Vibratory Signalling in Two Spider Species with Contrasting Web Architectures

Home , Giant house spider, Latrodectus hesperus, Latrodectus pallidus, Malthonica, Nephilengys, Neumania papillator

Vibratory signalling in two spider species with contrasting web architectures

by Samantha Vibert M.Sc., Université de Genève, 2002

B.Sc., Université de Genève, 2000

Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

in the Department of Biological Sciences Faculty of Science

 Samantha Vibert 2016 SIMON FRASER UNIVERSITY Summer 2016

Approval

Name: Samantha Vibert Degree: Doctor of Philosophy (Biological Sciences) Title: Vibratory signalling in two spider species with contrasting web architectures

Examining Committee: Chair: Michael Hart Professor

Gerhard Gries Senior Supervisor Professor

Bernard Roitberg Supervisor Professor Emeritus

Robert G. Bennett Supervisor Research Associate Royal British Columbia Museum

Staffan Lindgren Internal Examiner Professor Emeritus University of Northern British Columbia

Damian O. Elias External Examiner Associate Professor Department of Environmental Science, Policy and Management University of California, Berkeley

Date Defended/Approved: August 15, 2016

Abstract

Spiders provide a fascinating opportunity for the study of animal communication. Web- building spiders build their own signalling environments - the web is the medium that transmits vibrations from prey, predators and potential mates. However, we know little about how information is conveyed through different types of webs, or how spiders distinguish between different types of vibrations. In this thesis, I studied elements of vibratory communication in two species of spiders with contrasting web architecture: the western black widow, Latrodectus hesperus, which builds a tangle-web, and the hobo spider, Eratigena agrestis, which builds a funnel-web.

In chapter 2, I document formerly undescribed life history traits of E. agrestis, and conclude that life history traits are robust to differential predator and competitor densities across two study sites in British Columbia. In chapter 3, I present hitherto lacking quantitative descriptions of courtship behaviours in L. hesperus, revealing that web reduction by males correlates with reduced female aggression, and that it may improve mating success of courting males. In chapter 4, I describe how vibration frequencies are transmitted through the webs of L. hesperus and E. agrestis. I found little difference in propagation efficiency between longitudinal and transverse vibrations and that in both species vibration transmission is more variable within webs than between webs, suggesting that specific frequencies play a minor role in signalling. In chapter 5, I tested whether male courtship produces vibratory signals that differ from prey cues. I analysed vibrations produced by courting males and by two types of prey (flies and crickets) on the webs of L. hesperus and E. agrestis, and also played back male and prey vibrations through the webs of L. hesperus. Male vibrations differ more from those of prey in L. hesperus than in E. agrestis. This finding supports the hypothesis that L. hesperus males, faced with aggressive females, produce vibrations that prevent them from being mistaken for prey. The low-amplitude vibrations caused by abdominal tremulations of L. hesperus males may be linked with lowered female aggression.

Keywords: Araneae; courtship; vibration; communication; behaviour; ecology

iii

Dedication

To Gwylim, partner in life and to Catherine, partner in science.

Acknowledgements

My heartfelt gratitude goes to the following people:

My supervisor, Gerhard Gries – Your unwavering support, your scientific enthusiasm and your boundless optimism were all profoundly inspiring. Your faith that I should, could and would complete my degree was instrumental in helping me achieve this goal.

Regine Gries – Thank you for your kindness and for making the lab a gentle, organized workplace.

Bernie Roitberg – Thank you for your always insightful advice and for helping me broaden the scientific scope of my work.

Robb Bennett – Thank you for introducing me to the exciting world of spider research. From mentor and colleague, you quickly became a friend. Your generosity and hospitality contributed greatly to my feeling at home here in Canada.

Steve Takács – Thank you for sharing your technical expertise and for the many hours we spent trying to obtain recordings of spider courtship vibrations.

My friends and colleagues from the lab: Gagandeep Hehar, Adela Danci, Eloise Rowland, Kevin Lam, Tom Cowan, Nathan Woodbury, Kelly Ablard, Sebastian Ibarra and Vroni Lambinet – Thank you for the chats, the thoughtful feedback and the help.

Sean McCann – Thank you for all the conversations: I truly enjoy disagreeing with you! And of course, thank you for the wonderful photographs.

Maxence Salomon – Thank you for being such a good friend.

Catherine Scott – You have been a wonderful collaborator and friend, helping make my project better and a lot more fun! My thesis would not have been completed without your generous and expert support.

Table of Contents

Approval ...... ii Abstract ...... iii Dedication ...... iv Acknowledgements ...... v Table of Contents ...... vi List of Tables ...... ix List of Figures...... x

Chapter 1. Introduction ...... 1 References ...... 3

Chapter 2. Life history of the funnel-web spiders Eratigena agrestis and E. atrica (Araneae: Agelenidae) in the Pacific Northwest ...... 5 2.1. Abstract ...... 5 2.2. Introduction ...... 6 2.3. Methods ...... 7 2.3.1. Study sites ...... 7 2.3.2. Survey protocol...... 8 2.3.3. Variables measured ...... 8 2.4. Results ...... 10 2.5. Discussion ...... 12 2.5.1. Life cycles of E. agrestis vs. E. atrica ...... 12 2.5.2. Life history traits of E. agrestis and E. atrica in ecologically contrasting populations ...... 13 2.5.3. Potential implications of group living ...... 15 Acknowledgements ...... 16 References ...... 16

Chapter 3. Evidence that web reduction by western black widow males functions in sexual communication ...... 27 3.1. Abstract ...... 27 3.2. Résumé ...... 27 3.3. Main text ...... 28 Acknowledgements ...... 33 References ...... 34

Chapter 4. Vibration transmission through sheet webs of hobo spiders (Eratigena agrestis) and tangle webs of western black widow spiders (Latrodectus hesperus) ...... 38 4.1. Abstract ...... 38 4.2. Introduction ...... 39 4.3. Methods ...... 40

4.3.1. Study spiders ...... 40 4.3.2. General procedures ...... 41 Recordings ...... 41 Frequency sweeps ...... 41 Vibration playback devices ...... 41 Recording protocol ...... 42 4.3.3. Analysis ...... 44 Characterizing longitudinal and transverse vibration transmission through webs ...... 44 Consistency of vibration frequency transmission between and within webs ...... 45 4.4. Results ...... 46 4.4.1. Longitudinal and transverse vibration transmissions of webs ...... 46 4.4.2. Consistency of vibration frequency transmission between and within webs ...... 46 4.5. Discussion ...... 47 Acknowledgements ...... 49 References ...... 49

Chapter 5. A meal or a male: the ‘whispers’ of black widow males do not trigger a predatory response in females ...... 58 5.1. Abstract ...... 58 Introduction ...... 58 Results ...... 59 Conclusions ...... 59 5.2. Introduction ...... 59 5.3. Material and methods ...... 61 5.3.1. Characterization of prey and male vibrations on L. hesperus and T. agrestis webs ...... 61 (a) Study animals ...... 61 (b) Courtship behaviours ...... 62 (c) Recordings of web vibrations ...... 63 (d) Analyses of LDV and video recordings ...... 64 (e) Statistical analysis ...... 65 5.3.2. Vibration parameters L. hesperus females use to discern a prospective prey from a courting male ...... 66 (a) Study spiders ...... 66 (b) Test stimuli ...... 66 (c) Playback ...... 67 (d) Behavioural response of spiders to playback vibrations ...... 67 (e) Statistical analyses ...... 68 5.4. Results ...... 69 5.4.1. Characterization of prey and male vibrations on L. hesperus and T. agrestis webs ...... 69 5.4.2. Comparison of prey and male vibrations on L. hesperus and T. agrestis webs ...... 70 5.4.3. Vibration parameters triggering a predatory response in L. hesperus females ...... 71 5.5. Discussion ...... 72 5.6. Conclusions ...... 75 Acknowledgements ...... 75

vii

Competing interests ...... 76 Authors’ contributions ...... 76 References ...... 76

Chapter 6. Conclusion ...... 90 Appendix A. Supplementary material for chapter 4 ...... 92 Appendix B. Supplementary material for chapter 5 ...... 96 Additional files B1-B3 and B8 (video files) ...... 96 Additional file B4 ...... 97 Additional file B5 ...... 98 Additional file B6 ...... 99 Waveform quality and amplitude consistency of playback-induced vibrations ...... 99 Additional file B7 ...... 100

viii

List of Tables

Table 2.1 Summary of ecological characteristics of the study sites at Island View Beach (IVB) and Iona Beach Regional Park (IoB)...... 19 Table 2.2 Number of moults preceding maturity in laboratory-reared Eratigena agrestis individuals collected from Island View Beach...... 20 Table 5.1 Summary of parameters associated with males, prey, and background noise vibrations on webs of Latrodectus hesperus (top) and Tegenaria agrestis (bottom) ...... 81 Table 5.2 Results of logistic regression analysis of predatory responses of female Latrodectus hesperus to playbacks of low- or high- amplitude vibrations of house fly prey and conspecific males ...... 82

List of Figures

Figure 2.1 Map of the southern coast of British Columbia, Canada, with the locations of the field sites: Iona Beach (IoB) and Island View Beach (IVB). Relative abundance of Eratigena agrestis, E. atrica and Latrodectus hesperus spiders under driftwood logs at each site. Relative abundance is based on all spiders found between May 2005 and October 2006 at IVB, and between May 2006 and April 2007 at IoB. Black and white map from SimpleMappr (Shorthouse 2010); IVB map imagery © Google 2015; IoB map imagery © Google 2014...... 21 Figure 2.2 Mean temperature (°C) and total precipitation (mm) between May 2005 and April 2007 at Island View Beach (Vancouver Island) and Iona Beach (Greater Vancouver Regional District). Data were obtained from Environment Canada weather stations at the Vancouver International Airport and the Victoria International Airport, each situated less than 6 km from the relevant study site...... 22 Figure 2.3 Life cycles of Eratigena agrestis and of E. atrica at Island View Beach and Iona Beach. Each tick on the scales represents a month...... 23 Figure 2.4 Phenology of Eratigena agrestis and E. atrica at Island View Beach (IVB) and Iona Beach (IoB). The graphs represent the densities (numbers of individuals per m2 of available habitat, i.e. driftwood logs) of juveniles, adult females, subadult and adult males, and the number of egg sacs for each species and field site. Data were collected between July 2005 and June 2006 at IVB, and between May 2006 and April 2007 at IoB. The egg sacs spun by E. agrestis and E. atrica are very similar and could not be identified to species level...... 24 Figure 2.5 Changes over time in population densities of Eratigena agrestis and E. atrica juveniles of different sizes at Island View Beach (IVB) and Iona Beach (IoB). The total body length of all juveniles was assigned to one of five size categories: [1: 2-4 mm; 2: 4-6 mm; 3: 6-8 mm; 4: 8-10 mm; 5: > 10 mm]. Data were collected between July 2005 and June 2006 at IVB, and between May 2006 and April 2007 at IoB...... 25 Figure 2.6 Body length (mm) of adult Eratigena agrestis females at Island View Beach (IVB) and Iona Beach (IoB). The boxes depict the median, first and third quartiles. The whiskers represent the lowest and highest datum within the 1.5 interquartile range of the lower and upper quartile. Data were collected in 2005 at IVB and in 2006 at IoB...... 26

Figure 3.1 Summary of the courtship sequence for Latrodectus hesperus, indicating the time spent in each phase. Actions of males are in blue rectangles, and responses of females are in pink ovals. Distal courtship, proximal courtship, and copulation consist of cycles of multiple behavioural elements that are listed in the bottom left corner of each dashed box. Elements that did not occur in every mating trial are indicated as percentage of replicates in which they occurred. Quantitative data are based on 12 replicates for the distal phase and 11 for the proximal phase and for the copulation...... 36 Figure 3.2 A) Photograph of a male Latrodectus hesperus engaged in web reduction behaviour (note the dense rope of silk that is the result of his cutting, bundling, and wrapping with silk the section of the female's web which formerly covered the empty space in the bottom two thirds of the image). (B) Mean (+SD) number of instances of aggressive behaviour by female L. hesperus in the distal courtship phase towards males that did (n = 7), or did not (n = 5), engage in web reduction (U = 4.5, P < 0.05). (C) Mean (+SD) latency to female L. hesperusquiescence (time from the male's first movement on the web until his first contact with the female's abdomen) for males that did (n = 7), or did not (n = 5), engage in web reduction (U = 1, P < 0.01). (D) Correlation between time to female L. hesperus quiescence and time to copulation for males that copulated at least once (ρ = 0.688, P = 0.019, n = 11). One male that did engage in web reduction did not copulate within 4 hours and was excluded from the analysis...... 37 Figure 4.1 Web structures of Eratigena agrestis and Latrodectus hesperus. Schematic drawing illustrating a the sheet web of Eratigena agrestis, and b the tangle web of Latrodectus hesperus. Webs of E. agrestis consist of a two-dimensional sheet of silk with a funnel at one end serving as a retreat. Webs of L. hesperus consist of a dense three-dimensional tangle of threads. Glue-coated capture threads extend from the tangle to the ground. The stars indicate the locations where the spiders typically reside while waiting for prey. The position of the spiders indicates the location of their retreats. Illustrations reproduced from Vibert et al. (2014)...... 51 Figure 4.2 Representative piezo sweeps (a-d) and loudspeaker sweeps (e, f) at 1 mm from the input vibration. Frequency sweeps were generated by a piezoelectric disk or modified loudspeaker and recorded on webs of Eratigena agrestis (a, c, e) and Latrodectus hesperus (b, d, f) by a laser Doppler vibrometer 1 mm away from the source of the input vibration (see Fig. 4). The graphs show the velocity of vibrations (mm/s) over frequency (Hz). The beam of the vibrometer was either parallel or perpendicular to the plane of the webs to capture longitudinal vibrations (a, b) or transverse vibrations (c-f), respectively. Note the different scales of the y- axes: μm/s in a-d and mm/s in e and f...... 52

Figure 4.3 Experimental design for frequency sweeps. A 12-s frequency sweep from 0 to 500 Hz was transferred onto the web by a modified loudspeaker (S). The resulting transverse vibration was recorded by a laser Doppler vibrometer (LDV). The laser beam was set perpendicular to the plane of the web and focused on small squares (1 mm2) of reflective tape that were placed 1 mm (#0), 20 mm (#1 and #2), and 70 mm (#3 and #4) from the tip of the vibrating rod. Frequency sweeps were recorded seven times in a fixed sequence (see methods for details). Illustration modified from Vibert et al. (2014)...... 53 Figure 4.4 Comparison of transmission efficiencies for longitudinal and transverse web vibrations. Transmission efficiency of longitudinal and transverse vibrations were measured on webs of Eratigena agrestis (a, b) and Latrodectus hesperus (c, d) (n = 14 each) at distances of 20 mm and 70 mm from a piezoelectric vibrator generating the vibration. Mean transmission efficiency (solid line) and 95% confidence intervals (shaded band) are shown. Transmission efficiency of vibrations (dB) was measured by comparing vibrations recorded at a distance of 20 mm and 70 mm from the vibrator with those obtained just 1 mm away from it...... 54 Figure 4.5 Comparison of transmission efficiencies of longitudinal and transverse vibrations measured at two distances from the point of vibration input. Longitudinal transmission efficiencies of vibrations were measured on webs of Eratigena agrestis (a, b) and Latrodectus hesperus (c, d) (n = 14 each) at distances of 20 mm and 70 mm from a piezoelectric vibrator generating the vibration. Mean transmission efficiency (solid lines) and 95% confidence intervals (shaded bands) are shown. Transmission efficiency of vibrations (dB) was measured by comparing vibrations recorded at a distance of 20 mm and 70 mm from the vibrator with those obtained just 1 mm away from it...... 55 Figure 4.6 Comparison of transmission efficiencies of transverse vibrations measured at two locations equidistant to the point of vibration input. Mean transmission efficiencies (solid line) and standard deviations (grey band) were measured on webs of Eratigena agrestis (a-d) and Latrodectus hesperus (e-h) (n = 16 each). The transmission efficiencies of vibrations (dB) were measured by comparing vibrations recorded at a distance of 20 mm and 70 mm from the loudspeaker vibrator with those just 1 mm away from it...... 56

xii

Figure 4.7 Repeatability of transmission efficiency measurements of transverse vibrations. The repeatability of transmission efficiency measurements of transverse vibrations was calculated for recordings taken at a distance of 20 mm (black squares) and 70 mm (grey circles) from the point of the vibration input, using webs of Eratigena agrestis (a) and Latrodectus hesperus (b). Vibrations were caused by frequency sweeps ranging between 0–500 Hz. Repeatability values were calculated at 10-Hz intervals from 0 to 200 Hz, the range of dominant frequencies produced by prey and males. Values can vary between 0 and 1. A value near 0 indicates that most of the variance in transmission efficiency results from variations between the two measurements taken on the same web...... 57 Figure 5.1 Web structure of Latrodectus hesperus and Tegenaria agrestis. Schematic drawing illustrating (a) the tangle web of Latrodectus hesperus and (b) the sheet web of Tegenaria agrestis. L. hesperus webs consist of a dense three-dimensional tangle of threads. Glue-coated capture threads extend from the tangle to the ground. T. agrestis webs consist of a two-dimensional sheet of silk with a funnel at one end serving as a retreat. We recorded vibrations on empty webs at the active hunting location of the spider (marked by ★). The spiders indicate the position of the female’s retreat. L. hesperus web illustration modified from Blackledge et al. 2005...... 83 Figure 5.2 Prey vibrations on webs of Latrodectus hesperus and Tegenaria agrestis. Oscillograms depicting velocity [mm/s] over time [s] (upper panels) and frequency [Hz] (lower panels) of cricket and house fly vibrations recorded on empty webs of Latrodectus hesperus and Tegenaria agrestis. “a.e.” refers to root mean square amplitude envelope of the vibration and “a.m.f” refers to amplitude modulation factor measured between the lowest and the highest point of the amplitude envelope (in this example, a.m.f. = 120)...... 84 Figure 5.3 Playback design and original vibrations used during playback. (a) Schematic drawing of the experimental design for vibration playbacks. A looped male vibration and a looped house fly vibration were played back at low and high amplitude by a modified loudspeaker (S) placed in contact with the web 15 cm away from a Latrodectus hesperus female in her hunting position; (b) Original male abdominal tremulation vibration used to generate the input vibrations played back with the speaker. Oscillograms depict velocity [mm/s] over time [s] (upper panel) and frequency [Hz] (lower panel); (c) Original fly vibration used to generate the input vibrations played back with the speaker. Oscillograms depict velocity [mm/s] over time [s] (upper panel) and frequency [Hz] (lower panel)...... 85

xiii

Figure 5.4 Courtship vibrations of male Latrodectus hesperus. For each of the courtship vibrations including (1) abdominal tremulations, (2) walking and bundling silk, and (3) bundling silk and cutting displayed by Latrodectus hesperus males on empty webs of conspecific females, the upper panel depicts vibrations in the time domain, and the lower panel depicts vibrations in the frequency domain. The insert in (1) depicts the amplitude of abdominal tremulation (maximum baseline-to-peak amplitude = 0.7 mm/s) magnified 20 times...... 86 Figure 5.5 Courtship vibrations of male Tegenaria agrestis. For each of the courtship vibrations including (1) palp drumming, (2) walking, drumming and tapping, (3) jerks, walking, drumming and tapping, and (4) stretches produced by courting Tegenaria agrestis males on empty webs of conspecific females, the upper panel depicts vibrations in the time domain and the lower panel depicts vibration in the frequency domain. The insert in (1) depicts the amplitude of palp drumming (maximum baseline-to-peak amplitude = 0.3 mm/s) magnified 25 times...... 87 Figure 5.6 Comparison of vibration parameters associated with struggling prey and courting male spiders on webs of Latrodectus hesperus and Tegenaria agrestis. (a-1) and (b-1) Discriminant function analysis using dominant frequency, frequency bandwidth, root mean square amplitude and duration (transformed data) of vibrations produced by cricket and house fly prey and by males of Latrodectus hesperus and Tegenaria agrestis on empty L. hesperus webs (a-1) and T. agrestis webs (b-1) (n = 16 each). The inner circle shows the 95% confidence ellipse of each mean; the outer circle shows the normal 50% contours; (a-2 to a-4) and (b-2 to b-4) Boxplots of dominant frequency [Hz], root mean square (RMS) amplitude [mm/s] and duration [s] of vibrations produced by cricket and house fly prey and by males of L. hesperus and T. agrestis on empty L. hesperus webs (a-2 to a-4) and T. agrestis webs (b-2 to b-4) (n = 16 each). Median, mean, interquartile range (IQR) and outliers (untransformed data); whiskers = upper and lower data point values within 1.5 IQR; means with different letters are significantly different (Tukey’s HSD on transformed data, p < 0.05)...... 88 Figure 5.7 Response of female Latrodectus hesperus to playback of high- and low-amplitude prey and male vibrations. (a) Proportion of Latrodectus hesperus females responding aggressively to vibrations produced by house fly prey or conspecific males played back at high or low amplitude at a distance of 15 cm (n = 16 for each treatment). Whiskers = standard error; (b) Time [s] elapsed before females initiated a predatory response (number of responding females: n = 5 for prey/low, n = 7 for male/low, n = 14 for both prey/high and male/high)...... 89

xiv

Chapter 1.

Introduction

Communication involves the emission of a meaningful message between a sender and a receiver (Wilson, 1975). Communication signals of animals are transmitted through a signalling medium to reach the intended receiver. Ideally, the signals reach the receiver with a minimum loss of energy and without attenuation or degradation. Any signal alterations can be kept to a minimum by matching the properties of the signal with the properties of the signalling medium (Endler, 1992).

Spiders provide a fascinating system for the study of communication. Spiders are a large and diverse group of arthropods, comprising over 46000 described species, 3988 genera and 114 families (World spider catalog, 2016). One defining characteristic of spiders is the ability to produce silk webs. Webs are highly variable in shape and also serve many functions. These include protection against predators, prey capture, and courtship (Vollrath and Selden, 2007). Webs are the signalling medium that propagates vibrations from potential predators, prey or mates to the spider.

We currently have little understanding of the signalling properties of spider webs. Vibration transmission through webs has been studied in detail only in orb webs (Landolfa and Barth, 1996; Masters, 1984). Yet spider webs come in many forms (Blackledge et al., 2009; Foelix, 1996; Garrison et al., 2016). Orb webs are best known, but very common types also include sheet webs and cob webs, and these are the focus of the present thesis. I explore web-borne vibrations, as well as the behavioral context in which they are used, in Eratigena agrestis, which builds a funnel web (a type of sheet web) and Latrodectus hesperus, which builds a tangle web (a type of cob web).

E. agrestis (Walckenaer, 1802), the hobo spider, belongs to the family Agelenidae. It is widespread in Western Europe. It was introduced to the Pacific

Northwest of North America at the beginning of the 20th century (Crawford and Vest, 1989) and its range has since expanded (Vetter et al., 2003). The hobo spider's web is a horizontal silk sheet with a funnel-shaped retreat at one end. L. hesperus (Chamberlin and Ivie, 1935), the western black widow spider, is a Theridiidae. It is native to Western North America. Its web consists of a loose and irregular tangle of silk suspended over the substrate by supporting threads. Vertical sticky threads connect the tangle to the ground and trap passing arthropods (Foelix, 1996).

In chapter 2 I investigated the life history of Eratigena agrestis. Despite its widespread distribution in Western Europe and its invasive status in the Pacific Northwest, its ecology and natural history have not been described. This information is useful for understanding the distribution of this species and gives insight of potential use for conservation or management. For this thesis, this information is also critical for the design of behavioral studies and interpretation of results.

In chapter 3 I focussed on courtship interactions in Latrodectus hesperus. Males display on female webs. Courtship involves multiple behaviors and multiple signalling modalities: courting male behaviors generate vibrations, males deposit silk which may bear a chemical signal, and males also make contact with females during courtship's final phase. Accounts to date are incomplete (Kaston, 1970; Ross and Smith, 1979). I provide a quantitative description of the complex courtship sequence to fill this gap, and to infer the function of vibratory signals that I recorded and manipulated in chapter 5.

In chapter 4 I documented the web vibration transmission properties of E. agrestis and L. hesperus webs. The two species feature contrasting web architectures (funnel-web and tangle-web, respectively). They therefore provided an opportunity to ask whether web architecture influences vibration propagation. This complements previous research on web vibration that focussed on orb webs. A key feature of this study, given the complex and variable nature of the web types of our focal species, was to assess the variability of signal transmission, both within and between webs.

The findings of chapter 4 led me to ask which vibration parameters may convey information to a web's occupant. In chapter 5 I recorded the vibrations produced by courting males and two types of prey (cricket and fly) on the webs of E. agrestis and L.

hesperus. I asked whether the vibrations produced by males were similar to those of prey (which may maximise detection by females) or distinct (which may minimise female aggression). Male vibrations differed more from prey in L. hesperus than in E. agrestis. I then asked which vibration parameter(s) trigger a predatory response in females.

References

Blackledge, T. A., Scharff, N., Coddington, J. A., Szüts, T., Wenzel, J. W., Hayashi, C. Y., et al. (2009). Reconstructing web evolution and spider diversification in the molecular era. Proceedings of the National Academy of Sciences of the United States of America, 106(13), 5229-5234.

Crawford, R. and Vest, D. K. (1989). The hobo spider and other European house spiders, Burke Museum Educational Bulletin (Vol. 1, pp. 4).

Endler, J. A. (1992). Signals, signal conditions, and the direction of evolution. American Naturalist, 139, S125-S153.

Foelix, R. F. (1996). Biology of spiders (2nd ed.). Oxford: Oxford University Press.

Garrison, N. L., Rodriguez, J., Agnarsson, I., Coddington, J. A., Griswold, C. E., Hamilton, C. A., et al. (2016). Spider phylogenomics: untangling the Spider Tree of Life. PeerJ, 4, e1719.

Kaston, B. J. (1970). Comparative biology of American black widow spiders. Transactions of the San Diego Society of Natural History, 16, 33-82.

Landolfa, M. A. and Barth, F. G. (1996). Vibrations in the orb web of the spider Nephila clavipes: cues for discrimination and orientation. Journal of comparative physiology A: Neuroethology, sensory, neural, and behavioral physiology, 179(4), 493-508.

Masters, W. M. (1984). Vibrations in the orbwebs of Nuctenea sclopetaria (Araneidae) .1. Transmission through the web. Behavioral Ecology and Sociobiology, 15(3), 207-215.

Ross, K. and 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. Journal of Arachnology, 7(1), 69-77.

Vetter, R. S., Roe, A. H., Bennett, R. G., Baird, C. R., Royce, L. A., Lanier, W. T., et al. (2003). Distribution of the medically-implicated hobo spider (Araneae : Agelenidae) and a benign congener, Tegenaria duellica, in the United States and Canada. Journal of Medical Entomology, 40(2), 159-164.

Vollrath, F. and Selden, P. (2007). The role of behavior in the evolution of spiders, silks, and webs. Annual Review of Ecology Evolution and Systematics, 38, 819-846.

Wilson, E. O. (1975). Sociobiology: the new synthesis. Cambridge: Harvard University Press.

World Spider Catalog (2016). World Spider Catalog. Natural History Museum Bern, online at http://wsc.nmbe.ch, version 17.0, accessed on July 2 2016.

Chapter 2.

Life history of the funnel-web spiders Eratigena agrestis and E. atrica (Araneae: Agelenidae) in the Pacific Northwest

Samantha Vibert, Maxence Salomon, Catherine Scott, Gwylim S. Blackburn and Gerhard Gries

This chapter has been formatted for and submitted to The Canadian Entomologist

2.1. Abstract

The life history of the funnel-web spider Eratigena agrestis (Walckenaer) (Araneae: Agelenidae) is not well studied despite its widespread occurrence in Europe and its establishment and spread in the Pacific Northwest of North America since its introduction in the early 20th century. We document life-history traits of E. agrestis and another co-occurring funnel-web spider, Eratigena atrica (C. L. Koch), in two study sites in British Columbia, Canada. The most notable difference in phenology between the two Eratigena species was the timing of emergence: E. atrica spiderlings emerge in the fall whereas E. agrestis spiderlings emerge in the spring. Surprisingly, the contrasting densities of E. atrica in the two study sites and the presence of the black widow spider Latrodectus hesperus Chamberlin and Ivie (Araneae: Theridiidae) in one study site had little effect on the life history of E. agrestis. This unexpected finding may be explained by (i) low overall competition pressure in the study habitats, (ii) con- and heterospecifics exerting equivalent competition or predation pressures; and/or (iii) aggregations of heterospecifics providing benefits that offset costs associated with any competition.

2.2. Introduction

The life history of an individual is defined by traits that describe its growth, reproduction and survivorship. These traits, in turn, are affected by an individual's biotic and abiotic environment (i.e., temperature, humidity, food availability, the presence of competitors or predators). Interpreting the ecology, evolution or behaviour of a species requires knowledge of its life history. Yet, for many organisms detailed life-history accounts are lacking.

Here, we document the life history and phenology of two species of sheet-web spiders: the hobo spider Eratigena agrestis (Walckenaer) (Araneae: Agelenidae) (formerly Tegenaria agrestis; Bolzern et al. 2013) and the giant house spider Eratigena atrica (C. L. Koch), (formerly Tegenaria duellica; Bolzern et al. 2013), with a focus on E. agrestis. These two members of the family Agelenidae were introduced to the coastal Pacific Northwest of North America from Europe a century ago (Crawford and Vest 1989). Their range has since expanded across this region into a variety of habitats (Vetter et al. 2003). Eratigena atrica, in particular, is often found near or inside houses. Because specimens of both species are relatively large, brown and highly mobile, they are regularly misidentified as brown recluse spiders, Loxosceles reclusa (Gertsch and Mulaik) (Araneae: Sicariidae) (Bennett and Vetter 2004). Reports that E. agrestis bites can cause necrotic lesions have resulted in negative perception by the public, despite being unsubstantiated (McKeown et al. 2014). It is of interest to better understand the life cycle and life history of these spiders in the wild, both for their potential management and to inform further, more detailed, studies of their ecology and behaviour.

We recorded life history traits of E. agrestis and E. atrica in natural habitats in southern British Columbia, Canada. We surveyed two sites to assess variations in life history among populations of the two study species. The two study sites feature marked differences in the proportions of the two species, and of a third species that overlaps strongly in habitat use: the western black widow spider, Latrodectus hesperus Chamberlin and Ivie (Araneae: Theridiidae). Comparisons across study sites offer insight as to how varying heterospecific competition and predation pressures may affect life history.

2.3. Methods

2.3.1. Study sites

We studied populations of E. agrestis and E. atrica at two field sites separated by approximately 60 km in southern coastal British Columbia, Canada. The first site (48˚35' N, 123˚22' W; elevation: 3-4 m) was located at Island View Beach (referred to hereafter as "IVB") on the Saanich peninsula of southern Vancouver Island. The second site (49˚13' N, 123˚12' W; elevation: 0.5 m) was located at Iona Beach Regional Park ("IoB") on the mainland in the Greater Vancouver Regional District (Fig. 1). Both field sites are beach habitats featuring open sandy areas interspersed with logs of driftwood. Eratigena agrestis and E. atrica commonly build their funnel-sheet webs beneath these logs. Both E. agrestis and E. atrica are abundant at IVB, but E. agrestis predominates at IoB (Fig. 1). Latrodectus hesperus also builds cobwebs under driftwood logs at IVB and can prey upon adults and juveniles of both Eratigena species (Salomon 2011).

At both study sites, the sparse vegetation between logs consists of grasses, sedges, herbs and dwarf shrubs. Our study species were not observed to occupy this vegetation. Predominant plant species on IVB are Abronia latifolia, Ambrosia chamissonis, Carex macrocephala, Convolvulus soldanella, Festuca rubra, Grindelia integrifolia, Linaria genistifolia ssp. dalmatica, Lomatium nudicaule, Poa spp., Polygonum parochnychia and Rumex acetosella. Predominant plant species on IoB are Carex macrocephala, Elymus sp., Lathyrus japonicus, Poa sp. and Rumex acetosella, with patches of the moss Racomitrium canescens. Both sites are relatively windy, and are susceptible to flooding during winter storms, and to disturbance by humans.

Annual temperature profiles of IVB and IoB were similar over the course of the surveys at each site (Fig. 2). In 2006, the daily average temperatures for IVB and IoB were 9.7 ºC and 10.1 ºC, respectively, according to weather stations located within 6 km of each site at similar elevations (Environment Canada). IoB received more precipitation over the course of the study period (1199 mm) than IVB (883.3 mm).

2.3.2. Survey protocol

For one year at both study sites, we conducted monthly surveys of driftwood logs to document ecological and life-history traits of E. agrestis and E. atrica. At each site, we followed spider populations within study areas that were comparable in size and microhabitat characteristics (Table 1). Logistical constraints limited our survey to one year per site. At IVB, we surveyed from May 2005 to April 2006, at which point a major disturbance by vehicles compromised the integrity of the habitat and forced us to terminate the study. At IoB, we surveyed from May 2006 to April 2007. In January 2007, an extensive and persistent snow cover prevented a survey of the IoB site. The comparison among sites, therefore, is based on the assumption that the data reflect general annual patterns at each location (Fig. 2).

At each site, we measured the length and width of each log to assess the space available underneath. We turned over all logs that were at least 0.30 m long and detached from the ground substrate, and surveyed these logs for the presence of spiders. This length represents the minimum log length used by E. agrestis, E. atrica and L. hesperus (Salomon et al. 2010). At both sites, park visitors occasionally disturbed logs. These minor disturbances were typically restricted to only one or two logs. When logs were disturbed, we returned the affected logs to their original position, unless spiders had colonized the new underside. In the latter case, we surveyed the logs in their new position.

2.3.3. Variables measured

At each survey date and for each log, we recorded the following variables: (1) numbers of E. agrestis and E. atrica, (2) sex and age class of each individual [juvenile, subadult (i.e. penultimate instar), adult], (3) size (total body length) of each individual, (4) number of egg sacs, and (5) number of webs. We also recorded the number of L. hesperus and their cobwebs present under the logs. In spiders, total body length is an index of both age and feeding history (Toft and Wise 1999; Uetz et al. 2002). Using a glass vial, we carefully picked up each spider and measured total body length with callipers (the distance from the chelicerae to the tip of the abdomen) to the nearest 0.1 mm. We then returned the spiders to the exact location under the log where we had

originally found them. We identified all juveniles to species level based on somatic characters (Vetter and Antonelli 2002). To reliably compare growth patterns of juveniles across species and field sites, we grouped measurements of spiderlings into five size categories: 1 (2-4 mm), 2 (4-6 mm), 3 (6-8 mm), 4 (8-10 mm), and 5 (>10 mm). We confirmed the accuracy of species identifications by collecting 63 juveniles at IoB and by rearing them to adulthood in the laboratory. Voucher specimens will be deposited at the Royal British Columbia Museum, Victoria, Canada.

We collected 71 E. agrestis egg sacs at IVB in December 2005 to determine (i) the number of eggs in each sac, (ii) the timing of emergence, and (iii) the number of moults before the spiders reached adulthood. We focused on E. agrestis because E. atrica egg sacs were not available at the same time in the season. In a pilot study, we brought a dozen freshly deposited egg sacs indoors in September. Three months later no spiderlings had emerged from any of the sacs. This led us to suspect that E. agrestis egg sacs require an overwintering diapause. To maximize the chance of achieving hatching success, we randomly assigned each of the 71 egg sacs to one of two treatments. In the first treatment, we housed egg sacs outdoors until mid-February (2006), after which average temperatures increase (Fig. 2), and then moved them into a temperature-controlled growth chamber. There, we gradually increased the temperature over the course of 10 days from 10 ºC to 22 ºC, in increments of 2 ºC every two days. We changed night temperature in the same manner, but from 5 ºC to 16 ºC. In the second treatment, we kept egg sacs outdoors until spiderlings emerged. This allowed us to determine when emergence would occur under natural conditions. We monitored both treatment groups daily until emergence occurred, recording the date of emergence and the number of spiderlings for each egg sac. Immediately upon emergence, we randomly selected four spiderlings per egg sac and placed each one into its own plastic Petri dish (60 × 10 mm; n = 120). In two instances, the brood selected consisted of only three spiderlings, in which case we selected all three. Three times per week, we provided water and a diet of fruit flies (Drosophila melanogaster), house flies (Musca domestica), house crickets (Acheta domesticus) and mealworms (Tenebrio molitor), and noted the date of each spider moult. We retained all lab-reared spiders for use in experiments.

Two days after E. agrestis spiderlings first started to emerge in the laboratory, we transferred each brood to a separate Petri dish (150 × 25 mm). Each brood was large

and mobile, making it difficult to count the number of spiderlings. To ensure data accuracy, we counted each brood 3 times and calculated the mean number of spiderlings in each brood.

We compared the total body length of adult female Eratigena agrestis spiders across field sites (at least 10 females per site) in September, October, November and December.

2.4. Results

Over the 12-month survey at each site, we observed a total of 756 E. agrestis, 150 E. atrica and 0 L. hesperus underneath 124 logs at IoB, and a total of 538 E. agrestis, 588 E. atrica and 839 L. hesperus underneath 68 logs at IVB. Consequently, the proportions and population densities of the three study species differ among sites (Fig. 1; Table 1).

The egg-laying season of E. agrestis and E. atrica overlapped, beginning in August/September and lasting for 3 months (Fig. 3). Notably, the number of egg sacs doubled at both sites between September and December (Fig. 4). In both species, egg sacs were most numerous in December (Fig. 4). In the field, E. atrica spiderlings emerged in the fall, whereas those of E. agrestis emerged in the spring. At IoB, second- instar (i.e. just emerged) E. atrica spiderlings were present in large numbers from September to December, whereas second-instar E. agrestis spiderlings were present almost exclusively in July. At IVB, the number of second-instar E. atrica spiderlings (2-4 mm size category) began to increase in September and peaked in December (Fig. 5). Comparable data for second-instar E. agrestis spiderlings at IVB are not available due to uncertainties in the identification of small specimens during the first month of study at IVB. By the time we conducted the phenology survey at IoB, we were able to distinguish E. agrestis spiderlings from those of E. atrica.

However, data from the egg sacs we collected at IVB and reared in the laboratory showed differences in emergence date based on our treatments. Spiderlings emerged from egg sacs that were kept outdoors between May 31 and June 12 (i.e., >5

months after collection). Of the 38 eggs sacs we first kept outdoors and then moved indoors, spiderlings emerged from 26 sacs between March 17-20, less than four months after collection, and one month after we had moved them indoors. The contrasting emergence times of spiderlings in these two treatment groups implies that emergence is mediated, at least in part, by ambient temperature.

On average, 64.73 ± 24.76 (± SD) spiderlings emerged from the egg sacs reared indoors (range: 17-124). A mean of 60.52 ± 26.70 spiderlings emerged from the egg sacs left outdoors (range: 2-115).

We monitored moulting patterns for the 120 E. agrestis spiderlings that we housed singly immediately after emergence from the egg sac. The first moult took place within the egg sac after hatching and prior to emergence. Males matured after 8-12 moults and females after 7-9 moults (Table 2). Twenty-three males reached maturity, but did so over a protracted period (between July 2006 and April 2007), suggesting a potential delay imposed by conditions in captivity, probably a sub-optimal diet (Toft and Wise 1999). Similarly, 23 females reached maturity, completing their final moult in July and August 2006.

Developmental data of spiderlings are summarized in Figs. 4 and 5. For E. atrica at IVB, most size-1 spiderlings (2-4 mm) were present in December, indicating that they emerged the previous month (Fig. 5). Most size-2 (4-6 mm) juveniles were present from December to April (Fig. 5). Spiderlings in the larger size categories 3 (6-8 mm); 4 (8-10 mm); and 5 (> 10 mm) were present mainly from May to July, suggesting that they reach maturity after July. At IoB, E. atrica spiderlings were much less abundant but their phenology in general resembled that observed at IVB.

For E. agrestis at IVB, the density of spiders in the 4-6 mm, 6-8 mm and 8-10 mm size categories peaked in September, from December to April, and from March to May, respectively (Fig. 5). After May, most spiders presumably reached maturity. Phenology data of E. agrestis at IoB and IVB are highly comparable (Fig. 4 and 5). They showed that (i) size-1 juveniles (size 2-4 mm) peaked in frequency in July (indicative of emergence the previous month), (ii) size-2 (4-6 mm) spiderlings were most prevalent over the winter months, (iii) size-3 (6-8 mm) juveniles peaked in frequency through April

and May, (iv) size-4 (8-10 mm) spiderlings were found in the spring and early summer, and (v) mature spiders were present thereafter (Figs. 4 and 5).

At both field sites, E. agrestis males seemed to mature prior to conspecific females (Fig. 4). While we first observed adult E. agrestis males in July, with a peak abundance in August, adult females first appeared in August and reached their greatest abundance in October. By then, the number of adult E. agrestis males had already declined substantially. At IVB and IoB, the number of adult E. atrica males peaked in July, then declined sharply. At IoB, E. atrica females started maturing in August, with the highest numbers observed in October and November. Strangely, we did not observe a similar pattern at IVB. There, from a peak observed in July, the number of adult females tapered slowly throughout the following months.

Box plots comparing the body size of adult E. agrestis females collected from September to December (the only months where there were at least 10 adult females at both sites) revealed no body size differences between individuals from different field sites (Fig. 6).

2.5. Discussion

Overall, our phenological study of two closely related funnel-web spider species Eratigena agrestis and E. atrica in their natural habitat revealed that their life histories are similar except that spiderlings from each species emerge in different seasons. We also observed strong consistency in life-history traits across the two field sites. Below, we discuss the implications of these results and suggest avenues for future research.

2.5.1. Life cycles of E. agrestis vs. E. atrica

Both species have a similar life cycle at the study sites. Their mating seasons take place in late summer and early fall, with the first adult males appearing in July. Male abundance peaks in August then declines rapidly as mature adult males stop feeding. Eggs sacs are produced from September to November. In both species, there is an intriguing mismatch between the time when females mature and when adult males are

present: the maximum number of adult females is reached only following the decline of male numbers. This pattern may be consistent with a mating system in which females mate only once and males exhibit a strong preference for virgin females (Stoltz et al. 2007), and/or a system in which there is first-male sperm precedence. Males may also guard subadult females, as in some thomisid (Dodson and Beck 1993) and araneid spiders (Fahey and Elgar 1997). A detailed analysis of the mating behaviours of each species during the peak mating season would reveal which of these male behaviours is most prevalent. Furthermore, the fact that females mature late in the season raises the possibility that they may only reproduce during the next year’s mating season, provided they survive the winter.

The only major difference in life cycle between the two species concerns juveniles. Eratigena atrica spiderlings emerge from egg sacs deposited in early autumn after only a few weeks, whereas E. agrestis eggs overwinter and spiderlings emerge during the spring. This means that most E. agrestis juveniles mature the following year. As a possible consequence, there could be substantial differences in annual survival between the two species, provided that seasons vary across years. For example, under harsh winter conditions it is possible that the delayed emergence of E. agrestis maximizes survival. An interesting direction for future research at the two study sites would be to compare survival rates of E. agrestis and E. atrica in relation to variation in weather conditions.

2.5.2. Life history traits of E. agrestis and E. atrica in ecologically contrasting populations

Our study compared the life history traits of E. agrestis across two field sites that feature strongly contrasting densities of E. atrica and L. hesperus (Fig. 1), two species that are potential competitors or predators of E. agrestis. At IoB, L. hesperus is not present and the density of E. atrica is low. Conversely, at IVB, L. hesperus is abundant and E. agrestis and E. atrica are found at similar densities. As a consequence of these contrasting community compositions, it is conceivable that life history traits of E. agrestis differ based on the variation in potential competition and predation pressure. However, none of the factors we measured differed significantly between field sites. One potential explanation for this result is that competition between the two Eratigena species and

predation risks are low in our coastal habitats. However, several lines of evidence suggest otherwise.

First, all three species are generalist predators. In a previous study of L. hesperus at IVB, Salomon (2011) found that this species preys upon a broad range of arthropods, including E. agrestis and E. atrica spiders. Although the diet of E. agrestis and of E. atrica has not yet been specifically characterized, we have observed both species capturing a variety of arthropod prey, such as Diptera, Isopoda, Lepidoptera and Araneae. The prey spectra of these three spiders species appears to overlap broadly, even though L. hesperus is capable of subduing much larger prey compared to E. agrestis and E. atrica.

Second, the three study species occupy the same microhabitats: the underside of driftwood logs. In late summer, when spider numbers were at their peak, we frequently found all three species under a same log, with L. hesperus inhabiting >90 % of logs surveyed at IVB. Salomon et al. (2010) documented the rapid colonization of experimentally introduced habitat logs by all three species at IVB, providing strong evidence that naturally occurring logs are a limited resource. Lastly, remains of E. atrica and E. agrestis spiders beneath L. hesperus webs indicate that both Eratigena species are at least occasional prey of L. hesperus (Salomon 2011; this study).

An alternative explanation for the similarity in life-history characteristics across study sites is that conspecific and heterospecific individuals exert similar competition and predation pressures. In this case, community composition would have no bearing on life- history traits. This intriguing possibility suggests potential equivalence of community members in terms of their effect on the life-history traits we measured. However, net competition or predation pressures likely differed between the two sites because spider densities at IVB exceeded those at IoB four-fold (Table 1). Studying the costs of competition and predation for these spiders at IVB and IoB would provide estimates of resource availability and spider mortality. Such as study could be coupled with experimental manipulations of spider density or resource availability to highlight the potential effects of interspecific competition on life history.

Finally, it is possible that competition or predation alter life-history traits other than those measured here. However, we focused on several key traits such as phenology, growth and adult size, all of which are expected to be constrained by resource availability (Toft and Wise 1999; Oelbermann and Scheu 2002) or predation risk.

Future studies should also accurately quantify fertility by measuring the number of egg sacs produced and spiderling recruitment at each site. These traits are affected by feeding regimes (Spence et al. 1996), and may therefore vary in accordance with competition or predation pressures. The data presented in this study suggest that the life-history traits we document are not affected by strong differences in community composition of close competitors or predators.

2.5.3. Potential implications of group living

The high population densities of spiders sharing microhabitats at IVB are striking, considering that most spider species are highly territorial and aggressive toward other spiders (Whitehouse and Lubin 2005). The observed multi-species aggregations may be the result of habitat constraints, forcing individuals to build their webs in already occupied microhabitats. Despite the possibility for increased competition or predation risk in densely-populated microhabitats, such co-inhabitation might also be beneficial. Support for this possibility is evident in group-living species of spiders, where individuals spin a communal web and cooperate in prey capture. Indeed, group-living has been shown to confer individual benefits, including greater foraging and reproductive success (Whitehouse and Lubin 2005) and protection from predators (Uetz et al. 2002). It is not known whether E. agrestis and E. atrica spiders benefit from forming conspecific or heterospecific associations under driftwood logs. It is possible that the presence of several funnel-webs under a particular log increases prey retention under logs.

Aggregations may also form when spiders seeking suitable microhabitats use the presence of other spiders as an indicator of site quality (Schuck-Paim and Alonso 2001). Spider aggregations such as those encountered at our study sites may persist if benefits offset the costs of searching for and settling in a profitable microhabitat (Jakob 1991; Bilde et al. 2007). A study of microhabitat selection patterns at IVB and IoB would

provide explanations for (i) the high spider densities we observed in our study and (ii) the apparently stable E. agrestis life history across strongly varying community compositions.

Acknowledgements

Robb Bennett introduced us to the spider populations featured in the study. We are also grateful for the logistical support provided by the Bennett family. We thank Sean McCann for the photographs of the spiders. This research was conducted under a permit from the Greater Vancouver Regional District (IoB) and with the permission of the Tsawout First Nation (IVB). Funding was provided by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant and by an NSERC Industrial Research Chair to G.G., with Scotts Canada Ltd. as sponsor.

References

Bennett, R.G. and Vetter, R.S. 2004. An approach to spider bites. Erroneous attribution of dermonecrotic lesions to brown recluse or hobo spider bites in Canada. Canadian Family Physician, 50: 1098–1101.

Bilde, T., Coates, K.S., Birkhofer, K., Bird, T., Maklakov, A.A., Lubin, Y., and Aviles, L. 2007. Survival benefits select for group living in a social spider despite reproductive costs. Journal of Evolutionary Biology, 20: 2412–2426.

Bolzern, A., Burckhardt, D., and Hänggi, A. 2013. Phylogeny and taxonomy of European funnel-web spiders of the Tegenaria−Malthonica complex (Araneae: Agelenidae) based upon morphological and molecular data. Zoological Journal of the Linnean Society, 168: 723–848.

Crawford, R. and Vest, D.K. 1989. The hobo spider and other European house spiders, Burke Museum Educational Bulletin, 1: 1–4.

Dodson, G.N., and Beck, M.W. 1993. Pre-copulatory guarding of penultimate females by male crab spiders, Misumenoides formosipes. Animal Behaviour, 46: 951–959.

Fahey, F.B. and Elgar, A.M. 1997. Sexual cohabitation as mate-guarding in the leaf- curling spider Phonognatha graeffei Keyserling (Araneoidea, Araneae). Behavioral Ecology and Sociobiology, 40: 127–133.

Jakob, E.M. 1991. Costs and benefits of group living for pholcid spiderlings: losing food, saving silk. Animal Behaviour, 41: 711–722.

McKeown, N., Vetter, R.S., and Hendrickson, R.G. 2014. Verified spider bites in Oregon (USA) with the intent to assess hobo spider venom toxicity. Toxicon, 84: 51–55.

Oelbermann, K. and Scheu, S. 2002. Effects of prey type and mixed diets on survival, growth and development of a generalist predator, Pardosa lugubris (Araneae: Lycosidae). Basic and Applied Ecology, 3: 285–291.

Salomon, M. 2011. The natural diet of a polyphagous predator, Latrodectus hesperus (Araneae: Theridiidae), over one year. Journal of Arachnology, 39: 154–160.

Salomon, M., Vibert, S., and Bennett, R.G. 2010. Habitat use by western black widow spiders (Latrodectus hesperus) in coastal British Columbia: evidence of facultative group living. Canadian Journal of Zoology, 88: 334–346.

Schuck-Paim, C. and Alonso, W. J. 2001. Deciding where to settle: conspecific attraction and web site selection in the orb-web spider Nephilengys cruentata. Animal Behaviour, 62: 1007–1012.

Shorthouse, D.P. 2010. SimpleMappr, an online tool to produce publication-quality point maps. [Retrieved from http://www.simplemappr.net. Accessed May 21, 2016]

Spence, J.R., Zimmermann, M., and Wojcicki, J.P. 1996. Effects of food limitation and sexual cannibalism on reproductive output of the nursery web spider Dolomedes triton (Araneae: Pisauridae). Oikos, 75: 373–382.

Stoltz, J.A., McNeil, J.N., and Andrade, M.C.B. 2007. Males assess chemical signals to discriminate just-mated females from virgins in redback spiders. Animal Behaviour, 74: 1669–1674.

Toft, S. and Wise, H.D. 1999. Growth, development, and survival of a generalist predator fed single- and mixed-species diets of different quality. Oecologia, 119: 191–197.

Uetz, G.W., Boyle, J., Hieber, C.S., and Wilcox, R.S. 2002. Antipredator benefits of group living in colonial web-building spiders: the "early warning" effect. Animal Behaviour, 63: 445–452.

Uetz, G.W., Papke, R., and Kilinc, B. 2002. Influence of feeding regime on body size, body condition and a male secondary sexual character in Schizocosa ocreata wolf spiders (Araneae, Lycosidae): condition-dependence in a visual signaling trait. Journal of Arachnology, 30: 461–469.

Vetter, R.S. and Antonelli, A.L. 2002. How to identify (or misidentify) the hobo spider. Washington State University Cooperative Extension Pest Leaflet Series #116

Vetter, R.S., Roe, A.H., Bennett, R.G., Baird, C.R., Royce, L.A., Lanier, W.T., Antonelli, A.L., and Cushing, P.E. 2003. Distribution of the medically-implicated hobo spider (Araneae : Agelenidae) and a benign congener, Tegenaria duellica, in the United States and Canada. Journal of Medical Entomology, 40: 159–164.

Whitehouse, M.E.A. and Lubin, Y. 2005. The functions of societies and the evolution of group living: spider societies as a test case. Biological Reviews, 80: 347–361.

Table 2.1 Summary of ecological characteristics of the study sites at Island View Beach (IVB) and Iona Beach Regional Park (IoB). IVB IoB Approximate size of study site 900 m2 600 m2 Number of driftwood logs (i.e., potential 68b 124c spider microhabitats)a Total surface area available under the logsa 28.94 m2 35.87 m2 Mean (± SD) surface area of the logsa 0.43 (± 0.32) m2 d 0.29 (± 0.32) m2 Spider community composition: total number (A)e, % per site (B)e, spiders per log (C)f A B C A B C Eratigena agrestis 538 23.4 0.66 756 83.4 0.51 Eratigena atrica 588 29.2 0.72 150 16.6 0.10 Latrodectus hesperus 839 47.4 1.03 0 0.0 0.00 aAt the onset of the study; bTwo logs went missing, 14 were moved to new locations; cEight logs went missing, five were moved to new locations; dThe mean surface area under a log was greater at IVB than at IoB (Mann-Whitney U test: P < 0.0001) eTotal number and % per site over the survey period fAverage number of spiders per log over the survey period

Table 2.2 Number of moults preceding maturity in laboratory-reared Eratigena agrestis individuals collected from Island View Beach.

Number of adults Number of moults preceding maturity Females Males 7 4 0 8 11 2 9 8 3 10 0 5 11 0 8 12 0 5

24 C) ° 20 ● ● ● IVB ● ● ● ● ● 16 ● IoB ● ● ● 12 ● ● ● ● ● 8 ● ● ● ● ● ● ● ● ● 4 ● Mean temperatureMean ( 0 400

● 300

● 200 ●

●

● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● Total(mm) precipitaton ● ● ● 0 ● ● July July Apr. Apr. Oct. Oct. May May Jan. Jan. Feb. Feb. Mar. Mar. Aug. Aug. Nov. Nov. Dec. Dec. June June Sept. Sept. 2005 2006 2007 Time (months) Figure 2.2 Mean temperature (°C) and total precipitation (mm) between May 2005 and April 2007 at Island View Beach (Vancouver Island) and Iona Beach (Greater Vancouver Regional District). Data were obtained from Environment Canada weather stations at the Vancouver International Airport and the Victoria International Airport, each situated less than 6 km from the relevant study site.

Figure 2.3 Life cycles of Eratigena agrestis and of E. atrica at Island View Beach and Iona Beach. Each tick on the scales represents a month.

Figure 2.4 Phenology of Eratigena agrestis and E. atrica at Island View Beach (IVB) and Iona Beach (IoB). The graphs represent the densities (numbers of individuals per m2 of available habitat, i.e. driftwood logs) of juveniles, adult females, subadult and adult males, and the number of egg sacs for each species and field site. Data were collected between July 2005 and June 2006 at IVB, and between May 2006 and April 2007 at IoB. The egg sacs spun by E. agrestis and E. atrica are very similar and could not be identified to species level.

Figure 2.5 Changes over time in population densities of Eratigena agrestis and E. atrica juveniles of different sizes at Island View Beach (IVB) and Iona Beach (IoB). The total body length of all juveniles was assigned to one of five size categories: [1: 2-4 mm; 2: 4-6 mm; 3: 6-8 mm; 4: 8-10 mm; 5: > 10 mm]. Data were collected between July 2005 and June 2006 at IVB, and between May 2006 and April 2007 at IoB.

Figure 2.6 Body length (mm) of adult Eratigena agrestis females at Island View Beach (IVB) and Iona Beach (IoB). The boxes depict the median, first and third quartiles. The whiskers represent the lowest and highest datum within the 1.5 interquartile range of the lower and upper quartile. Data were collected in 2005 at IVB and in 2006 at IoB.

Chapter 3.

Evidence that web reduction by western black widow males functions in sexual communication

Catherine Scott, Samantha Vibert and Gerhard Gries

This paper has been published in The Canadian Entomologist, volume 144, issue 5 (2012), pages 672-278. DOI: http://dx.doi.org.proxy.lib.sfu.ca/10.4039/tce.2012.56

3.1. Abstract

A well-accepted function of courtship in sexually dimorphic and cannibalistic spiders is suppression of female predatory responses. We quantitatively analysed courtship in the western black widow, Latrodectus hesperus Chamberlin and Ivie (Araneae: Theridiidae), to determine the behavioural elements of the males’ courtship that are correlated with mating success and/or the females’ responses. The 58% of males that engaged in web reduction elicited fewer aggressive responses from females and induced female quiescence more quickly than did males not exhibiting web reduction behaviour. Our data suggest that web reduction by male L. hesperus functions in sexual communication, a context not previously explored.

3.2. Résumé

Chez les araignées qui présentent un dimorphisme et un cannibalisme sexuels, une fonction reconnue du comportement de cour est la suppression de la réponse de prédation de la femelle. Nous avons analysé quantitativement le comportement de cour chez la veuve noire de l'ouest, Latrodectus hesperus Chamberlin et Ivie (Araneae: Theridiidae), afin de déterminer les éléments du comportement de cour des mâles qui

sont corrélés avec leur succès de reproduction et/ou les réponses de la femelle. 58% des mâles se sont consacrés à la réduction de la toile; ils ont provoqué moins de réponses agressives de la part des femelles, et ont induit plus rapidement un état de transe chez la femelle, que les mâles qui ne se sont pas livrés à ce comportement de réduction de la toile. Nos données suggèrent que la réduction de la toile par les mâles L. hesperus joue un rôle dans la communication entre mâles et femelles, un contexte qui n'a pas été exploré jusqu’à présent.

3.3. Main text

Male western black widows, Latrodectus hesperus Chamberlin and Ivie are attracted to airborne pheromone emanating from females’ webs (Kasumovic and Andrade 2004). Contact with the web elicits courtship in males (Kaston 1970; Ross and Smith 1979). While elements of courtship have previously been described, a more detailed and quantitative report of the complete courtship sequence will allow the formulation of testable hypotheses about the function of individual behavioural elements, including web reduction. Our objectives were to (1) quantitatively analyse the courtship sequence in L. hesperus; and (2) reveal potential relationships between the pattern and duration of courtship behaviour by males (particularly web reduction) and their consequent mating success and the females’ responses.

Immature black widows were collected in summer 2009 from Island View Beach, on the Saanich peninsula of Vancouver Island, British Columbia, Canada (48° 35' N, 123° 22' W, elevation 3–4 m). Spiders were kept in the laboratory at 20–25°C on a reversed 12:12 hour (light:dark) light regime to facilitate experimentation during the spiders’ nocturnal activity phase. They were housed singly in Petri dishes (15 × 2.5 cm) and reared to adults on a diet of house flies and house crickets. Females were fed two days before testing them in mating trials to control for starvation level and minimise the likelihood that they would cannibalise courting males.

Just after moulting, mature females (n = 12) were placed in wood-frame boxes (30 × 30 × 20 cm) with clear plastic and mesh sides, and were allowed to build a web for 2–4 weeks. For mating trials, a virgin male 2–4 weeks post maturity was randomly

assigned to each female and introduced onto her web, which was illuminated by a 25 W incandescent lamp. Each trial was video recorded with a Canon XL2 camcorder until at least one copulation had occurred or >4 hours had elapsed. Males were left in the boxes and those that eventually died without being consumed by the female were removed and stored in ethanol. The size of preserved males was estimated by measuring the tibia- patella length of leg I. Video recordings were analysed using Jwatcher v1.0 (Blumstein et al. 2006). In addition to scoring the occurrence of courtship elements, we quantified female responses and the sequence of events as follows: (1) aggression of females was quantified by counting the number of times the male dropped off the web or walked away from the female immediately following her movement toward him; (2) latency to the female’s quiescence was measured as time from the male’s first movement on the web until his first contact with the female’s abdomen. Abdominal instead of leg-to-leg contact was chosen as an indicator of quiescence because it was always followed by the male mounting the motionless female; (3) latency to first copulation was recorded as the time from the male’s first movement on the web until his insertion of a pedipalp; and (4) copulation duration was recorded as the time from the male’s insertion of a pedipalp until he had fully withdrawn the embolus. Data were analysed by nonparametric statistics (Mann-Whitney U tests; Spearman rank correlations), using JMP 8.0 software (SAS Institute Inc. 2009).

Precopulatory courtship can be divided into distal and proximal phases (summarised in Fig. 1), based on observations of 12 independent mating trials. During the distal phase, the male actively explored the web, continuously trailing dragline silk. All males periodically paused and vibrated their abdomens dorsoventrally. Fifty-eight percent of males engaged in web reduction behaviour, spending 40–90% of the distal phase cutting threads of the female’s web, bundling the loose silk into thick ropes and balls and wrapping them extensively with silk (Fig. 2a). The remaining 42% of males did not visibly reduce the female’s web, but spent 0–7% of this phase occasionally cutting threads and laying down silk by pulling it from their spinnerets with legs IV, with the remaining time spent engaging in other elements of distal courtship. Eventually, the male approached the female, alternating between abdominal vibrations and walking slowly forward, until he made contact with one or more of her legs. If the female did not

respond with aggression, she either remained still or repeatedly twitched her abdomen dorsoventrally before becoming fully quiescent.

During the proximal phase, the male remained on the female’s abdomen or close by her on the web. Typically, each bout of abdominal vibration on the female was followed by a sudden violent shake of the male’s whole body. After the onset of quiescence, 75% of females occasionally moved, but all females remained completely motionless for 15–36 minutes (median = 22 minutes) before copulation occurred. Eleven males copulated at least once before 2.5 hours of courtship had elapsed; the remaining male failed to copulate within 4 hours, even though the female remained motionless for > 2 hours.

Males that engaged in web reduction experienced less female aggression than males that did not (0–6 instances of aggression, median = 3 instances, n = 7, versus 2– 53 instances, median = 19 instances, n = 5; U = 4.5, P < 0.05, Fig. 2b). Web-reducing males also induced female quiescence more quickly (10–62 minutes, median = 28 minutes, n = 7, versus 54–88 minutes, median = 72, n = 5; U = 1, P < 0.01, Fig. 2c). Female aggression and time to female quiescence were not significantly correlated (ρ = 0.428, P = 0.165, n = 12). Time to female quiescence was positively correlated with time to first copulation in the 11 trials in which at least one copulation occurred (ρ = 0.688, P = 0.019, n = 11, Fig. 2d). Copulation duration was not correlated with courtship duration (ρ = 0.007, P = 0.984, n = 11).

Male size was not correlated with female aggression (ρ = -0.195, P = 0.589, n = 10), time to female quiescence (ρ = 0.115, P = 0.751, n = 10), or percentage of the distal phase spent engaging in web reduction behaviour (ρ = 0.600, P = 0.067, n = 10). Two males that did not engage in web reduction were cannibalised after mating and thus were not included in statistical analyses. The percentage of the distal phase that males spent engaging in web reduction behaviour was negatively correlated with their age post moult to mature adults (ρ = -0.604, P = 0.038, n = 12), but was not correlated with the age of females post moult to mature adults, which is equivalent to the age of their webs (ρ = 0.343, P = 0.275, n = 12).

Our quantitative analysis of the courtship sequence in L. hesperus indicates that the female’s sustained quiescence is required for copulation to occur. All but one male copulated at least once, but variation in the latency to copulation suggests that differences in male strategies, male quality, or female receptivity affect the length of courtship required for successful mating. The strong relationship between web reduction by males and the females’ responses (Fig. 2) suggests that web reduction in L. hesperus plays a role in sexual communication, possibly in addition to, or instead of, lessening male-male competition. Web reduction behaviour could transmit, or improve transmission of, vibratory and/or chemical signals that function to suppress female aggression, induce female quiescence, or convey information about male quality for mate choice by females.

Our study adds to current knowledge of courtship behaviour in L. hesperus in that we quantify the sequence of behavioural elements, note the percentage of males engaging in web reduction behaviour, and provide evidence for an alternative function of web reduction. We report for the first time the 15–36 minute period of sustained female quiescence that consistently precedes copulation. Elements of proximal courtship other than the wrapping of the female with silk have been largely overlooked elsewhere. The male’s repeated bouts of abdomen vibrations followed by violent shakes while courting on the female’s abdomen have not been previously mentioned. We speculate that the male’s shakes may ‘test’ the female’s responsiveness, as a means to ensure that she is fully quiescent. That males vibrate their abdomens throughout both phases of courtship hints that this behaviour might transmit an essential signal. The extent to which vibratory, chemical, and tactile signals are involved in inducing and maintaining female quiescence warrants further study.

Web-reducing male L. hesperus experienced less aggression from females and induced their quiescence more quickly than did males not engaging in web reduction. Accelerated quiescence induction seems to ultimately minimise the duration of precopulatory courtship. The females’ ostensibly positive responses to web-reducing males are somewhat surprising because web reduction incurs costs in the form of lost resources (silk), diminished opportunity for prey-capture, and web reconstruction and repair (Schneider and Lubin 1998). Our results suggest that web reduction by males

may improve their mating success by lessening aggression and accelerating induction of quiescence in females.

Lowering male-male competition is the only experimentally supported function of web reduction behaviour in spiders. This function was demonstrated in several linyphiids, in which only females that remain virgins for 7–10 days after their adult moult incorporate pheromone into their silk. Males reduce the web size of these highly receptive virgin females by > 90% before mating, thus minimising pheromone emission from the silk and rendering the web less attractive to other males. Mated females then rebuild their webs void of pheromone (Rovner 1968; Watson 1986; Schulz and Toft 1993). Similar changes in pheromone production by female L. hasselti Thorell (Andrade and Kasumovic 2005; Stoltz et al. 2007) suggest that web reduction by conspecific males may serve the same function as in linyphiids but no data are reported regarding the percentage of males engaging in web reduction or the extent to which webs are reduced. Web modifications may otherwise affect male-male competition by reducing both access routes to the female and the area a male must defend against rivals (Rovner 1968; Forster 1995). Male L. hesperus equally respond to webs of juvenile females, and adult virgin and mated females (Ross and Smith 1979), and males in our study typically reduced the size of virgin females’ webs by much less than 50%. Whether or not web reduction in Latrodectus spp. lessens male-male competition remains to be investigated.

Several alternative explanations for the function of web reduction are consistent with the results of our study. Web reduction might alter signal transmission characteristics of the web (Berendonck 2003), improving both the signal-to-noise ratio of the male’s vibratory message and the female’s ability to receive it. Web reduction behaviour could directly convey signals to the female via the high-amplitude web vibrations produced when silk threads are cut. It might also facilitate the female’s assessment of a male’s quality based on the time and energy he invests in web reduction, vibrations he produces during silk-cutting, or the amount of pheromone- impregnated silk he deposits (Anava and Lubin 1993; Berendonck 2003; Harari et al. 2009). We need to test experimentally whether any of these mechanisms explain the observed responses of females to web-reducing males in our study.

Not all Latrodectus males engage in web reduction (69% of L. revivensis Shulov, Anava and Lubin 1993; 58% of L. hesperus, this study). Age-related differences in the pheromone content of females’ webs are unlikely to explain this variation in courtship behaviour in L. hesperus. All females were virgins 2–4 weeks past their final moult, and we found no relationship between the age of females (equivalent to the age of webs) and the propensity of males to engage in web reduction. Instead, we propose that a signal or cue from a receptive female, which might be her stillness, elicits web reduction behaviour, whereas a cue from a less receptive female – possibly aggressiveness – elicits an alternate pattern of courtship. This concept is consistent with results of a recent study suggesting that the behaviour of L. hesperus females has a greater effect on the courtship of males than do chemical cues on the web (Johnson et al. 2011). Alternatively, it is conceivable that only high-quality males may engage in web reduction behaviour. Male size as a possible indicator of male fitness was not correlated with web reduction in our study, but the age of mature males was negatively correlated with the percentage of distal courtship they spent engaging in web reduction. As mature males do not forage, their body condition deteriorates over time, possibly affecting their ability to engage in energetically costly web reduction behaviour. Having shown here that web reduction is associated with accelerated induction of quiescence and less aggression in females, we will now test whether the occurrence of this behaviour is male- or female- driven.

Acknowledgements

We thank Sean McCann for the photograph and Stevo DeMuth for graphical illustrations. Funding was provided by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant and by an NSERC Industrial Research Chair to G. G., with Contech Enterprises, SC Johnson Canada, and Global Forest Science as sponsors.

References

Anava, A. and 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.B. and Kasumovic, M.M. 2005. Terminal investment strategies and male mate choice: extreme tests of Bateman. Integrative and Comparative Biology, 45: 838–847.

Berendonck, B. 2003. Reproductive strategies in Latrodectus revivensis (Araneae; Theridiidae): functional morphology and sexual cannibalism. Thesis, Heinrich- Heine-Universität, Düsseldorf, Germany.

Blumstein, D.T., Daniel, J.C., and Evans, C.S. 2006. JWatcher 1.0 [online]. Available from http://www.jwatcher.ucla.edu [accessed 11 March 2012].

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

Forster, L.M. 1995. The behavioural ecology of Latrodectus hasselti (Thorell), the Australian redback spider (Araneae: Theridiidae): a review. Records of the Western Australia Museum, Supplement 52: 13–24.

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

Johnson, J.C., Trubl, P., Blackmore, V., and 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.

Kaston, B.J. 1970. Comparative biology of American black widow spiders. Transactions of the San Diego Society of Natural History, 16: 33–82.

Kasumovic, M.M. and Andrade, M.C.B. 2004. Discrimination of airborne pheromones by mate-searching black widow males: species- and population-specific responses. Canadian Journal of Zoology, 82: 1027–1034.

Robinson, M.H. 1982. Courtship and mating behavior in spiders. Annual Review of Entomology, 27: 1–20.

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

Rovner, J.S. 1968. Territoriality in the sheetweb spider, Linyphia triangularis (Clerck) (Araneidae, Linyphiidae). Zeitschrift für Tierpsychologie, 7: 69–77.

SAS Institute Inc. 2009. JMP 8.0. SAS Institute Inc., Cary, North Carolina.

Schulz, S. and Toft, S. 1993. Identification of a sex pheromone from a spider. Science, 260: 1635–1637.

Schneider, J.M. and Lubin, Y. 1998. Intersexual conflict in spiders. Oikos, 83: 496–506.

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

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

Stoltz, J.A., McNeil, J.N., and Andrade, M.C.B. 2007. Males assess chemical signals to discriminate just-mated females from virgins in redback spiders. Animal Behaviour, 74: 1669–1674.

Uhl, G. and Elias, D.O. 2011. Communication. In Spider behaviour: flexibility and versatility. Edited by M.S. Herberstein. Cambridge University Press, Cambridge, UK. pp. 127–189.

Watson, P.J. 1986. Transmission of a female sex pheromone thwarted by males in the spider Linyphia litigiosa (Linyphiidae). Science, 233: 219–221.

Figure 3.2 A) Photograph of a male Latrodectus hesperus engaged in web reduction behaviour (note the dense rope of silk that is the result of his cutting, bundling, and wrapping with silk the section of the female's web which formerly covered the empty space in the bottom two thirds of the image). (B) Mean (+SD) number of instances of aggressive behaviour by female L. hesperus in the distal courtship phase towards males that did (n = 7), or did not (n = 5), engage in web reduction (U = 4.5, P < 0.05). (C) Mean (+SD) latency to female L. hesperusquiescence (time from the male's first movement on the web until his first contact with the female's abdomen) for males that did (n = 7), or did not (n = 5), engage in web reduction (U = 1, P < 0.01). (D) Correlation between time to female L. hesperus quiescence and time to copulation for males that copulated at least once (ρ = 0.688, P = 0.019, n = 11). One male that did engage in web reduction did not copulate within 4 hours and was excluded from the analysis.

Chapter 4.

Vibration transmission through sheet webs of hobo spiders (Eratigena agrestis) and tangle webs of western black widow spiders (Latrodectus hesperus)

Samantha Vibert, Catherine Scott and Gerhard Gries

This chapter has been formatted for and submitted to Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology

4.1. Abstract

Web-building spiders construct their own vibratory signalling environments. Web architecture should affect signal design, and vice versa, such that vibratory signals are transmitted with a minimum of attenuation and degradation. However, the web is the medium through which a spider senses both vibratory signals from courting males and cues produced by captured prey. Moreover, webs function not only in vibration transmission, but also in defense from predators and the elements. These multiple functions may impose conflicting selection pressures on web design. We investigated vibration transmission efficiency and accuracy through two web types with contrasting architectures: sheet webs of Eratigena agrestis (Agelenidae) and tangle webs of Latrodectus hesperus (Theridiidae). We measured vibration transmission efficiencies by playing frequency sweeps through webs with a piezoelectric vibrator and a loudspeaker, recording the resulting web vibrations at several locations on each web using a laser Doppler vibrometer. Transmission efficiencies through both web types were highly variable, with within-web variation greater than among-web variation. There was little difference in transmission efficiencies of longitudinal and transverse vibrations. The inconsistent transmission of specific frequencies through webs suggests that parameters

other than frequency are most important in allowing these spiders to distinguish between vibrations of prey and courting males.

4.2. Introduction

Communication signals of animals are transmitted through a signalling medium to reach the intended receiver. Ideally, the signals reach the receiver with a minimum loss of energy and without attenuation or degradation. Any signal alterations can be kept to a minimum by matching the properties of the signal with the properties of the signalling medium (Endler 1992).

Web-building spiders build their own signalling environment and thus provide a fascinating system to study interactions between signals and the signalling medium (Witt 1975; Maklakov et al. 2003; Vollrath and Selden 2007). Being able to alter the signalling medium, web-building spiders may be able to optimize signal transmission. Yet, spider webs provide not only a medium for sexual signalling, with males courting on females' webs, they also serve for prey-detection and prey-capture, and as shelter from the elements and predators. In response to different evolutionary pressures, spiders have evolved to produce very different types of webs, and to use their webs in different ways (Vollrath and Selden 2007).

Using their webs as an extension of their sensory systems, spiders are quickly alerted to the presence of any prey, predator, competitor or prospective mate that contacts the web and thus causes web vibrations. However, only orb webs have been studied in detail for vibration transmission (Masters 1984; Landolfa and Barth 1996), so variation in transmission properties across contrasting web architectures is not known. Here, we propose to study sheet and tangle webs for their signal transmission characteristics. This study will advance our understanding as to how web architecture affects the transmission of vibrations.

Our study species are the hobo spider, Eratigena agrestis (Agelenidae), which builds a sheet or funnel web (Fig. 1, a), and the western black widow spider, Latrodectus hesperus (Theridiidae), which builds a tangle web (Fig. 1, b). The mostly 2-dimensional

web of E. agrestis consists of a funnel-shaped retreat from which extends a dense sheet of multiple layers of silk threads. The 3-dimensional tangle web of L. hesperus consists of loose threads that are suspended above substrate from supporting threads, with sticky capture threads connecting the tangle to the ground.

These two types of web are highly variable depending on the age and size of the spider, her nutritional state, the web age and the geometry of the surrounding microhabitat (Blackledge and Zevenbergen 2007; S.V. pers. obs.). Parameters such as web size, thread density, number of web attachment points, and overall web shape, all vary. The high degree of variability of these web parameters invokes doubtswhether sheet and tangle webs can serve as a medium for reliable and consistent transmission of vibratory signals or cues.

Our objectives were (1) to characterize longitudinal and transverse vibration transmission of sheet and tangle webs, and (2) to investigate the consistency of frequency transmission both within and between webs.

4.3. Methods

4.3.1. Study spiders

We collected juvenile black widow and hobo spiders from Island View Beach, on the Saanich peninsula of Vancouver Island, British Columbia, Canada. We housed spiders singly in large Petri dishes (100 × 15 cm) under a photoperiod of 12 L : 12 D. We fed juvenile spiders twice a week, and adults once a week, either crickets (Acheta domestica) or house flies (Musca domestica), providing water ad libitum. To facilitate web-spinning, we confined randomly selected virgin females, 10 to 21 days post maturity, inside wood-framed boxes (20 × 20 × 15 cm for hobo spiders; 30 × 30 × 20 cm for black widow spiders) for 10 to 15 days. The mean (SD) masses of adult female hobo spiders and black widow spiders were 186 (37) mg and 226 (41) mg, respectively.

4.3.2. General procedures

Recordings

We recorded web vibrations inside a sound-dampened room. To keep extraneous vibratory noise to a minimum (see Fig. S1 for recordings of background noise), we also placed the entire recording apparatus on a concrete table covered with a thick rubber mat. We recorded web vibrations using a laser Doppler vibrometer (LDV) (Polytec OFV-2500 with OFV-534 sensor head; Polytec GmbH, Waldbronn, Germany), the associated Polytec data acquisition system VIB-E-220, and the VibSoft 4.8 data acquisition software. In sequence, we recorded vibrations at bandwidths of 0 to 1000 Hz (objective 1) and 0 to 500 Hz (objective 2), using a frequency resolution of 78.125 mHz, and a vibrometer range of 10 mm/s/V. All individual recordings lasted 12.8 s, the longest possible data acquisition period.

Frequency sweeps

We used software developed with LabView Graphical Programming for Instrumentation (National Instruments Corporation) to generate two frequency sweeps. The computer generating the sweeps was connected to an amplifier (Creek OBH-21SE) that, in turn, was connected to a piezoelectric vibrator or modified loudspeaker as described below. Both sweeps were constant-amplitude and 12.8 s in duration. The starting frequency was 0 Hz and the end frequency was 1000 Hz (when using the piezoelectric vibrator) or 500 Hz (when using the loudspeaker). Both the sound volume of the computer and the volume of the amplifier were on maximum during playback of sweeps.

Vibration playback devices

The piezoelectric vibrator consisted of a piezoelectric diaphragm 44 mm in diameter and 0.23 mm wide (part no. CEB-44D06, CUI Inc., Tualitin, OR, USA) that we cut into a pointed wedge shape and attached to a metal dowel. The resonant frequency of the piezoelectric element was 0.6 kHz and the resonant impedance was 1000 Ω. We clamped the dowel to an adjustable stand such that the piezoelectric vibrator was suspended pointing down. The vibrator was oriented in the same way for recording of both longitudinal and transverse vibration of the webs. Because many individual silk

threads interconnect in the sheet of an E. agrestis web and in the upper tangle of a black widow webs, vibrations of the piezo in a single direction (always perpendicular to the sheet or upper tangle) produced vibrations that propagated in multiple orientations through the web, including those that we measured (i.e. longitudinal and transverse). Here and throughout the paper, when we refer to longitudinal and transverse vibrations, we refer to vibrations of the sheet or tangle web as a whole, rather than of individual threads.

Vibration-inducing struggling prey or courting males do not generate vibrations > 200 Hz on either E. agrestis or L. hesperus webs (Vibert et al. 2014). We therefore decided to restrict subsequent frequency sweeps to a biologically relevant range of < 500 Hz. As the piezoelectric vibrator produced low-magnitude vibrations below 200 Hz (maximal velocity: 1.1 μm/s on E. agrestis webs and 1.9 μm/s on L. hesperus webs), we further decided to deploy a vibrator that produced high-magnitude vibrations at low frequencies (maximal velocity: 2.85 mm/s on E. agrestis webs and 4.57 mm/s on L. hesperus webs). The vibration-inducing apparatus consisted of a modified unenclosed loudspeaker (12 Ω; diameter 14 cm), with its cone removed and a metal rod (1 mm thick × 18 mm long) attached to the center of the dust cap. We attached the loudspeaker to an adjustable stand so that it could be suspended over the webs.

We used the adjustable stand to position the vibration playback device such that the tip of the piezo element, or the vibrating rod, was facing downward, 1 cm above the topmost threads. Using a small jack, we then raised the box containing the web until the tip of the vibrator was in contact with the web causing a slight tensing of the threads. We positioned the tip of the playback device at the entrance to the funnel for E. agrestis, and at the top of the tangle in the middle of the web for L. hesperus; these are the positions where spiders await prey during their active hunting period.

Recording protocol

Prior to recordings, we removed each female spider from her 10- to 21-day-old web. If a web was damaged during the removal of the female, we returned her to the web for another one to two days to effect repairs. Onto each empty web, we positioned small squares (1 mm², mass [mean ± SE]: 0. 9 mg ± 0.3 mg) of reflective tape (Polytec

Retroreflective Sheeting; Polytec GmbH, Waldbronn, Germany). When using the piezoelectric vibrator, we placed three reflective squares in a straight line at a distance of 1 mm, 20 mm and 70 mm, respectively, from the tip of the vibrator. When using the loudspeaker, we placed five reflective squares as follows: one square (#0 in Fig. 3) just 1 mm from the tip of the rod and two squares each at a distance of 20 mm (#1 and #2) and 70 mm (#3 and #4) from the tip of the vibrator on two lines (Fig. 3). The orientation of these lines varied somewhat from web to web and was guided by the shape of the web, the location of the retreat, and the areas of the web that were most dense (to facilitate placement of reflective squares). Because E. agrestis webs are sheet-like, we attached reflective squares at the intersection of many individual threads, whereas on L. hesperus webs we placed squares at an intersection of three or more threads. We then placed the web-containing box on the vibration-proof table and focussed the LDV beam on the piece of reflective tape closest (1 mm) to the vibrator.

For recordings using the piezoelectric vibrator, we started the 1- to 1000-Hz frequency sweep and the LDV recording of the resultant web vibration simultaneously. Once the sweep was completed, we repeated the procedure focussing on the pieces of reflective tape 20 mm and 70 mm from the piezoelectric vibrator. For 28 E. agrestis webs and 28 L. hesperus webs, we obtained three recordings (one each at a distance of 1 mm, 20 mm and 70 mm from the piezoelectric vibrator), with 14 webs assigned to recordings of longitudinal vibrations and 14 webs assigned to recordings of transverse vibrations. To record transverse vibrations, we oriented the LDV beam at a 90° angle to the plane of the web. To record longitudinal vibrations, we oriented the LDV beam as parallel as possible to the plane of the web. For these recordings, the squares of reflective tape were also oriented at approximately 90° to the plane of the web.

We did not measure transverse vibrations in the plane of the webs (lateral vibrations) because we assumed the results would have been similar to those we obtained with longitudinal vibration measurements. In the upper tangle of the widow web, silk threads of all orientations are interconnected in a 3-dimensional “sheet” approximately parallel to the substrate, and in the sheet of the hobo web, silk threads are oriented in all directions within a roughly 2-dimensional “plane”. Any measurement taken in the “plane” of an E. agrestis web, or parallel to the 3-dimensional upper tangle

of a L. hesperus web, would include both lateral and longitudinal vibrations, thus we expected that taking a second set of measurements would have yielded similar results.

For recordings using the loudspeaker, we began with the beam of the LDV focussed on square #0. In order to record transverse vibrations, we oriented the beam of the LDV at a 90° angle to the plane of the web. We simultaneously triggered the 1- to 500-Hz frequency sweep and the LDV recording of the resultant web vibration. Once we had completed a sweep, we focussed the laser on the next square. From each web, we obtained seven recordings, focussing the beam of the LDV in a fixed sequence on the following reflective squares: #0 (1 mm), #1 (20 mm), #3 (70 mm), #0 (1 mm), #2 (20 mm), #4 (70 mm) and again #0 (1 mm). If the vibration profile at 1 mm differed among subsequent recordings on the same web, we began a new series of recordings. We then used the mean of the three recordings as a baseline for calibrating the 20-mm and 70- mm recordings as described below. We used a total of 16 individual webs of each species. Here we recorded only transverse vibrations because the transmission of longitudinal and transverse vibrations was similar when using the piezoelectric vibrator.

4.3.3. Analysis

Characterizing longitudinal and transverse vibration transmission through webs

From the VibSoft 4.8 data acquisition software, we obtained a measure of a web's velocity (measured in mm/s) at 1-Hz intervals in response to the 1- to 1000-Hz sweep played through the piezoelectric vibrator. For each web, we used the vibration measurement curve obtained at the 1-mm recording point as the baseline against which we measured the vibration transmission efficiency of the web. For the 20-mm and 70- mm recording positions, we calculated the transmission efficiency curve by dividing a web’s velocity at the respective measurement point by the baseline velocity measured at the 1-mm recording position, at 1-Hz intervals, and then by taking the log10 of the resulting ratio.

Consistency of vibration frequency transmission between and within webs

From the VibSoft 4.8 data acquisition software, we obtained a measure of the amplitude of a web's velocity (measured in mm/s) at 1-Hz intervals in response to the 1- to 500-Hz sweep played through the modified loudspeaker. For each web, we averaged the data of the three square #0-recordings and used these mean data as the baseline against which we estimated the transmission efficiency of the web. For recordings at squares #1, #2, #3, and #4, we calculated the transmission efficiency curve by dividing the velocity at the respective square by the baseline velocity (mean of three recordings at square #0), at 1-Hz intervals and then by taking the log10 of the resulting ratio. We then calculated the repeatability of each transmission efficiency measurement using within-web and among-web variance components ( and ) (Nosil and Crespi 2006). We calculated , an estimate of , using the group mean square (MS) from the ANOVA results (Table S2). In the equation below, n is the number of measurements taken on each web (here, n = 2).

Repeatability values can vary between 0 and 1. A repeatability value near 0 indicates that nearly all the variance in transmission efficiency results from differences between separate measurements taken on the same web. A repeatability value near 1 indicates that repeated measurements on the same web result in very similar transmission efficiency profiles. We calculated repeatability values at 10-Hz intervals between 10 and 200 Hz, the range of dominant frequencies produced by prey and males (Vibert et al. 2014). We calculated the repeatability for the distances 20 mm (squares #1 and #2) and 70 mm (squares #3 and #4) from the point of the input vibration (Fig. 3).

4.4. Results

4.4.1. Longitudinal and transverse vibration transmissions of webs

Webs of both E. agrestis and L. hesperus exhibited similar transmission efficiencies of longitudinal and transverse vibrations (Fig. 4a-d). Longitudinal vibrations tended to be more greatly attenuated than transverse vibrations, except at 70 mm in E. agrestis webs, where the reverse was true. The transmission efficiency tended to be greater at 20 mm than at 70 mm in both web types, but the difference was much more pronounced in E. agrestis webs (Fig. 5a-d), indicating that vibrations attenuate differently over distance in E. agrestis and L. hesperus webs. Furthermore, in both web types there was a tendency for transmission loss to increase with frequency up to about 300 Hz, after which it became relatively consistent. There are large differences in transmission efficiency among webs of either species (Table A1), as indicated by the large standard deviations of the data. It seems that the highly variable and complex web structure leads to great differences in vibration transmission profiles. On average, input vibrations of < 50 Hz cause the least mean attenuation. In both species, this low-frequency range overlaps with that of vibrations produced by struggling prey and courting males that generate vibrations mostly between 10 and 75 Hz (Vibert et al., 2014).

4.4.2. Consistency of vibration frequency transmission between and within webs

The transmission efficiency of transverse vibrations caused by 1- to 500-Hz frequency sweeps was highly variable in both E. agrestis (Fig. 6a-d) and L. hesperus (Fig. 6e-h). To determine whether this variability was driven mostly by between-web or within-web variation, we calculated the repeatability of measurements at 10-Hz intervals between 10 and 200 Hz (Table A2). For webs of E. agrestis, the repeatability was on average 0.23 and 0.26 for data collected at a distance of 20 mm and 70 mm, respectively, from the input vibration. For 87.5% of all frequencies tested, the repeatability of measurements was < 0.5 (Fig. 7a). For L. hesperus webs, the repeatability was on average 0.25 and 0.39 for data collected at a distance of 20 mm and 70 mm, respectively, from the input vibration. For 85% of all vibration frequencies tested, the repeatability of measurements was < 0.5 (Fig. 7b). These results show that most of the total variance in transmission efficiency resulted from variation within the same web.

4.5. Discussion

Our data can be summarized as follows: (1) the sheet webs of E. agrestis and tangle webs of L. hesperus revealed large variability in transmission efficiency across the frequency range of the input vibrations; (2) transmission loss is quite variable even within a single web, with the result that transmission profiles obtained at different locations equidistant from the point of the input vibration can be very dissimilar; (3) within-web rather than between-web variation plays a larger role in explaining the variability among transverse vibrations (suggesting that the same is likely true for longitudinal vibrations); and (4) there is little difference in transmission efficiency between longitudinal and transverse vibrations.

Our findings that both sheet and tangle webs have similar transmission efficiencies for longitudinal and transverse vibrations contrast with findings reported for orb webs. This is in part because we measured longitudinal and transverse vibrations of entire webs rather than through individual silk threads, as has been done for orb webs. In webs of Nuctenea sclopetaria, longitudinal vibrations attenuate less than transverse ones (Masters and Markl 1981). The same is true in webs of Nephila clavipes: longitudinal vibrations attenuate five times less than transverse ones (Landolfa and Barth 1996). That sheet and tangle webs have similar transmission efficiencies for longitudinal and transverse vibrations is likely due to the architecture of these types of webs. The tangle of a tangle web is formed by silk threads deposited in all directions, from vertical to horizontal orientations, with multiple points of contact between threads. A sheet web is composed of mostly horizontal threads criss-crossing each other and deposited on top of each other until a dense mat is formed. Given the density of nodes, or points of contact, in this type of web, longitudinal and transverse web vibrations probably combine in a complex manner. This may explain why on E. agrestis webs longitudinal vibrations were more greatly attenuated than transverse vibrations at 20 mm, but the opposite was true at 70 mm (Fig. 4a,b).

Vibrations can be characterized by envelope (amplitude modulation), spectral (frequency) and temporal patterns (duration, periodicity of repeating elements). Information can be conveyed by one or all of these parameters (Bradbury and Vehrencamp 1998). Frequency has been shown to be of significance in birds (Podos

2001), amphibians (Ryan et al. 1990), and insects (Rebar et al. 2009). For E. agrestis and L. hesperus, specific frequencies are probably not of great relevance, as they are not transmitted consistently through the web. However, lower frequencies in the range that includes the dominant frequencies of male and prey vibrations (Vibert et al. 2014) tended to be transmitted with the least attenuation through both web types. Thus, transmission frequency may be important to these spiders in that it must fall into a certain range in order to transmit information. In L. hesperus, the amplitude of a male courtship vibration determines whether or not a female will respond with aggression (Vibert et al. 2014). Temporal patterns may also play an important role, as they do in other spider systems in which spiders communicate via substrate-borne vibrations. For instance, both spectral and temporal properties of male signals affect responses of female spiders in the Ctenidae (Schüch and Barth 1990, Schmitt et al. 1994), Lycosidae (Gibson and Uetz 2008) and Salticidae (Sivalinghem et al. 2010).

The sheet and tangle webs we tested here exhibit a striking degree of variability in the transmission efficiency of vibratory cues or signals, and both types of webs are apparently inferior media for consistent transmission of specific frequencies. Our input vibrations were greatly attenuated and large differences in transmission frequency profiles were noticeable both within and between webs. These phenomena reflect the complexity and variability of web architecture. Contrary to orb webs, E. agrestis sheet webs and L. hesperus tangle webs are not uniform structures. The density of threads and their orientation, degree of tension, number of connections, and distance to anchor points all vary greatly from one area of the web to another, and undoubtedly affect a web’s transmission characteristics. Tangle and sheet webs may be the product of potentially conflicting adaptive imperatives. Superior transmission characteristics of web vibrations may have been compromised in favour of prey capture. For a hunting spider to decide whether to treat the source of a web vibration as prey, she may need to know only the amplitude of that vibration. To then effectively capture the prey, the spider may need to know only the location of the prey on the web. Both types of information are likely sensed by E. agrestis and L. hesperus, even if some vibration characteristics are transmitted inconsistently.

Frequency sweeps are useful to obtain an overall view of the transmission properties of a medium (Masters 1984; Landolfa and Barth 1996) based on controlled

input vibrations. Greater insight into the relevance of these properties for the spider receiving the vibrations transmitted through the web could be gained by examining the transmission of specific vibrations such as those generated by courting males, struggling prey, wind, or debris hitting the web. One could study, in particular, the parameters of courtship signals or prey cues that are transmitted with the highest fidelity, and whether certain areas of the web transmit the information that is relevant to the receiving spider better than others.

Acknowledgements

We thank Stephen Takács for making the frequency sweeps, Stephen DeMuth for some illustrations, and G. Barth and two anonymous reviewers for meticulous reviews and constructive comments. Funding was provided by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant and by an NSERC Industrial Research Chair to G.G., with Scotts Canada Ltd. as the industrial sponsor.

References

Blackledge TA, Zevenbergen JM (2007) Condition-dependent spider web architecture in the western black widow, Latrodectus hesperus. Anim Behav 73:855–864 doi: 10.1016/j.anbehav.2006.10.014

Bradbury JW, Vehrencamp L (1998) Principles of animal communication. Sinauer Associates, Sunderland.

Endler JA (1992) Signals, signal conditions, and the direction of evolution. Am Nat 139:S125–S153 stable url: http://www.jstor.org/stable/2462431

Gibson JS, Uetz GW (2008) Seismic communication and mate choice in wolf spiders: components of male seismic signals and mating success. Anim Behav 75:1253– 1262 doi: 10.1016/j.anbehav.2007.09.026

Landolfa MA, Barth FG (1996) Vibrations in the orb web of the spider Nephila clavipes: cues for discrimination and orientation. J Comp Physiol A 179:493–508 doi: 10.1007/BF00192316

Maklakov AA, Bilde T, Lubin Y (2003) Vibratory courtship in a web-building spider: signalling quality or stimulating the female? Anim Behav 66:623–630 doi: 10.1006/anbe.2003.2245

Masters WM (1984) Vibrations in the orbwebs of Nuctenea sclopetaria (Araneidae). Behav Ecol Sociobiol 15:207–215 doi: 10.1007/BF00292977

Masters WM, Markl H (1981) Vibration signal transmission in spider orb webs. Science 213:363–365 doi: 10.1126/science.213.4505.363

Nosil P, Crespi BJ (2006) Experimental evidence that predation promotes divergence in adaptive radiation. Proc Nat Acad Sci USA 103:9090–9095 doi: 10.1073pnas.0601575103

Podos J (2001) Correlated evolution of morphology and vocal signal structure in Darwin's finches. Nature 409:185–188 doi: 10.1038/35051570

Rebar D, Bailey NW, Zuk M (2009) Courtship song's role during female mate choice in the field cricket Teleogryllus oceanicus. Behav Ecol 20:1307–1314 doi: 10.1093/beheco/arp143

Ryan MJ, Fox JH, Wilczynski W, Rand AS (1990) Sexual selection for sensory exploitation in the frog Physalaemus pustulosus. Nature 343:66–67 doi: 10.1038/343066a0

Schmitt A, Schuster M, Barth FG (1994) Vibratory communication in a wandering spider, Cupiennius getazi: female and male preferences for features of the conspecific male's releaser. Anim Behav 48:1155–1171 doi: 10.1006/anbe.1994.1348

Schüch W, Barth FG (1990) Vibratory communication in a spider: female responses to synthetic male vibrations. J Comp Physiol A 166:817–826 doi: 10.1007/BF00187328

Sivalinghem S, Kasumovic MM, Mason AC, Andrade MCB, Elias DO (2010) Vibratory communication in the jumping spider Phidippus clarus: polyandry, male courtship signals, and mating success. Behav Ecol arq150:1–7 doi: 10.1093/beheco/arq150

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. Front Zool 11:4 doi: 10.1186/1742-9994-11-4

Vollrath F, Selden P (2007) The role of behavior in the evolution of spiders, silks, and webs. Annu Rev Ecol Evol Syst 38:819–846 stable url: http://www.jstor.org/stable/30033881

Witt PN (1975) The web as a means of communication. Biosci Commun 1:7-23.

Figure 4.1 Web structures of Eratigena agrestis and Latrodectus hesperus. Schematic drawing illustrating a the sheet web of Eratigena agrestis, and b the tangle web of Latrodectus hesperus. Webs of E. agrestis consist of a two-dimensional sheet of silk with a funnel at one end serving as a retreat. Webs of L. hesperus consist of a dense three-dimensional tangle of threads. Glue-coated capture threads extend from the tangle to the ground. The stars indicate the locations where the spiders typically reside while waiting for prey. The position of the spiders indicates the location of their retreats. Illustrations reproduced from Vibert et al. (2014).

Figure 4.2 Representative piezo sweeps (a-d) and loudspeaker sweeps (e, f) at 1 mm from the input vibration. Frequency sweeps were generated by a piezoelectric disk or modified loudspeaker and recorded on webs of Eratigena agrestis (a, c, e) and Latrodectus hesperus (b, d, f) by a laser Doppler vibrometer 1 mm away from the source of the input vibration (see Fig. 4). The graphs show the velocity of vibrations (mm/s) over frequency (Hz). The beam of the vibrometer was either parallel or perpendicular to the plane of the webs to capture longitudinal vibrations (a, b) or transverse vibrations (c-f), respectively. Note the different scales of the y-axes: μm/s in a-d and mm/s in e and f.

LDV

#3 #1 #0 #2

10 E. agrestis longitudinal − 20 mm a E. agrestis longitudinal − 70 mm b 5 E. agrestis transverse − 20 mm E. agrestis transverse − 70 mm 0

−5

−10

−15

−20

−25

Transmission(dB) efficiency −30

−35 10 L. hesperus longitudinal − 20 mm c L. hesperus longitudinal − 70 mm d 5 L. hesperus transverse − 20 mm L. hesperus transverse − 70 mm 0

−5

−10

−15

−20

−25

Transmission(dB) efficiency −30

−35 0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000 Frequency (Hz) Frequency (Hz)

Figure 4.4 Comparison of transmission efficiencies for longitudinal and transverse web vibrations. Transmission efficiency of longitudinal and transverse vibrations were measured on webs of Eratigena agrestis (a, b) and Latrodectus hesperus (c, d) (n = 14 each) at distances of 20 mm and 70 mm from a piezoelectric vibrator generating the vibration. Mean transmission efficiency (solid line) and 95% confidence intervals (shaded band) are shown. Transmission efficiency of vibrations (dB) was measured by comparing vibrations recorded at a distance of 20 mm and 70 mm from the vibrator with those obtained just 1 mm away from it.

10 E. agrestis longitudinal − 20 mm a E. agrestis transverse − 20 mm b 5 E. agrestis longitudinal − 70 mm E. agrestis transverse − 70 mm 0

−5

−10

−15

−20

−25

Transmission(dB) efficiency −30

−35 10 L. hesperus longitudinal − 20 mm c L. hesperus transverse − 20 mm d 5 L. hesperus longitudinal − 70 mm L. hesperus transverse − 70 mm 0

−5

−10

−15

−20

−25

Transmission(dB) efficiency −30

−35 0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000 Frequency (Hz) Frequency (Hz)

Figure 4.5 Comparison of transmission efficiencies of longitudinal and transverse vibrations measured at two distances from the point of vibration input. Longitudinal transmission efficiencies of vibrations were measured on webs of Eratigena agrestis (a, b) and Latrodectus hesperus (c, d) (n = 14 each) at distances of 20 mm and 70 mm from a piezoelectric vibrator generating the vibration. Mean transmission efficiency (solid lines) and 95% confidence intervals (shaded bands) are shown. Transmission efficiency of vibrations (dB) was measured by comparing vibrations recorded at a distance of 20 mm and 70 mm from the vibrator with those obtained just 1 mm away from it.

10 E. agrestis transverse − 20 mm 1 a E. agrestis transverse − 70 mm 1 b 5 0 −5 −10 −15 −20 −25 −30 Transmission(dB) efficiency −35 10 E. agrestis transverse − 20 mm 2 c E. agrestis transverse − 70 mm 2 d 5 0 −5 −10 −15 −20 −25 −30 Transmission efficiency (dB) −35 10 L. hesperus transverse − 20 mm 1 e L. hesperus transverse − 70 mm 1 f 5 0 −5 −10 −15 −20 −25 −30 Transmission efficiency (dB) −35 10 L. hesperus transverse − 20 mm 2 g L. hesperus transverse − 70 mm 2 h 5 0 −5 −10 −15 −20 −25 −30 Transmission(dB) efficiency −35 0 50 100 150 200 250 300 350 400 450 500 0 50 100 150 200 250 300 350 400 450 500 Frequency (Hz) Frequency (Hz)

Figure 4.6 Comparison of transmission efficiencies of transverse vibrations measured at two locations equidistant to the point of vibration input. Mean transmission efficiencies (solid line) and standard deviations (grey band) were measured on webs of Eratigena agrestis (a-d) and Latrodectus hesperus (e-h) (n = 16 each). The transmission efficiencies of vibrations (dB) were measured by comparing vibrations recorded at a distance of 20 mm and 70 mm from the loudspeaker vibrator with those just 1 mm away from it.

1 0.9 E. agrestis a 20 mm 0.8 ● 70 mm 0.7 0.6 ●

0.5 ● ● ● 0.4 ● Repeatability ● ● ● 0.3 ● ● ● ● 0.2 ● ● ● ● 0.1 ● ● 0 ● ● 1 0.9 L. hesperus b 0.8 ● 0.7 ● 0.6 ●

● ● 0.5 ● ● ● ● ● 0.4 ● ● ● Repeatability ● 0.3 ● ● ● ● 0.2 0.1 ●

0 ● 0 20 40 60 80 100 120 140 160 180 200 Frequency (Hz)

Chapter 5.

A meal or a male: the ‘whispers’ of black widow males do not trigger a predatory response in females

Samantha Vibert, Catherine Scott and Gerhard Gries

This paper has been published in Frontiers in Zoology, issue 11 (2014) DOI:10.1186/1742-9994-11-4

5.1. Abstract

Introduction

Female spiders are fine-tuned to detect and quickly respond to prey vibrations, presenting a challenge to courting males who must attract a female’s attention but not be mistaken for prey. This is likely particularly important at the onset of courtship when a male enters a female’s web. In web-dwelling spiders, little is known about how males solve this conundrum, or about their courtship signals. Here we used laser Doppler vibrometry to study the vibrations produced by males and prey (house flies and crickets) on tangle webs of the western black widow Latrodectus hesperus and on sheet webs of the hobo spider Tegenaria agrestis. We recorded the vibrations at the location typically occupied by a hunting female spider. We compared the vibrations produced by males and prey in terms of their waveform, dominant frequency, frequency bandwidth, amplitude and duration. We also played back recorded male and prey vibrations through the webs of female L. hesperus to determine the vibratory parameters that trigger a predatory response in females.

Results

We found overlap in waveform between male and prey vibrations in both L. hesperus and T. agrestis. In both species, male vibrations were continuous, of long duration (on average 6.35 s for T. agrestis and 9.31 s for L. hesperus), and lacked complex temporal patterning such as repeated motifs or syllables. Prey vibrations were shorter (1.38 - 2.59 s), sporadic and often percussive. Based on the parameters measured, courtship signals of male L. hesperus differed more markedly from prey cues than did those of T. agrestis. Courtship vibrations of L. hesperus males differed from prey vibrations in terms of dominant frequency, amplitude and duration. Vibrations of T. agrestis males differed from prey in terms of duration only. During a playback experiment, L. hesperus females did not respond aggressively to low-amplitude vibrations irrespective of whether the playback recording was from a prey or a male.

Conclusions

Unlike courtship signals of other spider species, the courtship signals of L. hesperus and T. agrestis males do not have complex temporal patterning. The low- amplitude ‘whispers’ of L. hesperus males at the onset of courtship are less likely to trigger a predatory response in females than the high-amplitude vibrations of struggling prey.

5.2. Introduction

Signals are shaped by the sensory system of the receiver and properties of the signal transmission channel. This “sensory drive” defines constraints imposed by each parameter on others [1,2]. In guppies, for instance, the light environment has shaped the visual system of the fish which, in turn, constrains the colour of males and their display behaviour toward females [3].

In web-dwelling spiders, the web is both a prey-capturing device and the signalling environment through which males transmit vibratory courtship signals [4,5,6,7,8]. This presents a challenge because most spiders are predatory and highly aggressive. Males are at risk of being treated as prey when they enter a female’s web

and start signalling their presence. Indeed, sexual cannibalism in web-dwelling spiders has been widely documented [9,10,11,12]. This risk potentially constrains male courtship strategies.

Given that pre-copulatory cannibalism is never to a male’s advantage [11], a courting male must draw the female’s attention but minimize the risk of being attacked and consumed. In some species, males have evolved efficient strategies that help avoid cannibalism, such as cutting threads of the female’s web to limit her movement, courting from a mating thread [9, 13], or attempting to mate with a moulting female [14].

The signal transmission properties of webs likely exert a strong influence on the male’s signalling strategy. Both the web and the female’s sensory system are fine-tuned to detect prey vibrations [15,16]. Males signalling with prey-like vibrations during courtship may be readily detected by females but may then face a predatory attack. Males avoiding prey-like vibrations may maximize their survival by clearly advertising themselves as potential mates. As yet little is known about how males signal their presence when they enter a female’s web [17,18], or about vibrations that entangled and struggling prey produce [15,19].

In our study we focus on the onset of courtship, when a male enters a female’s web. The identity-signalling challenge he faces is expected to occur in this early phase of courtship. Our study addresses web vibrations from the perspective of the female spider as the receiver of vibratory signals or cues, detecting vibrations from all areas of her web and from various sources. We document vibrations as they reach the female’s location after transmission through the web, rather than at the source, prior to transmission. We chose two species of web-dwelling spiders: the western black widow spider, Latrodectus hesperus Chamberlin and Ivie (Araneae: Theridiidae) which produces a tangle web, and the hobo spider, Tegenaria agrestis Walckenaer (Araneae: Agelenidae) which produces a sheet web. In both species courtship is lengthy (2–3 h for L. hesperus and 0.5-1.0 h for T. agrestis). The male’s courtship display takes place on the female’s web and comprises repeated behavioural elements that cause distinctive vibrations. In L. hesperus, females are much larger than males and exhibit aggression toward males in the early phase of courtship [20]. In T. agrestis, females are only slightly

larger than males and are seldom aggressive towards them prior to copulation (S. Vibert, unpublished data).

Vibrations can be characterized by envelope (amplitude modulation), spectral (frequency) and temporal patterns (duration, periodicity of repeating elements). Information can be conveyed by all of these parameters [21]. To determine the parameter(s) triggering a behavioural response in a receiver, playback experiments have been widely used across taxa, including spiders [22,23], katydids [24], and tree frogs [25].

Our objectives were to (1) characterize vibrations produced by prey [house flies (Musca domestica); house crickets (Acheta domesticus)] and male spiders (L. hesperus, T. agrestis) during the first phase of courtship, at the female’s location; (2) determine whether male vibratory courtship signals differ from prey vibratory cues; and (3) determine the vibration parameter(s) that trigger a predatory response in females.

5.3. Material and methods

5.3.1. Characterization of prey and male vibrations on L. hesperus and T. agrestis webs

(a) Study animals

We chose our two study species because they are locally abundant and easily reared. We collected juvenile L. hesperus and T. agrestis from Island View Beach (48° 35' N, 123° 22' W, elevation 3–4 m), on the Saanich peninsula of Vancouver Island, British Columbia, Canada. We housed spiders individually in large Petri dishes (15 × 2.5 cm) at 20–25°C on a reversed 12:12 h (light:dark) light regime to facilitate experimentation during the spiders’ nocturnal activity phase. We raised spiders to adults on a diet of laboratory-reared house crickets and house flies, with water ad libitum. Like many spiders, L. hesperus and T. agrestis are generalist predators. They feed on various prey, both flying and crawling. We recorded vibrations produced by house crickets and flies in order to capture some of the diversity of prey vibrations to which L. hesperus and T. agrestis respond so that we could compare transmission characteristics

of the same prey vibrations on two types of webs. Because adult T. agrestis females do not attack prey larger than themselves (S. Vibert, personal observation), we used nymphal 3-week-old crickets (mean mass: 27.1 mg (7.3 SD; n = 25)) and adult house flies (mean mass: 15.6 mg (4.3 SD; n = 25)) in our experiments. Both T. agrestis and L. hesperus females feed readily on these prey items and have been successfully reared on such a diet. Ten days post maturity, we placed virgin L. hesperus and T. agrestis females singly inside wood-framed boxes (30 × 30 × 20 cm and 15 × 20 × 15 cm, respectively) and allowed them to spin a web for 10 to 15 days. We tested a total of 27 and 18 webs in L. hesperus and T. agrestis trials, respectively, using virgin male spiders 7–10 days post maturity (mean mass (SD) of L. hesperus males: 22.6 mg (4.1); n = 21; of T. agrestis males: 154.0 mg (31.2); n = 17).

(b) Courtship behaviours

Both L. hesperus and T. agrestis males engage in the first (distal) phase of courtship in the absence of a female. Courtship of L. hesperus males consists of bouts of abdominal tremulations (dorso-ventral oscillations of the abdomen while stationary; see Additional file B1) and exploration of the web. Moreover, some males also cut some of the web’s threads, and bundle cut sections with their own silk [20]. Courting T. agrestis males engage in an extensive exploration of the female web. Their walking is always coupled with drumming with the pedipalps, tapping the web with the first pair of legs, and depositing silk (see Additional file B2 for a video recording). Occasionally, after several minutes of exploration, a male stands still and slowly drums with his pedipalps for a few seconds. Thereafter, he sometimes exhibits a “jerk”, or rapid contraction of all legs, and then immediately resumes walking while tapping and drumming (see Additional file B3 for a video recording). The jerks are a common part of the later, or proximal, phase of the courtship and are usually performed in close proximity to the female (S. Vibert, unpublished data), but are sometimes exhibited on an empty web. Additionally, a few of the males we observed stopped moving after bouts of walking while tapping and drumming and, while stationary, slowly contracted and relaxed all legs four or five times in succession (stretches).

We recorded web vibrations caused by prey or courting male spiders inside a sound-attenuated room on a concrete table to minimize extraneous acoustic or vibratory noise. Recordings employed a laser Doppler vibrometer (LDV; Polytec OFV-2500 with OFV-534 sensor head) and data acquisition software VIB-E-220 and VibSoft 4.8 (all products of Polytec Inc., Irvine, CA). Preliminary recordings with a 0 Hz to 2 kHz bandwidth did not reveal any prey or male vibrations with a dominant frequency ≥ 500 Hz. Thereafter, we acquired data with a 0 Hz to 1 kHz bandwidth and a frequency resolution of 78.125 mHz, applying no filtering. We limited all individual recordings to 12.8 s — the longest possible recording time under these settings. During tests with a male spider, we obtained simultaneous video recordings with a Canon FS100 camcorder (Canon USA, Lake Success, NY, USA).

Prior to LDV recordings, we removed female spiders from their 10-15 day old webs. If we damaged a web in the process, we returned the spider to her web for 1–2 days to effect repairs. For recordings, we placed a small square (1 mm2; mean weight: 0.9 mg; n = 25) of reflective tape (Polytec Retroreflective Sheeting, Polytec, Inc.) on an empty web, at the top of the densest area of the tangle in front of the retreat (L. hesperus) and at the entrance to the funnel (T. agrestis) (Figure 1). These are the respective positions where spiders await prey. We then placed the box containing the web on a vibration-proof table, and focused the LDV beam on the reflective tape, at a 90° angle to the plane of the web in order to record transversal vibrations. Although we knew about the relevance of longitudinal and lateral web vibrations in other spider species [19], we restricted our measurements to transversal vibrations for technical reasons. The complex, 3-D structure of L. hesperus webs and the sheet-like nature of T. agrestis webs made it too difficult to position our equipment for recordings of lateral and longitudinal vibrations.

Before we introduced a prey or a male onto an empty web, we obtained a 12.8 s recording of background noise (the waveform and frequency spectrum of a representative background noise recording for each web type are presented in Additional file B4). Once we had dropped a 3-week-old cricket, or a house fly, onto a web, we commenced recordings of web vibrations as soon as the prey moved. We

allowed prey to move freely within the enclosure containing the web. Most crickets quickly disentangled themselves from the web and then dropped to the bottom of the enclosure. Thereafter, many of the recordings captured vibrations produced by crickets coming into contact with capture threads. Thus, we recorded the vibrations that a waiting female would receive, after their transmission from all areas of the web. We recorded clips of 12.8 s in succession as long as the prey was moving. We terminated each test after 30 min or after 50 recordings. We report the mean number of vibration recordings obtained for each prey type in Additional file B5.

To record web vibrations caused by a courting male spider, we removed the male from his Petri dish and placed him in a 15 ml Falcon test tube for 2 h. Using forceps, we then gently placed the male on a randomly assigned web, starting concurrent LDV and video recordings as soon as he initiated courtship. We terminated each recording session after 30 min or when we had obtained 50 recordings. We limited our analyses to six randomly selected recordings per individual (see Additional file B5 for the number of recordings obtained for each individual prey or male).

(d) Analyses of LDV and video recordings

We exported the 12.8 s LDV recordings from VibSoft as WAV files, using the software Raven Lite 1.0 (Bioacoustics Research Program, Cornell Lab of Ornithology, Ithaca, NY) to determine recordings containing at least one ‘event’. We measured the maximum peak-to-baseline amplitude levels of background recordings and determined an amplitude threshold below which prey or male vibrations were indistinguishable from noise. We fixed this threshold at 75 μm/s. We define an ‘event’ as a prey or male vibration with an amplitude > 75 μm/s and lasting ≥ 0.2135 s. We ignored events ≤ 0.2135 s because reliable spectrograms could not be generated. We deemed an event to have ended when its amplitude dropped to below-threshold levels for at least 0.5 s. For each replicate of the cricket, fly, and male stimuli, we randomly selected six recordings with at least one ‘event’, and for each ‘event’ we measured six variables: (1) duration; (2) maximum peak-to-baseline amplitude, (3) root mean square (RMS) amplitude, (4) amplitude modulation factor (AMF), (5) dominant frequency, and (6) bandwidth. We note that vibrations of L. hesperus and T. agrestis males sometimes continued for more than 12.8 s (the maximum recording time), and that we could not include these lengthy

vibrations in quantitative analyses. Nonetheless, our recordings were sufficiently long to detect a significant difference between the males’ vibrations and the much briefer prey vibrations (see Results). We calculated dominant frequency using a fast Fourier transform (FFT). We measured bandwidth as the range of frequencies with an amplitude above a threshold of ¼ the amplitude of the dominant frequency. For each replicate and for each quantitative variable, we calculated a mean value based on all events contained in the six recordings analysed. We report the mean number of events measured within six randomly selected recordings for each fly, cricket and male replicate in Additional file B5. We measured the RMS amplitude envelope for each vibration, using 0.2 s intervals [26]. As a measure of the “percussiveness” of recorded vibrations, we then calculated an amplitude modulation factor (AMF) by dividing the maximum envelope amplitude value by the minimum value. Low AMFs correspond to vibrations with small changes in amplitude, whereas high AMFs correspond to vibrations with large changes in amplitude (Figure 2).

We reviewed video recordings of the male’s courtship behaviour, which we acquired concurrently with LDV recordings, with Windows Movie Maker 6.0 (Microsoft Corporation). Matching the time frames between LDV and video recordings allowed us to link specific vibrations with the specific male behaviours listed above.

(e) Statistical analysis

To achieve normality and equality of variance of data, we subjected data to a Box-Cox transformation prior to analyses. For each species, we conducted a linear discriminant analysis to test whether recorded vibrations could be reliably assigned to their source (fly, cricket, or male) based on the vibratory parameters measured (dominant frequency, frequency bandwidth, RMS amplitude, and duration). We then conducted univariate ANOVAs for individual response variables, followed by Tukey’s HSD post-hoc analyses. We used JMP 8.0 (SAS Institute Inc.) for all statistical analyses.

5.3.2. Vibration parameters L. hesperus females use to discern a prospective prey from a courting male

(a) Study spiders

We reared L. hesperus as described above. We kept virgin females ≥10 days post maturity singly inside wood-framed boxes (30 × 30 × 20 cm) and allowed them to spin a web for 21 days during which time they received four house flies per week. We did not feed spiders for seven days prior to testing to increase the probability that females would respond to a prey stimulus.

(b) Test stimuli

From the vibrations obtained in the previous section we selected a prey and a male vibration for testing the females’ behavioural responses. The percussive type prey vibration (AMF = 34) had been generated by a house fly. The male vibration corresponded to a male’s abdominal tremulation (see Appendix B, Additional file B1 for a video recording of a male’s tremulation), with a rather constant waveform (AMF = 3). We use the term “waveform” to describe or refer to amplitude changes of a vibration over time. The prey vibration had a dominant frequency of 40 Hz with a secondary peak of 67 Hz. The male vibration had a dominant frequency of 36 Hz with a secondary peak of 60 Hz (Figure 3b,c). These two dominant frequency values are intermediate between the mean dominant frequency of vibrations produced by house flies (28.57 Hz), crickets (31.82 Hz), and by L. hesperus males (52.34 Hz). For bioassays, we extracted a 5.2 s segment of the selected prey or male vibration with Raven Lite 1.0 from original 12.8 s LDV recordings, and then looped it for 5 min of continuous playback in order to give females ample time to respond. As most females responded within 10 s (see Results), before two repetitions of the 5.2 s segment were complete, it is not likely that the looping of test stimuli altered the females’ responses.

Based on data obtained in the previous section, the mean maximum amplitude of male abdominal tremulations was 1.85 mm/s (2.35 SD; n = 17), whereas the mean maximum amplitude of fly vibrations was 21.25 mm/s (15.71 SD; n = 16). We modified amplitude levels of test stimuli with the “Amplify” function in Raven Lite 1.0 to produce a ‘low’-amplitude test stimulus equivalent to the mean amplitude of male abdominal

tremulations (‘low’ mean maximum amplitude = 1.99 mm/s), and to produce a ‘high’- amplitude test stimulus equivalent to half the mean amplitude of fly vibrations (‘high’ mean maximum amplitude = 9.47 mm/s). Equipment limitations did not allow playback of vibrations at a higher amplitude. We measured the stimulus amplitude levels during a calibration experiment described in Appendix B, Additional file B6. We report details pertaining to the quality and consistency of playback test stimuli in terms of waveform and amplitude in Additional files B6 and B7.

The playback apparatus (see Figure 3a) consisted of a modified unenclosed loudspeaker (12 Ω; 14 cm diam.) with its cone removed, and a metal rod (180 × 1 mm) attached to the centre of the dust cap. We attached the loudspeaker to an adjustable stand so that it was perpendicular to the plane of the upper tangle of the web, and the tip of the rod could make contact with several silken strands. We connected the speaker to an amplifier (Creek OBH-21SE) which we plugged into the headphone jack of a laptop computer (Toshiba Satellite, Pentium 4, 2.66GHz processor; operating Windows XP version 2002). On the laptop, we opened the looped ‘prey’ and ‘male’ vibration playback files with Windows Media Player 11.0, and played them through the speaker, resulting in vertical movements of the rod.

(d) Behavioural response of spiders to playback vibrations

Prior to testing, we inspected each web for the position of the spider. If she was not in her active hunting position (see above), we postponed testing until the following dark phase. If she was in a hunting position, we chose a location on the upper part of the tangle, 15 cm from her, for the input of playback vibrations. During the initial distal phase of courtship, males spend most of their time on the top part of the web and court far away from the female, usually at a distance between 10 and 30 cm. We selected a location on the web where a male would be likely to court and that was both accessible to the playback apparatus and connected to the female by a dense tangle of threads. We positioned the loudspeaker above the chosen location and put the rod in contact with the web, imposing minimal tension on the threads contacted. During playbacks, contact between the threads and the rod was maintained by this slight tension and the adherence of the silk. If the spider moved during positioning of the rod, we postponed

testing for at least 1 h. Once the rod was in place, we started simultaneously the vibration playback (Windows Media Player 11.0) and behaviour-scoring software (JWatcher 1.0, [49]).

Upon entering a female’s web, a courting L. hesperus male invariably engages in lengthy and repeated bouts of abdominal tremulations. During the early (distal) phase of courtship, females are typically immobile and do not display any response to a courting male. When approached by a male, they sometimes respond by twitching their abdomen [20,27]. In contrast, females respond to the presence of a struggling prey on their web by rapidly moving toward the prey (S. Vibert, personal observation). We recorded the time of the spider’s first predatory response, which we defined as a forward motion of more than 1 cm toward the source of the vibration (see Appendix B, Additional file B8 for a video recording of a female’s response). Spiders that did not move at all, readjusted the position of their legs, oriented toward the rod without forward motion, or twitched their abdomen, were all scored as non-responders. We stopped the playback after 5 min or once the spider had reached the rod, whichever came first. We tested each of 64 spiders only once so that each of four treatments entailing ‘low’- or ‘high’-amplitude vibrations of prey or males (prey/low, prey/high, male/low, male/high) was replicated 16 times.

(e) Statistical analyses

We used a nominal logistic regression to test the effect of amplitude or waveform of playback vibrations, or interaction between these parameters, on the females’ predatory responses. We then used the latency of the females’ responses to conduct a survival analysis [28], with non-responders right-censored at 5 min, to determine whether response times of females within the ‘low’-amplitude level and the ‘high’- amplitude level differed based on waveform. We used JMP 8.0 for all analyses.

5.4. Results

5.4.1. Characterization of prey and male vibrations on L. hesperus and T. agrestis webs

Parameters of prey and male vibrations are compiled in Table 1. On L. hesperus webs (Figure 1), cricket and fly vibrations were typically brief and percussive, with rapid and strong changes in amplitude (= high amplitude modulation factor (AMF); Figure 2). RMS amplitudes of fly vibrations were on average 3 times greater than those of cricket vibrations. Courting L. hesperus males, in contrast, typically produced continuous instead of intermittent vibrations that often persisted throughout the 12.8 s recording period (Figure 4). When cutting web threads, males produced brief, high-amplitude vibrations that resembled those of prey, but before and after cutting threads they typically produced continuous vibrations by walking or bundling silk. Their walking on webs and bundling silk produced sustained vibrations with varying amplitude (moderate AMF) but with no complex temporal pattern. Only four males engaged in cutting threads and bundling silk. Abdominal tremulation produced unique signals of very low and fairly constant amplitude (low AMF), and continuous duration (Figure 4).

On T. agrestis webs (Figure 1), cricket and fly vibrations resembled those on L. hesperus webs (Figure 2), with fly vibrations on average of greater amplitude than cricket vibrations. Some fly vibrations also contained a high-frequency component (~ 200 Hz) corresponding to wing beats. Courting T. agrestis males produced four distinct types of vibrations: (1) drumming with their pedipalps produced continuous, low-amplitude and low-amplitude-modulation vibrations unique to males; (2) walking on webs while pedipalp-drumming and tapping with the first pair of legs produced sustained vibrations of varying amplitude (high AMF); (3) jerks (see methods) produced brief and highly percussive types of vibrations that resembled those of prey but were always followed by continuous vibrations associated with walking on the web while drumming and tapping; and (4) stretches (see methods) produced a distinct temporal pattern of four or five percussive vibrations which were always preceded and followed by silence (Figure 5).

5.4.2. Comparison of prey and male vibrations on L. hesperus and T. agrestis webs

For webs of both L. hesperus and T. agrestis females, linear discriminant analyses revealed significant variation in dominant frequency, bandwidth, RMS amplitude, and duration of vibrations produced by prey and courting males (L. hesperus:

Wilks’ lambda = 0.05, F8,84 = 35.5, p < 0.001; T. agrestis: Wilks’ lambda = 0.19,

F8,84 = 13.29, p < 0.001). There was only slight overlap in the vibration parameters from each source (Figure 6). In L. hesperus, only 12.5% of vibrations were misclassified, and no male vibrations were misclassified as prey. In T. agrestis, only 16.7% of vibrations were misclassified; two male vibrations were misclassified as fly vibrations, and three fly vibrations were misclassified as male vibrations. On L. hesperus webs, the source of vibrations had a significant effect on the dominant frequency (F2,45 = 5.8, p = 0.0057),

RMS amplitude (F2,45 = 18.83, p = 0.0001), and the duration (F2,45 = 128.27, p < 0.0001) of vibrations. Post-hoc analyses revealed that vibrations of L. hesperus males have a mean dominant frequency twice as high as that of crickets and flies, while the mean duration of male vibrations is four to five times longer than those of flies or crickets. The amplitude of male L. hesperus vibrations was not significantly different from that of cricket vibrations, but was lower than that of fly vibrations (Figure 6).

On T. agrestis webs, the source of vibrations had a significant effect on the amplitude (F2,45 = 7.57, p = 0.0015) and the duration (F2,45 = 52.52, p < 0.0001) of vibrations but not on their dominant frequency (F2,45 = 0.35, p = 0.71). Post-hoc analyses showed that vibrations of T. agrestis males last on average two to three times longer than those of flies or crickets. The amplitude of male vibrations was not significantly different from that of either fly or cricket vibrations (Figure 6).

We did not perform ANOVAs on bandwidth because we found this variable to be highly correlated with dominant frequency for crickets, flies and males in both species (L. hesperus: Pearson correlation coefficient r = 0.76, p < 0.0001, n = 48; T. agrestis: r = 0.65, p < 0.0001, n = 48).

5.4.3. Vibration parameters triggering a predatory response in L. hesperus females

We tested the response of L. hesperus females to playback of prey and male vibrations presented at low and high amplitude. The amplitude, but not the waveform, of vibrations had a significant effect on the behavioural response of L. hesperus females

2 2 (amplitude: χ 1,64 = 14.41, p = 0.0001; waveform: χ 1,64 = 0.17, p = 0.679). Far fewer females responded to the ‘low’ amplitude stimulus (‘prey’ waveform: 31.25%; ‘male’ waveform: 43.75%) than to the ‘high’ amplitude stimulus (‘prey’ and ‘male’ waveform: 87.5%). The interaction term between amplitude and waveform was not significant

2 (χ 1,64 = 0.17, p = 0.679). The proportion of females that exhibited a predatory response to playbacks of low- or high-amplitude prey cues and low- or high-amplitude male signals is shown in Figure 7a, and results of the logistic regression analysis for the full factorial

2 model (χ 3,64 = 18.762, p = 0.0003) are presented in Table 2.

Among the females exposed to playback recordings of ‘low’ amplitude ‘prey’ or ‘male’ vibrations, seven females remained immobile. In response to the ‘low prey’ stimulus, two females twitched their abdomen and four females moved their legs or oriented toward the source of the vibration. In response to the ‘low male’ stimulus, one female displayed an abdomen twitch and two females adjusted the position of their legs or oriented toward the source of the vibration. The only two females who did not exhibit a predatory response to the playback of a ‘high’ amplitude ‘prey’ or ‘male’ vibration remained immobile.

For females that responded to the playback stimulus, the latency of their response to the four test stimuli (see Materials and Methods) is reported in Figure 7b. Results of the survival analysis show that there was no significant difference between the response times of females exposed to playback of low-amplitude prey or male

2 vibrations (Wilcoxon test; χ 1,32 = 0.2921, p = 0.589), and no significant difference between the response times of females to playback of high-amplitude prey or male

2 vibrations (Wilcoxon test; χ 1,32 = 2.3105, p = 0.129). Most females responded very quickly to both high-amplitude stimuli and started moving toward the source of the vibration in less than 10 s. Far fewer females responded to the low-amplitude treatments, and they did so more slowly.

5.5. Discussion

We have (1) characterized vibratory cues of house fly and house cricket prey and vibratory signals of L. hesperus and T. agrestis males; (2) determined that vibratory courtship signals of males differ from prey vibratory cues; and (3) ascertained the vibration parameter(s) triggering a predatory response in females.

On both L. hesperus and T. agrestis webs, cricket and fly vibrations were similar: short, sporadic, and on average with high amplitude modulation. Most vibrations of L. hesperus and T. agrestis males were continuous, lengthy, and lacking a complex temporal structure. Vibrations of L. hesperus males differed from prey in terms of duration and dominant frequency. Male vibrations were of lower amplitude than fly, but not cricket, vibrations. Vibrations of T. agrestis males differed from prey in terms of duration only. During the playback experiment, significantly fewer L. hesperus females responded aggressively to low-amplitude vibrations, irrespective of whether these stimuli were recorded vibrations of prey or male spiders, suggesting that the likelihood of a predatory response depends on the amplitude but not the waveform of incoming vibrations. Below we discuss the implications of these findings for male signal function and signalling constraints.

The absence of complex temporal patterns in most courtship vibrations of L. hesperus and T. agrestis males is in stark contrast to observations in other spiders. For example, female Cupiennius getazi use the duration and structure of male-produced syllables to identify conspecific males [29]. Males of the wolf spider Lycosa tarentula fasciiventris produce courtship vibrations that comprise series of repeating syllables followed by pauses at regular intervals [30]. The temporal structure of courtship vibrations produced by male Schizocosa ocreata is linked to female mate choice [31]. Finally, the vibratory courtship displays of 11 species of jumping spiders (Salticidae) within the Habronattus coecatus clade are complex and comprise up to 20 distinct elements organized in motifs [32]. All of the above examples refer to wandering spiders whose courtship takes place on plant stalks, leaf litter, or the ground. Similarly, courting males of the orb-weaver Argiope keyserlingi produce vibrations with repeated, pulse-like characteristics [18]. This distinct temporal patterning may be well transmitted because A.

keyserlingi males court on a single silk thread. The resulting vibrations are quite different from the ones reported in our study, but the abdominal tremulation of L. hesperus males and the shuddering of A. keyserlingi males are very similar types of behaviour. The absence of temporally complex signalling in L. hesperus, and its scant presence in T. agrestis, is curious. Based on their rate and amplitude modulation, tremulations can produce signals with a lot of information [33] but abdominal tremulations of male L. hesperus generated uniform waveforms that can be described solely on the basis of their amplitude and frequency. Future work is needed to reveal whether sheet and tangle webs impose constraints on the temporal complexity of signals.

The transmission properties of a medium impose constraints on the characteristics of signals [1,34]. Contrary to orb webs, L. hesperus tangle webs and T. agrestis sheet webs are not uniform structures. The density of threads and their orientation, degree of tension, number of connections and distance to anchor points all vary greatly from one area of the web to another, and likely affect the transmission characteristics of the webs. When we explored the propagation properties of transversal vibrations on L. hesperus and T. agrestis webs (using frequency sweeps from 0 to 500 Hz), we found great variability both within and between webs in both types of webs (S. Vibert, unpublished data). Within a single web, transmission profiles obtained at different locations were sometimes very dissimilar. Similarly, playback of recorded prey vibrations of known dominant frequency revealed that frequency was not well transmitted across L. hesperus webs (S. Vibert, unpublished data).

There are several plausible explanations for the difference between male and prey vibrations. Vibrations on webs during courtship interactions might communicate species identity and help females distinguish between con- and heterospecific males or between conspecific males and potential prey. Alternatively, vibrations of a male might communicate his identity, quality, or current location.

Our results suggest that the low-amplitude vibrations produced by L. hesperus males reduce the probability of being attacked by females during courtship. Female attack rate was twice as low when prey or male vibrations were played back at the low amplitude of male abdominal tremulations than at the high amplitude of prey vibrations. We also observed females twitching their abdomen dorso-ventrally in response to three

low-amplitude playbacks. In a previous study [20], 75% of L. hesperus females displayed “twitching” during advanced stages of courtship, whereas no female ever displayed twitching in response to live prey (S. Vibert, pers. obs.). Our results do suggest that L. hesperus males must “whisper” during courtship, but the potential information content and sexual function of these whispers are yet be studied. It would be particularly interesting to investigate whether T. agrestis females respond differently to vibrations of varying duration, the one parameter in which vibrations of males differed from those of prey. While prey vibrations were intermittent, vibrations of L. hesperus and T. agrestis males were continuous, which may be another determinant factor for the females’ predatory responses.

The function of L. hesperus male vibratory signals is not likely to advertise male quality. Whenever males deploy acoustic signals that broadcast their quality, females prefer loud (high-amplitude) signals, as has been demonstrated in gray tree frogs [35], túngara frogs [36], katydids [37], wax moths [38], and passerine birds [39,40]. Whether the amplitude of vibratory signals produced by courting spider males is indicative of their quality as prospective mates, or whether it serves another function, has hardly been studied. Large males of the funnel-web spider Agelenopsis aperta are more likely to achieve mating success [6], possibly because they produce louder signals, as has been shown for airborne signals in the toad Bufo americanus [41]. Similarly, male Schizocosa ocreata wolf spiders producing higher-amplitude signals were more successful at securing a mate [31]. In the wandering spider Cupiennius, however, the amplitude of signals seems of no relevance to females [42,43,44]. The quiet songs of birds exemplify a signalling display characterized by low amplitude; quiet songs prevent eavesdropping from competitors or rivals in contexts of territorial disputes or mating interactions [45]. The courtship display of L. hesperus might well represent a novel context in which males must signal at low amplitude to avoid triggering a female predatory response.

A reduction of female aggressiveness is often cited as one of the possible functions of male courtship in spiders but few studies have tested this experimentally. Many adaptations may function to avoid or reduce female aggressiveness. Behavioural adaptations include approaching a female while she is feeding [46], mate binding [47], or inducing a quiescent state [6]. We suggest that courtship signals of L. hesperus and T. agrestis males that differ markedly from prey vibrations might represent another

adaptation in males facing large and aggressive females. Conversely, in species where females are not aggressive towards males, it may be adaptive for courting males to take advantage of the females’ sensory systems being tuned for prey cues by producing prey-like vibrations, as has been demonstrated in the water mite Neumania papillator [48].

5.6. Conclusions

Silk production is one of the most fascinating innovations of spiders, aiding in many aspects of their natural history. The use of webs in prey capture is well documented. Less studied is how males communicate through the females’ webs during courtship displays. We present an exploratory study of vibratory courtship signals on tangle and sheet webs. In both L. hesperus and T. agrestis, we found that some parameters of male courtship signals contrast with those of prey cues. In L. hesperus, one of these parameters seems to facilitate male courtship in that L. hesperus females are less likely to attack in response to the characteristic low-amplitude vibrations of L. hesperus males than in response to the high-amplitude vibrations of prey. Many other aspects are yet to be investigated in future studies. They include signal transmission properties of the highly complex and variable tangle and sheet webs, their potential constraints on male signal design, the information content of male signals, and their role in eliciting a sexual response from females.

Acknowledgements

We thank Stephen Takács for technical expertise, Bernard Roitberg and Gwylim Blackburn for comments on the manuscript, Ummat Somjee for help with data acquisition, Stephen DeMuth for graphical illustrations, and Sean McCann for the cover photograph. Funding was provided by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant and by an NSERC Industrial Research Chair to G.G., with Contech Enterprises, SC Johnson Canada, and Global Forest Science as sponsors.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

SV conceived of the study and drafted the manuscript. SV and CS designed the study, collected the study animals, performed the research and statistical analyses. CS and GG critically revised the manuscript. All authors read and approved the final manuscript.

References

1. Endler JA: Signals, signal conditions, and the direction of evolution. Am Nat. 1992, 139: S125-S153. 10.1086/285308.

2. Endler JA, Basolo AL: Sensory ecology, receiver biases and sexual selection. Trends Ecol Evol. 1998, 13: 415-420. 10.1016/S0169-5347(98)01471-2.PubMed

3. Endler JA: Predation, light-intensity and courtship behavior in Poecilia reticulate (Pisces, Poeciliidae). Anim Behav. 1987, 35: 1376-1385. 10.1016/S0003- 3472(87)80010-6.

4. Foelix RF: Biology of spiders. 1996, Oxford: Oxford University Press

5. Maklakov AA, Bilde T, Lubin Y: Vibratory courtship in a web-building spider: signalling quality or stimulating the female? Anim Behav. 2003, 66: 623-630. 10.1006/anbe.2003.2245.

6. Singer F, Riechert SE, Xu H, Morris AW, Becker E, Hale JA, Noureddine MA: Analysis of courtship success in the funnel-web spider Agelenopsis aperta. Behaviour. 2000, 137: 93-117. 10.1163/156853900501890.

7. Vollrath F, Selden P: The role of behavior in the evolution of spiders, silks, and webs. Annu Rev Ecol Evol Syst. 2007, 38: 819-846. 10.1146/annurev.ecolsys.37.091305.110221.

8. Witt PN: Web as a means of communication. Biosci Commun. 1975, 1: 7-23.

9. Elgar MA: Sexual cannibalism, size dimorphism, and courtship behavior in orb- weaving spiders (Araneidae). Evolution. 1991, 45: 444-448. 10.2307/2409679.

10. Elgar MA, Schneider JM: Evolutionary significance of sexual cannibalism. Adv Study Behav. 2004, 34: 135-163.

11. Newman JA, Elgar MA: Sexual cannibalism in orb-weaving spiders - an economic model. Am Nat. 1991, 138: 1372-1395. 10.1086/285292.

12. Schneider JM, Lubin Y: Intersexual conflict in spiders. Oikos. 1998, 83: 496-506. 10.2307/3546677.

13. Elgar MA, Nash DR: Sexual cannibalism in the garden spider Araneus diadematus. Anim Behav. 1988, 36: 1511-1517. 10.1016/S0003-3472(88)80221-5.

14. Robinson MH: Courtship and mating behavior in spiders. Annu Rev Entomol. 1982, 27: 1-20. 10.1146/annurev.en.27.010182.000245.

15. Masters WM: Vibrations in the orbwebs of Nuctenea sclopetaria (Araneidae). 2. Prey and wind signals and the spiders response threshold. Behav Ecol Sociobiol. 1984, 15: 217-223. 10.1007/BF00292978.

16. Watanabe T: Web tuning of an orb-web spider, Octonoba sybotides, regulates prey- catching behaviour. Proc R Soc B. 2000, 267: 565-569. 10.1098/rspb.2000.1038.

17. Wignall AE, Taylor PW: Assassin bug uses aggressive mimicry to lure spider prey. Proc R Soc B. 2011, 278: 1427-1433. 10.1098/rspb.2010.2060.

18. Wignall AE, Herberstein ME: The influence of vibratory courtship on female mating behaviour in orb-web spiders (Argiope keyserlingi, Karsch 1878). PLoS One. 2013, 8: e53057. http://dx.doi.org/10.1371/journal.pone.0053057

19. Landolfa MA, Barth FG: Vibrations in the orb web of the spider Nephila clavipes: cues for discrimination and orientation. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 1996, 179: 493-508.

20. Scott C, Vibert S, Gries G: Evidence that web reduction by western black widow males functions in sexual communication. Can Entomol. 2012, 144: 672-678. 10.4039/tce.2012.56.

21. Bradbury JW, Vehrencamp SL: Principles of animal communication. 1998, Sunderland: Sinauer Associates

22. Parri S, Alatalo RV, Kotiaho JS, Mappes J, Rivero A: Sexual selection in the wolf spider Hygrolycosa rubrofasciata: female preference for drum duration and pulse rate. Behav Ecol. 2002, 13: 615-621. 10.1093/beheco/13.5.615.

23. Schmitt A, Friedel T, Barth FG: Importance of pause between spider courtship vibrations and general problems using synthetic stimuli in behavioral studies. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 1993, 172: 707-714.

24. De Luca PA, Morris GK: Courtship communication in meadow katydids: female preference for large male vibrations. Behaviour. 1998, 135: 777-794. 10.1163/156853998792640422.

25. Warkentin KM, Caldwell MS, McDaniel JG: Temporal pattern cues in vibrational risk assessment by embryos of the red-eyed treefrog, Agalychnis callidryas. J Exp Biol. 2006, 209: 1376-1384. 10.1242/jeb.02150.

26. Caetano M, Rodet X: Improved estimation of the amplitude envelope of time-domain signals using true envelope cepstral smoothing. Acoustics, Speech and Signal Processing (ICASSP), 2011 IEEE International Conference; 22–27 May. 2011, 4244-4247.

27. Ross K, Smith RL: Aspects of the courtship behavior of the black widow spider, Latrodectus hesperus (Araneae, Theridiidae), with evidence for the existence of a contact sex pheromone. J Arachnol. 1979, 7: 69-77.

28. Moya-Larano J, Wise DH: Survival regression analysis: a powerful tool for evaluating fighting and assessment. Anim Behav. 2000, 60: 307-313. 10.1006/anbe.2000.1495.

29. Schmitt A, Schuster M, Barth FG: Vibratory communication in a wandering spider, Cupiennius getazi - Female and male preferences for features of the conspecific males releaser. Anim Behav. 1994, 48: 1155-1171. 10.1006/anbe.1994.1348.

30. Fernandez-Montraveta C, Schmitt A: Substrate-borne vibrations produced by male Lycosa tarentula fasciiventris (Araneae, Lycosidae) during courtship and agonistic interactions. Ethology. 1994, 97: 81-93.

31. Gibson JS, Uetz GW: Seismic communication and mate choice in wolf spiders: components of male seismic signals and mating success. Anim Behav. 2008, 75: 1253-1262. 10.1016/j.anbehav.2007.09.026.

32. Elias DO, Maddison WP, Peckmezian C, Girard MB, Mason AC: Orchestrating the score: complex multimodal courtship in the Habronattus coecatus group of Habronattus jumping spiders (Araneae: Salticidae). Biol J Linn Soc. 2012, 105: 522-547. 10.1111/j.1095-8312.2011.01817.x.

33. Uhl G, Elias DO: Communication. Spider behaviour: versatility and flexibility. Edited by: Heberstein ME. 2011, Cambridge: Cambridge University Press, 127-189.

34. Ey E, Fischer J: The "acoustic adaptation hypothesis" - a review of the evidence from birds, anurans and mammals. Bioacoustics. 2009, 19: 21-48. 10.1080/09524622.2009.9753613.

35. Beckers OM, Schul J: Phonotaxis in Hyla versicolor (Anura, Hylidae): the effect of absolute call amplitude. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 2004, 190: 869-876.

36. Akre KL, Ryan MJ: Proximity-dependent response to variably complex mating signals in tùngara frogs (Physalaemus pustulosus). Ethology. 2010, 116: 1138- 1145. 10.1111/j.1439-0310.2010.01825.x.

37. Galliart PL, Shaw KC: The effect of variation in parameters of the male calling song of the katydid, Amblycorypha parvipennis (Orthoptera: Tettigonndae), on female phonotaxis and phonoresponse. J Insect Behav. 1996, 9: 841-855. 10.1007/BF02208973.

38. Jang YW, Greenfield MD: Ultrasonic communication and sexual selection in wax moths: female choice based on energy and asynchrony of male signals. Anim Behav. 1996, 51: 1095-1106. 10.1006/anbe.1996.0111.

39. Pasteau M, Nagle L, Kreutzer M: Preferences and predispositions of female canaries (Serinus canaria) for loud intensity of male sexy phrases. Biol J Linn Soc. 2009, 96: 808-814. 10.1111/j.1095-8312.2008.01136.x.

40. Ritschard M, Riebel K, Brumm H: Female zebra finches prefer high-amplitude song. Anim Behav. 2010, 79: 877-883. 10.1016/j.anbehav.2009.12.038.

41. Gerhardt HC: Sound pressure levels and radiation patterns of the vocalizations of some North American frogs and toads. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 1975, 102: 1-12. 10.1007/BF00657481.

42. Baurecht D, Barth FG: Vibratory communication in spiders 1. Representation of male courtship signals by female vibration receptor. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 1992, 171: 231-243.

43. Baurecht D, Barth FG: Vibratory communication in spiders 2. Representation of parameters contained in synthetic male courtship signals by female vibration receptor. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 1993, 173: 309-319. 10.1007/BF00212695.

44. Schuch W, Barth FG: Vibratory communication in a spider - Female responses to synthetic male vibrations. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 1990, 166: 817-826.

45. Dabelsteen T, McGregor PK, Lampe HM, Langmore NE, Holland JO: Quiet songs in birds: an overlooked phenomenon. Bioacoustics. 1998, 9: 89-105. 10.1080/09524622.1998.9753385.

46. Fromhage L, Schneider JM: Safer sex with feeding females: sexual conflict in a cannibalistic spider. Behav Ecol. 2005, 16: 377-382.

47. Zhang SC, Kuntner M, Li DQ: Mate binding: male adaptation to sexual conflict in the golden orb-web spider (Nephilidae: Nephila pilipes). Anim Behav. 2011, 82: 1299-1304. 10.1016/j.anbehav.2011.09.010.

48. Proctor HC: Courtship in the water mite Neumania papillator - Males capitalize on female adaptations for predation. Anim Behav. 1991, 42: 589-598. 10.1016/S0003-3472(05)80242-8.

Table 5.1 Summary of parameters associated with males, prey, and background noise vibrations on webs of Latrodectus hesperus (top) and Tegenaria agrestis (bottom)

Male Prey Background

Abdomen Walking Bundling silk Cutting All Cricket Fly tremulation

Dominant frequency (Hz) 43.38 (26.78) 55.38 (31.65) 36.58 (16.9) 36.6 (16.77) 52.34 (25.28) 31.82 (16.29) 28.57 (17.41) 32.63 (16.41)

Bandwidth (Hz) 43.38 (29.42) 44.73 (39.85) 44.96 (71.56) 4.1 (2.79) 74.8 (75.5) 45.4 (34.57) 54.73 (53.02)

RMS amplitude (mm/s) 0.46 (0.61) 1.53 (0.94) 1.32 (0.87) 3.73 (3.09) 0.60 (0.34) 0.88 (0.83) 3.27 (2.59) 0.03 (0.01)

Max amplitude (mm/s) 1.85 (2.35) 14.44 (11.73) 6.04 (4.05) 17.91 (10.66) 6.27 (4.42) 7.58 (8.71) 21.25 (15.72) 0.10 (0.04)

A.M.F 3.91 (1.20) 14.69 (12.08) 8.12 (5.55) 11.50 (6.75) 9.65 (9.55) 33.78 (33.16) 51.56 (48.07)

Duration (s) 6.45 (3.0) 8.06 (4.12) 8.59 (4.31) 1.0 (0.04) 9.31 (2.43) 1.38 (1.0) 2.01 (0.8)

n 17 18 4 4 16 16 16 48

Walking, drumming Palp drumming Jerk Stretch All Cricket Fly tapping

Dominant frequency (Hz) 32.83 (19.91) 40.64 (18.56) 50.88 (8.85) 54.46 (25.79) 44.31 (20.08) 45.88 (28.11) 40.57 (39.01) 34.14 (20.58)

Band-width (Hz) 92.63 (63.35) 71.38 (18.75) 63.35 (18.96) 107.31 (113.6) 65.34 (21.54) 105.18 (85.82) 54.39 (38.77)

RMS amplitude (mm/s) 0.21 (0.16) 2.19 (1.98) 3.52 (4.63) 4.05 (3.88) 0.86 (0.58) 0.42 (0.30) 1.81 (2.22) 0.02 (0.01)

Max amplitude (mm/s) 1.36 (1.04) 20.37 (19.41) 19.13 (25.91) 28.54 (30.11) 8.27 (5.86) 2.92 (2.21) 7.06 (6.0) 0.05 (0.02)

A.M.F 5.37 (2.91) 29.95 (17.88) 10.34 (3.72) 7.49 (5.46) 14.49 (14.85) 23.04 (34.13) 34.29 (42.0)

Duration (s) 7.25 (3.39) 9.62 (3.17) 1.58 (1.20) 1.43 (1.24) 6.35 (2.73) 1.44 (0.47) 2.59 (1.12)

n 10 13 6 9 16 16 16 38

Values are means with standard deviations in parentheses.

Table 5.2 Results of logistic regression analysis of predatory responses of female Latrodectus hesperus to playbacks of low- or high-amplitude vibrations of house fly prey and conspecific males

Model - LogLikelihood DF χ2 p

Difference 9.381 3 18.762 0.0003

Full 32.959

Reduced 42.340

R2 (U) 0.222

Predictor β SE β χ 2 p Odds ratio

Constant - 0.713 0.325 4.82 0.028 N/A

Amplitude 1.233 0.325 14.41 0.0001 1.308

Waveform - 0.134 0.325 0.17 0.679 0.085

Amplitude*Waveform - 0.134 0.325 0.17 0.679 N/A The levels of the factors amplitude and waveform were coded as follows: low = 0, high = 1; male = 0, prey = 1.

Figure 5.1 Web structure of Latrodectus hesperus and Tegenaria agrestis. Schematic drawing illustrating (a) the tangle web of Latrodectus hesperus and (b) the sheet web of Tegenaria agrestis. L. hesperus webs consist of a dense three-dimensional tangle of threads. Glue- coated capture threads extend from the tangle to the ground. T. agrestis webs consist of a two-dimensional sheet of silk with a funnel at one end serving as a retreat. We recorded vibrations on empty webs at the active hunting location of the spider (marked by ★). The spiders indicate the position of the female’s retreat. L. hesperus web illustration modified from Blackledge et al. 2005.

Figure 5.2 Prey vibrations on webs of Latrodectus hesperus and Tegenaria agrestis. Oscillograms depicting velocity [mm/s] over time [s] (upper panels) and frequency [Hz] (lower panels) of cricket and house fly vibrations recorded on empty webs of Latrodectus hesperus and Tegenaria agrestis. “a.e.” refers to root mean square amplitude envelope of the vibration and “a.m.f” refers to amplitude modulation factor measured between the lowest and the highest point of the amplitude envelope (in this example, a.m.f. = 120).

Figure 5.3 Playback design and original vibrations used during playback. (a) Schematic drawing of the experimental design for vibration playbacks. A looped male vibration and a looped house fly vibration were played back at low and high amplitude by a modified loudspeaker (S) placed in contact with the web 15 cm away from a Latrodectus hesperus female in her hunting position; (b) Original male abdominal tremulation vibration used to generate the input vibrations played back with the speaker. Oscillograms depict velocity [mm/s] over time [s] (upper panel) and frequency [Hz] (lower panel); (c) Original fly vibration used to generate the input vibrations played back with the speaker. Oscillograms depict velocity [mm/s] over time [s] (upper panel) and frequency [Hz] (lower panel).

Figure 5.5 Courtship vibrations of male Tegenaria agrestis. For each of the courtship vibrations including (1) palp drumming, (2) walking, drumming and tapping, (3) jerks, walking, drumming and tapping, and (4) stretches produced by courting Tegenaria agrestis males on empty webs of conspecific females, the upper panel depicts vibrations in the time domain and the lower panel depicts vibration in the frequency domain. The insert in (1) depicts the amplitude of palp drumming (maximum baseline-to-peak amplitude = 0.3 mm/s) magnified 25 times.

Figure 5.6 Comparison of vibration parameters associated with struggling prey and courting male spiders on webs of Latrodectus hesperus and Tegenaria agrestis. (a-1) and (b-1) Discriminant function analysis using dominant frequency, frequency bandwidth, root mean square amplitude and duration (transformed data) of vibrations produced by cricket and house fly prey and by males of Latrodectus hesperus and Tegenaria agrestis on empty L. hesperus webs (a-1) and T. agrestis webs (b-1) (n = 16 each). The inner circle shows the 95% confidence ellipse of each mean; the outer circle shows the normal 50% contours; (a-2 to a-4) and (b-2 to b-4) Boxplots of dominant frequency [Hz], root mean square (RMS) amplitude [mm/s] and duration [s] of vibrations produced by cricket and house fly prey and by males of L. hesperus and T. agrestis on empty L. hesperus webs (a-2 to a-4) and T. agrestis webs (b-2 to b-4) (n = 16 each). Median, mean, interquartile range (IQR) and outliers (untransformed data); whiskers = upper and lower data point values within 1.5 IQR; means with different letters are significantly different (Tukey’s HSD on transformed data, p < 0.05).

Figure 5.7 Response of female Latrodectus hesperus to playback of high- and low-amplitude prey and male vibrations. (a) Proportion of Latrodectus hesperus females responding aggressively to vibrations produced by house fly prey or conspecific males played back at high or low amplitude at a distance of 15 cm (n = 16 for each treatment). Whiskers = standard error; (b) Time [s] elapsed before females initiated a predatory response (number of responding females: n = 5 for prey/low, n = 7 for male/low, n = 14 for both prey/high and male/high).

Chapter 6.

Conclusion

In my thesis, I set out to investigate aspects of the vibratory courtship signaling in two spider species, Eratigena agrestis and Latrodectus hesperus. With reference to the study goals outlined in "Chapter 1: Introduction", my findings are as follows:

In Chapter 2, I found that E. agrestis spiderlings emerged in the spring and mature in two years. Contrasting densities of the congener E. atrica in the two study sites, and the presence of L. hesperus in only one study site, had no obvious effects on the life history of E. agrestis. This unexpected finding may be explained by (i) low overall competition pressure in the study habitats, (ii) con- and heterospecifics exerting equivalent competition or predation pressures; and/or (iii) aggregations of heterospecifics providing benefits that offset costs associated with any competition.

Observations of courtship behavior (Chapter 3) revealed that males engaging in web reduction elicited fewer aggressive responses from females and induced female quiescence more quickly than did males not exhibiting web reduction behaviour. This suggests that web reduction by male L. hesperus functions in sexual communication, a context not previously explored. I also documented male vibratory signalling in numerous aspects of courtship behavior, suggesting that vibratory signalling is important in sexual communication between male and female L. hesperus.

Documentation of signal parameters (Chapter 4) showed that transmission efficiencies are highly variable in the sheet webs of E. agrestis and the tangle webs of L. hesperus. Within-web variation is greater than among-web variation. There was little difference in transmission efficiencies of longitudinal and transverse vibrations. The inconsistent transmission of specific frequencies through webs suggests that parameters

other than frequency are most important in allowing these spiders to distinguish between vibrations of prey and courting males.

As concluded from experiments in Chapter 5, courtship signals of males differed more markedly from prey cues in L. hesperus than in T. agrestis. Unlike courtship signals of other spider species, the courtship signals of L. hesperus and T. agrestis males do not have complex temporal patterning. In L. hesperus, females did not respond aggressively to low-amplitude vibrations, irrespective of whether the playback recording was from a prey or a male. This suggests that the low-amplitude ‘whispers’ of L. hesperus males at the onset of courtship are less likely to trigger a predatory response in females than the high-amplitude vibrations of struggling prey.

Overall, my work indicates that vibrations are integral to courtship signalling in both of the focal species. My work provides insight to the transmission of vibrations through spider webs in general, revealing specific aspects of vibrations that may be important for communication. Differences between the focal species in web architecture and aspects of signal transmission suggest different trade-offs between communication efficiency and avoiding female aggression. To gain a general sense of how web architecture affects vibration transmission, it would be useful to explore vibratory signalling in other spider species that feature sheet and tangle web architectures. Consideration of parameters other than those studied here, such as temporal signal variation, may also provide new insights to the vibratory signal function.

Appendix A.

Supplementary material for chapter 4

Table A1. Transmission efficiencies of the two web types. Transmission efficiencies (TE) of longitudinal (L) and transverse (T) vibrations caused by 12-s, 0- to 1000-Hz frequency sweeps generated by a piezoelectric vibrator. Vibrations were transmitted through webs of Eratigena agrestis and Latrodectus hesperus, and measured by a laser Doppler vibrometer at a distance of 20 mm and 70 mm from the point of the input vibration. See materials and methods in the main text for complete details. Species Type of Distance from Mean TE Median TE Min. SD Max. SD vibration input vibration (dB) (dB)

20 mm -9.36 -10.16 3.54 10.07 La 70 mm -15.40 -16.34 2.50 12.02 20 mm -6.79 -7.04 2.9 8.64 a E.agrestis T 70 mm -18.78 -19.72 2.79 10.91

20 mm -8.53 -9.19 3.29 9.71 La 70 mm -12.59 -13.56 3.19 10.77 20 mm -6.98 -7.37 2.63 9.94 Ta L. hesperus L. 70 mm -9.56 -9.51 3.45 10.85 a n = 14 webs

Table A2 Repeatability values of transmission efficiency measurements of Eratigena agrestis and Latrodectus hesperus web vibrations calculated using within-web and among-web variance components (see text for details).

Species Distance from input vibration Frequency (Hz) F-ratio Repeatability 10 4.43 0.63 20 3.08 0.51 30 2.24 0.38 40 3.18 0.52 50 2.01 0.34 60 1.59 0.23 70 1.62 0.24 80 1.38 0.16 90 1.49 0.20 100 0.71 0 20 mm 110 0.62 0 120 3.33 0.54 130 0.91 0 140 2.19 0.37 150 1.30 0.13 160 1.63 0.24 170 0.82 0 180 1.06 0.03 190 0.74 0 200 0.75 0 10 2.90 0.49 E.agrestis 20 4.19 0.61 30 2.01 0.33 40 3.02 0.50 50 2.08 0.35 60 1.40 0.17 70 2.60 0.44 80 1.95 0.32 90 1.83 0.29 100 1.29 0.13 70 mm 110 1.18 0.08 120 0.60 0 130 1.96 0.32 140 1.06 0.03 150 0.93 0 160 1.75 0.27 170 1.44 0.18 180 1.66 0.25 190 1.48 0.19 200 1.72 0.27

Table A2 continued

Species Distance from input vibration Frequency (Hz) F-ratio Repeatability 10 0.86 0 20 2.92 0.49 30 1.77 0.28 40 1.73 0.27 50 0.72 0 60 2.22 0.38 70 1.93 0.32 80 1.84 0.30 90 1.89 0.31 100 1.39 0.16 20 mm 110 3.21 0.53 120 2.34 0.40 130 0.48 0 140 2.60 0.44 150 2.63 0.45 160 3.30 0.54 170 1.21 0.10 180 1.21 0.09 190 0.86 0 200 0.81 0 10 2.58 0.44 L.hesperus 20 1.80 0.29 30 6.17 0.72 40 2.09 0.35 50 2.57 0.44 60 2.06 0.35 70 1.68 0.25 80 3.10 0.51 90 1.77 0.28 100 1.25 0.11 70 mm 110 0.94 0 120 2.94 0.49 130 2.79 0.47 140 2.08 0.35 150 1.61 0.24 160 5.25 0.68 170 4.12 0.61 180 3.11 0.51 190 1.87 0.30 200 2.90 0.49 Repeatability values vary between 0 and 1. A value near 0 indicates that most of the variance in transmission efficiency results from differences between two measurements, each set of the two measurements taken on the same web at a distance of 20 mm or 70 mm from the point of the input vibration (a frequency sweep generated by a modified loudspeaker; see Fig. 4).

Fig. A1. Background noise recordings on Latrodectus hesperus and Eratigena agrestis webs. Representative recordings of background noise associated with empty webs, with the laser Doppler vibrometer measuring transverse vibrations at the location where the female spider normally sits while waiting for prey (stars in Fig. 1 or #1 in Fig. 3). Oscillograms depict velocity [mm/s] over time [s] (upper panel) and velocity [μm/s] over frequency [Hz] (lower panel).

Appendix B. Supplementary material for chapter 5

Additional files B1-B3 and B8 (video files)

Table B1 List of video files associated with chapter 5 Video File name Description of video Additional file B1 Video_B1.wmv Abdominal tremulation of black widow male Additional file B2 Video_B2.wmv Walking, drumming and tapping of hobo spider male Additional file B3 Video_B3.wmv Jerk of hobo spider male Additional file B8 Video_B8.wmv Aggressive response of a black widow female to a male high- amplitude playback vibration Videos can be downloaded from the SFU Library website, or accessed on the Frontiers in Zoology website

Additional file B4

Figure B4 Background noise recordings on Latrodectus hesperus and Tegenaria agrestis webs. Representative recordings of background noise associated with empty webs. Oscillograms depict velocity [mm/s] over time [s] (upper panel) and velocity [μm/s] over frequency [Hz] (lower panel).

Additional file B5

Table B5 Number of recordings and number of events within 6 randomly selected recordings for each replicate. Mean, standard deviation (SD), min. and max. number (#) of 12.8-s laser Doppler vibrometer recordings of (i) vibrations produced by courting males, house flies, and crickets on empty webs of female Latrodectus hesperus (Lh) and Tegenaria agrestis (Ta) and (ii) ‘events’ measured within six randomly selected recordings for each male, fly, and cricket replicate. ‘Events’ are segments of a recording with a vibration amplitude > 2 × background amplitude. Recordings per replicate Mean SD Min. # Max. # Male 32.1 11.0 16 52 Lh Fly 17.6 8.1 6 38 Cricket 17.9 6.4 6 26 Male 24.7 11.7 6 46 Ta Fly 12.9 3.9 6 20 Cricket 15.9 1.7 14 20 Events per replicate within 6 recordings Mean SD Min. # Max. # Male 7.9 1.7 6 11 Lh Fly 12.0 4.0 7 22 Cricket 10.9 3.6 7 17 Male 10.4 3.5 7 17 Ta Fly 12.6 4.6 6 25 Cricket 12.7 4.0 7 21 Number of replicates for males, flies, and crickets = 16 for both Lh and Ta.

Additional file B6

Waveform quality and amplitude consistency of playback-induced vibrations To assess waveform fidelity after transmission through the web as well as amplitude consistency of playback-induced vibrations, we played back the ‘prey’ and ‘male’ vibrations at ‘low’ and ‘high’ levels on nine 21-day-old empty female webs and recorded the resulting playback-induced vibrations at a distance of 15 cm from the input location of the playback vibrations (Additional file B7a). We then measured the peak-to-baseline amplitude of the playback-induced vibrations. At the ‘low’ level, prey and male playback vibrations induced a mean peak-to-baseline vibration amplitude of 2.14 mm/s (1.35 S.D.) and 1.83 mm/s (1.39 S.D.), respectively. At the ‘high’ level, male and prey playback vibrations induced a mean vibration amplitude of 8.32 mm/s (2.80 S.D.) and 10.61 mm/s (4.53 S.D.), respectively. Equipment limitations and increasing distortion of vibration waveforms with increasing amplification prevented us from raising the high level to the desired amplitude of 21 mm/s (mean amplitude of fly vibrations). We conducted a two- way ANOVA with maximum (peak-to-peak) amplitude as the response variable and waveform (‘prey’ or ‘male’) and amplitude level (‘high’ or ‘low’) as fixed effect factors, using web number as a blocking factor. Amplitude differed significantly between the high-and low-amplitude level treatments (F3,24 = 94.59, p < 0.0001) but not between the prey and male treatments (F3,24 = 1.96, p = 0.17). There was no significant interaction between waveform and amplitude level (F3,24 = 3.25, p = 0.08) (see Additional file B7-d). These measurements indicate that playback-induced vibration amplitudes were consistently transmitted at a distance of 15 cm from the input location. To determine whether the waveform of vibrations is conserved after transmission through 15 cm of web we measured the amplitude modulation factor (AMF) of the resulting waveforms. Before proceeding with statistical analysis, we log-transformed the data to meet the assumption of equal variance. We conducted a two-way ANOVA with amplitude modulation factor (AMF) as the response variable and waveform (‘prey’ or ‘male’) and amplitude level (‘high’ or ‘low’) as fixed effect factors; web number was used as a blocking factor. The back-transformed marginal mean AMF values for the ‘male’ and ‘prey’ waveforms were 3.46 (SE 1.13) and 16.76 (SE 1.13), respectively. AMF differed significantly between the ‘male’ and 'prey’ waveforms (F3,24 = 95.79, p < 0.0001). There was also a significant difference in AMF between the ‘low’ and ‘high’ amplitude levels (F3,24 = 6.38, p = 0.019). There was no significant interaction between waveform and amplitude level (F3,24 = 0.10, p = 0.76) (see Additional file B7-e). Although the waveform of playback-induced vibrations was not transmitted with perfect fidelity through 15 cm of web (as reflected by the wide range of AMF values for the ‘prey’ waveform), we observed very little overlap in terms of AMF between our ‘male’ and ‘prey’ treatment. Most (88 %) ‘male’ playback-induced AMFs ranged between 2 and 6, and all ‘prey’ playback-induced AMFs were > 7.5. This indicates that the unique characteristics of the ‘prey’ and ‘male’ vibrations were consistently conserved in the playback-induced vibrations.

Additional file B7

100

Figure B7 Playback design, input and playback-induced vibrations, and boxplots of the root mean square amplitude and amplitude modulation factor of the playback-induced vibrations. (a) Schematic drawing of the experimental design for vibration playbacks. A looped abdomen tremulation vibration of a male Latrodectus hesperus and a looped house fly vibration were played back on empty webs at low and high amplitude by a modified loudspeaker (S) (see main text for details), and transmission of playback-induced vibrations were recorded by a laser Doppler vibrometer (LDV) at a distance of 15cm. The star (★) indicates the recording location; (b-1 and b-2) Oscillograms of the looped male input vibration played back at low amplitude level, and a well-transmitted low-amplitude male playback-induced vibration; (c-1 and c-2) Oscillograms of the looped prey input vibration played back at high amplitude level, and a well-transmitted high-amplitude prey playback-induced vibration; note the difference in amplitude between input vibrations and playback-induced vibrations; (d) Box plots of the maximum amplitude of playback-induced prey and male vibrations at high- and low-amplitude level (n = 9 webs) recorded 15 cm away from the location of the input vibration. Median, mean and interquartile range (IQR); whiskers = upper and lower data point values within 1.5 IQR; (e) Box plots of the amplitude modulation factor of playback-induced prey and male vibrations at high-and low-amplitude level (n = 9 webs) recorded 15 cm away from the location of the input vibration. Median, mean and IQR; whiskers = upper and lower data point values within 1.5 IQR

101