University of Cincinnati

UNIVERSITY OF CINCINNATI

Date:______

I, ______, hereby submit this work as part of the requirements for the degree of: in:

It is entitled:

This work and its defense approved by:

Chair: ______

Seismic Communication in a Wolf Spider

A thesis submitted to the

Division of Research and Advanced Studies Of the University of Cincinnati

In partial fulfillment of the requirements for the degree of

MASTER’S OF SCIENCE (M.S.)

In the department of Biological Sciences of the McMicken College of Arts and Sciences

2005

Jeremy S. Gibson

B.S. Northern Kentucky University, 2000

Committee:

Dr. George W. Uetz, Chair Dr. Elke Buschbeck Dr. Kenneth Petren ii Abstract

I investigated the importance of the seismic component, substratum-borne vibrations, of the

multimodal courtship display in the wolf spider Schizocosa ocreata (Hentz) (Araneae:

Lycosidae). It is currently known that the visual signaling component of male multimodal

courtship displays conveys condition-dependent information, and that females can use this signal alone in mate choice decisions. I found that isolated seismic signals are also used in mate

choice, as females preferred males that were louder, higher pitched and with shorter signaling

pulses. Results also showed that male seismic signals are dependent on current condition and may convey information about male size and body condition. Seismic signals and visual signals are likely redundant, although some aspects of seismic signals may convey different information, supporting both the redundant and multiple messages hypotheses.

ii Acknowledgements

I would foremost like to thank Dr. George Uetz, my graduate advisor for all of his

support throughout this adventure. His time investment in my mentoring has meant more to me than I can possibly convey in words, but I will attempt to covey some of my thoughts and feelings. I know I haven’t told him often enough how much I have appreciated his patience and understanding concerning the several issues I faced while matriculating through UC’s graduate program. He has not only made me a better thinker and scientist he has also helped me to vastly improve my scientific writing. George, thank you so much!

My graduate research committee which was made up of Dr. Ken Petren and Dr. Elke

Buschbeck were also very instrumental in my success at UC. They were always willing to supply feedback and support when ever I asked. They made and continued to make wonderful suggestions and positive feedback on all portions of my research. Their dedication to understand and help me definitely added to my success.

Other faculty at UC also helped at various stages of my research progress, despite not having an actual connection with my research. Dr. Theresa Culley was always willing to answer statistics-related questions along with helping me setup SAS to run repeated measures logistic regression analysis. Dr. Steven Pelikan further supported some of my statistical inquires and helped me isolated the appropriate statistics for parts of my research. Thank you both for your contributions!

There are numerous peers and undergraduates to thank for helping me through my graduate experience at UC. I would particularly like to extend a very large thank you to my

iii fellow graduate students, Casey Harris, Julee Johns, Matt Klooster, Anne Lohrey, Jenai Milliser,

Dr. J. Andrew Roberts, and Kerri Wrinn; they were instrumental in my mental health throughout

this process. Not only did several of these people help in spider care and rearing a couple of

them even helped setup spiders for experiments. Thanks guys and gals! Spider care and

collection would have been a nightmare without the help of some very dedicated undergraduates,

Stephanie Doherty, Shanquala Pruitt, Melissa Salpietra and Erin White. I would particularly like

to thank Shanquala Pruitt for all of her assistance when the majority of this research was

conducted. Her work ethic and timeliness was absolutely invaluable to the success of this

project. Without her assistance in spider care and rearing a large portion of this research would

have been lost. Thanks so much!

A very special “thank you” goes to my family. They have shown me unconditional

support and love through this process, despite many of them having no real understanding of the

trials and tribulations I faced. I would definitely like to thank my Mother, Carole Warner, who showed me that no matter how big I dreamt, with dedication and the desire to succeed it was reachable. Thank you so much for all of your years of hard work to provide me with everything and more! I love you! I’d also like to thank my Grandmother, Madeline Goose, whom would go to mars and back if she had to, just for me. I love you too, thank you!

One special person deserves a heartfelt thank you for having to put up with me day in and day out, Keena Cole, my wife. Thank you for all of your support both mental and financial!

You have been my foundation, unwavering and unbreakable throughout! Thank you for being so wonderful!

iv Table of Contents

Abstract i Acknowledgements iii List of Tables 2 List of Figures 3

Chapter Page

I. Introduction 5 • Study Organism 5 • Research Objectives / Hypotheses 7 • References 9

II. Male Seismic Courtship Communication and Female Receptivity 11 in a Wolf Spider (Araneae: Lycosidae) • Abstract 12 • Introduction 13 • Methods 16 • Results 21 • Discussion 23 • Acknowledgements 25 • References 27 • Tables and figures 32

III. Effect of Rearing Environment and Feeding on Dynamic and 39 Static Attributes of Male Seismic Communication • Abstract 40 • Introduction 41 • Methods 43 • Results 48 • Discussion 52 • Acknowledgements 55 • References 57 • Tables and figures 61

IV. Conclusion and Future Direction 72

1 List of Tables and Figures

Tables Page

Table 2.1: Stepwise multiple logistic regression analysis for attributes of 35 percussive strikes alone. A lack of fit value close to 1.0 indicates a strong statistical result.

Table 2.2: Stepwise multiple logistic regression analysis for attributes of 35 seismic pulses. A lack of fit value close to 1.0 indicates a strong statistical result.

Table 2.3: Stepwise multiple logistic regression analysis for the combined 36 significant attributes for both percussive strikes and seismic pulses. A lack of fit value close to 1.0 indicates a strong statistical result.

Table 2.4: Stepwise multiple logistic regression analysis for attributes of male 36 courtship behavior. A lack of fit value close to 1.0 indicates a strong statistical result.

Table 2.5: Pearson’s pairwise correlation analysis results for seismic attributes 37 on aspects of male morphology and condition. r2(p-value)

Table 2.6: Seismic signal attribute levels of highly successful (3 receptive 38 females), successful (>0 receptive females) and unsuccessful males (0 receptive females). The letter “S” stands for a static trait while “D” means dynamic.

Table 3.1: Pearson pairwise correlation matrix results between male seismic 70 attributes and aspects of male size, condition and traits. “S” stands for static traits, while “D” stands for dynamic traits. All significant values have been highlighted; while values that maybe approaching significance are underlined and italicized.

Table 3.2: ANOVA table comparing laboratory-reared (LR), field-caught 71 (FC), field-caught-fed (FC-F), and field-caught-starved (FC-S) spiders. Post-hoc Tukey tests are given (under treatment group), same letters indicate statistical similarity while different letters indicate statistical difference.

2 Figures Page

Figure 2.1: A male S. ocreata, with conspicuous tufts. 32

Figure 2.2: Cue isolation apparatus: a) female chamber, b) male chamber, c) 32 visual barrier, d) point of recording.

Figure 2.3: Laser Doppler vibrometer setup with cue isolation apparatus. 33

Figure 2.4: Representative spectra of a sample of male seismic courtship. A) 34 Time signal (waveform) highlighting the two fundamentally different signal components. B) Close up view illustrating percussive strike temporal measurements. C) Close up view of seismic pulse temporal measurements. D) FFT-spectrogram taken from ‘B’ illustrating amplitude and frequency.

Figure 3.1: Recording platform: a) male chamber, b) point of recording 61

Figure 3.2: Comparison of laboratory-reared and field-caught spiders by 62 rearing environment for; (A) male size (cephalothorax width), (B) male mass, and (C) male body condition index (residuals of a regression analysis of mass by cephalothorax width)

Figure 3.3: Comparison of behavioral rates for the two components of jerky- 63 tapping between laboratory-reared and field-caught spiders.

Figure 3.4: Comparison of two dynamic seismic signal attributes between 64 laboratory-reared and field-caught spiders, (A) percussive strike peak amplitude and (B) seismic pulse duration.

Figure 3.5: Comparison of two static seismic signal attributes between 65 Laboratory-reared and field-caught spiders, percussive strike peak frequency and percussive strike maximum frequency.

Figure 3.6: Comparison of dynamic male traits over time (initial = time 0 and 66 final = after 14 days) between starved and fed field-caught male spiders. (A) mean percent change in male mass. (B) mean percent change in male body condition index. p-value of <0.0001 is indicated by ***.

Figure 3.7: Comparison of static seismic signal attributes over time (initial = 67 time 0 and final = after 14 days) between field-caught fed and starved spiders. (A) Mean percussive strike peak frequency repeated measures. (B) Mean percussive strike maximum frequency.

3 Figures Page

Figure 3.8: Comparison of dynamic seismic signal attributes over time (initial 68 = time 0 and final = after 14 days) between field-caught fed and starved spiders. (A) Mean percussive strike peak amplitude. (B) Mean seismic pulse duration. p-value of <0.005 is indicated by **.

Figure 3.9: Comparison of dynamic behavioral components over time (initial = 69 time 0 and final = after 14 days) between field-caught fed and starved spiders. (A) Mean rate of cheliceral strikes. (B) Mean rate of double-tapping. P-value of <0.001 indicated by **, <0.0001 indicated by ***.

4 Chapter I

Introduction

Multimodal communication (involving more than one sensory mode - e.g. chemical, acoustic, seismic, visual, and/or tactile) is common throughout the animal kingdom (Bradbury &

Vehrencamp 1998; Partan & Marler 1999). The use of multimodal displays can be seen in a variety of different contexts, but its role in courtship has been the focus of intense study (Hebets

& Papaj 2005). Female mate choice can drive the evolution of male courtship displays, as females are influenced by signals that indicate male quality (Andersson 1994). The question of whether male courtship signals in multiple sensory modes convey the same (redundant) or different (multiple) information about male quality is central to understanding the adaptive function and evolution of multimodal communication (Moller & Pomiankowski 1993; Omland

1996; Scheffer et al. 1996). To address this question, complex multi-sensory signals must be broken down into components and studied separately as well as together (Boake 2002). In the following studies, I use a wolf spider (Araneae: Lycosidae) model system to investigate questions about multimodal communication and sexual selection.

Study organism

The brush-legged wolf spider, Schizocosa ocreata (Hentz), is common in deciduous forests throughout the eastern United States; inhabiting the forest floor leaf litter of this region

(Cady 1984). The life cycle of S. ocreata begins in summer when the young are hatched and disperse from the mother’s abdomen. These spiderlings (juvenile spiders) grow until autumn, overwinter under the litter, then emerge and mature in late April as temperatures begin to warm.

5 The breeding season of S. ocreata is short, ranging from late April to June, with males maturing approximately 1 week prior to females (Roberts, Taylor & Uetz unpublished). Males typically search for mates once reaching adulthood, and if possible, will mate with more than one female; females will typically only mate once (Norton & Uetz 2005). During the breeding season the sex ratio ranges from male-biased to relatively equal, but as the season continues it later becomes skewed toward females (Roberts, Taylor & Uetz unpublished). Females persist into summer, when they create eggsacs and the cycle is repeated.

It is likely that due to the complex nature of their leaf litter habitat, S. ocreata rely on an assortment of sensory modalities to guide their behavior, including: tactile, chemical, seismic and visual modes. These various sensory modalities are not only vital during prey capture, but also play a key role in reproductive behavior. Males exhibit courtship in response to chemical cues from females (Tietjen 1977; Tietjen & Rovner 1982), and contact with these cues seems more important than other sensory cues such as visual presence of the female (Gibson & Uetz in prep). The courtship display of males contains both visual and seismic signals performed simultaneously (Stratton & Uetz 1981, 1983, 1986; Scheffer et al. 1996; Uetz 2000; Uetz &

Roberts 2002). Studies have demonstrated female choice based on isolated male visual signals, such as male foreleg tufts which serve as condition-indicating visual signals (Uetz et al. 2002;

Uetz & Roberts 2002). Little is currently known about the seismic signal of male courtship largely due to technical problems in signal detection and recording. Currently it is known that females can use isolated male seismic signals to distinguish heterospecifics from conspecifics

(Stratton & Uetz 1981, 1983) and these signals may also be used in mate choice (Scheffer et al.

1996). Despite the amount of research concerning multimodal communication in this species, questions remain. Do courtship signals of S. ocreata in different modes (visual and seismic)

6 convey different (multiple message hypothesis) or the same (redundant message hypothesis) information about male quality (Moller & Pomiankowski 1993; Omland 1996)? It is also not understood whether seismic signals are condition-dependent, as are the male secondary sexual visual signals (e.g. foreleg tufts). The research presented here focuses on these questions, to fill in the gap of our understanding on this well studied model system.

Objectives

This research focused on the seismic component of male S. ocreata courtship display, using computer-assisted laser Doppler vibrometry to record and analyze male signals.

Quantitative analyses of male S. ocreata seismic communication was used to investigate questions about its role in female mate choice and as an indicator of body condition.

Obj. 1 Female receptivity based on isolated male seismic communication (Chapter II)

Here I focus on two hypotheses: 1) females use male seismic signals in mate choice

decisions; and 2) male seismic signal attributes correlate with aspects of male morphology and

therefore may convey information. To test the above hypotheses, I will determine those seismic

signal attributes that best predict female receptivity, and ascertain whether or not male seismic

signals convey information about body condition.

7 Obj. 2 Effects of rearing environment and feeding on dynamic and static attributes of male

seismic communication (Chapter III)

Obj. 2a Here I test the hypothesis that rearing environment affects male seismic

communication. I predict that consistently fed laboratory-reared spiders will have an

overall higher quality seismic signal compared to spiders collected from the field as

adults. I will also examine associations between male seismic signal attributes and male

morphology to further substantiate the potential for information transfer via seismic

communication.

Obj. 2b I will also test the hypothesis that the differences observed between laboratory-

reared and field-collected spiders are primarily due to food availability. I predict that

dynamic signaling attributes of field-collected spiders fed regularly should recover to a

level more indicative of spiders that have had continual access to food as compared to

field-collected spiders that are starved. Again, I will examine correlative associations

between seismic signal attributes and male morphology.

8 Reference Cited

Andersson, M.B. 1994. Sexual Selection. Princeton, New Jersey: Princeton University Press.

Boake, C.R.B. 2002. Sexual signaling and speciation, a microevolutionary perspective. Genetica, 116:205-214.

Bradbury, J.W. & Vehrencamp, S.L. 1998. Principles of animal communication. Sunderland, Mass.: Sinauer Associates.

Cady, A. B. 1984. Microhabitat selection and locomotor activity of Schizocosa ocreata (Walckenaer)(Araneae: Lycosidae). Journal of Arachnology, 11:297:307.

Hebets, E.A. & Papaj, D.R. 2005. Complex signal function: developing a framework of testable hypotheses. Behavioral ecology and sociobiology, 57:197-214.

McClintock, W.J. & Uetz, G.W. 1996. Female choice and pre-existing bias: Visual cues during courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour, 52:167- 181.

Moller, A.P. & Pomiankowski, A. 1993. Why have birds got multiple sexual ornaments? Behavior Ecology and Sociobiology, 32:167-176.

Norton, S. & Uetz, G. 2005. Mating frequency in Schizocosa ocreata (Hentz) wolf spiders: evidence for a mating system with female monandry and male polygyny. Journal of Arachnology, 33:16-24.

Omland, K.E. 1996. Female mallard mating preferences for multiple male ornaments I Natural variation. Behavioral Ecology & Sociobiology, 39:353-360.

Partan, S. & Marler, P. 1999. Communication Goes Multimodal. Science, 283:1272-1273.

Scheffer, S.J., Uetz, G.W. & Stratton, G.E. 1996. Sexual selection, male morphology, and the efficacy of courtship signaling in two wolf spiders (Araneae: Lycosidae). Behavior Ecology and Sociobiology, 38:17-23.

Stratton, G.E. & Uetz, G.W. 1981. Acoustic communication and reproductive isolation in two species of wolf spiders. Science, 214:575-577.

Stratton, G.E. & Uetz, G.W. 1983. Communication via substratum-coupled stidulation and reproduction isolation in wolf spiders (Araneae; Lycosidae). Animal Behaviour, 31:164-172.

Stratton, G.E. & Uetz, G.W. 1986. The inheritance of courtship behavior and its role as a reproductive isolating mechanism in two species of Schizocosa wolf spiders. Evolution, 40:129-141.

9 Tietjen, W. 1977. Dragline-following by male lycosid spiders. Psyche, 84:165-178.

Tietjen, W. & Rovner, J. 1982. Chemical communication in lycosids and other spiders. In: Spider Communication: Mechanisms and Ecological Significance (Ed. By Witt, N. & Rovner, J.), pp. 249-279. Princeton, New Jersey: Princeton University Press.

Uetz, G.W. 2000. Signals and multi-modal signaling in spider communication. In: Animal Signals: Signaling and signal design in animal communication (Ed. By Espmark, Y., Amundsen, T., and Rosenqvist, G.), pp. 387-405. Trondheim, Norway: Tapir Publishers.

Uetz, G.W. & Roberts, J.A. 2002. Multisensory cues and multimodal communication in spiders: insights from video/audio playback studies. Brain, Behavior, and Evolution, 59:222-230.

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

10 Chapter 2

Male Seismic Courtship Cues and Female Mate Choice in a Wolf Spider

(Araneae: Lycosidae)

Jeremy S. Gibson and George W. Uetz

Department of Biological Sciences

University of Cincinnati

P.O. Box 210006

Cincinnati, OH 45221-0006

Formatted for submission to Animal Behaviour

11 Abstract

The brush-legged wolf spider Schizocosa ocreata (Hentz) (Araneae: Lycosidae) has proven to be an excellent model organism for the study of sexual selection and communication; due in part because males use complex multimodal signals (visual and vibratory [seismic]) during courtship displays. The visual component of male courtship is well described for this species and has been shown to serve as a condition-dependent signal females may use to evaluate potential mates.

Here, we present the results of female choice trials based on isolated male seismic signals to identify which aspects females might use to evaluate males and to investigate whether visual and seismic signals are redundant (i.e., ‘backup’ signals) or convey multiple messages (different information). Females exhibited receptive behavior when exposed to the seismic component of male courtship alone. The attributes best predicting female receptivity were: percussive strike

(peak amplitude, peak frequency, maximum frequency) and seismic pulse duration. Moreover, two behavioral attributes associated with the production of seismic signals cheliceral strike rate and double-tap rate, were also good predictors of female receptivity. Males that elicited receptive behavior (of females) were in better condition than males that did not, suggesting the possibility of information on male quality in the seismic channel of communication.

Correlations between aspects of male size and mass with attributes of seismic communication further substantiated this hypothesis. These correlations also revealed considerable overlap in the information contained within both the visual and seismic channels of complex multimodal signaling behavior. Given this information overlap, we conclude that seismic and visual signals are to a large extent redundant and serve as ‘backup’ signals. The use of redundant signaling may be adaptive in this species as it inhabits structurally complex forest leaf litter environments where signals can be attenuated rapidly.

12 Introduction

Multimodal communication (involving more than one sensory mode, e.g. chemical, acoustical, vibratory, visual, electrical, and/or tactile) is observable throughout the animal kingdom, and occurs in a variety of different contexts, including courtship, mating, competitive interactions and predator deterrence (Partan & Marler 1999). Despite the multitude of different types of animal interactions in which communication can occur, its role in courtship has been under intense study for many years. A more recent focus on courtship signals is in part due to interest in sexual selection, particularly the role of exaggerated male traits and displays and the potential information contained therein (Johnstone 1995). Originally, it was suggested that the primary function of courtship displays was to avoid species misidentification; however, it has become increasingly evident that signals convey other information as well, such as body condition (Mappes et al. 1996; Omland 1996; Uetz et al. 2002; Jawor & Bretwisch 2004) and parasite resistance (see review by Moller et al. 1999). It is also evident that different types of signal components in a multimodal display may communicate different information (see review by Hebets & Papaj 2005; Partan & Marler 2005).

Several hypotheses have been proposed to explain the role of each sensory mode of a multimodal signal, and their interactions (Hasson 1989; Moller & Pomiankowski 1993; Holland

& Rice 1998; Andersson et al. 2002; Candolin 2003; Hebets & Papaj 2005). Among these hypotheses recently revisited and/or proposed by Hebets & Papaj (2005), the “redundant message” and “multiple message” hypotheses have received the most attention. The redundant message hypothesis or “back-up” signal hypothesis (Moller & Pomiankowski 1993; Johnstone

1996) suggests that multiple sensory modes, or signal components of the same mode, contain the same or similar information. Hence, the different sensory modes serve as backup to each other.

13 Zuk et al. (1992) suggest that redundant signals may also act as natural gauges for which inconsistent traits can be compared. For example, when male jungle fowl had a single signaling trait exaggeratedly manipulated, females favored unmanipulated traits (Zuk et al. 1992; Zuk et al.

1990).

In contrast, the multiple message hypothesis (Moller & Pomiankowski 1993) predicts that each sensory mode, or each component signal of the same mode, contains different information

(e.g., different aspects of male quality). This was demonstrated by Jawor and Breitwisch (2004) with four different visual ornaments (e.g., Redness of breast plumage, bills and black face mask size) of male cardinals. They discovered that each visual ornament correlated with different aspects of the male such as body size, body condition, paternal care and reproductive success.

Arthropods are often used to study communication (Ewing 1989; Greenfield 2002), as their unique sensory organs and genetically “pre-programmed” behaviors allow insight into the function and evolution of neural and behavioral mechanisms. Spiders are excellent arthropod models for the study of male-female communication, as selection from potential cannibalism favors species-specific signals in a variety of modes (Witt & Rovner 1982; Uetz 2000; Barth

2002; Uetz & Roberts 2002; Persons & Uetz 2005). Spider communication has been suggested to function in agonistic encounters (Rovner 1968; Aspey 1977a, b), species recognition (Stratton

& Uetz 1981, 1983, 1986), female stimulation and/or pacification (Maklakov et al. 2003), courtship (Uetz 2000; Barth 2002; Rypstra et al. 2003) and mate choice (Parri et al. 1997, 2002;

Rivero et al. 2000).

Wolf spiders (Araneae: Lycosidae) are an excellent model system for the study of multimodal communication (Uetz & Roberts 2002). Males of the genus Schizocosa produce courtship displays that are either unimodal or bimodal (Miller et al. 1998; Hebets & Uetz 1999,

14 2000). The courtship displays produced by the males of all Schizocosa species studied so far contain vibratory (seismic) signals, but the use of visual signals is usually associated with the presence of ornamentation (Hebets & Uetz 1999, 2000). The research presented herein focuses on male Schizocosa ocreata (Hentz), which perform a variety of courtship behaviors consisting of visual displays and seismic displays that are behaviorally coupled (full description below).

Previous research has demonstrated that female S. ocreata distinguish conspecific and heterospecific males based on either visual or seismic signals alone or together (Stratton & Uetz

1983, 1986; Scheffer et al. 1996). Moreover, several studies focused on isolated male visual displays have shown its importance in courtship (McClintock & Uetz 1996; Uetz & Roberts

2002), as females can use the visual signal alone to make mating decisions. Male foreleg tuft size and symmetry (components of the conspicuous visual display) vary with feeding history and thereby serve as a condition-indicating visual signal in female mate choice (Uetz et al. 2002;

Uetz & Roberts 2002). Previous work has also shown that females rely almost equally on visual and seismic signals during mate choice (Scheffer et al. 1996). Despite equal reliance on both visual and seismic displays, questions still remain about the purpose of multimodal communication in this species. Here we use this well studied species to address how the seismic component of multimodal courtship signaling influences female receptivity and whether or not seismic signals convey information related to male quality. We will also investigate the relationship between seismic signals and visual signals in regard to the ‘redundant message’ and

‘multiple message’ hypotheses.

Methods

Spider collection and housing

Spiders were collected at the Cincinnati Nature Center (CNC) Rowe Woods facility

(Clermont Co. OH) as juveniles (to ensure virginity), and brought back to the laboratory for rearing and study. Each spider was housed individually in a plastic container (9 cm diameter, 6 cm height) with access to water ad libitum, and maintained under controlled conditions

(13L:11D, 22°C, ~ 70% RH). Crickets (Acheta domestica) were provided twice weekly as appropriate for spider size (no larger than the size of the spider’s cephalothroax).

Description of S. ocreata seismic and visual courtship behaviors

The primary component of male courtship behavior in S. ocreata is a complex multimodal display called “jerky-tapping”, which contains both visual and seismic signals produced in parallel (signals produced simultaneously) and/or serially (signals produced in a series) with one another (for original description, see Stratton & Uetz 1983). Jerky-tapping can be broken down into two major behavioral components - double-tapping and body bounces with cheliceral strikes - each of which contain elements of the seismic signal. Double-tapping, the primary visual signal, involves the raising and lowering of the first pair of legs, which are adorned with darkly pigmented bristles (tufts) located on the tibiae (Figure 2.1). During double- tapping, pulses of stridulation (seismic pulses) are produced by an organ located at the tibio- tarsal joint of the pedipalps (Rovner 1975). Cheliceral strikes are identified by an up and down

“bounce” motion of the entire body, usually accompanied by the chelicerae striking the substratum (Uetz & Denterlein 1979; Miller et al. 1998). While cheliceral strikes are being

16 performed, a percussive strike and seismic pulses (both of which are components of the seismic signal) are being produced simultaneously.

Experimental design

To ensure that females were exposed only to male seismic communication during each trial, a cue isolation apparatus was used (Figure 2.2). Males and females were placed in bottomless containers on a shared substrate but were visually blocked from each other by an opaque barrier.

In this configuration, seismic communication could be conveyed via the shared substratum, but all visual displays were prevented by the barrier placed between the pair.

Unmated adult females (7 to 29 days post maturity) were chosen at random and placed in the cue isolation apparatus alone, 12 hours prior to experimentation, to deposit silk (and chemical cues: e.g., pheromones) as they moved around the enclosure. Previous studies have shown that male S. ocreata will initiate and maintain courtship while in contact with female silk (Uetz &

Denterlein 1979; Stratton & Uetz 1981; Roberts & Uetz 2002). Experimentation began when the female was moved to an adjacent chamber and a randomly chosen male was placed in the chamber with her silk.

Male seismic signals were recorded using a laser Doppler vibrometer (LDV) (model:

PVD-100; Polytech PI) at a sampling rate of 12.5 kHz, a 4 channel analyzer (Model: OR24; Oros

Inc.) and a Dell Inspiron laptop PC (Figure 3). Recorded male seismic signals were analyzed using Avisoft Pro (ver 4.32; detailed below – Analysis of Seismic Signals). Additionally, the behaviors of both males and females were recorded using a digital video-camcorder (Canon XL-

1) for later analysis (detailed below – Analysis of Behaviors). Each male (N = 20) was paired

17 with three different females (N = 60) and this procedure was repeated for the Fall of 2003 and the Spring of 2004.

Analysis of Seismic Signals

Using Avisoft Pro, all recorded male seismic signals were resampled (256 bits) at a rate of 6 kHz (frequency resolution of 3 kHz) and had amplitude (dB) increased by a constant 20 dB.

To improve the signal to noise ratio, a time domain IIR high pass filter at 0.1kHz (settings: filter type Butterworth at an order of 8) was also applied. Relative amplitude was standardized by using the default 0 dB = 1 Volt calibration settings for all measurements. Spectrograms were created with an FFT length of 256 points with a spectral overlap of 93.75% (Flattop window,

100% window size). These settings allowed for a frequency resolution of 23Hz and a temporal resolution of 2.67ms. Two separate analyses were performed to measure the fundamentally different signal components: percussive strikes and seismic pulses (Figure 4A). However, all signals were analyzed for maximum frequency (Hz), peak frequency (frequency at highest amplitude; Hz), peak amplitude (dB), rate (#/time) duration (seconds) and interval (seconds)

(Figure 2.4B-D).

For percussive strike measurements, a randomly chosen one minute segment from each

5min trial was analyzed. Percussive strikes, which are of relatively higher amplitude than seismic pulses, could be identified using Avisoft’s automatic signal locator (settings: Threshold -

35dB amplitude, hold time 0.02ms). Signals identified by Avisoft’s automatic signal locator were verified by audio playback; improperly identified signals were removed from the data set.

Seismic pulses (pulses of stridulation) were of low amplitude and could not be identified via Avisoft's automatic signal locator. Hence, identification was conducted manually from a

18 randomly chosen 20 sec segment from the percussive strike segment using Avisoft’s magic cursor (settings: snap distance 10, threshold -50dB, start/end threshold -10dB relative to peak).

Seismic pulses were identified by playback and spectrogram signature.

Analysis of Behaviors

Behavioral analyses were conducted using digitally recorded video of live trials scored with The Observer (ver 5.0; Noldus). All trials were scored twice, once for male courtship behavior and a second time for the presence or absence of female receptivity behaviors (Stratton

& Uetz 1983, Scheffer et al.1996). Female S. ocreata perform three primary receptive behaviors: slow pivot (turning on an axis 90º to 360º), settle (lowering of the cephalothroax to the substrate) and tandem leg extend (often in association with settle, forelegs are straightened and extended, resting on the substrate). These behaviors have been associated with female mate choice and receptivity in previous studies (Uetz & Denterlein 1979; Stratton & Uetz 1981, 1983,

1986; McClintock & Uetz 1996; Scheffer et al. 1996; Uetz 2000; Uetz & Roberts 2002), and it has been shown that female receptivity behavior usually precedes successful copulations

(Montgomery 1903; Uetz & Denterlein 1970; Delaney 1997).

Male mating success was determined by the presence or absence of female receptivity behavior: males were scored as successful if they received any positive receptivity responses from a female; males that were unable to elicit a receptivity response from a female were scored as unsuccessful. Male success was summed across three trials; for example, if a male was successful in 2 out of 3 trials in eliciting a receptive response from females, he received a summed receptivity score of 2.

19 Male courtship behavior was also scored using digitally recorded video with The

Observer (ver. 5.0; Noldus). The primary male courtship display behavior, jerky-tapping, and its components cheliceral strike and double-tapping were the focus of this analysis. To assess male courtship vigor, the rate of each of these behaviors was used in the final analysis.

Male physical traits

After studies were conducted, spiders were euthanized using CO2 and preserved in vials

containing 70% ethanol. To assess male condition, a variety of potential condition-indicating

traits were measured, including: mass (mg), cephalothorax width (mm [CW]) and tuft size

(mm2). One picture for each trait (except for mass) was digitized using a digital camera (Pixera)

stereomicroscope (Wild M5) combination. Pictures for each trait were measured 3 times using

Image Tools (UTHSCSA, Ver 2.00); the averages of those measurements were then used in all

later statistical tests and calculations. A body condition index (BCI) was calculated from the

residuals of a regression analysis of mass on CW as in Jakob et al. (1996). Likewise, relative

male tuft size was scaled to body size (CW) using the same method.

Data Analysis

Variation in female receptivity was taken into account by repeated male use with

different females (3 females per 1 male). Due to repeated male use, measured variables were

averaged across trials. This method provided us with a conservative measure of male trial-to-

trial variation, while also accounting for female choice variation. Male success however,

eliciting receptive behavior from females, was added together for a composite success score (4

levels – 0,1,2,3). To test which seismic signal attributes were best at predicting female

20 receptivity we used multiple logistic regression analysis (as in Hardy & Field 1998; Taylor et al in press). Using a backward elimination process we removed all highly non-significant variables

(p = > 0.1), which left those attributes that best predicted female receptivity. To determine the possibility of information content contained within male seismic signals we used Pearson’s correlations; a correlative relationship between signal attributes and male traits would signify the possibility of information transfer. Variables that violated assumptions of normality were transformed before performing linear analyses.

Results

Seismic signals and receptivity

Isolated seismic signals from courting males were sufficient to elicit receptivity responses from females. Moreover, a large portion of males were able to elicit receptivity from at least one female (88%). When attributes of each seismic element (percussive strikes and seismic pulses) were considered, both elements had attributes that predicted receptivity. For both seismic pulses and percussive strikes, the peak frequency, maximum frequency and duration of each attribute were part of the ‘elements’ final multiple logistic regression model (Table 2.1, 2.2). The logistic model for seismic pulses differed from percussive strikes, in that the coefficient of variation for element interval and duration were also important in predicting female receptivity (Table 2.2).

To address how the significant attributes from both elements interacted on female receptivity, they were combined into a final multiple logistic regression analysis. In the final model, all attributes of seismic pulses became non-significant except for duration (Table 2.3). In the behavioral analysis of courtship vigor, multiple logistic regression analysis was performed on jerky-tapping and its components; this analysis revealed that the rate of cheliceral strikes and

21 double-tapping behavior also predicted female receptivity (Table 2.4). Highly successful males

(i.e., males that received receptivity responses from all three females) had seismic signals that were high in frequency, high in intensity, short in seismic pulse duration and very vigorous courtship behavior (Table 2.6).

Seismic signals and male condition

Males that received no receptivity responses from females exhibited significantly lower condition than males that received positive responses (t test: t58 = -2.041, p = 0.045). Hence,

male body condition was a good predictor of female receptivity (logistic regression: X2 = 7.86, df

= 3, p = 0.048). Despite body condition being a good predictor of female receptivity, only one

aspect of male courtship, behavioral rate of double-tapping, showed a relationship with it (Table

2.5). However, the covariates of the body condition index (male size and mass) did have

correlative associations with some seismic elements. Specifically, male size and mass were

negatively correlated with percussive strike frequency (maximum and peak) but positively

correlated with its duration (Table 2.5). No other association was found with male size. Though

male size did not correlate with any other attributes, male mass did; both behavioral components

of jerky-tapping (cheliceral strikes and double-tapping) were negatively correlated with male

mass. The visual trait, tuft area, was negatively correlated with various attributes, including:

percussive strike (maximum frequency and duration), seismic pulse (maximum and peak frequency) and cheliceral strike (rate).

22 Discussion

It is evident from our results that isolated male seismic signals are sufficient in eliciting female receptivity. This finding is consistent with other studies of this species (Scheffer et al.

1996; Hebets & Uetz 1999; Stratton & Uetz 1983, 1986). Unlike previous studies however, we identified which aspects of male seismic signals influenced female receptivity; females seemed to prefer males with seismic signals that were louder, lower in frequency and with a shorter duration seismic pulses (Table 2.6). Moreover, it also appears that females chose males on the basis of body condition, as males that were in poorer condition were less likely to receive receptivity responses from females. These results suggest that females are capable of choosing mates that are in better condition on the information gained via male seismic signaling alone.

These findings parallel results found for the visual signals of S. ocreata. Females exposed to manipulated video male visual signals (e.g. tufts) consistently chose males with larger tufts

(McClintock & Uetz 1996; Scheffer et al. 1996; Uetz & Roberts 2002; Uetz & Norton unpublished). Moreover, tuft size has been shown to correlate with male size to reflect feeding history and therefore serve as a condition-dependent trait (Uetz et al. 1999). Since females seem capable of choosing males in better condition via either seismic or visual signals in isolation, aspects of those signals must contain sufficient relevant information.

The presence of male quality information within male seismic signals was substantiated by the correlation of several seismic attributes with male size and condition (Table 2.5). For example, as male size increased, percussive strike peak frequency and maximum frequency decreased. This could be potentially explained by the relationship between size and pitch

(frequency): as size increases, the surfaces responsible for producing percussive strikes would also increase, and this would result in signals of lower frequency. However, this relationship

23 between size and pitch for seismic communication does not necessarily follow the same rules as in acoustical signals (Cocroft & Rodriguez 2005), as even very small animals (i.e. spiders) can create very low frequency signals via a substrate.

The negative relationship found between double tapping and body condition implies that males may be able to compensate for smaller size by increasing courtship vigor for short periods of time. Since courtship is energetically costly, prolonged bouts of courtship may not be possible for spiders in poorer condition. In the field, males may follow female drag lines for extended periods of time, all the while producing courtship displays (personal observations).

This extended search time combined with extended courtship times may be more energetically rigorous for males than the brief length of time represented by experimental trials in this study.

In the research presented here, spiders were confined to a small arena and trials lasted for only a short period of time; hence, it is also possible that males overcompensated during trials. Despite a negative correlation with current condition, traits and/or ornaments that are fixed at adulthood

(static condition indicators, i.e. cephalothorax width and tuft size) would not be affected by current nutritional gains/losses. Therefore, the fundamental frequency of percussive strikes should not change in relation to current condition (addressed in later chapter(s)). However, current condition would affect dynamic courtship/signal element attributes such as rate, duration, and amplitude.

Our results suggest that seismic signals and visual signals contain similar information

(i.e., male size and condition) and as a result can be considered to be redundant information.

While the contents of seismic and visual signals are not exactly the same, they both serve as indicators of male mate quality. Body size and mass, which are correlated with seismic attributes, are both components of the body condition index, which has also been shown to be

24 positively correlated with tuft size (Uetz et al. 2002). This overlap in information content of both courtship signals thus supports the redundant signaling hypothesis (Moller & Pomiankowski

1993).

Signal redundancy in this species may be adaptive due to the structural complexity of the leaf litter habitat, which can cause serious communication difficulties for small invertebrates. As

Scheffer et al. (1996) illustrated, by using a standard point source, seismic signals that are broadcast through the heterogeneous leaf litter environment traveled a maximum distance of

<20cm from the point source. Moreover, this distance would be likely to vary due to the structural complexity of leaf litter. In order for seismic communication to be effective, the signal needs to be sent across an unbroken chain of overlapping leaves, or to have the intended receiver situated on the same leaf. Hence, the reliance on seismic signals alone for courtship could be detrimental. Signal redundancy in this species prevents miscommunication, which could result in sexual cannibalism by females (Persons & Uetz 2005) and/or hybridization with closely- related species (Stratton & Uetz 1986) both of which reduce the fitness of males and females

(Stratton 1997; Orr 2001).

Acknowledgements

This research represents a portion of a thesis submitted as part of the requirements for the degree of Master of Science at the University of Cincinnati. We are grateful to the National

Science Foundation (Grants IBN-9906446 and IBN-0239164/0238854) for supporting this research along with other support from the American Arachnological Society and from the

University of Cincinnati, Department of Biological Sciences. We thank Elke Buschbeck, Ken

Petren, J. Andrew Roberts and Theresa Culley for their assistance at various stages of this

25 project. We also thank Steven Pelikan for statistical consulting and Raimund Specht for his expertise and help with Avisoft. We extend a very well deserved thank you to Keena Cole,

Shanquala Pruitt, Matt Klooster, Kerri Wrinn, Jenai Milliser, Casey Harris, Melissa Salpietra,

Stephanie Doherty, Erin White and Anne Lohrey for all their help with spider care, spider rearing and moral support.

26 References cited

Andersson, M.B. 1994. Sexual Selection. Princeton, New Jersey: Princeton University Press.

Andersson, S., Pryke, S., Ornborg, J., Lawes, M. & Andersson, M. 2002. Multiple receivers, multiple ornaments, and a trade-off between agonistic and epigamic signaling in a widowbird. American Naturalist, 160:683-691.

Aspey, W.P. 1977a. Wolf spider sociobiology: I. agonistic display and dominance-subordinance relationships in adult male Schizocosa crassipes. Behaviour, 62:103-141.

Aspey, W.P. 1977b. Wolf spider sociobiology: II. density parameters influencing agonistic behavior in Schizocosa crassipes. Behaviour, 62:142-163.

Bailey, W.J. & Haythornthwaite, S. 1998. Risk of calling by the field cricket Teleogryllus oceanicus; potential predation by Australian long-eared bats. Journal of Zoology, 244:505- 513.

Barth, F.G. 2002. A spider’s world: senses and behavior. Berlin, N.Y.: Springer.

Boake, C.R.B. 2002. Sexual signaling and speciation, a microevolutionary perspective. Genetica, 116:205-214.

Candolin, U. 2003. The use of multiple cues in mate choice. Biological Review, 78:575-595.

Cocroft, R.B. & Rodriguez, R.L. 2005. The behavioral ecology of insect vibrational communication. BioScience, 55:323-334.

Delaney, K. 1998. Masters Thesis, University of Cincinnati.

Ewing, A.W. 1989. Arthropod bioacoustics: neurobiology and behaviour. Ithaca, N.Y.: Comstock Publishing Associates.

Foelix, R.F. 1996. Biology of spiders. Oxford, N.Y.: Oxford University Press.

Greenfield, M.D. 2002. Signalers and receivers: mechanisms and evolution of arthropod communication. Oxford, N.Y.: Oxford University Press.

Hardy, I.C. & Field, S.A. 1998. Logistic analysis of animal contests. Animal Behaviour, 56:787- 792.

Hasson, O. 1989. Amplifiers and the handicap principle in sexual selection: a different emphasis. Proceedings of the Royal Society of London B, 234:383-406.

Haynes, K.F. & Yeargan, K.V. 1999. Exploitation of intraspecific communication systems: Illicit signalers and receivers. Annals of the Entomological Society of America, 92:960-970.

Hebets, E.A. & Uetz, G.W. 1999. Female responses to isolated signals from multimodal male courtship displays in the Wolf spider genus Schizocosa (Araneae: Lycosidae). Animal Behaviour, 57:865-872.

Hebets, E.A. & Uetz, G.W. 2000. Leg ornamentation and the efficacy of courtship display in four species of wolf spider (Araneae: Lycosidae). Behavioural Ecology and Sociobiology, 47:280-286.

Hebets, E.A. & Papaj, D.R. 2005. Complex signal function: developing a framework of testable hypotheses. Behavioral ecology and sociobiology, 57:197-214.

Holland, B. & Rice, W. 1998. Perspective: chase-away sexual selection: antagonistic seduction versus resistance. Evolution, 52:1-7.

Jakob, E.M., Marshall, S.D. & Uetz, G.W. 1996. Estimating fitness: A comparison of body condition indices. Oikos, 77: 61-67.

Jawor, J.M. & Breitwisch, R. 2004. Multiple ornaments in male Northern Cardinals, Cardinalis cardinalis, as indicators of condition. Ethology, 110:113-126.

Johnstone, R.A. 1995. Sexual selection, honest advertisement in the handicap principle: reviewing the evidence. Biological Review, 70:1-65.

Johnstone, R.A. 1996. Multiple displays in animal communication: ‘backup signals’ and ‘multiple messages’. Philosophical Transactions: Biological Sciences, 351:329-338.

Kaston, B. 1972. How to know the spiders. Dubuque, Iowa: Wm. C. Brown Company Publishers.

Klappert, K. & Reinhold, K. 2003. Acoustic preference functions and sexual selection on the male calling song in the grasshopper Chorthippus biguttulus. Animal Behaviour, 65:225-233.

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

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

Mappes, J.M., Alatalo, R.V., Kotiaho, J. & Parri, S. 1996. Viability costs of Condition-dependent sexual male display in a drumming Wolf spider. Proceedings: Biological Sciences, 263:785- 789.

28 McClintock, W.J. & Uetz, G.W. 1996. Female choice and pre-existing bias: Visual cues during courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour, 52:167- 181.

Moller, A.P. & Pomiankowski, A. 1993. Why have birds got multiple sexual ornaments? Behavior Ecology and Sociobiology, 32:167-176.

Moller, A.P., Christe, P. & Lux, E. 1999. Parasitism, post immune function, and sexual selection. Quarterly Review of Biology, 74:1-21.

Montgomery, T.H. Jr. 1903. Studies on the habitats of spiders, particularly those of the mating period. Proc. Acad. Nat. Sci. Phil., 43:131-138.

Omland, K.E. 1996. Female mallard mating preferences for multiple male ornaments I Natural variation. Behavioral Ecology & Sociobiology, 39:353-360.

Orr, M. 2001. Masters Thesis, University of Cincinnati.

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

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

Partan, S. & Marler, P. 1999. Communication Goes Multimodal. Science, 283:1272-1273.

Partan, S. & Marler, P. 2005. Issues in the Classification of multimodal communication signals. American Naturalist, 166:231-245.

Persons, M.H. & Uetz, G.W. 2005. Sexual cannibalism and mate choice decisions in wolf spiders: influence of male size and secondary sexual characters. Animal Behavour, 69:83-94.

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

Roberts, J.A. 2003. Dissertation, University of Cincinnati.

Rovner, J.S. 1968. An analysis of display in the lycosid spider Lycosa rabida Walkenaer. Animal Behaviour, 20:133-128.

Rovner, J.S. 1975. Sound production by neartic wolf spiders: a substratum-coupled stridulating mechanism. Science, 190:1309-1310.

Rypstra, A.L., Wieg, C., Walker, S.E. & Persons, M.H. 2003. Mutual mate assessment in wolf spiders: differences in the cues used by males and females. Ethology, 109:315-325.

Scheuber, H.S., Jacot, A. & Brinkhof, M.W.G. 2003. Condition dependence of a multicomponent sexual signal in the field cricket Gryllus campestris. Animal Behaviour, 65:721-727.

Simmons, L.W. 1995. Correlates of male quality in the field cricket, Gryllus campestris L.: age, size, and symmetry determine pairing success in field populations. Behavioral Ecology, 6:376-381.

Smith, M.J. & Roberts, J.D. 2003. Call structure may affect male mating success in the quacking frog Georgiana (Anura: Myobatrachidae). Behavior Evolution and Sociobiology, 53:221- 226.

Stratton, G. E. 1997. Investigation of species divergence and reproductive isolation of Schizocosa stridulans (Araneae: Lycosidae) from Illinois. Bulletin of the British Arachnological Society, 10:313-321.

Stratton, G.E. & Uetz, G.W. 1981. Acoustic communication and reproductive isolation in two species of wolf spiders. Science, 214:575-577.

Stratton, G.E. & Uetz, G.W. 1983. Communication via substratum-coupled stidulation and reproduction isolation in wolf spiders (Araneae; Lycosidae). Animal Behaviour, 31:164-172.

Stratton, G.E. & Uetz, G.W. 1986. The inheritance of courtship behavior and its role as a reproductive isolating mechanism in two species of Schizocosa wolf spiders. Evolution, 40:129-141.

Uetz, G.W. & Denterlein, G. 1979. Courtship behavior, habitat, and reproductive isolation in Schizocosa rovneri Uetz and Dondale (Araneae: lycosidae). Journal of Arachnology, 7:121- 128.

Uetz, G.W. & Roberts, J.A. 2002. Multisensory cues and multimodal communication in spiders: insights from video/audio playback studies. Brain, Behavior, and Evolution, 59:222-230.

Witt, P.N. & Rovner J.S. 1982. Spider communication: mechanisms and ecological significance. Princeton, N.J.: Princeton University Press.

Zuk, M. & Kolluru, G.R. 1998. Exploitation of sexual signals by predators and parasitoids. Quarterly Review of Biology, 73:415-438.

Zuk, M., Johnson, K., Thornhill, R. & Ligon, J. 1990. Mechanisms of female choice and red jungle fell. Evolution, 44:477-485.

Zuk, M., Ligon, J. & Thronhill, R. 1992. Effects of experimental manipulation of male secondary sex characters on female mate preference in red jungle fell. Animal Behaviour, 44:999-1006.

Figure 2.1: A male S. ocreata, with conspicuous tufts.

Figure 2.2: Cue isolation apparatus: a) female chamber, b) male chamber, c) visual barrier, d) point of recording.

Figure 2.3: Laser Doppler vibrometer setup with cue isolation apparatus.

Figure 2.4: Representative spectra of a sample of male seismic courtship. A) Time signal (waveform) highlighting the two fundamentally different signal components. B) Close up view illustrating percussive strike temporal measurements. C) Close up view of seismic pulse temporal measurements. D) FFT-spectrogram taken from ‘B’ illustrating amplitude and frequency.

Table 2.1: Stepwise multiple logistic regression analysis for attributes of percussive strikes alone. A lack of fit value close to 1.0 indicates a strong statistical result. x2 df p Percussive Bounces 51.641 15 <0.0001 Source Eliminated predictors Interval 4.909 3 0.688 Coef. Var. Interval 0.959 3 0.811 Coef. Var. Duration 1.479 3 0.687 Rate 2.624 3 0.453 Final model Season 30.755 3 <0.0001 peak frequency 7.155 3 0.067 peak Amplitude 23.455 3 <0.0001 max frequency 33.999 3 <0.0001 duration 10.059 3 0.018 Lack of fit 48.663 102 >0.999

Table 2.2: Stepwise multiple logistic regression analysis for attributes of seismic pulses. A lack of fit value close to 1.0 indicates a strong statistical result.

x2 df p Stridulation 45.396 15 <0.0001 Source Eliminated predictors Rate 0.487 3 0.917 Season 0.739 3 0.864 Peak amplitude 1.019 3 0.797 Interval 3.153 3 0.369 Final model Coef. Var. Interval 16.466 3 0.0009 Duration 9.352 3 0.0250 Coef. Var. Duration 19.569 3 0.0002 Peak Frequency 17.872 3 0.0005 Max Frequency 18.745 3 0.0003 Lack of fit 54.909 102 >0.999

Table 2.3: Stepwise multiple logistic regression analysis for the combined significant attributes for both percussive strikes and seismic pulses. A lack of fit value close to 1.0 indicates a strong statistical result. x2 df p Seismic signal attributes 49.617 15 <0.0001 Source Eliminated predictors Coef. Var. stridulation interval 0.698 3 0.874 Coef. Var. stridulation duration 0.999 3 0.802 Percussive bounce duration 2.808 3 0.422 Stridulation peak frequency 4.356 3 0.225 Stridulation max frequency 4.375 3 0.224 Final model Season 26.112 3 <0.0001 Percussive bounce peak frequency 14.163 3 0.0027 Percussive bounce peak Amplitude 16.193 3 0.0010 Percussive bounce max frequency 26.521 3 <0.0001 Stridulation duration 8.035 3 0.0453 Lack of fit 50.688 102 >0.999

Table 2.4: Stepwise multiple logistic regression analysis for attributes of male courtship behavior. A lack of fit value close to 1.0 indicates a strong statistical result. x2 df p Seismic signal rates 30.937 9 0.0003 Source Eliminated predictors Jerky – tap rate 4.001 3 0.261 Arching rate 4.001 3 0.261 Final model Season 10.381 3 0.0156 Cheliceral strike rate 9.539 3 0.0229 Double – tap rate 8.337 3 0.0395 Lack of fit 34.683 108 0.9986

Table 2.5: Pearson’s pairwise correlation analysis results for seismic attributes on aspects of male morphology and condition. r2(p-value)

Signal Attribute BCI Mass CW Tuft area Element Percussive Peak frequency -0.12(0.46) -0.36(0.02) -0.32(0.05) -0.23(0.15) strike Max frequency -0.06(0.69) -0.31(0.06) -0.40(0.01) -0.37(0.02) Peak amplitude 0.04(0.8) 0.02(0.91) -0.04(0.8) 0.01(0.96) Duration 0.18(0.22) 0.38(0.02) 0.35(0.03) 0.29(0.06)^

Seismic Pulses Interval -0.11(0.94) 0.04(0.83) 0.09(0.59) 0.22(0.17) Duration -0.05(0.78) 0.19(0.25) 0.23(0.15) -0.16(0.33) Cf.V. Duration 0.02(0.88) 0.14(0.38) 0.2(0.21) -0.06(0.7) Peak frequency 0.17(0.29) -0.08(0.62) -0.25(0.12) -0.34(0.03)^ Max frequency 0.18(0.28) -0.08(0.62) -0.25(0.11) -0.34(0.03)^

Behavioral Cheliceral Strike -0.24(0.13) -0.35(0.02) -0.26(0.1) -0.31(0.05)^ Double-tap -0.37(0.02) -0.40(0.01) -0.16(0.34) -0.15(0.34)^

Table 2.6: Seismic signal attribute levels of highly successful (3 receptive females), successful (>0 receptive females) and unsuccessful males (0 receptive females). Highly Successful Seismic Signal Successful Males Unsuccessful Males Males Element Mean SE Mean SE Mean SE Percussive strike 306 256 226 32 12 32 peak frequency (Hz) (Hz) (Hz)

Percussive strike 1,826 1,428 1,090 225 89 225 maximum frequency (Hz) (Hz) (Hz)

Percussive strike -24.27 -25.22 -26.09 0.97 0.36 0.97 peak amplitude (dB) (dB) (dB)

Seismic pulse 0.052 (s) 0.002 0.056 (s) 0.002 0.066 (s) 0.002 duration

Behavioral

Element

Cheliceral strike 23 17 14 2.8 1.2 2.8 rate (#/min) (#/min) (#/min)

59 45 46 Double-tap rate 5.4 2 5.4 (#/min) (#/min) (#/min)

38 Chapter 3

Effect of Rearing Environment and Food Availability on Seismic Signaling in Male Wolf

Spiders (Araneae: Lycosidae)

Jeremy S. Gibson and George W. Uetz

Department of Biological Sciences

University of Cincinnati

P.O. Box 210006

Cincinnati, OH 45221-0006

Formatted for submission to Animal Behaviour

Abstract

In the wolf spider Schizocosa ocreata (Hentz) (Araneae: Lycosidae) male visual signals

(i.e., foreleg tufts) are a condition-dependent secondary sexual characteristic, largely reflecting the feeding regime experienced while developing. Here we test if male seismic signals are also condition-dependent, using two different experimental approaches. The first approach involved comparing spiders from the same cohort raised in different environments (natural (field-caught -

FC) and artificial (laboratory-reared - LR)). The second approach was a food limitation experiment, where field-collected spiders were either fed or starved for three weeks. Our results from the rearing environment experiment show that LR spiders were in better condition and performed courtship more vigorously than did FC spiders. Seismic attributes also differed such that LR spiders produced signals that were lower in amplitude, higher in pitch and shorter in duration than FC spiders. Results from the food limitation experiment show that starved and fed spiders were not statistically different for static seismic signal attributes (i.e., frequency), which suggests that static signal attributes likely reflect nutrition before adulthood. Starved and fed spiders were statistically different for some dynamic seismic signaling traits (i.e., courtship vigor). However, two dynamic attributes (i.e., percussive strike peak amplitude and seismic pulse duration) were not over all significant but a trend of differentiation was found between starved and fed spiders. Changes in dynamic traits of well-fed field-collected spiders were such that several of their dynamic traits resembled levels exhibited by laboratory-reared spiders. This further supports the hypothesis that natural wolf spider populations exist under food limited conditions and that food limitation may explain a large portion of the variability seen in seismic signaling attributes.

Introduction

Sexual displays that contain multiple signals or components, called multimodal displays

(Partan & Marler 1999, 2005; Hebets & Papaj 2005), are made up of signals from a variety of different modalities, such as: chemical, mechanical (seismic, acoustic), electrical, and visual. It has become more evident that multimodal signals not only convey information, but possibly convey different types of information (Moller & Pomiankowski 1993; Johnstone 1996; Candolin

2003). This has become even more apparent in multimodal displays containing dynamic and static signal components (Gerhardt 1991, 1994). Static and dynamic signaling attributes have been studied in a wide variety of animals: fish (Rosenthal et al. 1996; Kodric-Brown & Nicoletto

2001), frogs (Gerhardt 1991, 1994; Prohl 2003), grasshoppers (Klappert & Reinhold 2003), crickets (Shaw & Herlihy 2000; Scheuber et al. 2003a), and wolf spiders (Rivero et al. 2000;

Parri et al. 2002; Uetz et al. 2002). The distinction between static and dynamic signals can be based on how signals are created and/or the amount of variation. Dynamic signals (i.e. signaling rate, signaling duration) are produced in a way that may change in response to intrinsic/extrinsic factors in the short-term. Static signals (i.e. carrier frequency), however, are most often produced by structures that do not change in response to intrinsic/extrinsic factors. Due to these differences, it has been proposed that static and dynamic signals may reflect different aspects of male quality (Moller & Pomiankowski 1993). Hence, it is likely that different information may be conveyed via these different channels. In this study, we were interested in the effects of rearing environment and food availability on dynamic and static attributes of the seismic communication of a wolf spider.

41 Arthropods such as spiders have proven to be excellent models in the study of communication (Witt & Rovner 1982; Uetz 2000; Barth 2002; Uetz & Roberts 2002; Huber

2005). Wolf spiders (Araneae: Lycosidae) particularly Schizocosa ocreata (Hentz) have proven to be excellent models for the study of multimodal communication (Uetz and Roberts 2002).

The courtship communication of S. ocreata is a complex multimodal display called ’jerky- tapping’ (Stratton and Uetz 1983). Jerky-tapping consists of signals being produced in two different sensory modalities - visual and seismic (Uetz & Denterlein 1979; Stratton & Uetz 1981,

1983). Jerky-tapping is produced in such a way that both the visual and seismic signals are produced in parallel and/or serially with one another. Females, when presented with male signals in these two modalities either together or in isolation, can distinguish between heterospecifics and conspecifics (Stratton & Uetz 1983, 1986; Scheffer et al. 1996).

Additionally, several studies have shown that females can choose males based solely on the visual component of male courtship, which has been shown to be condition-dependent

(McClintock & Uetz 1996; Uetz & Roberts 2002; Uetz et al. 2002). Until recently, the seismic component of male courtship had been shown only to function in species recognition (Stratton &

Uetz 1981, 1983), although there was some evidence that females may use the seismic component in mate choice (Scheffer et al. 1996). Recently, Gibson and Uetz (Chapter II) showed that females can use male seismic signals alone when making mate choice decisions. In that study, they isolated four seismic signal attributes (percussive strike peak amplitude, peak frequency and maximum frequency; seismic pulse duration) and two behavioral attributes (rate of cheliceral strikes and double-taps) that predicted female receptivity. Moreover, the males chosen by females were in better condition than males that received no receptivity responses

(Chapter II).

42 The attributes outlined by Gibson and Uetz (Chapter II) can be categorized on the basis of how they are produced. Percussive strike peak frequency (Hz) and maximum frequency (Hz) are produced via the chelicerae impacting the substrate (for full description see chapter II). Since the chelicerae are a feature of the adult male’s exoskeleton, which does not change in response to intrinsic/extrinsic factors, percussive strike peak frequency and maximum frequency are likely static signals. However, percussive strike peak amplitude, seismic pulse duration and both behavioral attributes (rate of cheliceral strikes and double-taps), are produced in a way in which intrinsic/extrinsic factors may have an effect. Percussive strike peak amplitude (or intensity) and seismic pulse duration are likely a product of male size, mass, body condition, muscle control and available energy (calories). In this case, these attributes are likely dynamic signals. Due to the fundamental differences between these signaling attributes, different factors experienced while developing may impact each signaling category differently. Hence, the research presented here has two goals: 1) to test the effects of rearing environment on dynamic/static signaling attributes; and 2) to test whether manipulation of body condition via food availability impacts dynamic/static signal components. By rearing spiders from the same cohort in different environments, and by subjecting adult spiders to different feeding régimes, we are able to tease apart the potential differences in information content of these seismic signals.

Methods

Rearing environment

Spiders from the same 2003-2004 cohort were collected from the Cincinnati Nature

Center (CNC) Rowe Woods facility (Clermont County, Ohio). Two different rearing groups were established: laboratory-reared (collected in Autumn) and field-caught (collected in Spring).

43 Laboratory-reared (LR; n=23) spiders were raised from egg sac to adulthood in the laboratory under controlled conditions (13L:11D, 22ºC, ~ 70% RH). These spiders were the offspring of females collected from the field (CNC) in the spring of 2003. Adult females captured from the field were housed individually in plastic containers (9 cm diameter, 6 cm height) with a 1 inch piece of garden hose which served as a retreat. Females that produced egg sacs were monitored daily for spiderlings (newly hatched juvenile spiders). After hatching, spiderlings rode on their mother’s abdomen for seven days; after which they were disbursed and placed in individual containers (plastic specimen containers; 5 cm diameter, 7 cm height) with a piece of moistened dental wick and 3 to 4 Collembola (springtails). As spiderlings aged, their diet was changed to accommodate for their increased size and consisted of springtails, fruit flies

(vestigial wing Drosophila melanogaster) and crickets (Acheta domestica).

Field-caught (FC; n=25) spiders were collected as adults from the field (CNC) in the spring of 2004 and brought back to the laboratory for study. To preserve the spider’s ‘natural condition’, trials were conducted the same day as collection. Upon completion of experiments, spiders were humanely sacrificed with CO2 anesthesia.

Food availability and body condition

To test the effect of manipulated food availability and body condition on male seismic

signals and behavioral rates, we collected adult male S. ocreata (N = 40) from the field (CNC) in

the spring of 2004. In order to preserve their field condition, initial recordings and mass

measurements were taken the same day spiders were collected. After initial recordings, males

were randomly divided into two treatment groups - satiated (n = 20) and starved (n = 20). A

subsequent trial occurred 14 days after the initial trial. Satiated spiders were fed 2 crickets

44 (Acheta domestica) twice weekly for the duration of this study and had access to water ad libitum. Starved spiders did not receive any prey throughout the testing period but had access to water ad libitum. Spiders of both treatment groups were housed individually in plastic containers (9 cm diameter, 6 cm height) and maintained under controlled conditions (13L:11D,

22°C, ~ 70% RH).

Recording male seismic communication

Seismic signals of courting males were recorded from a recording platform (Figure 3.1).

Courtship was induced by placing males onto silk (chemical cues, e.g. pheromones) that was laid down 12 hours prior to experimentation by randomly chosen adult virgin females. Seismic signals were recorded using a laser Doppler vibrometer (LDV) (model: PVD-100; Polytech PI) at a sampling rate of 12.5 kHz, a 4 channel analyzer (Model: OR24; Oros Inc.) and a Dell

Inspiron laptop PC. Recorded male seismic signals were analyzed using Avisoft Pro (ver 4.32; detailed below – Analysis of Seismic Signals). Additionally, the behaviors of males were recorded using a digital video-camcorder (Canon XL-1) for later analysis (detailed below –

Analysis of Behaviors).

Analysis of seismic signals

Using Avisoft Pro, all recorded male seismic signals were resampled (256 bits) down to a rate of 6 kHz (frequency resolution of 3 kHz) and had amplitude (dB) increased by a constant 20 dB. To improve the signal to noise ratio, a time domain IIR high pass filter at 0.1kHz (settings: filter type Butterworth at an order of 8) was also applied. Relative amplitude was standardized by using the default 0 dB = 1 Volt calibration settings for all measurements. Spectrograms were

45 created with an FFT length of 256 points with a spectral overlap of 93.75% (Flattop window,

100% window size). These settings allowed for a frequency resolution of 23Hz and a temporal resolution of 2.67ms. Male seismic signals were analyzed for attributes that were previously found to be associated with female receptivity (Chapter II) such as: percussive strike peak and maximum frequency (static; Hz), percussive strike peak amplitude (dynamic; dB), and seismic pulse duration (dynamic; seconds). Separate analyses were performed to measure the fundamentally different signal components: percussive strikes and seismic pulses (Chapter II -

Figure 2.4).

For percussive strike measurements, a randomly chosen one minute segment from each

5min trial was analyzed. Percussive strikes, which are of relatively higher amplitude than seismic pulses, could be identified using Avisoft’s automatic signal locator (settings: Threshold -

35dB amplitude, hold time 0.02ms). Signals identified by Avisoft’s automatic signal locator were verified by audio playback; improperly identified signals were removed from the data set.

Seismic pulses were of low amplitude and could not be identified via Avisoft's automatic signal locator; their identification was conducted manually from a randomly chosen 20 sec segment from the percussive strike segment using Avisoft’s “magic cursor” (settings: snap distance 10, threshold -50dB, start/end threshold -10dB relative to peak). Seismic pulses were identified by playback and spectrogram signature.

46 Analysis of behaviors

Male courtship behaviors were analyzed using digitally recorded video of live trials scored with The Observer (ver 5.0; Noldus). Here we focused on the rate of the two components of jerky-tapping - cheliceral strikes and double-tapping - as they have been associated with female choice (ChapterII).

Male trait measurements

After studies were conducted, spiders were euthanized using CO2 and preserved in vials

containing 70% ethanol. To assess male condition, a variety of potential condition-indicating

traits were measured, including: mass (mg), cephalothorax width (mm [CW]) and tuft size

(mm2). One picture for each trait (except for mass) was digitized using a digital camera (Pixera)

stereomicroscope (Wild M5) combination. Pictures for each trait were measured 3 times using

Image Tools (UTHSCSA, Ver 2.00); the averages of those measurements were then used in all

later statistical tests and calculations. A body condition index (dynamic; BCI) was calculated

from the residuals of a regression analysis of mass on CW as in Jakob et al. (1996). Likewise,

relative male tuft size was scaled to body size (static; CW) using the same method.

Data analysis

To test how rearing environment affected dynamic and static attributes of male seismic

signals we used One-way ANOVA with rearing environment as the independent variable. To

address whether or not seismic communication contains information about the sender, we used

Pearson’s pairwise correlations between male physical trait measurements and measured seismic

47 signal attributes. Variables that violated normality were transformed to better fit the assumptions of linear statistics.

To address the change in seismic signal attributes over time caused by the change in food availability we compared measurements from the initial trial to the final trial using matched pairs

ANOVA. Here, each seismic signal attribute served as a dependent variable while starved or fed treatments as the grouping factor. We also used Pearson’s pairwise correlation matrices between male size/body condition and aspects of male seismic signal attributes to identify the potential for information content contained within signals. Variables that violated the normality assumptions of linear statistics were transformed.

Results

Rearing environment

Adult male size measured by cephalothorax width (CW), a static feature of the exoskeleton, varied significantly with rearing environment (ANOVA: F1,49 = 11.1425, p =

0.0016), with spiders raised in the laboratory (LR) being larger than FC spiders (Figure 3.2A).

Corroborating the above, we also found that LR spiders had significantly greater mass (ANOVA:

F1,51 = 31.1712, p = <0.0001). Analysis of mass-CW regression residuals confirmed that LR

males were in better – body – condition (ANOVA: F1,51 = 15.0360, p = 0.0003) than FC spiders

(Figure 3.2B & 3.2C). Similarly courtship vigor, a dynamic behavioral trait, which could serve

as an indirect measure of condition showed a similar trend to the above; LR spiders were much

more vigorous when performing cheliceral strikes (ANOVA: F1,51 = 27.8236, p = <0.0001) and

double-taps (ANOVA: F1,46 = 50.4779, p = <0.0001) than FC spiders (Figure 3.3A & 3.3B).

Dynamic seismic signal attributes also varied significantly between LR and FC spiders.

48 Laboratory-reared spiders had a lower intensity (amplitude) percussive strike compared to FC spiders (ANOVA: F1,49 = 68.6504, p = <0.0001) (Figure 3.4A). Furthermore, LR spiders seismic

pulses were shorter in duration than FC spiders (ANOVA: F1,45 = 14.6705, p = 0.0004) (Figure

3.4B). We found that for the static seismic signal attribute percussive strike peak frequency, LR

and FC spiders did not statistically differ (ANOVA: F1,49 = 2.8949, p = 0.0952). However, LR

spiders had a significantly higher percussive strike maximum frequency than FC spiders

(ANOVA: F1,47 = 10.6710, p = 0.0020) (Figure 3.5).

Pearson’s pairwise analyses revealed correlations between aspects of male seismic signal

attributes, male size and male body condition. Specifically, both components of jerky-tapping

(cheliceral strike rate and double-tap rate) were positively correlated with male condition

(cheliceral strike rate - r2 = 0.35; p = 0.01 / double-tap rate - r2 = 0.31; p = 0.02) and male mass

(cheliceral strike rate - r2 = 0.33; p = 0.01 / double-tap rate - r2 = 0.28; p = 0.04). We also found

that scaled tuft area correlated negatively with two different signaling attributes, cheliceral strike

rate (r2 = -0.29; p = 0.04) and percussive strike peak frequency (r2 = -0.30; p = 0.03); but

correlated positively with percussive strike peak amplitude (r2 = 0.50; p = 0.0002).

Food availability and body condition

Over the course of this experiment, 9 out of 40 spiders died; but death was not due to

treatment group (df = 1; X2 = 0.144; p = 0.705). Hence, the spiders that died were excluded from

later statistical analyses (15 fed and 16 starved, n = 31). Due to the amount of variation in mass

and body condition (dynamic measurements), for adult male spiders collected from the field, we

used percent change as a means of standardization. Males that were starved had a significantly

lower mass than fed spiders (matched pairs ANOVA; F28 = 23.6585, p = <0.0001; Figure 3.6A).

49 Furthermore, spiders that were starved were in poorer condition than spiders that were fed

(matched pairs ANOVA; F30 = 44.4767, p = <0.0001; Figure 3.6B). For both mass and body

condition, there was a highly significant change from initial to final trials (mass – matched pairs

ANOVA; F28= 66.2705, p = <0.0001; body condition – F30= 69.4691, p = <0.0001).

We found no significant effect based on feeding treatment (fed and starved) for either of

the static seismic signal attributes, percussive strike peak frequency (matched pairs ANOVA; F28

= 0.0447, p = 0.8341; Figure 3.7A) and maximum frequency (matched pairs ANOVA; F28=

0.2491, p = 0.6218; Figure 3.7B). However, for both dynamic seismic signal attributes, a trend of divergence between fed and starved spiders was found. For percussive strike peak amplitude, despite no overall model significance (matched pair ANOVA; F28= 2.5956, p = 0.1188), there

was a significant change between their final and initial measurements (matched pairs ANOVA;

F28 = 6.4460, p = 0.0172; Figure 3.8A). Moreover, a post-hoc t-test showed a significant

difference between fed and starved spiders at the final measurement for percussive strike peak

amplitude (t-test; t27 = 3.153, p = 0.0039). Seismic pulse duration showed a similar trend as

percussive strike peak amplitude, with no overall model significance (matched pairs ANOVA;

F24 = 0.1863, p = 0.6700) and with the initial and the final trials approaching significance

(matched paris ANOVA; F24 = 3.4319, p = 0.0768; Figure 3.8B). Unlike percussive strike peak

amplitude, a post-hoc t-test showed no statistical difference between fed and starved spiders at

the final measurement (t-test; t23= 1.499, p = 0.1475) for seismic pulses. Male courtship vigor

(measured by the behavioral components of jerky tapping) showed a distinct separation between

fed and starved spiders for cheliceral strikes (matched pairs ANOVA; F30 = 3.8782, p = 0.0585;

Figure 3.9A) and double-taps (matched pairs ANOVA; F29= 10.9994, p = 0.0025; Figure 3.9B).

Both behavioral components had significant changes from the initial to final measurements

50 (cheliceral strike – matched pairs ANOVA; F30= 19.0619, p = 0.0001; double-taps – F29= 9.1879, p = 0.0052).

We used Pearson pairwise correlations between male size/condition and seismic signal attributes to identify the potential for information content contained within males seismic signals.

All of the dynamic seismic signaling attributes we measured had significant correlations with male body condition and mass (Table 3.1). Percussive strike peak amplitude, was the only dynamic signaling trait to also correlate with male size (r2 = 0.48; p = <0.001).

Rearing environment and food availability comparison

After analyzing the two different experiments presented above, we decided to combine

them for an overall comparison, giving us 4 treatment groups (LR, FC, FC-fed, & FC-starved).

This was justified as all spiders were from the same cohort and in essence all shared the same

independent variables of food availability / food limitation.

Rearing environment (LR & FC) significantly affected all dynamic attributes we

measured (one-way ANOVA; Table 3.2). Laboratory reared and FC-fed spiders were not statistically different (post-hoc Tukey tests) for any measured dynamic attribute except for one, percussive strike peak amplitude (Table 3.2). As with the dynamic traits, static traits were also affected by rearing environment. Laboratory reared and FC-starved spiders were statistically different (post-hoc Tukey tests) for all static seismic signal attributes but overlaps were found between LR, FC, and FC-fed spiders (Table 3.2).

51 Discussion

Rearing environment

It is clear from our results that rearing environment affected several aspects of male S. ocreata. Overall, laboratory-reared (LR) spiders were larger, in better condition, performed courtship behavior more vigorously, and had a seismic display more indicative of a successfully

‘chosen’ male, (based on results from Chapter II; see Table 2.6 for seismic attributes of a successful male) than field-caught (FC) spiders. Since all spiders used in this experiment were from the same cohort but developed in either natural or artificial conditions, both subpopulations were under different selection pressures.

It is likely that natural selection and / or artificial selection could have had an impact on our experimental groups. Field-caught spiders were collected during the natural breeding season, which means they had survived several naturally selective forces (e.g. overwintering, food limitation and predation) during that time. Hence, field-caught spiders represent individuals that

“succeeded” at surviving until the breeding season and therefore might reflect higher genetic quality. Laboratory-reared spiders were not subjected to any “natural” selective forces and as a result may represent a wider range of possible phenotypes that would have normally been removed from the population. Moreover, several studies have shown that environmental experience can shape the development of the central nervous system (Bailey & Kandel 1993;

Heisenberg et al. 1995; Barth et al. 1997; Hoikka & Suvanto 1999; Dukas & Mooers 2003). The central nervous system has been shown to be plastic and capable of responding to environmental heterogeneity during development (mammals and birds, a review by: Rosenzweig & Bennet

1996). However, very little is currently known about the development and plasticity of spider nervous systems particularly in response to environmental heterogeneity.

52 New environments can produce conditions to which animals do not react as they would normally. This effect, “neophobia”, could have possibly created the courtship vigor differences we observed between our two groups. Despite the behavioral differences observed, it is not likely that neophobic effects could alter physical attributes (i.e. frequency) of male seismic signals. Field-caught spiders raised in natural conditions developed under prolonged food- limited conditions (Miyashita 1968, Anderson 1974; as cited in Moya-Larano 2003); developing under these conditions may result in a lasting reduction in overall attainable condition (Metcalfe

& Monaghan 2001). Many animals have some ability to compensate for poor rearing conditions, but the costs of compensation are not clearly understood (Metcalfe & Monaghan 2001). Spiders raised throughout their lives in food-limited conditions make trade-offs between vital processes and less important processes (Jespersen & Toft 2003). These trade-offs usually have lasting and often costly side effects that usually do not reveal themselves until late in adulthood, often reducing lifespan (Metcalfe & Monaghan 2001).

Food availability and body condition

Our results clearly demonstrate an effect of feeding regime on seismic communication, but this effect appears limited to dynamic seismic signal attributes. Seismic signal attributes that were static (i.e. frequency), were not affected by changes in body condition in response to being fed or starved. However, dynamic signal attributes (i.e. amplitude and courtship vigor) were affected by food availability. These results suggest that food limitation / food availability may explain a large portion of the differences we saw in the rearing environment study. Furthermore, these results also confirm that our initial designations for the various seismic signal attributes were correct; attributes categorized as dynamic were affected by changes in body condition

53 while those attributes categorized as static were not. Since these designations are in part based on how/where the signal is produced, dynamic and static signals may reflect different information.

Dynamic seismic signals were correlated with changes in male current body condition and mass; however static signal attributes were not (Table 3.1). This supports a hypothesis of non-redundant signaling (“multiple messages”, Moller & Pomiankowski 1993) based on signals containing static and dynamic components. It is feasible that static and dynamic signaling attributes may convey “snapshots” of various developmental stages. For example, in the field cricket Gryllus campestris, juvenile foraging success directly affects adult body size and harp size (fixed (static) acoustic signaling organ) (Scheuber et al. 2003a). Subsequently, males with smaller harps produce calls of higher frequency (pitch) than males with larger harps. While the harp (static) reflects conditions experienced as a juvenile, daily calling rate and chirp rate (both dynamic) are affected by current nutritional gains and/or losses (current body condition)

(Scheuber et al. 2003b). Females may use these various signal components (multicomponent sexual signals) during mate choice differently and to varying degrees. It is evident that in S. ocreata, females may choose males on the basis of visual cues or seismic cues when each is presented in isolation, and it is also evident that visual and seismic signals contain similar information (Chapter II). The similarity in information contained within the visual and seismic signals may change due to environmental influences. It was shown that static seismic signals correlated with aspects of male size and mass in laboratory-maintained spiders (Chapter II) but the lack of correlations in this study suggests these relationships may change based on developmental conditions. Due to the complexity of potential environmental interactions more controlled studies should be done to isolate what factors affect which category of signaling

54 (dynamic / static) the most. The differences we found for these two basic categorical (static and dynamic) features of male signals may reflect a suite of available information which females may use during mate selection.

Rearing environment and food availability comparison

A comparison of both experiments showed that food limitation had a strong influence in creating the differences observed between LR and FC spiders; but only for dynamic attributes.

Our comparisons of static attributes unfortunately did not reveal any clear relationship other than they were also affected by rearing environment. These comparisons suggest that FC spiders are food limited in their natural environment and as a result are in poorer overall condition when compared to well-fed spiders. But our results also show that it is possible, if food availability increases, that FC spiders can improve their condition to a level similar of well fed spiders.

Acknowledgements

This research represents a portion of a thesis submitted as part of the requirements for the degree of Master of Science at the University of Cincinnati. We are grateful to the National

Science Foundation (Grants IBN-9906446 and IBN-0239164/0238854) for supporting this research along with other support from the American Arachnological Society and from the

University of Cincinnati, Department of Biological Sciences. We thank Elke Buschbeck, Ken

Petren, J. Andrew Roberts and Theresa Culley for their assistance at various stages of this project. We also thank Steven Pelikan for statistical consulting and Raimund Specht for his expertise and help with Avisoft. We extend a very well deserved thank you to Keena Cole,

Shanquala Pruitt, Matt Klooster, Kerri Wrinn, Jenai Milliser, Casey Harris, Melissa Salpietra,

55 Stephanie Doherty, Erin White and Anne Lohrey for all their help with spider care, spider rearing and moral support.

56 References Cited

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

Barth, F.G. 2002. A spider’s world: senses and behavior. Berlin, N.Y.: Springer.

Barth, M., Hirsch, H., Meinertzhagen, I. & Heisenberg, M. 1997. Experience-dependent development plasticity in the optic lobe of Drosophila melanogaster. Journal of Neuroscience, 17:1493-1504.

Bailey, C. & Kandel, E. 1993. Structural changes accompanying memory storage. Annual Review of Physiology, 55:397–426.

Candolin, U. 2003. The use of multiple cues in mate choice. Biological Review, 78:575-595.

Carducci, J. & Jakob, E. 2000. Rearing environment affects behaviour of jumping spiders. Animal Behaviour, 59:39-46.

Dukas, R. & Mooers, A. 2003. Environmental enrichment improves mating success in fruit flies. Animal Behaviour, 66:741-749.

Gerhardt, H. C. 1991. Female mate choice in treefrogs: static and dynamic acoustic criteria. Animal Behaviour, 42:615-635.

Gerhardt, H. C. 1994. The evolution of the vocalization in frogs and toads. Annual Review of Ecology and Systematics, 25:293-324.

Gibson, R. & Bradbury, J. 1985. Sexual selection in lekking sage grouse: phenotypic correlates of male mating success. Behavioural Ecology and Sociobiology, 18:117:123.

Gibson, R., Bradbury, J. & Vehrencamp, S. 1991. Mate choice in lekking sage grouse revisited: the roles of vocal display, female site fidelity, and copying. Behavioral Ecology, 2:165–180.

Gibson, R. 1996. Female choice in sage grouse: the roles of attraction and active comparison. Behavioural Ecology and Sociobiology, 39:55-59.

Hebets, E.A. & Papaj, D.R. 2005. Complex signal function: developing a framework of testable hypotheses. Behavioral ecology and sociobiology, 57:197-214.

Heisenberg, M., Heusipp, M. & Wanke, C. 1995. Structural plasticity in the Drosophila brain. Journal of Neuroscience, 15:1951–1960.

Hoikkala, A. & Suvanto, L. 1999. Male courtship song frequency as an indicator of male mating success in Drosophila Montana. Journal of Insect Behavior, 12:599-609.

57 Huber, B. 2005. Sexual selection research on spiders: progress and biases. Biological Review, 80:363-385.

Jakob, E.M., Marshall, S.D. & Uetz, G.W. 1996. Estimating fitness: A comparison of body condition indices. Oikos, 77: 61-67.

Jespersen, L & Toft, S. 2003. Compensatory growth following early nutritional stress in the wolf spider Pardosa prativaga. Functional Ecology, 17:737-746.

Johnstone, R.A. 1996. Multiple displays in animal communication: ‘backup signals’ and ‘multiple messages’. Philosophical Transactions: Biological Sciences, 351:329-338.

Klappert, K. & Reinhold, K. 2003. Acoustic preference functions and sexual selection on the male calling song in the grasshopper Chorthippus biguttulus. Animal Behaviour, 65:225-233.

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

McClintock, W.J. & Uetz, G.W. 1996. Female choice and pre-existing bias: Visual cues during courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour, 52:167- 181.

Metcalfe, N. & Monaghan, P. 2001. Compensation for a bad start: grow now, pay later? Trends in Ecology and Evolution, 16:254-260.

Miyashita, K. 1968. Quantitative feeding biology of Lycosa T- insignita BOES. et STR. (Araneae: Lycosidae). B. Natl. I. Agr. Sci. C 22: 329–343.

Moller, A.P. & Pomiankowski, A. 1993. Why have birds got multiple sexual ornaments? Behavior Ecology and Sociobiology, 32:167-176.

Moya-Larano, J., Orta-Ocana, J., Barrientos, J., Bach, C. & Wise, D. 2003. Intriguing compensation by adult female spiders for food limitation experienced as juveniles. Oikos, 101:539:548.

Partan, S. & Marler, P. 1999. Communication Goes Multimodal. Science, 283:1272-1273.

Partan, S. & Marler, P. 2005. Issues in the classification of multimodal communication signals. The American Naturalist, 166(2):231-245.

58 Prohl, H. 2003. Variation in Male Calling Behaviour and Relation to Male Mating Success in the Strawberry Poison Frog (Dendrobates pumilio). Ethology, 109:273-290.

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

Rosenthal, G., Evens, C. & Miller, W. 1996. Female preference for dynamic traits in the green swordtail Xiphophorus helleri. Animal Behaviour, 51:811-820.

Rosenzweig, M. & Bennett, E. 1996. Psychobiology of plasticity: effects of training and experience on brain and behavior. Behavior and Brain Research, 78:57–65.

Scheuber, H., Jacot, A. & Brinkhof, M. 2003a. The effect of past condition on a multicomponent sexual signal. Proceedings: Biological Sciences, 270:1779-1784.

Scheuber, H.S., Jacot, A. & Brinkhof, M.W.G. 2003b. Condition dependence of a multicomponent sexual signal in the field cricket Gryllus campestris. Animal Behaviour, 65:721-727.

Shaw, K. & Herlihy, D. 2000. Acoustic preference functions and song variability in the Hawaiian cricket Laupala cerasina. Proceedings: Biology Sciences, 267:577-584.

Stratton, G.E. & Uetz, G.W. 1981. Acoustic communication and reproductive isolation in two species of wolf spiders. Science, 214:575-577.

Stratton, G.E. & Uetz, G.W. 1983. Communication via substratum-coupled stidulation and reproduction isolation in wolf spiders (Araneae; Lycosidae). Animal Behaviour, 31:164-172.

Stratton, G.E. & Uetz, G.W. 1986. The inheritance of courtship behavior and its role as a reproductive isolating mechanism in two species of Schizocosa wolf spiders. Evolution, 40:129-141.

Uetz, G.W. & Denterlein, G. 1979. Courtship behavior, habitat, and reproductive isolation in Schizocosa rovneri Uetz and Dondale (Araneae: lycosidae). Journal of Arachnology, 7:121- 128.

Uetz, G.W. & Roberts, J.A. 2002. Multisensory cues and multimodal communication in spiders: insights from video/audio playback studies. Brain, Behavior, and Evolution, 59:222-230.

Witt, P.N. & Rovner J.S. 1982. Spider communication: mechanisms and ecological significance. Princeton, N.J.: Princeton University Press.

Wiley, R. 1973. The strut display of male sage grouse: a “fixed” action pattern. Behaviour, 47:129–152.

Figure 3.1: Recording platform: a) male chamber, b) point of recording

) 4.6 A 4.5

4.4

4.3 Mean 4.2

4.1

cephalothorax width (mm 4 Laboratory Reared Field Caught Rearing Environment

0.08 B 0.07 0.06 0.05 0.04 0.03

Mean mass (g) Mean mass 0.02 0.01 0 Laboratory Reared Field Caught Rearing Environment

0.01 C 0.008

0.006

0.004

0.002

-0.002

Mean body condition index -0.004 Laboratory Reared Field Caught Rearing Environment

Figure 3.2: Comparison of laboratory-reared and field-caught spiders by rearing environment for; (A) male size (cephalothorax width), (B) male mass, and (C) male body condition index (residuals of a regression analysis of mass by cephalothorax width)

) 50 45 Cheliceral strike 40 Double-tap 35 30 25 20 15 10 5

Mean behavioral rate (#/min rate behavioral Mean 0 Laboratory Reared Field Caught Rearing Environment

Figure 3.3: Comparison of behavioral rates for the two components of jerky- tapping between laboratory-reared and field-caught spiders.

-35 A -30

-25

-20

-15

-10

Mean amplitude (dB) -5

0 Laboratory Reared Field Caught Rearing Environment

0.07 B 0.06

0.05

0.04

0.03

0.02

Mean duration (sec) 0.01

0 Laboratory Reared Field Caught Rearing Environment

Figure 3.4: Comparison of two dynamic seismic signal attributes between laboratory-reared and field-caught spiders, (A) percussive strike peak amplitude and (B) seismic pulse duration.

700 Peak frequency 600 Maximum frequency

500

400

300

200

Mean frequency (Hz) 100

0 Laboratory Reared Field Caught Rearing Environment

Figure 3.5: Comparison of two static seismic signal attributes between Laboratory- reared and field-caught spiders, percussive strike peak frequency and percussive strike maximum frequency.

1.5 A 1

0.5

-0.5

-1 Fed Starved Mean percent change in mass -1.5 Initial Final *** Time

200 B 150 100 50 0 -50 -100

condition index -150 -200 Fed -250 Mean percent change in body Starved -300 Initial Final *** Time

Figure 3.6: Comparison of dynamic male traits over time (initial = time 0 and final = after 14 days) between starved and fed field-caught male spiders. (A) mean percent change in male mass. (B) mean percent change in male body condition index. p-value of <0.0001 is indicated by ***.

300 A 250

200

150

100 frequency (Hz) frequency

Fed 50 Starved Mean percussive strike peak strike percussive Mean

0 Intial Final Time

700 m B 600

500

400

300

frequency (Hz) 200 Fed 100 Starved Mean percussive strike maximu 0 Intial Final Time

Figure 3.7: Comparison of static seismic signal attributes over time (initial = time 0 and final = after 14 days) between field-caught fed and starved spiders. (A) Mean percussive strike peak frequency repeated measures. (B) Mean percussive strike maximum frequency.

-16 A k

-18

-20

-22

-24 amplitude (dB) -26

-28 Fed Mean percussive strike pea Starved -30 Initial Final ** Time

) 0.08 B 0.075

0.07

0.065

0.06

0.055 Fed 0.05 Starved 0.045

Mean seismic pulse duration (Sec. 0.04 Initial Final Time

Figure 3.8: Comparison of dynamic seismic signal attributes over time (initial = time 0 and final = after 14 days) between field-caught fed and starved spiders. (A) Mean percussive strike peak amplitude. (B) Mean seismic pulse duration. p-value of <0.005 is indicated by **.

20 A Fed

s 18 Starved 16

(#/min) 8

Mean rate of cheliceral strike 2

0 Initial Final** Time

B 60 Fed Starved 50

***

Mean rate of double-tapping (#/min) 0 Initial Final Time

Figure 3.9: Comparison dynamic behavioral components over time (initial = time 0 and final = after 14 days) between field-caught fed and starved spiders. (A) Mean rate of cheliceral strikes. (B) Mean rate of double-tapping. P-value of <0.001 indicated by **, <0.0001 indicated by ***.

Table 3.1: Pearson pairwise correlation matrix results between male seismic attributes and aspects of male size, condition and traits. “S” stands for static traits, while “D” stands for dynamic traits. All significant values have been highlighted; while values that maybe approaching significance are underlined and italicized. Body Signal Mass CW Tuft Area Attribute Condition Element r2(p-value) r2(p-value) r2(p-value) r2(p-value) Peak Percussive Strike S 0.05(0.7) 0.02(0.87) -0.02(0.86) 0.17(0.19) Frequency S Max Frequency -0.02(0.88) -0.11(0.41) -0.14(0.3) -0.11(0.39)

D Peak Amplitude 0.35(0.007) 0.58(<0.001) 0.48(<0.001) 0.14(0.3) Seismic pulse D Duration -0.24(0.09) -0.25(0.08) -0.12(0.39) -0.04(0.78) Behaviors Vigor Cheliceral D Rate 0.37(0.003) 0.14(0.28) -0.18(0.15) 0.09(0.47) Strike Double-tap D Rate 0.38(0.003) 0.25(0.05) -0.03(0.8) 0.02(0.91)

Table 3.2: ANOVA table comparing laboratory-reared (LR), field-caught (FC), field- caught-fed (FC-F), and field-caught-starved (FC-S) spiders. Post-hoc Tukey tests are given (under treatment group), same letters indicate statistical similarity while different letters indicate statistical difference.

Treatment Group Dynamic Attributes f-value df p-value LR FC FC-F FC-S

Body Condition 20.7564 3,80 <0.0001 a b a b

Mass 25.3220 3,80 <0.0001 a b a c

Cheliceral strike 16.1060 3,80 <0.0001 a b a b

Double-tap 14.7841 3,79 <0.0001 a b a b

Percussive strike peak 24.2015 3,79 <0.0001 a b b b amplitude

Seismic pulse duration 6.5619 3,75 0.0005 a b a,b b

Static Attributes

Cephalothorax width 4.8872 3,80 0.0036 a b a,b b

Percussive strike peak 3.4071 3,79 0.0216 a a,b a,b b frequency Percussive strike maximum 7.5214 3,79 0.0002 a a,b b b frequency

71 Chapter 4

Conclusion and Future Directions

The research presented here fills a significant gap in our knowledge of the importance of seismic communication for wolf spiders. I conclude that females use male seismic signals in mate choice decisions and male seismic signals convey information about the sender.

Furthermore, it is evident that like visual signals male seismic signals are condition-dependent, but may reflect different conditions experienced at different developmental stages. Hence, the information content of the dynamic seismic signals overlaps with dynamic visual signals (i.e. leg waving behavior – double-tapping) and supports the redundant signaling hypothesis. However, since I did not find this overlap for static seismic signal attributes, differences in information content between dynamic and static signals suggests the possibility of multiple messages.

Future research on S. ocreata communication should focus on questions concerning which aspects of male courtship communication are the most important components during mating. Answering such questions may help us to understand how the constraints of the environment shape the reliance on a particular mode of communication over others. Is there a link between female choice and rearing environment? As suggested by others and supported by results presented here, wolf spiders are naturally food limited in the field. The food limited conditions experienced by male spiders can have serious impacts on their seismic communication, producing males that have lower quality courtship displays. If females in food limited conditions maintain higher standards (as seen in Chapter II) for mates, instead of matching standards to environmental conditions, males maybe under strong sexual selection.

Since the environment wolf spiders experience is continuously changing over the course of their breeding season, does the reliance on a particular signaling mode by females change over

72 the season? If females change their reliance on a particular signaling modality, it may affect the selection pressures experienced by males. For example, early in the breeding season when leaf litter is thick, females may rely more on seismic rather than visual signals. However, as the breeding season continues, the complexity of the leaf litter is reduced due to decomposition. In this situation, of lowered habitat complexity, visual obstructions are reduced possibly favoring visual communication. These changes in the environment may have serious impacts on the selection pressures experienced by males. If this trend is present, do males that develop late in the breeding season invest more energy into producing larger tufts (visual signal) rather than on attributes related to seismic communication? By answering the above questions we will begin to be able to better resolve the question of why multimodal communication has evolved in S. ocreata. By having a greater understanding of how multimodal communication evolved in this system we may be able to extrapolate better how it has evolved in many other systems.