al., 2002) and can sense very small vibrations propagating through the substrate or web (Masters and Markl, 1981; Walcott, 1963). In general they show high pass characteristics and are most sensitive to vibrations above ~200 Hz (Barth and Geethabali, 1982; Blickhan and Barth, 1985). Paradoxically, in the lower frequency range (<200 Hz) observed in all web-borne vibrations (Barth and Geethabali, 1982; Blickhan and Barth, 1985) including those we have measured (Figure S4.10), all lyriform organs studied to date show lower sensitivity and a flat frequency response. It is only further upstream, in the central nervous system, that some degree of frequency discrimination is observed (Speck-Hergenröder and Barth, 1987).

Before the nervous system becomes involved, however, mechanical input to the lyriform organs is shaped by the mechanics of the spider’s body and its environment. In earlier studies, measurements using precise and innovative techniques, often developed by the authors themselves, have shown that web tension and the coupling of the leg to web have strong effects

122 123 on vibrations delivered to the leg (Walcott, 1963). Even older work had indicated that the mechanics of the spider’s body could modulate the vibrations delivered to the joints (Liesenfeld, 1956, 1961). It is increasingly recognized that perception is a cognitive process that occurs through the interaction of the nervous system, the body, and the environment in which the body is situated (Alva, 2009; Chiel et al., 2009). Even animals with relatively simple nervous systems, like spiders, can act as adaptive, problem-solving agents that use their bodies and environment to modulate perception, thus demonstrating embodied or extended cognition (Japyassú and Laland, 2017; Keijzer, 2017). As a first step towards a more complete picture of spider vibration perception, we decided to exploit the sensitivity of modern measurement techniques and the power of current modelling tools to test how the mechanics of the spider’s body, arising from its natural posture on the web would affect its perception (Barth and Geethabali, 1982; Blickhan and Barth, 1985).

In web-dwelling spiders, vibration perception becomes particularly rich and interesting given that both the environment and the body can be manipulated by the spider (Blackledge and Zevenbergen, 2007; Elias and Mason, 2011; Japyassú and Laland, 2017; Tso et al., 2005). Spiders effectively develop and alter their environment to suit their perceptual needs, via their web (Elias and Mason, 2011; Kaplan, 2012; Zevenbergen et al., 2008). A great deal of research now has considered the effect and contribution of the web as an extension of spider perception (Eberhard, 1990; Mortimer et al., 2016; Vibert et al., 2016, 2014; Zevenbergen et al., 2008). The role of the other player in the system - the reconfigurable, dynamic, and flexible spider body - has, with a few exceptions (Blickhan and Barth, 1985; Walcott, 1963), remained largely unstudied. Here we explore the role of the body of the female black-widow spider (Latrodectus hesperus) in vibration perception. Specifically, we consider how vibrations are transmitted to different joints in a female black widow spider, freely suspended on its own web, and how posture affects this spatial pattern of vibration.

Since the primary mechanical input to lyriform organs results from the bending of leg joints (Barth and Geethabali, 1982; Blickhan and Barth, 1985; French et al., 2002), it would be ideal to study joint bending directly using strain sensors. However, given the size of black-widow spider legs and strain amplitudes that we expect, it is difficult to manufacture strain sensors small enough not to disrupt natural body mechanics. Laser Doppler vibrometry (LDV), however, allows contact-free measurement of movement at near picometer sensitivity (Johansmann et al.,

124

2005) and we can infer joint bending from the relative motion of different leg segments under the reasonable assumption that these behave as rigid bodies at the low frequencies under consideration. We exploit the sensitivity of LDV, in combination with multi-body dynamic modelling (MBD), to study the effects of the body, its size, and posture on the tuning of leg joints in the female black-widow spider.

4.3 Materials and Methods 4.3.1 Body Dynamics: Vibrometry

Adult female L. hesperus were allowed to build webs on custom wooden frames made of a base (length x width x height cm3 ~ 12 x 12 x 1.5) with posts (diameter x height = 0.8 x 11 cm) mounted at each corner (Figure S4.1). All females were given at least two weeks to construct webs. Experiments were carried out on unrestrained females and replicate numbers are provided for each experiment. Females were briefly removed from the webs, anaesthetized with CO2, and the ventral surface of all leg-segments (avoiding joints), cephalothorax, and abdomen were spot- painted with white nail polish to enhance reflectivity. Additionally, we placed a few reflective glass beads on each painted spot (45-63 µm diameters; Polytec Inc., P-RETRO Retroreflective Glass Beads) to further enhance reflectivity. Females were reintroduced to their webs and measurements were made only after recovery from anesthesia.

To excite vibrations in the web, a small neodymium disc magnet (0.25 mm diameter x 0.25 mm thickness, 14 mg mass) and was mounted on small strip of velcro (hooks side)and suspended on the web at a randomly selected position. The weight and size (total mass 20 mg) of this assembly resembled that of a male black widow spider, or small prey, producing the same local tension in the web. An electromagnet powered by a power amplifier (Brüel and Kjær, Type 2718) was used to drive this assembly. The electromagnet was held 1-3 cm above the web (Figure S4.1) and a 25 ms pulse was used to attract and then release the permanent magnet. The motion of the permanent magnet assembly was monitored using a single point fibre-optic vibrometer (Polytec OFV 511 with 3001 controller) positioned above the web and at a 45° angle to the xy plane (Figure S4.1). The signal was adjusted for a peak velocity of ~20 mm/s.

Vibration velocities (z direction) of different parts of the spider were measured using a scanning laser Doppler vibrometer (Polytec PSV 400, OFV 505 scanning head) via front-silvered mirrors

125 positioned at a 45° angle above the web (Figure S4.1). We monitored signal quality and included only high quality signals in our analysis (Figure S4.2). Vibrations were measured as velocity rather than displacement to account for kinetic energy dissipation (½mv2) within the system, whether translational or rotational (Figure S4.3). Time-domain measurements were collected through the Polytec Vibrometer software (Version 9.1.1) and A/D card (National Instruments PCI 6110, sampling rate 25.6 kHz). A total of 2.56 s of data averaged over 10 measurements was collected per scan point.

We completed full-body scans of 10 females in order to characterize vibrational behaviour of the whole spider. Vibrational modes were identified by mapping the transfer function between body and magnet vibrations for all measurement points. Peaks in transfer functions were identified, and modes determined from the spatial map. To study the effect of size, 5 females with below average body mass were chosen (Figures S4.6 & S4.7) and their vibrational behaviour was tested either by conducting scans over the complete body or along the leg 1 and abdomen transect depicted in figure 1. After this, a small metal washer (95 mg) was glued to the abdomen of each female, simulating the effect of a large meal (increase in mass). The vibrational behaviour of the female was then retested under these conditions. Weighted small females did not exceed the average mass for black widow females, and thus did not place an unnatural strain on the web. By beginning with females of a range of weights, we could also test a range of percent changes in female density. We also measured the vibrational behaviour of 6 females in two postures, the ‘neutral’ and the ‘crouch’ posture (Figure S4.9). In the crouch posture, however only three positions on the body were accessible to the LDV laser and comparisons between the modeled and real postures are made using these measurements.

4.3.2 Body Dynamics: Modelling

A simulation of transmission of vibrations within the spider body was developed using a multibody dynamics approach with the Simscape Multibody package (v4.9 Matlab R2016b, further details on the multi-body approach in Figure S4.4). This approach treats the spider’s body segments as rigid solids connected by joints which allow relative motion between the connected segments. Spider legs are partial hydrostats. Leg extension is achieved by increasing internal pressure; retraction is driven by muscles. Simple rotational-spring-mass-dashpot models, such as used by the MBD approach, have been previously applied to spider leg joints (Landkammer et

126 al., 2016; Zentner, 2013). In this system, the main rotational stiffness and damping are provided by muscle, and internal extension force by pressure (Landkammer et al., 2016; Zentner, 2013). Each leg segment (distal to proximal: tarsus, meta-tarsus, tibia, patella, femur, trochanter, and coxa), the cephalothorax and abdomen were treated as rigid bodies. The size and position of each rigid body in different postures was approximated from photographs. The density of spider cuticle is taken to be 1060 kg/m3 (Zentner, 2013). The abdominal mass of the model female was 464 mg which compares with the average mass of adult females (Johnson et al., 2011).

Based on previous studies of spider joint mechanics (Gasparetto et al., 2008; Landkammer et al., 2016), all inter-segmental joints within the leg were modeled as revolute joints. All joints rotate in the xz plane except the patella-tibia joint which rotates in the xy plane. The coxa- cephalothorax joint, and the joint between the cephalothorax and abdomen were modeled as a ball and socket joints (Gasparetto et al., 2008; Landkammer et al., 2016). For a first approximation, and to avoid overfitting the model, we assumed that all joints have the same stiffness, but frictional damping would depend on the surface area of the joint (Figure S4.4). We use the joint stiffness and damping from Phrixotrichus roseus (Zentner, 2013) as a starting point and vary these two parameters to fit vibrometry data (Figures 4.2 & S4.5). Implementing a joint stiffness of 30e-5 N*m/rad and damping of 1e-6 x segment radius N*m/(rad/s) replicated the mechanical behaviour of black widow females and is within the range suggested by data from other spiders (Barth, 2001; Zentner, 2013). An analysis of sensitivity to these parameter choices was made and the model outputs were found to be robust to the expected range of variation in these parameters (Figures S4.5 & S4.6).

To model different postures adopted by females, we defined the observed leg joint angles in those postures as equilibrium positions, and allowed the model to reach equilibrium before testing its vibrational behaviour. Such postural equilibria have been posited in Aplysia with evidence suggesting these are neuromechanical “set-points” achieved by coalitions of muscles (Chiel et al., 2009). Spider leg joints have limited ranges of motion, and show some deflection- dependent stiffness, largely with lateral deflection (Blickhan, 1986), against the bend of the joint. In axial rotation, as in our model, joints are very linear (Blickhan, 1986) and only change at very large deflections (>80°) (Barth, 2001). This stiffness change is likely due to recruitment of other tissues as additional spring elements (cuticle, or intersegmental contact). We have not included this change in the current model. Similarly, we implemented a module that includes the stiffness

127 and damping forces (Jönsson et al., 2005) during contact between the hind legs and abdomen, but these are not in play in simulations reported here.

We did not explicitly model the web. The local attachment of spider to web is a claw from which the spider is suspended, and which is actively controlled and repositioned by the female. This was modeled as a ball and socket joint with no stiffness or damping. We capture the local elastic behaviour of the silk itself by treating it as a prismatic joint which allows motion only in the z direction. Values for stiffness and damping of the web were taken from measurements of webs. The permanent magnet assembly was excited as described before, and an FFT of the displacement response was calculated at a frequency resolution of 0.4 Hz using a rectangular window. We tested 5 webs at 5 positions each, fitted a simple harmonic oscillator model to the response of the web, and calculated the local spring constant and damping of the web in the z direction. The webs had a mean spring constant of 0.31 ± 0.29 N/m (mean + SE, n=5 webs) and a mean damping constant of 0.06 ± 0.01 N/(m/s) (mean + SE, n=5 webs). These values are commensurate with the measured Young’s moduli of Latrodectus silk (Koski et al., 2013).

To simulate the forces experienced by the spider on the web, we applied forces only to the tips of the model spider legs. Two types of forces were used to excite the model to study its frequency response. The first is a force similar to that experienced by the real spider in the experiment (Figures 4.1 & 4.2). We took the derivative of the velocity waveform measured from the tarsus, to calculate its acceleration, which will be proportional to the local force. This waveform (French et al., 2002) was multiplied with a constant calibrated to the output velocity amplitude of the tarsus and applied to leg tips in the spider model. Both the waveforms and spectra of the model response were compared to the measured behaviour of a real spider (Figures 4.1 & 4.2). To study the complete vibration response of the spider’s body from the model, a more idealized force waveform is used, a force impulse of duration 0.1 ms and amplitude 5 N.

4.3.3 Body Dynamics: Modelling

We compared previously recorded courtship vibrations of male L. hesperus to vibrations generated by struggling prey on the web (using prey items that readily elicit attack by female widows). Specifically, we used male abdominal tremulation signals produced near the female, as these signals were the most distinct and conspicuous of their repertoire. Data were collected from 16 males, 5 crickets (Acheta domesticus), 5 black ants (Camponotus pennsylvanicus, collected on

128 campus) and 5 mealworms (Tenebrio molitor) released on webs. All prey were released ~7 cm away from the female. Web vibrations were measured near the females’ refuge in order to estimate the signal reaching the female after transmission through the web, rather than the local vibrations produced near the source. Vibration data were collected using a laser Doppler vibrometer (LDV; Polytec PDV 100; 20 mm/s peak measurement range; 0.5 Hz - 22.0 kHz frequency range; 24 bits). Vibration signals were recorded from a tiny reflective tape (~1mm2) that was placed carefully on web-silk near the entrance of females’ refuge, and stored on a data recorder (Sound Devices 722; 16 bits; 48.0 kHz sampling rate). For each of the prey item 1s of data were analyzed and for each male at least 5 sections of data of 1 second each. We calculated the power spectral density using Welch’s method at a resolution of 2 Hz for every section of data. For males, the average power spectrum for each male is presented along with the population average (Figure S4.10A); for the three prey species, individual data are presented with the population average (Figure S4.10B, C, & D).

4.4 Results 4.4.1 Frequency Segregation

A vibrational input was produced by suspending a permanent magnet (of similar mass to a male spider) on the web, and by driving it with an impulsive force generated by an electromagnet. The vibrations produced in the magnet assembly were periodic (Figure 4.1A) and included a range of frequencies (Figure 4.2A). Measurements were focused on the long foreleg (leg 1) and the abdomen since these were the easiest to measure. The resulting vibrations in the spider body were also periodic, but different body parts vibrated at different frequencies (Figure 4.1A). The vibrations observed in distal leg segments contained higher frequencies than those in the vibrations of proximal leg segments and the abdomen (Figure 4.1A). Spectral analysis showed that distal segments had higher vibration velocities above ~30 Hz, than proximal segments (Figure 4.2A).

A transfer function analysis was used to remove the effect of the vibration transmission through the web, and to concentrate on the biomechanics of the spider body. Vibrations are introduced into each leg of the spider at the point of contact with the web - the tip of the most distal segment, the tarsus. We therefore calculated a transfer function of the vibration of each segment with respect to that of the tarsal segment (Figure 4.2B). These leg tip transfer functions allow us

129 to estimate the decay of vibrations of different frequencies after they are introduced into the spider body and as they travel through the body. The transfer functions indicate that frequencies above ~30 Hz decay substantially as they transmit through the spider’s body. The largest decreases in high frequency vibration velocities occur at the joints between the metatarsus and the tibia, the tibia and the patella; and then between the patella and the femur (Figure 4.2B). Thus, at these higher frequencies, each distal segment moves more than the next proximal segment, suggesting that the joints between these segments are being bent. A similar pattern is evident for lower frequencies, but mainly between the femur, trochanter/coxa, and the abdomen (Figure 4.2B), indicating that these more proximal joints are bending at lower frequencies.

4.4.2 Modelling Whole-spider Vibrational Mechanics

We simulated vibration transmission in the spider body using multi-body dynamics (MBD) modelling. When driven by forces that mimic our experimental stimuli (see methods), the model shows similar vibrational behaviour to the real spider (Figure 4.1B). The waveform and vibration levels are similar in corresponding body segments. As observed in experimental measurements, more distal segments had higher vibration frequencies than more proximal segments (Figure 4.1B). Spectral analysis of the model shows a very close match to the behaviour of the real spider, both in levels and in the overall frequency behaviour (Figure 4.2A & C).

Transfer function analysis of the relative motion of leg and body segments also suggests that the model captures intersegmental vibrational behaviour (Figure 4.2B & D). Transfer functions predict both the decay of frequencies higher than ~30 Hz as they travel through the body of the spider, and an increase in motion at low frequencies. The model, like the real female, predicts bending at the distal joints, between the metatarsus, tibia, patella and the femur, at high frequencies. At low frequencies, it predicts bending at more proximal joints, between the femur, trochanter/coxa and abdomen.

Figure 4.1: Vibration transmission through spider (A) compared to the model (B). Vibration velocity waveforms for different body segments are presented (position indicated by coloured circle). In both spider and model distal leg segments show higher frequency motion than proximal segments or abdomen, which move primarily at low frequencies - high frequencies are dissipated more distally than are low frequencies. (Red traces show motion of the magnet assembly in spider measurements, and force applied to leg tips in the model, which approximates the real force on the spider legs.)

130

Figure 4.2: Vibration spectra of measured (A) and modeled (C) spiders, showing frequency segregation in vibration transmission through body segments. In both the real and model spider, all segments show similar levels of motion below 25 Hz and decreasing motion in proximal leg segments and abdomen at higher frequencies. Transfer functions of different body segments relative to the tarsus (location of vibration input) estimate the dissipation of different frequencies through the body in measured (B) and modeled (D) spiders.

131

Our model has one important free parameter, joint stiffness, which was tuned to match the vibrational behaviour of the different segments of leg 1 (see methods, Figure S4.5). Other features of the model, such as the geometry and density of body segments, are either from measurements or from the literature. This model captured the main features of the vibrational behaviour of each segment (Figures 4.1 & 4.2), including the intersegmental decay of frequencies along the leg (Figure 4.2). Our results also show that the observed segregation of low and high frequencies along the leg was not strongly dependent on the specific value of this or other parameters (Figure S4.5). Thus, we believe the MBD model developed here is sufficient for predicting the segregation of vibrations of different frequencies within the long foreleg of female black widow spiders.

There were some differences between the model and the real spider, but these can be explained by the variation in locations of vibration measurement between the model and the real spider, and the subtle variability in the posture of female spiders which cannot be matched perfectly in the model (see later sections, Figure S4.7).

4.4.3 Full Body Vibrational Mode

We further test the accuracy of the model, by considering its predictions over the entire spider body. The local peaks in the intersegmental transfer functions show two modes of vibration, one between 10 and 20 Hz and another between 40 and 60 Hz (Figure 4.2D). The model predicts a first mode where the abdomen and proximal segments of all legs move relative to the distal segments (Figure 4.3A) i.e. proximal joints undergo bending motions, whereas distal joints do not. The model also predicts a second mode where these relationships are reversed (Figure 4.3B), distal segments move more than proximal segments, and bending is experienced at the leg tips.

Scans that reconstruct the measured motion of the entire body of the spider confirm that these two modes can be observed in the expected frequency bands in real spiders (Figure 4.3C & D; frequency of mode 1: 15.62 ± 3.66 Hz, mode 2: 55.98 ± 11.45 Hz, mean ± SD, N=10). At low frequencies, we observe motion primarily near the abdomen, implying bending at proximal joints, and at higher frequencies, we observe motion at the leg tips only, implying bending at the distal joints. The relative amplitudes of the two modes as observed are similar to the model. In real measurements, however, the relative amplitudes can be highly variable (Figure 4.3D). Very

132 133 high amplitude motion is often observed at some leg tips. These high amplitude motions are likely the result of local variations in silk stiffness, or discontinuities in the heterogeneous structure of black widow web, which we do not incorporate in our model. Nonetheless, the model which has been tuned on the behaviour of the long foreleg captures reasonably well the vibrational behaviour of the complete spider body and therefore can be used to predict its mechanical behaviour in other contexts.

4.4.4 Body Size Effects

The majority of female mass is in the abdomen, which therefore represents the major inertial component governing the vibrational behaviour of the spider body. There is considerable variation in abdomen size and therefore mass among individuals; additionally the mass of the abdomen can undergo large changes after feeding but may also increase in density (Blackledge and Zevenbergen, 2007; Johnson et al., 2011). Indeed, in black widow spiders, leg length does not vary a great deal and most of the change in spider mass can be explained by changed in abdominal size and volume. We used the model to investigate the possibility that this individual variation could affect the frequency segregation we observed. First, we varied the size of the modeled abdomen which changed overall mass, then we changed the density causing additional changes in mass (see methods and Figure S4.4).

The model predicted only minor changes in frequency segregation due to variation in abdominal size and density (Figures 4.4A-D & S4.4). We tested these predictions by measuring frequency segregation in the bodies of real females of different masses (Figures 4.4C-D & S4.5; frequency (mode 1) = 13.47± 2.39 Hz, frequency (mode 2) = 49.14 ± 16.09 Hz, mean ± SD, N=5) and by simulating the effect of a large meal by adding a small mass (95 mg) to their abdomens externally (Figures 4.4E-F & S4.5; frequency (mode 1) = 13.23± 5.20 Hz, frequency (mode 2) = 45.47± 13.89 Hz, mean ± SD, N=5). The vibrational behaviour of the females corroborated the predictions of the model and did not vary greatly with initial abdominal size or change after a mass was added (Figures 4.4G-H & S4.5; paired t-test for frequency (mode 1): P=0.44; paired t- test for frequency (mode 2) = 0.17, N=5).

Figure 4.3: Full body vibration modes predicted by model (A&B) and measured in real spiders (C&D). The model predicts two main modes, which were also observed in vibration measurements from real spiders. Mode 1 (A-predicted, C-observed) is characterized by high amplitudes proximally (near the abdomen), while distal points remain relatively motionless. Mode 2 (B-predicted, D-observed) is the reverse, with higher amplitudes distally and relatively motionless abdomen. Modeled velocity amplitudes of body segments (A&B) are depicted in 3D schematic showing maximum deflection relative to resting, with colour-coded amplitude level. Measured vibration amplitudes are depicted by surface plots of the transfer function between each measurement location and the input velocity at the vibrating magnet (source).

134

Figure 4.4: Abdominal size and density have little effect on the spatial segregation of vibration frequencies. (A-D) The model output is not significantly changed even with abdominal mass is as low as (A) 20% and (B) 40% of the average mass (changing abdominal size alone in the model). Similarly an increase in the density of the abdomen, such as after a large meal, does not significantly change vibrational behaviour; (C) increase of 100% and (D) increase of 54%. Model predictions were verified with real animals, of similar weights, (E) 22% and (F) 37% showing similar vibrational behaviour. Weighting the abdomens to simulate a large meal with a mass increase of (G) 90%, or (H) 54% also did not change vibrational behaviour greatly, and in fact in one of these cases (F,H) increased rather than decreased the modal frequency suggesting that body mass was not the principle determinant of the change observed here.

135

Thus, the vibrational dynamics of the black-widow spider body are robust to natural variation in body size. This might underlie their ability to correctly segregate and discriminate between incoming vibrational signals as has been suggested before by their ability to recognize courting males despite changes in feeding status (Andrade, 1998; Vibert et al., 2014).

4.4.5 Posture Effects

Posture is another obvious variation in the spider body that might affect the mechanics of vibration. In general, female black-widows, like most web-dwelling spiders, are sit-and-wait predators (Elias and Mason, 2011; Hódar and Sánchez-Piñero, 2002) and remain motionless on the web for extended periods. Nevertheless, their posture on the web is variable, and three distinct postures are commonly observed: the “neutral posture” (legs extended, body horizontal); the “lowered-abdomen posture” (legs extended, body angled away from the web in the antero- posterior direction); and the “crouching posture” (legs retracted) (Figure 4.5). Neutral posture is the most commonly observed. Lowered-abdomen is commonly observed during courtship when it indicates receptivity to a male (M. Andrade, personal communication). The crouching posture is usually adopted in the refuge part of the web and is more usual when the female is hungry or after a large vibrational disturbance on the web, (e.g. wind). In each of these postures, the abdomen, the inertial element of the spider body, is supported by legs held at different angles, which results in different leg spans on the web and different distances from the web surface (Figure 4.5).

We used the model to investigate the effect of posture on vibration transmission and segregation through the female body. In verifying the model, we used transfer functions to infer bending at leg joints with the female in neutral posture. We repeated these calculations for the other postures (Figure S4.8). After running the simulations, we experimentally tested the model’s prediction in the crouch posture (Figure S4.9). As in neutral posture, there is a reasonable match between transfer functions predicted by the model and those observed from real animals (Figure S4.9). However, since it is joint bending that drives slit sensilla compression and expansion (Blickhan and Barth, 1985; Landkammer et al., 2016; Schaber et al., 2012; Speck and Barth, 1982), the main quantity of interest is joint bending, which we can obtain directly only from the models. The model thus enables us to directly calculate bending velocity spectra and enables an estimation of the mechanical sensitivity of leg joints (Figure 4.5).

136 137

The most striking conclusion from joint sensitivity spectra is that most spider leg joints are tuned. Interestingly, this tuning is not seen in peripheral neurophysiological measurements, which in most spiders has been observed to be flat below ~100-1000 Hz (Barth and Geethabali, 1982; Blickhan and Barth, 1985). This suggests that, although the receptors can be broadly tuned, the mechanical input to the joints shows frequency tuning. Therefore, the stimuli reaching the receptors are transformed by the body mechanics of the spider, and body mechanics can play an important role during frequency discrimination.

Other insights arise from this ‘embodied’ approach to vibrational sensing, especially concerning the effects of posture on joint tuning. We next considered the pattern of sensitivity at each joint as determined by bending amplitude and frequency tuning.

Mechanical joint sensitivity:

The tarsus-metatarsus joint is well known to be important in vibration sensing in spiders (Hergenröder and Barth, 1983) since it is not bent significantly during locomotion, unlike the tibia-metatarsus or femur-patella joints. In all three postures, this joint shows very low tuning and has a nearly flat bending spectrum. However, it also experiences the largest amplitude bending velocities, and has the highest mechanical vibration sensitivity. Among the four leg pairs, the highest sensitivity is surprisingly observed in leg 3 in all postures (Figure 4.5C-F), rather than the long fore-leg (leg 1) which is often used by the female in exploratory behaviour. Nonetheless, the tarsus-metatarsus joint also tends to be the joint with the highest overall bending velocity in all legs in the neutral and lowered abdomen posture. The bending amplitude at each leg differs, which suggests that the legs may achieve a degree of range fractionation. Our results suggest that the tarsus-metatarsus joint may be specialized for detailed sensing of small amplitude web-vibrations in the context of prey capture and courtship but also for faithfully representing the full frequency range of a signal.

Figure 4.5: Posture affects joint tuning. (A) Legs are numbered from the front to the back. (B) Joints are indicated by the two segments they connect (distal-proximal). Bending plane is indicated on the figure. Lyriform organs are stimulated by joint bending. (C-F) Modeled amplitude spectra for bending velocity at all joints in different postures (left): (C) neutral posture, (D) lowered-abdomen posture (E) crouch posture, and (F) a hybrid posture with right foreleg extended from the crouch posture. In symmetric postures only data from the left legs are presented. For the crouch posture data from both sides are presented. The grey shading in the spectra indicates the frequency band of male signalling (Figure S4.10).

138

Mechanical joint tuning:

All other joints are tuned, either uni- or bimodally in the frequency range of interest. As expected from studying intersegmental transfer functions (Figure 4.2) and the full body vibration patterns, we found that in the neutral posture, the modal peaks occur either at ~20 Hz or ~60 Hz. The low frequency mode causes some bending at almost all joints, whereas the high frequency mode is only observed at the distal joints. The highest tuned motion occurs at the distal tibia-metatarsus joint, and the proximal trochanter-femur joint. The frequencies that the joints are tuned to, in this posture, form a large component of male courtship signals (Figures 4.5 & S4.10). In the lowered- abdomen posture, usually adopted during courtship, tuning remains similar, and is marked by increased mechanical sensitivity at the low frequency mode in legs 1 and 2, suggesting increased ‘attention’ to low frequency portion of male signals.

Male courtship frequency range overlaps with that of large prey in the low frequency end of the spectrum (Figure S4.10, (Landolfa and Barth, 1996)) suggesting that both these postures allow females to simultaneously pay attention to prey and male signals. Some prey signals, particularly of large prey items such as crickets include frequencies even lower than this range and it is likely these signals are sensed mainly in the metatarsus-tarsus joint.

Crouch postures:

The most dramatic changes in mechanical sensitivity spectra occur in the crouch posture. The crouch posture reduces the leg span of the spider on the web and brings the abdomen closer to the attachments on the web. This has the dual effect of severely reducing the mechanical sensitivity of most tarsus-metatarsus joints and shifting joint tuning upwards, the first peak now appearing at ~50 Hz compared to ~18 Hz in the neutral posture. In the crouch posture, tuning moves to a range that overlaps the vibration frequencies usually produced by much smaller prey such as ants (Figures 4.5E & S4.10). This suggests a possible mechanism whereby a hungry female could “tune” her peripheral vibration sensors to be more attentive to smaller rather than larger prey. More crucially, given the reduced mechanical sensitivity to all signals between 10 and 50Hz, a female in this posture is insensitive to the majority of male courtship signals, and may either ignore them or misidentify them. Thus there might be a sensory deficit only in this posture which may contribute to the observed increase in cannibalism in hungry females (Andrade, 1998; Johnson et al., 2011). Additionally, in this posture, the female would not sense

139 140 the very low frequency vibrations caused by environmental disturbances such as wind, potentially preventing receptor habituation and preserving sensitivity to other signals in other frequency ranges. When tested in the transfer function domain, we find that the predictions of the model are well-matched by measurements from real females in neutral and crouched postures (Figure S4.9, N=6).

Hybrid postures:

Given the highly irregular structure of a cobweb, females are rarely as symmetrically positioned as our models. Additionally, females in one posture are also known to partially adopt another posture, such as extending a single leg from a crouch. So next, we examined whether such a hybrid posture could combine the mechanical sensitivity of each individual state simultaneously. A single extended leg, for example, could allow vibrations to be detected more sensitively from a specific part of the web. An analysis of joint bending spectra bears out this intuition. The extended leg, which alone has a ‘neutral’ leg-extended posture with respect to the abdominal mass, regains some of the mechanical sensitivity it had in that posture (Figure 4.5). The main changes are seen in the bending amplitude of the first two joints in the extended leg: the mechanical sensitivity of the tarsus-metatarsus joint increases and the frequency tuning of the tibia-metatarsus joint lowers (Figure 4.5). The other legs and their joints, on the other hand, largely retain the tuning of the ‘crouch’ posture. This shows that both abdominal mass and leg position affect vibrational input to slit sensillae and that at least for the long front legs, their sensitivity may be configured somewhat independently of the rest of the body. This is consistent with the way the spiders seem to use these legs in a sensory role.

4.5 Discussion 4.5.1 Simplified Body and Web Mechanics

In the model presented here, we have only considered the interactions between segments that are directly articulated by a joint and made the simplifying assumption that joint stiffness is uniform throughout the body. While we have tested the sensitivity of the model to some of these parameters, and found it to be relatively robust (Figures S4.5 & S4.6), we know that this does not fully describe the joints of spiders where stiffness can change with posture (Barth, 2001) and viscoelastic properties affect the vibration transmission through the leg (McConney et al., 2007).

141

Additionally, the assumption of revolute joints may be an over-simplification (Landkammer et al., 2016), and there may be other intersegmental interactions that occur due to contact between disconnected body parts which we have not considered.

Spiders in general (Japyassú and Laland, 2017), and black-widow spiders in particular (Blackledge and Zevenbergen, 2007; Zevenbergen et al., 2008), are known to change the structure of their webs and silk in response to their condition and sexual status. This has been suggested to be a form of extended perception rather than the embodied perception that we have presented in our work (Japyassú and Laland, 2017; Kaplan, 2012). In effect, we have taken a simplified approach to studying spider body mechanics, and have used internal transfer functions to isolate the effect of the body from that of the web. Despite the simplifications, the model we have developed captures reasonably well the overall vibrational behaviour observed from real spiders. The model also predicts significant postural effects, which we have tested against measurement from real spiders in different postures and found to be reasonable (Figure S4.9). Thus, we suggest that web structure and joint stiffness may permit the spider a degree of additional control over perceptual processes and the postural effects which we identify here would serve as conservative estimates.

4.5.2 Multifunctional Sensors

Many organisms control proprioceptive feedback via descending neuronal systems which typically refine the output of locomotor behaviour (Tuthill and Azim, 2018). In our work here, we attempt to close the sensorimotor loop and show how sensory input can be modulated through locomotor and behavioural mechanisms. Our results demonstrate how both the levels and tuning of the mechanical inputs to spider lyriform organs can be reconfigured simply by changing posture. Reconfigurable mechanical tuning may reflect a need for slit receptors (i.e. lyriform organs) to precondition the stimulus for the nervous system. Control over the mechanical input to the nervous system would allow sensors to be multifunctional; i.e. function in different behavioural contexts, such as locomotion, predation, and mate perception (Chiel et al., 2009), each requiring different amplitude and frequency sensitivities. Locomotion generates larger leg bending motions than web vibrations would and at much lower frequencies (Blickhan and Barth, 1985). It is very likely that lyriform organs are activated during locomotion, and may be used to provide feedback and proprioceptive control as they are in the large wandering spider,

142

Cupiennius salei (Barth, 2001; Blickhan and Barth, 1985; Seyfarth, 1978). Similarly, prey caught in the web are likely to make large struggling movements and produce high amplitude vibrations (Vibert et al., 2014) at a range of frequencies, likely depending on prey size (Figure S4.10). The lowest amplitude vibrations are likely male signals (Vibert et al., 2014), which may contain frequency-dependent information crucial to determining male identity and quality. Our results suggest that female posture affects both joint bending amplitude and frequency tuning, and this property may be deployed by females in different behavioural contexts. Posture could thus be thought of as a saliency filter, a form of selective attention (Nityananda, 2016), that operates in the space of signal characteristics rather than physical space itself. It would be interesting in the future to examine the mechanics and peripheral tuning of the posture adopted by females in each of these contexts in greater detail.

4.5.3 Sensory Complexity and Intentionality

Using a dynamic systems approach it has been suggested that each posture in an animal’s repertoire is a state of neuro-mechanical equilibrium, and that animals can transition from one posture to another depending on the sensory feedback they receive (Chiel et al., 2009). Different postures may also be viewed as points from which locomotory programs may be initiated, with some being more likely than others. An interesting possibility from this kind of description is that static postures can be thought of as equilibria. Some equilibria will be closer to each other than others in “posture space”. In the black widow system, for instance, joint tuning suggests that the lowered-abdomen posture allows the female to be more sensitive to male signals. Therefore the dynamics of bodies may have evolved such that they can transition preferentially between relevant postures i.e. equilibria (Chiel et al., 2009). For instance, the lowered-abdomen posture might both allow the female to perceive the male better, but also allow her to transition into the mating posture. Similarly, while the crouch posture allows the female to perceive even small prey items, it might also allow her to transition into an attacking posture more easily.

Additionally, the transitional motions between equilibria can also be mapped as existing in joint angle space. Since transitional postures will be small continuous changes in joint positions, which define their position in this multi-dimensional space, they are expected to lie on a continuous but bounded surface within this space, i.e. on a manifold. Postural manifolds are routinely used in machine vision to identify postures visualized from different angles, and in

143 robotics, to plan the actions in a multi-limbed and jointed body (Datta et al., 2009; Gu, 2000; Lee and Elgammal, 2007). Recently some evidence has been found for neural manifolds controlling motion in biological systems (Gallego et al., 2017). Our data suggest that posture can profoundly affect the sensory system of an animal, and that transitional states between postures take on transitional sensory properties. Thus, just as one can posit a postural manifold, one can consider a sensory manifold for in a spider’s body. The body, with varying posture, behaves as a sensor with dynamic properties that preconditions vibrational input to the many joints in the spider body. Therefore, interpreting the amplitude and spectral information in incoming sensory data should pose a significant challenge requiring feedback and mapping between the postural and sensory manifolds. The more complex the mechanics of the physical body, in terms of number of joints and sensors producing sensory inputs and their possible sensory states, the more complex the task of interpreting incoming sensory data is likely to be. This tight loop of interaction between perception and behaviour required by such mapping may well be what explains the surprising cognitive complexity observed in spiders (Cross and Jackson, 2017, 2005; Japyassú and Laland, 2017).

4.6 Authors’ Contribution Statement

Natasha Mhatre and Senthurran Sivalinghem collected data and analyzed results. Natasha Mhatre carried out the modeling. All authors contributed to the design of the experiments and the writing of the manuscript.

4.7 Acknowledgements

Natasha Mhatre acknowledges a Wissenschaftskolleg zu Berlin fellowship in developing expertise required to use the modelling tools used here. We thank Ed Yong for commentary on a preprint of this paper suggesting the distinction between spatial and informational selective attention.

4.8 References

Alva, N., 2009. Out of our heads: Why you are not your brain, and other lessons from the biology of consciousness. Macmillan.

Andrade, M.C.B., 1998. Female hunger can explain variation in cannibalistic behavior despite male sacrifice in redback spiders. Behavioral Ecology 9, 33–42.

Barth, F.G., 2004. Spider mechanoreceptors. Current Opinion in Neurobiology 14, 415–422.

Barth, F.G., 2001. A spider’s world: Senses and Behaviour. Springer-Verlag, Berlin Heidelberg.

Barth, F.G., Geethabali, 1982. Spider vibration receptors: Threshold curves of individual slits in the metatarsal lyriform organ. Journal of Comparative Physiology A 148, 175–185.

Blackledge, T.A., Zevenbergen, J.M., 2007. Condition-dependent spider web architecture in the western black widow, Latrodectus hesperus. Animal Behaviour 73, 855–864.

Blickhan, R., 1986. Stiffness of an arthropod leg joint. Journal of Biomechanics 19, 375–384.

Blickhan, R., Barth, F.G., 1985a. Strains in the exoskeleton of spiders. Journal of Comparative Physiology A 157, 115–147. https://doi.org/10.1007/BF00611101

Chiel, H.J., Ting, L.H., Ekeberg, O., Hartmann, M.J.Z., 2009. The brain in its body: Motor control and sensing in a biomechanical context. Journal of Neuroscience 29, 12807– 12814.

Cross, F.R., Jackson, R.R., 2017. Representation of different exact numbers of prey by a spider-eating predator. Interface Focus 7, 1–16.

Cross, F.R., Jackson, R.R., 2005. Spider heuristics. Behavioural Processes 69, 125–127.

Datta, A., Sheikh, Y., Kanade, T., 2009. Modeling the product manifold of posture and motion. 2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops 2009 1034–1041.

Eberhard, W.G., 1990. Function and phylogeny of spider webs. Annual Review of Ecology and Systematics 21, 341–372.

Elias, D.O., Mason, A.C., 2011. Signaling in variable environments: substrate-borne signaling mechanisms and communication behavior in spiders, in: The Use of Vibrations in Communication: Properties, Mechanisms and Function across Taxa. Kerala: Research Signpost.

French, A.S., Torkkeli, P.H., Seyfarth, E.A., 2002. From stress and strain to spikes: Mechanotransduction in spider slit sensilla. Journal of Comparative Physiology A 188, 739–752.

144 145

Gallego, J.Á., Perich, M.G., Miller, L.E., Solla, S.A., 2017. Neural manifolds for the control of movement. Neuron 94, 978–984.

Gasparetto, A., Vidoni, R., Seidl, T., 2008. Kinematic study of the spider system in a biomimetic perspective. 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 3077–3082.

Gu, E.Y.L., 2000. Configuration manifolds and their applications to robot dynamic modeling and control. IEEE Transactions on Robotics and Automation 16, 517–527.

Hergenröder, R., Barth, F.G., 1983. The release of attack and escape behavior by vibratory stimuli in a wandering spider (Cupiennim salei Keys.). Journal of Comparative Physiology A 152, 347–359.

Hódar, J. A., Sánchez-Piñero, F., 2002. Feeding habits of the black widow spider Latrodectus lilianae (Araneae: Theridiidae) in an arid zone of south-east Spain. Journal of Zoology 257, 101–109.

Hößl, B., Böhm, H.J., Rammerstorfer, F.G., Barth, F.G., 2007. Finite element modeling of arachnid slit sensilla - I. The mechanical significance of different slit arrays. Journal of Comparative Physiology A 193, 445–459.

Japyassú, H.F., Laland, K.N., 2017. Extended spider cognition. Animal Cognition.

Johansmann, M., Siegmund, G., Pineda, M., 2005. Targeting the limits of laser Doppler vibrometry. Proc. IDEMA 1–12.

Johnson, J.C., Trubl, P., Blackmore, V., Miles, L., 2011. Male black widows court well-fed females more than starved females: Silken cues indicate sexual cannibalism risk. Animal Behaviour 82, 383–390.

Jönsson, A, Bathelt, J., Broman, G., 2005. Implications of modelling one-dimensional impact by using a spring and damper element. Proceedings of the Institution of Mechanical Engineers, Part K: Journal of Multi-body Dynamics 219, 299–305.

Kaplan, D.M., 2012. How to demarcate the boundaries of cognition. Biology and Philosophy 27, 545–570.

Keijzer, F.A., 2017. Evolutionary convergence and biologically embodied cognition. Interface Focus 7, 20160123.

Koski, K.J., Akhenblit, P., McKiernan, K., Yarger, J.L., 2013. Non-invasive determination of the complete elastic moduli of spider silks. Nature Materials 12, 262–267.

146

Landkammer, S., Winter, F., Schneider, D., Hornfeck, R., 2016. Biomimetic spider leg Joints: A review from biomechanical research to compliant robotic actuators. Robotics 5, 15.

Landolfa, M.A., Barth, F.G., 1996. Vibrations in the orb web of the spider Nephila clavipes: cues for discrimination and orientation. Journal of Comparative Physiology A 179, 493– 508.

Lee, C.S., Elgammal, A., 2007. Modeling view and posture manifolds for tracking. Proceedings of the IEEE International Conference on Computer Vision 1–8.

Liesenfeld, F., 1956. Untersuchungen am netz und uber den erschutterungssin on Zygiella x- notata (Cl)*(Araneidae). Zeitschrift fur Vergleichende Physiologie 38, 563–592.

Liesenfeld, F.J., 1961. Über leistung und sitz des erschütterungssinnes von netzspinnen. Biol Zentralbl 80, 465–475.

Masters, W.M., Markl, H., 1981. Vibration signal transmission in spider orb webs. Science 213, 363–365.

McConney, M.E., Schaber, C.F., Julian, M.D., Barth, F.G., Tsukruk, V. V., 2007. Viscoelastic nanoscale properties of cuticle contribute to the high-pass properties of spider vibration receptor (Cupiennius salei Keys). Journal of the Royal Society Interface 4, 1135–43.

Mortimer, B., Soler, A., Siviour, C.R., Zaera, R., Vollrath, F., 2016. Tuning the instrument: sonic properties in the spider’s web. Journal of the Royal Society Interface 13, 20160341.

Nityananda, V., 2016. Attention-like processes in insects. Proceedings Biological sciences / The Royal Society 283, 20161986.

Schaber, C.F., Gorb, S.N., Barth, F.G., 2012. Force transformation in spider strain sensors: white light interferometry. Journal of the Royal Society Interface 9, 1254–1264.

Seyfarth, E.A., 1978. Lyriform slit sense organs and muscle reflexes in the spider leg. Journal of Comparative Physiology A 125, 45–57.

Speck, J., Barth, F.G., 1982. Vibration sensitivity of pretarsal slit sensilla in the spider leg. Journal of Comparative Physiology A 148, 187–194.

Speck-Hergenröder, J., Barth, F.G., 1987. Tuning of vibration sensitive neurons in the central nervous system of a wandering spider, Cupiennius salei Keys. Journal of Comparative Physiology A 160, 467–475.

Tso, I.-M., Wu, H.-C., Hwang, I.-R., 2005. Giant wood spider Nephila pilipes alters silk protein in response to prey variation. Journal of Experimental Biology 208, 1053–1061.

Tuthill, J.C., Azim, E., 2018. Proprioception. Current Biology 28, R194–R203.

147

Vibert, S., Scott, C., Gries, G., 2014. A meal or a male: the “whispers” of black widow males do not trigger a predatory response in females. Frontiers in Zoology 11, 4.

Walcott, C., 1963. The effect of the web on vibration sensitivity in the spider, Achaearanea Tepidariorum (Koch). Journal of Experimental Biology 40, 595–611.

Zentner, L., 2013. Modelling and application of the hydraulic spider leg mechanism, in: Nentwig, W. (Ed.), Spider Ecophysiology. Springer Berlin Heidelberg, pp. 451–462.

4.9 Supplementary Materials

Figure S4.1: A schematic of the experimental setup. The drawing is not to scale.

148 149

Figure S4.2: Signal quality during vibrometry. Both spider cuticle and web silk have low reflectivity and it is crucial to monitor signal quality during vibrometry in order to get reliable results.

For this reason, we made our measurements in the time rather than in the FFT domain. We maintained an averaged waveform and inspected the waveform for smoothness and to detect noisy transient signals. It is worth noting that where the reflectance of the vibrating surface is low, the noise floor increases considerably (see 1st panel) and the noise may in fact incorrectly show ‘vibration’ levels higher than the real signals. If only an FFT analysis is carried out, without an inspection of the raw waveforms, the spectra would simply be of the noise, which would not be classified as such without a coherence analysis.

To ensure the reliability of our signals, we monitored the signal quality measured by the scanning LDV at all times using the inbuilt quality channel which reflects the level of incoming reflected laser signal. We found that when the quality signal remained above 1V on average, the velocity outputs had smooth and continuous waveforms and did not show noisy transients. Signals below this quality level were deemed unreliable and not used in analysis.

150

Figure S4.3: The relationship between angular velocity and displacement is predictable and depicted here. Velocity for the same level of displacement increases linearly with angular frequency, i.e. V(w)= w*D(w). We use angular velocity since it enables a better understanding of where kinetic energy is dissipated in the body of the spider. (Kinetic energy varies with velocity squared).

151

Figure S4.4: Multibody dynamics (MBD) is based on simple Newtonian classical mechanics, with contributions from Euler which describe the dynamical behaviour of rigid bodies. In the early 20th century, there was an interest in understanding the dynamics of complicated assemblies with multiple moving parts like gyroscopic systems and biomechanical systems like human locomotion. Multibody systems dynamics, which was developed as a response to this need, is a mathematical formalism to describe such systems that is easily coded into a computer algorithm. MBD can describe the dynamic mechanical behaviour of any mechanical system made of rigid bodies connected by ideal joints in any arbitrary topology.

For each rigid body in an MBD assembly, Newton-Euler equations can describe the complete translational and rotational dynamics of the body and are shown above (Schiehlen, 1990). The inertia of each body is represented by mi and the tensor Ii orients the inertia with respect to the centre of mass of each body. All forces and torques on the body, such as those caused by external forces applied to the full assembly, and those due to the weight of different components, the e springs, dampers, supports and joint reactions are all included in the applied force vector fi and e torque vector li . Torques are with respect to the centre of mass. The full formalism generates comprehensive equations for the full assembly, known as the Kane equations (further details in Schiehlen, 1997) and these are implemented in computer algorithms such as Matlab Simscape multibody which we use. These models can be used to describe the motion of mechanical assemblies, including biological systems ranging from jointed organisms to molecules (Schmidt et al 2008).

In the mechanical assembly we have constructed here to describe the black widow spider body, the only elements are rigid bodies with either (A) revolute or (B) ball and socket joints. Each rigid body has a position in space, and one or more positions where it is connected by a joint to another rigid body or to the fixed ‘ceiling’. Since each body and joint has a position, each joint

152 effectively has a starting angle. Each joint is fully described by a spring constant and a damping coefficient. No further parameters are needed.

In our model, to reduce the number of required assumptions, we use a single spring constant across the assembly (unless otherwise indicated as in the parameter sensitivity analyses); the damping coefficient is proportional to the body segment diameter. (C) The positions of the joint type used are shown on one leg of the spider above; revolute joints are shown by a semi-circle and the plane of rotation given. Ball and socket joints are indicated by a cartoon hemi-sphere; stiffness and damping are equal in all directions.

While, no body is fully rigid in all force regimes and no joint is perfectly ideal, this method is an excellent first approximation. Particularly because in the force regimes considered here the spider body undergoes translational and rotational vibrational motions of very low magnitude. At this level of motion and the bending at the joints is expected to be linear and body segments are expected to be rigid.

References:

Schiehlen, W. O., 1990. Multibody systems handbook. Berlin: Springer-Verlag.

Schmidt, J. P., Delp, S. L., Sherman, M. A., Taylor, C. A., Pande, V. S., Altman, R. B., 2008. The simbios national center: Systems biology in motion. Proceedings of the IEEE 96(8), 1266-1280.

153

Figure S4.5: Model stiffness sensitivity analyses. The model was tested for its sensitivity to the stiffness parameters used. (A) We have used the same stiffness for all model joints and the main effect of the decreasing or increasing the joint stiffness (JtK) is a frequency shift in the transfer function spectra pattern. We test the model at stiffnesses ranging from half the best-fit stiffness to twice the best-fit stiffness. Lower stiffnesses lead to a lower peak in the transfer function and higher stiffnesses lead to a higher peak. (B) We also tested the effect of variation in joint stiffness and the results of 10 runs are plotted here. We found that joint stiffness variation had relatively little effect on the overall pattern of the transfer-function spectra. The stiffness of each joint in the model was selected from a distribution with a mean of 30e-5 N.m/rad and a standard deviation of 15% i.e. 4.5e-5 N.m/rad (inset). (C) We also tested the effect of variation of silk stiffness on transfer function spectra. The spring constant of the all attachments to web-silk (SilkK) were decreased and increased by 1 standard deviation and we found the effects on intersegmental transfer functions to be negligible. (Abd: abdomen; Cx: coxa; Trc: trochanter; Fem: femur; Pat: patella; Tib: tibia; Met: metatarsus; Tar: tarsus).

154

Figure S4.6: Body size sensitivity analyses. (A, B) The size and mass of a female spider can and does change in many ways. (C) One possibility is that different spiders are simply allometrically smaller and larger than each other, i.e. their spatial dimensions change by the same factor. We tested the effect of allometric size and found that larger spiders had lower peaks in their transfer function spectra but the change was small despite large changes in size and effective mass. In real black widow females however, leg length does not change a great deal (<1%) even when large changes in mass (50-200% average mass) are observed [23], implying that the change in mass occurs either through a change in abdominal size or density. (D) We varied the size of the abdomen in the model to mimic the body size variation observed in real females and found effects similar to that described before, larger females had lower frequency peaks than smaller females, however, the absolute change in frequency segregation in the spider body was small. (E) We also tested females with abdomens of different densities and observed similar model behaviour. This shows that variations in other parameters such as mean joint stiffness or changes caused by posture exceeded any variation that may be caused by size. (Abd: abdomen; Cx: coxa; Trc: trochanter; Fem: femur; Pat: patella; Tib: tibia; Met: metatarsus; Tar: tarsus).

155

Figure S4.7: Body size variation has little effect on frequency segregation in the spider body. The model predicts that, despite large changes in abdomen mass, little change will be observed in the frequency segregation behaviour of the spider body, as indicated by a transfer function between the tarsus and other segments. We tested females of a range of body sizes and found that the peak transfer frequency variations did not change systematically with size. The first row of transfer functions are from 5 different females of increasing mass. The second row of transfer functions are from the same females after a large mass (95 mg) was added to their abdomen. Frequency segregation behaviour does not change systematically in either row of transfer functions. A lack of systemic change in frequency segregation indicates that rather than mass, a different parameter explains the variations we observe in vibrational behaviour. We believe that both variations in average joint stiffness, and changes caused by posture exceed any change that may be caused by size.

156

Figure S4.8: Posture has a strong effect on frequency segregation in the spider body. The segregation of frequencies in the spider body is different in the four postures investigated (A) the normal leg-extended posture, (B) the lowered-abdomen posture, (C) the crouch and (D) the hybrid crouch posture with a single extended foreleg. The main difference between the first two postures is an increased motion in proximal segments to low frequency vibrations. The main response in the crouch posture is a much lowered response to low frequencies across all legs. Differences between the hybrid and the complete crouch position can be observed but their effect is clearest in the joint bending spectra. Some of the transfer functions measured from real females resemble the hybrid transfer functions more closely than those of the symmetric leg- extended posture which seems reasonable considering that completely symmetrical postures would be difficult to achieve on the web which provides a very sparse and irregular substrate. (Abd: abdomen; Cx: coxa; Trc: trochanter; Fem: femur; Pat: patella; Tib: tibia; Met: metatarsus; Tar: tarsus).

157

Figure S4.9: We tested model predictions by measuring six females in both the neutral and crouched posture. In crouched posture, legs are retracted close to the body. In this posture, we could not reliably measure from leg segments, only from leg tips, coxa and abdomen. We compare the predictions of the model for transfer functions between leg tips, coxae and abdomen with measured transfer functions. (A) The model predicts that at frequencies above ~27 Hz, coxal motion is attenuated in the neutral posture compared to the crouched posture. (B) The effect at the abdomen is similar but of smaller magnitude. We find that the measured behaviour of the (C) coxa and (D) the abdomen above ~27 Hz are similar to that predicted by the model. The lines indicate the means from 6 females and the shaded region depicts the standard deviation around the mean. We make a statistical comparison by summing the transfer function in frequency range 27-200 Hz and find the measured transfer function ratio to be higher in the crouch posture (coxa: 1800±979, N=6; abdomen=1230±890, N=6) compared to the neutral posture (coxa: 829±539, N=6; abdomen=720±521, N=6; paired t-test: coxa: P=0.065; abdomen: P=0.066).

Figure S4.10: Male L. hesperus courtship and prey vibration power density spectra. (A) The courtship signals of males, who weigh 14 mg on average, mainly occupy the band between 23 and 61 Hz. The signals of certain potential prey items, which may differ in size such as (B) ants, (C) crickets (Acheta domesticus) and (D) mealworms (Tenebrio malitor) appear to reflect their mass, smaller prey items having higher frequencies and larger prey items having lower frequencies. In the species tested here, signals largely remain outside the frequency band occupied by males.

158

A pre-print version of this study was published in BioRxiv (Mhatre et al., 2018), and is reprinted here.

Mhatre, N., Sivalinghem, S., Mason, A.C., 2018. Posture controls mechanical tuning in the black widow spider mechanosensory system. BioRxiv p, 484238.

159 160

Chapter 5 General Discussion

Very little is known about vibratory communication in web-building spiders (Herberstein et al., 2014). Despite numerous observations of courtship vibratory behaviours in many web-building spiders (Herberstein et al., 2014; Uhl and Elias, 2011), only four studies have quantitatively examined the web-borne vibration signals involved (Singer et al., 2000; Suter and Renkes, 1984; Vibert et al., 2014; Wignall and Herberstein, 2013). This oversight leaves a significant knowledge gap in our understanding of this mode of communication in spiders. In the genus Latrodectus, males across different species have been consistently documented to display various vibratory behaviours during courtship and vibratory signaling may be common in this genus (Chapter 1: Table 1.2). The overarching aim of my dissertation was to use an integrative approach to investigate aspects of vibratory communication in the western black widow spider (Latrodectus hesperus). Using L. hesperus as my focal organism, I specifically addressed four main research questions: 1) Do the different vibratory behaviours correspond to characteristically distinct or similar signals? 2) How are males generating different signals on female webs? 3) Do courtship vibration signals contain any information about males, and are vibrations signals important for male mating success? 4) What happens to the vibrations transmitted along female legs, before being detected by vibration sense organs? I conducted a series of experiments, as well as developed novel techniques, to quantitatively examine these facets of vibratory communication, from sender to receiver, to tackle some of the knowledge gaps. Below I discuss the implications of my research findings in relation to the four main questions addressed.

5.1 Research Implications and Future Directions 5.1.1 Males Produce Multiple Vibration Signals

Male L. hesperus produced three characteristically distinct web-borne vibration signals during courtship and copulation. Given that similar vibratory behaviours were also observed in other Latrodectus species (Chapter 1: Table 1.2), it is likely that other species within this genus also communicate using multiple vibratory signals. The three signals characterized in this thesis for L. hesperus males may only be part of a larger signaling repertoire, as courting males consistently engaged in wrapping females with his own silk (chemical signals; multimodal signaling), and

161 various tactile behaviours that can also generate vibrations (e.g. leg tapping and palpal boxing/drumming) (Ross and Smith, 1979). Future studies need to characterize these signals to better understand the signals involved during courtship, and their significance. I have demonstrated the feasibility of recording vibrations directly from female bodies (Chapter 4), and this technique can be used to record male tactile signals during courtship. In addition to male signals, I have also recorded and characterized female abdominal twitch vibrations produced during courtship interactions (Figure 5.1). These female vibrations were rare (N = 5 females), but have been previously described in L. hesperus (Ross and Smith, 1979). It is not clear why or when females will produce these signals, but some anecdotal observations suggest that in some L. hesperus populations these signals may be part of male-female vibrational duetting behaviour (Catherine Scott, personal communications). This presents an additional avenue for deeper exploration.

Black widow spiders are interesting organisms as there is considerable variation in life history traits, reproductive tactics and mating behaviours, not only across species, but across populations within species (Andrade, 1996; Garb et al., 2004; Kasumovic and Andrade, 2006; Stoltz et al., 2009, 2008; Stoltz and Andrade, 2010). This diversity among Latrodectus species makes them ideal for future comparative studies on vibratory communication, and the role of vibration signals in different behavioural and mating contexts (e.g. aggressive/competitive interactions, monogamous vs. polygynous mating systems, group-living social environments, etc). Even within L. hesperus, there are both genetic and behavioural variations between coastal and inland populations (Garb et al., 2004; Charmaine Condy and Maydianne Andrade, personal communications). Future comparative studies examining vibration signals across different populations and/or species, and overlaying them with aspects of life history traits and/or reproductive tactics, can provide valuable insights into signal diversity and evolution, and allows for developing testable hypotheses regarding the functional role of vibration signals in this genus, and possible web-building spiders in general.

Nevertheless, placing my results in context with recent work by Wignall and Herberstein (2013), who also demonstrated that in the orb-web spider, Argiope kerserlingi, males communicate with several distinct web-borne vibration signals during courtship, it is becoming more evident that many web-building spiders may communicate with multiple vibration signals. The use of multiple signals among web-building spiders had been implied in the literature, but now there are

162 emerging evidence to support this. This raises the question: what selection pressure(s) is/are driving the evolution of multiple signals in web-building spiders?

Web-building spiders construct the substrate on which they communicate, and are thus considered substrate specialists (Elias and Mason, 2011). Given that web-building spiders have control over the substrate and signals, both can be optimized for effective information transfer (Elias and Mason, 2011). The sensory drive hypothesis predicts that selection should favour signals with characteristics that allow effective transmission and detection leading to a signal- substrate match (Basolo and Endler, 1995; Cummings and Endler, 2018; Endler, 1992; Endler and Basolo, 1998). Therefore, substrate specialization can impose strong constraints on signal evolution. However, webs are remarkably effective substrates for transmitting vibrations (Mortimer, 2019; Mortimer et al., 2019, 2016; Vibert et al., 2016). For example, Vibert et al. (2016) showed that female L. hesperus cobwebs passed low frequencies with little attenuation (high efficacy). Therefore, rather than increased constraints, webs can relax constraints on signal transmission, which may promote evolution of novel signals and signal diversity (see Rosenthal et al., 2019). As mentioned in the introduction chapter, male web-dwelling spiders are widely known to employ a multitude of vibration signaling behaviours during courtship. It would be interesting for future studies to examine the relationship between signal transmission efficacy of different web-types and signal repertoire and characteristics.

Evolution of multiple signals may also be driven by female preference for signal novelty/complexity (Elias et al., 2006). In the jumping spider, Habronattus pugillis, females showed preference for complex signals of foreign males compared to less complex local males (Elias et al., 2006). It is not clear whether this would also be the case for many web-dwelling spider species. However, there is emerging evidence showing that different L. hesperus populations, along the west coast of North America, as well as between coastal and inland populations, vary greatly in life history traits, behaviour, and genetics (Charmaine Condy, personal communication). These varying populations may be ripe for examining variations in males vibratory signaling and female preference.

In addition to having multiple signals in their repertoire, my research also showed that L hesperus males occasionally transitioned between loose/haphazard displays of different signal elements to more temporally organized stereotyped sequences (structured signaling). This

163 additional level of nascent signal complexity in L. hesperus male signals, is a novel discovery from my research, and suggests that vibratory communication in L. hesperus may be more complex than previously thought (Herberstein et al., 2014). Previously, L. hesperus male signals were thought to lack any noticeable patterning (Vibert et al., 2014). However, while Vibert et al. (2014) primarily focused on male signaling during the early stages of male-female interactions, my research showed that in the later stages, and during copulation, these complex structured signals are more prevalent (Chapter 2). Furthermore, my research also demonstrated that these composite multicomponent displays have important consequences for male mating success (discussed in section 1.3 below). This non-random active sequencing of individual signal components is common in birdsongs (Alger et al., 2016; Chatfield and Lemon, 1970; Große Ruse et al., 2016; Kershenbaum, 2014; Weiss et al., 2014), but is less known or common in arthropods. To date, active non-random sequencing of signal components has only been described in a few species of jumping spiders (Elias et al., 2012). Comparatively, web-building spiders were considered relatively ‘simple’ signalers (Herberstein et al., 2014). Despite this, however, many descriptions of vibratory behaviours in the literature had eluded to possible multicomponent displays in other web-building species. For example, in the funnel-web spider Thelechoris karschi, males were described to employ a multitude of vibratory behaviours during courtship, including “quivering”, “twitching”, “body jerking”, and “tapping” (Coyle and O’Shields, 1990). Coyle and O’Shields (1990) noted that signaling bouts that combined all these different behaviours were common during courtship. Likewise, in the orb-web spider, Leucauge mariana, Eberhard and Huber (1998) examined the patterns of occurrence of male vibratory behaviours and showed that quite often males grouped different vibratory behaviours together. Similarly, Lubin (1986) described stereotyped sequences of vibratory behaviours in the cobweb spider, Achaearaneae wau, where male signaling bouts started with abdominal vibrations and followed by bursts of web-plucks (“twanging”). These examples from the literature, along with the novel results of my research, suggest that previous assumptions regarding the relative simplicity of web-building spider signaling may be premature, and further research is needed to determine the level of signal complexity in web-dwelling spiders.

Figure 5.1: Oscillogram (top panel) and mean power spectrum (bottom panel [thick black line]) of female abdominal twitch signals. Thin blue lines show means of individual females (N=5). Female abdominal twitch vibrations were low-frequency narrow-band signals. The mean (SD) duration, peak frequency, peak power, peak amplitude, and RMS amplitude of female abdominal twitch vibrations were 0.33 (0.1), 34.47 (19.52), 98.68 (8.84), 0.16 (0.14), and 0.04 (0.03), respectively.

164

5.1.2 Males Use Different Signal Production Mechanisms

Synchronized high-speed video and laser Doppler vibrometry analysis in chapter two of this thesis demonstrated that the three vibration signals of L. hesperus males were produced by independent mechanisms. Male bounce and web-pluck signals represent percussive signaling, while abdominal tremulation signals are tremulatory. Do date, the mechanisms underlying vibratory signal production in web-building spiders has only been examined for the sheet-web spider Frontinella pyramitela (Suter and Renkes, 1984). In F. pyramitela, males use their abdomen to produce both tremulation (“fast abdomen flexion”) and percussive (“abdomen flexion”) signals during courtship (Suter and Renkes, 1984). My research, along with previous work by Suter and Renkes (1984) demonstrate that male web-building spiders can use quite varied mechanisms to generate different signal types. However, previous qualitative descriptions of vibratory behaviours in web-building spiders make comparisons between species difficult. For example, in L. hesperus, male bounce signaling, shares similarities to the “web-flexion” behaviour in Agelenopsis aperta (Singer and Riechert, 1995), “twitching” behaviour in Thelechoris karschi (Coyle and O’Shields, 1990), and the “8-leg flexion” in Frontinella pyramitela (Suter and Renkes, 1984). Examining the details of how signals are produced is crucial for understanding the function of signals and potential physical and/or physiological constraints on signal evolution.

Although my research provides insights into the body structures involved during signal productions, the finer details of how signals are produced remains unknown. For example, abdominal tremulations in the wandering spider, Cupiennius getazi, are produced by contractions of two different muscles (muscle 81 and 85) attached to the pedacil, which corresponds to upward and downward movements of the abdomen, respectively (Dierkes and Barth, 1995). Although it is not yet known whether abdominal tremulations are generated by analogous muscle contractions in L. hesperus, it is possible that the physiological and physical properties of possible muscles can affect signal characteristics (e.g. frequency), and in turn signal function. For example, in L. hesperus, it is possible that properties of such muscles could limit how variable these signals are between males. This could in turn affect the function of these signals. Thus, abdominal tremulation signals may be less suitable for conveying information about male quality, and better suited for conveying information about mate identity, given that they may be

165 166 fairly similar across males. My research showed that male abdominal tremulation signal characteristics were not correlated with male traits (Chapter 2). However, Vibert et al. (2014) previously demonstrated that male abdominal tremulation signals allowed females to distinguished mates from prey. Further studies examining signaling mechanisms at a finer scale are needed to determine how the movements of different body structures are generated during vibratory signaling in L. hesperus, as well as in other web-building spiders.

5.1.3 Male Signals are Informative and Important for Male Mating Success

In chapter three, I demonstrated that males’ structured multicomponent signals are important for male mating success. Specifically, males that transitioned from loose unstructured signaling to structured signaling more often mated sooner (Chapter 3). It is evident from my research that these structured signals play an important role during courtship. It is possible that structured signaling is more energetically intense than unstructured signaling (De Luca et al., 2015). However, at present, the energetic costs of vibratory signaling are not known. To determine whether producing structured multicomponent signals is more energetically intense, future studies should compare differences in metabolic rates during structured and unstructured signaling.

Although structured signaling was shown to be important, the role of unstructured signaling, at present, is not clear. During unstructured signaling, male engage in various other behaviours not examined in this thesis; including, web-modification, adding silk on webs and females, and different tactile behaviours (also see: Scott et al., 2012). It is possible that these other behaviours are equally important for male mating success, and needs to be examined in more detail.

Multicomponent vibratory signaling is common in many cursorial spiders (Elias et al., 2012; Hebets et al., 2013; Herberstein et al., 2014; Uetz et al., 2016, 2013), and studies have shown that individual signal components can convey information about males (Gibson and Uetz, 2008; Schüch and Barth, 1990, 1985), and are necessary for mating success (Elias et al., 2005; Hebets et al., 2013). However, very little is known about female perception of individual vibration components. In the wandering spider, Cupiennius salei, male tremulation and percussive signals stimulate different slits of female Lyriform organs, suggesting that females can process these signals in parallel (Baurecht and Barth, 1992). It is important for future studies to examine neural

167 responses to different male signals to determine whether females can discriminate between signal components.

Results from this thesis also suggest that the different signal components may convey different information to females (multiple message hypothesis). Male bounce and web-pluck signal intensities were correlated with male mass/size (Chapter 2), and abdominal tremulation signals were shown to function for mate identity (Vibert et al., 2014). It would be interesting to see whether these signals have similar functions in other Latrodectus species. Abdominal tremulation and bounce (or “jerking”) behaviours have commonly been described in different Latrodectus species (Chapter 1: Table 1.2). Whether these signals have similar functions across species remains to be tested.

5.1.4 Female Body Mechanics Affect Vibration Tuning of Leg Joints

In chapter four, my research showed that female legs exert influence on vibration transmission, and thus may play a vital role during perception. My finding that more distal leg segments respond to higher frequency vibrations, while proximal segments respond to lower frequencies are consistent with previous studies examining vibration transmission across the legs of other arthropods (e.g. fiddler crabs (Aicher et al., 1983), cave crickets (Stritih Peljhan and Strauß, 2018; Stritih-Peljhan et al., 2019), and wandering spiders (Dierkes and Barth, 1995). These consistent low-pass characteristics seen across different arthropods is likely due to the inherent nature of how waves travel through bent structures (truss structures) in general (i.e. leg-joints in animals) (Guo, 1995, 1994). Nevertheless, this may allow organisms to have greater control over selective attention within their signal space. This is seen from the results in chapter four where female posture was shown to affect joint bending, and in turn frequency tuning. Similarly, in the cave cricket, Troglophiles neglectus, leg position was shown to affect frequency tuning (Stritih Peljhan and Strauß, 2018). Individuals can actively tune in and/or tune out of different frequency ranges via leg extension and flexion, which can have great behavioural consequences. In L. hesperus, leg posture may allow, for example, hungry females to focus their attention to higher frequencies of smaller prey.

The mechanical segregation of low and high frequencies along L. hesperus female legs may allow females to discriminate between mates and prey. In L. hesperus, male vibratory signals

168 have relatively lower frequencies compared to various prey vibrations (Chapter 4: Figure S4.10), and females were previously shown to use these signals (i.e. abdominal tremulation signals) to distinguish mates from prey (Vibert et al., 2014). Frequency segregation may also be important for females to discriminate prey from vibration noise. In nature, female webs are susceptible to vibrational disturbances from wind, vegetation, and anthropogenic noises (Wignall et al., 2011; Wu and Elias, 2014). Preliminary research examining prey capture behaviour under various intensities of wind noise conditions showed that L. hesperus females are quite adept at accurately detecting and localizing prey (unpublished). Future neurophysiological studies in L. hesperus females are needed to further compare peripheral neural tuning with the mechanical tuning seen in this thesis.

My research also showed that higher frequencies decayed substantially as vibrations travelled from the metatarsus to the tibia of female legs. This highlights the particular function of the metatarsal Lyriform organ for prey detection and localization (Barth and Geethabali, 1982; Klärner and Barth, 1982; Landolfa and Barth, 1996). The metatarsal Lyriform organ (HS10 organ) is located dorsally on the distal end of the metatarsus at the tarsus-metatarsus joint (Barth and Geethabali, 1982; Morley et al., 2016) (Figures 5.2, 5.3, & 5.4). Much of our current understanding of vibration perception in spiders comes from extensive research on the neurophysiology and biomechanics of the HS10 organ, specifically in the wandering spider, Cupiennius salei (Barth and Geethabali, 1982; Erko et al., 2015; Hößl et al., 2009, 2007, 2006; McConney et al., 2007; Molina et al., 2009; Schaber et al., 2012). The location of the HS10 at the most distal joint make them well suited for prey localization. It is possible that vibrations arriving at females can stimulate the eight HS10 sensors with temporal and amplitude variations, which can in turn provide information about the location and distance of the vibration source. We can examine this by looking at the vibration behaviour of female bodies. Preliminary research examining the motion of female bodies (i.e. phase and magnitude differences) has shown that both phase and magnitude differences can provide directional information (unpublished).

Figure 5.2: Metatarsal Lyriform organ (HS10) of the western black widow spider (Latrodectus hesperus). Scanning electron micrographs of L. hesperus (A) female and (B) male HS10 organ. (C) Longitudinal section through the tarsus-metatarsus joint. The HS10 organ is located dorsally at the distal end of the metatarsus (MT), at the tarsus-metatarsus joint. Adjacent to the HS10 organ is the cuticular pad (P). The slits of the HS10 organ are oriented perpendicularly to the long axis of the leg. Individual slits are innervated by two bipolar sensory cells located distally. The leg nerve (N) runs through the middle of the leg. Web-borne vibrations transmitted along the leg causes the tarsus (T) to bend and strike the cuticular pad, which in turn causes the slits of the HS10 organ to compress, resulting in nerve response.

169

Figure 5.3: Structural details of the metatarsal Lyriform organ (HS10) of Latrodectus hesperus. Top and bottom panels show longitudinal section through two different HS10 organs, respectively. Each slit of the HS10 organ is bounded by an outer (o) and inner (i) membrane, and two cuticular lamellae (L). The outer (o) and inner (i) membranes form the coupling cylinder (Cc). A vessel (v) is attached to the inner (i) membrane of the coupling cylinder (Cc). A tubule (T) runs through the vessel (v) and into the coupling cylinder (Cc) and attaches to the cuticular lamellae (L) (bottom panel). The tubule is an extension of the neurilemma (n), which covers the dendritic bulb (Db) of a sensory cell (Sc). The dendritic bulb (Db) is located near the vessel (v), while the sensory cell (Sc) is located further up the leg. Thin filaments originating from dendritic bulb (Db) extend into the tubule and attach to either the inner or outer membrane (not shown). Each slit is innervated by two bipolar sensory cells.

170

Figure 5.4: Transmission electron micrograph of Latrodectus hesperus HS10 organ slits. (Cc: coupling cylinder; L: cuticular lamellae; i: inner membrane of coupling cylinder; o: outer membrane of coupling cylinder; v: vessel; n: neurilemma).

171

5.2 Conclusion

In this dissertation, I have shown that vibratory communication in the western black widow spider comprise multiple signals organized into complex displays that play a functional role during male-female interactions. I combined behavioural experiments with laser Doppler vibrometry, and high-speed video analysis to provide one of the first detailed examinations of vibratory communication in a web-building spider. The knowledge gained from this thesis provides insights into nascent signal complexities in a web-building spider, previously thought to be ‘simple’ signalers, and lays the foundation for future studies on vibratory communication in black widow spiders, and other web-building spiders. Additionally, the novel techniques developed in this thesis – using electromagnets to stimulate female webs without contact, and measure vibrations directly from female bodies – can be applicable to a wide range of future behavioural, biomechanical, and neurophysiological studies. Research into vibratory communication in web-building studies is still in its early years, and there remains a lot more work to be done. It is my hope that the work conducted in this thesis will promote greater research interest in this underappreciated and understudied area of animal communication.

172

5.3 References

Alger, S.J., Larget, B.R., Riters, L.V., 2016. A novel statistical method for behaviour sequence analysis and its application to birdsong. Animal Behaviour 116, 181–193.

Andrade, M.C.B., 1996. Sexual selection for male sacrifice in the Australian redback spider. Science 271, 70–72.

Barth, F.G., Geethabali., 1982. Spider vibration receptors: Threshold curves of individual slits in the metatarsal lyriform organ. Journal of Comparative Physiology A 148, 175–185.

Basolo, A.L., Endler, J.A., 1995. Sensory biases and the evolution of sensory systems. Trends in Ecology & Evolution 10, 489.

Baurecht, D., Barth, F.G., 1992. Vibratory communication in spiders. Journal of Comparative Physiology A 171, 231–243.

Chatfield, C., Lemon, R.E., 1970. Analysing sequences of behavioural events. Journal of Theoretical Biology 29, 427–445.

Coyle, F.A., O’Shields, T.C., 1990. Courtship and mating behavior of Thelechoris Karschi (Araneae, Dipluridae), an African funnel-web spider. The Journal of Arachnology 18, 281–296.

Cummings, M.E., Endler, J.A., 2018. 25 Years of sensory drive: The evidence and its watery bias. Current Zoology 64, 471–484.

Elias, D.O., Hebets, E.A., Hoy, R.R., 2006. Female preference for complex/novel signals in a spider. Behavioral Ecology 17, 765–771.

173 174

Endler, J.A., 1992. Signals, signal conditions, and the direction of evolution. The American Naturalist 139, S125–S153.

Endler, J.A., Basolo, A.L., 1998. Sensory ecology, receiver biases and sexual selection. Trends in Ecology & Evolution 13, 415–420.

Erko, M., Younes-Metzler, O., Rack, A., Zaslansky, P., Young, S.L., Milliron, G., Chyasnavichyus, M., Barth, F.G., Fratzl, P., Tsukruk, V., 2015. Micro-and nano- structural details of a spider’s filter for substrate vibrations: relevance for low-frequency signal transmission. Journal of the Royal Society Interface 12, 20141111.

Gibson, J.S., Uetz, G.W., 2008. Seismic communication and mate choice in wolf spiders: Components of male seismic signals and mating success. Animal Behaviour 75, 1253– 1262.

Große Ruse, M., Hasselquist, D., Hansson, B., Tarka, M., Sandsten, M., 2016. Automated analysis of song structure in complex birdsongs. Animal Behaviour 112, 39–51.

Guo, Y.P., 1995. Flexural wave transmission through angled structural joints. The Journal of the Acoustical Society of America 97, 289–297.

Guo, Y.P., 1994. Diffraction of flexural waves at structural joints. The Journal of the Acoustical Society of America 95, 1426–1434.

Herberstein, M.E., Wignall, A.E., Hebets, E.A., Schneider, J.M., 2014. Dangerous mating systems: Signal complexity, signal content and neural capacity in spiders. Neuroscience

175

& Biobehavioral Reviews, Beyond Sexual Selection: the evolution of sex differences from brain to behavior 46, 509–518.

Kasumovic, M.M., Andrade, M.C.B., 2006. Male development tracks rapidly shifting sexual versus natural selection pressures. Current Biology 16, R242–R243.

Kershenbaum, A., 2014. Entropy rate as a measure of animal vocal complexity. Bioacoustics 23, 195–208.

Klärner, D., Barth, F.G., 1982. Vibratory signals and prey capture in orb-weaving spiders (Zygiella x-notata, Nephila clavipes; Araneidae). Journal of Comparative Physiology A 148, 445–455.

Landolfa, M.A., Barth, F.G., 1996. Vibrations in the orb web of the spider Nephila clavipes: cues for discrimination and orientation. Journal of Comparative Physiology A 179, 493– 508.

Lubin, Y.D., 1986. Courtship and alternative mating tactics in a social spider. The Journal of Arachnology 14, 239–257.

McConney, M.E., Schaber, C.F., Julian, M.D., Barth, F.G., Tsukruk, V.V., 2007. Viscoelastic nanoscale properties of cuticle contribute to the high-pass properties of spider vibration receptor (Cupiennius salei Keys). Journal of The Royal Society Interface 4, 1135–1143.

Morley, E.L., Sivalinghem, S., Mason, A.C., 2016. Developmental morphology of a lyriform organ in the western black widow (Latrodectus hesperus). Zoomorphology 135, 433–440.

Mortimer, B., 2019. A spider’s vibration landscape: Adaptations to promote vibrational information transfer in orb webs. Integrative and Comparative Biology 59, 1636–1645.

176

Mortimer, B., Soler, A., Siviour, C.R., Zaera, R., Vollrath, F., 2016. Tuning the instrument: Sonic properties in the spider’s web. Journal of The Royal Society Interface 13, 20160341.

Rosenthal, M.F., Hebets, E.A., Kessler, B., McGinley, R., Elias, D.O., 2019. The effects of microhabitat specialization on mating communication in a wolf spider. Behavioral Ecology 30, 1398–1405.

Schaber, C.F., Gorb, S.N., Barth, F.G., 2012. Force transformation in spider strain sensors: white light interferometry. Journal of The Royal Society Interface 9, 1254–1264.

Schüch, W., Barth, F.G., 1990. Vibratory communication in a spider: female responses to synthetic male vibrations. Journal of Comparative Physiology A 166, 817–826.

Schüch, W., Barth, F.G., 1985. Temporal patterns in the vibratory courtship signals of the wandering spider Cupiennius salei Keys. Behavioral Ecology and Sociobiology 16, 263– 271.

Scott, C., Vibert, S., Gries, G., 2012. Evidence that web reduction by western black widow males functions in sexual communication. The Canadian Entomologist 144, 672–678.

Singer, F., Riechert, S., Xu, H., Morris, A., Becker, E., Hale, J., Noureddine, M., 2000. Analysis of courtship success in the funnel-web spider Agelenopsis aperta. Behaviour 137, 93–117.

Singer, F., Riechert, S.E., 1995. Mating system and mating success of the desert spider Agelenopsis aperta. Behavioral Ecology and Sociobiology 36, 313–322.

Stoltz, J.A., Andrade, M.C.B., 2010. Female’s courtship threshold allows intruding males to mate with reduced effort. Proceedings of the Royal Society B: Biological Sciences 277, 585–592.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2009. Male courtship effort determines female response to competing rivals in redback spiders. Animal Behaviour 77, 79–85.

Stoltz, J.A., Elias, D.O., Andrade, M.C.B., 2008. Females reward courtship by competing males in a cannibalistic spider. Behavioral Ecology and Sociobiology 62, 689–697.

Stritih Peljhan, N., Strauß, J., 2018. The mechanical leg response to vibration stimuli in cave crickets and implications for vibrosensory organ functions. Journal of Comparative Physiology A 204, 687–702.

177

Stritih-Peljhan, N., Rühr, P.T., Buh, B., Strauß, J., 2019. Low-frequency vibration transmission and mechanosensory detection in the legs of cave crickets. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology 233, 89–96.

Suter, R.B., Renkes, G., 1984. The courtship of Frontinella Pyramitela (Araneae, Linyphiidae): Patterns, vibrations and functions. The Journal of Arachnology 12, 37–54.

Uetz, G.W., Clark, D.L., Roberts, J.A., 2016. Multimodal communication in wolf spiders (Lycosidae)—An emerging model for study, in: Advances in the Study of Behavior. Elsevier, pp. 117–159.

Uhl, G., Elias, D., 2011. Spider behaviour: flexibility and versability. Cambridge University Press, Cambridge Virant-Doberlet M, Cokl C (2004) Vibrational communication in insects. Neotropical Entomology 33, 121134.

Vibert, S., Scott, C., Gries, G., 2014. A meal or a male: the ‘whispers’ of black widow males do not trigger a predatory response in females. Frontiers in Zoology 11, 4.

Wignall, A.E., Herberstein, M.E., 2013. The influence of vibratory courtship on female mating behaviour in orb-web spiders (Argiope keyserlingi, Karsch 1878). PLoS One 8.

Wignall, A.E., Jackson, R.R., Wilcox, R.S., Taylor, P.W., 2011. Exploitation of environmental noise by an araneophagic assassin bug. Animal behaviour 82, 1037–1042.

Wu, C.-H., Elias, D.O., 2014. Vibratory noise in anthropogenic habitats and its effect on prey detection in a web-building spider. Animal Behaviour 90, 47–56.