dear damian elias, - cornell universitydear damian elias, here are the electronic proofs of your...

Damian Elias,Dear Here are the electronic proofs of your article.

• You can submit your corrections online or by fax. Together with your proof corrections you must return the Copyright Transfer Statement to complete the proof process.

• Print out the proof. (If you do not already have Acrobat Reader, just download it from http://www.adobe.com.)

• Check the metadata sheet to make sure that the header information, especially author names and the corresponding affiliations, are correctly shown.

• Check that the text is complete and that all figures, tables and their legends are included. Also check the accuracy of special characters, equations, and electronic supplementary material if applicable. If necessary refer to the Edited manuscript.

• The publication of inaccurate data such as dosages and units can have serious consequences. Please take particular care that all such details are correct.

• Please do not make changes that involve only matters of style. We have generally introduced forms that follow the journal’s style. Substantial changes in content, e.g., new results, corrected values, title and authorship are not allowed without the approval of the responsible editor. In such a case, please contact the Editorial Office and return his/her consent together with the proof.

• For online submission please insert your corrections in the online correction form [available in your eProof webpage]. Always indicate the line number to which the correction refers.

• For fax submission, please ensure that your corrections are clearly legible. Use a fine black pen and write the correction in the margin, not too close to the edge of the page.

• The cover sheets (including the Copyright Transfer Statement and the Offprint Order Form) can either be scanned and sent electronically or sent by fax.

• If we do not receive your corrections within 48 hours, we will send you a reminder.

Please note

Your article will be published Online First approximately one week after receipt of your corrected proofs. This is the official first publication citable with the DOI. Further changes are, therefore, not possible.

After online publication, subscribers (personal/institutional) to this journal will have access to the complete article via the DOI using the URL: http://dx.doi.org/[DOI].

If you would like to know when your article has been published online, take advantage of our free alert service. For registration and further information go to: http://www.springerlink.com.

Due to the electronic nature of the procedure, the manuscript and the original figures will only be returned to you on special request. When you return your corrections, please inform us if you would like to have these documents returned.

The printed version will follow in a forthcoming issue.

http://www.springerlink.com

SHEELA. S

SHEELA. S,

Journal of Comparative Physiology A DOI: 10.1007/s00359-006-0116-7Measuring and quantifying dynamic visual signals in jumping spiders

Damian Elias · Bruce Land · Andrew Mason · Ronald Hoy

Springer-Verlag

123 Fax to: +44 870 622 1325 (UK)/ +44 870 762 8807(UK)

From:

6&7, 5th Street, Radhakrishnan Salai, Chennai, Tamil Nadu, India – 600 004 Re:

Authors:

I. Permission to publish Dear I have checked the proofs of my article and

� I have no corrections. The article is ready to be published without changes.

� I have a few corrections. I am enclosing the following pages:

� I have made many corrections. Enclosed is the complete article.

II. Offprint order

� Offprint order enclosed � I do not wish to order offprints Remarks: Date / signature __________________________________________________________________________

III. Copyright Transfer Statement (sign only if not submitted previously)

The copyright to this article is transferred to ___________________________________________________ (for U.S. government employees: to the extent transferable) effective if and when the article is accepted for publication. The author warrants that his/her contribution is original and that he/she has full power to make this grant. The author signs for and accepts responsibility for releasing this material on behalf of any and all co-authors. The copyright transfer covers the exclusive right to reproduce and distribute the article, including reprints, translations, photographic reproductions, microform, electronic form (offline, online) or any other reproductions of similar nature. An author may self-archive an author-created version of his/her article on his/her own website and his/her institution’s repository, including his/her final version; however he/she may not use the publisher’s PDF version which is posted on www.springerlink.com. Furthermore, the author may only post his/her version provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer’s website. The link must be accompanied by the following text: “The original publication is available at www.springerlink.com.” The author is requested to use the appropriate DOI for the article (go to the Linking Options in the article, then to OpenURL and use the link with the DOI). Articles disseminated via www.springerlink.com are indexed, abstracted and referenced by many abstracting and information services, bibliographic networks, subscription agencies, library networks, and consortia. After submission of this agreement signed by the corresponding author, changes of authorship or in the order of the authors listed will not be accepted by Springer. Date / Author’s signature ___________________________________________________________________

Journal: Journal of Comparative Physiology A10.1007/s00359-006-0116-7

Damian EliasDepartment of Life Sciences, Integrative Behaviour and NeuroscienceUniversity of Toronto at Scarborough1265 Military TrailScarborough, Toronto, M1C 1A4, Canada

Damian EliasDepartment of Life Sciences, Integrative Behaviour and NeuroscienceUniversity of Toronto at Scarborough1265 Military TrailScarborough, Toronto, M1C 1A4, Canada

Offprint Order Form

• To determine if your journal provides free offprints, please check the journal's instructions to authors.

• If you do not return this order form, we assume that you do not wish to order offprints.

• If you order offprints after the issue has gone to press, costs are much higher. Therefore, we can supply offprints only in quantities of 300 or more after this time.

• For orders involving more than 500 copies, please ask the production editor for a quotation.

Please enter my order for:

Pages 1-4 1-4 5-8 5-8 9-12 9-12 13-16 13-16 17-20 17-20 21-24 21-24 25-28 25-28 29-32 29-32

Copies EUR USD EUR USD EUR USD EUR USD EUR USD EUR USD EUR USD EUR USD

❏ 50 250,00 275,00 300,00 330,00 370,00 405,00 430,00 475,00 500,00 550,00 525,00 575,00 575,00 630,00 610,00 670,00

❏ 100 300,00 330,00 365,00 405,00 465,00 510,00 525,00 580,00 625,00 685,00 655,00 720,00 715,00 785,00 765,00 840,00

❏ 200 400,00 440,00 525,00 575,00 645,00 710,00 740,00 815,00 860,00 945,00 925,00 1,015,00 1,005,00 1,105,00 1,080,00 1,190,00

❏ 300 500,00 550,00 680,00 750,00 825,00 910,00 955,00 1,050,00 1,095,00 1,205,00 1,190,00 1,310,00 1,295,00 1,425,00 1,390,00 1,530,00

❏ 400 610,00 670,00 855,00 940,00 1,025,00 1,130,00 1,195,00 1,315,00 1,360,00 1,495,00 1,485,00 1,635,00 1,615,00 1,775,00 1,740,00 1,915,00

❏ 500 720,00 790,00 1,025,00 1,130,00 1,225,00 1,350,00 1,430,00 1,575,00 1,625,00 1,785,00 1,780,00 1,960,00 1,930,00 2,125,00 2,090,00 2,300,00

Orders will only be processed if a credit card number has been provided. For German authors, payment by direct debit is also possible.

I wish to be charged in ❏ Euro ❏ USD Prices include surface mail postage and handling. Customers in EU countries who are not registered for VAT should add VAT at the rate applicable in their country. VAT registration number (EU countries only): __________________________________________ Please charge my credit card For authors resident in Germany: payment by ❏ Eurocard/Access/MasterCard direct debit: ❏ American Express I authorize Springer-Verlag to debit the amount ❏ Visa/Barclaycard/Americard owed from my bank account at the due time. Number (incl. check digits): Account no.: ________________________________ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Bank code: _________________________________ Valid until: _ _ / _ _ Bank: _____________________________________ Date/Signature: ________________________________ Date/Signature: _____________________________

Send receipt to: Ship offprints to:

❏ ❏

❏ ___________________________________ ❏ ___________________________________

_____________________________________ _____________________________________ _____________________________________ _____________________________________ _____________________________________ _____________________________________ _____________________________________ _____________________________________ _____________________________________ _____________________________________

Ogvcfcvc"qh"vjg"ctvkeng"vjcv"yknn"dg"xkuwcnk¦gf"kp"QpnkpgHktuv

CtvkengVkvng Ogcuwtkpi"cpf"swcpvkh{kpi"f{pcoke"xkuwcn"ukipcnu"kp"lworkpi"urkfgtu

Lqwtpcn"Pcog Lqwtpcn"qh"Eqorctcvkxg"Rj{ukqnqi{"C

Eqttgurqpfkpi"Cwvjqt Hcokn{"Pcog GnkcuRctvkeng

Ikxgp"Pcog Fcokcp"Q0Uwhhkz

Qticpk¦cvkqp Eqtpgnn"Wpkxgtukv{

Fkxkukqp Fgrctvogpv"qh"Pgwtqdkqnqi{"cpf"Dgjcxkqt

Cfftguu Uggng{"I0"Owff"Jcnn."36:75."Kvjcec."P[."WUC

Qticpk¦cvkqp Wpkxgtukv{"qh"Vqtqpvq"cv"Uectdqtqwij

Fkxkukqp Fgrctvogpv"qh"Nkhg"Uekgpegu."Kpvgitcvkxg"Dgjcxkqwt"cpf"Pgwtquekgpeg

Cfftguu 3487"Oknkvct{"Vtckn."O3E"3C6."Uectdqtqwij."Vqtqpvq."QP."Ecpcfc

Gockn fqg4Beqtpgnn0gfw

Cwvjqt Hcokn{"Pcog NcpfRctvkeng

Ikxgp"Pcog Dtweg"T0Uwhhkz




Gockn dtn6Beqtpgnn0gfw

Cwvjqt Hcokn{"Pcog OcuqpRctvkeng

Ikxgp"Pcog Cpftgy"E0Uwhhkz

Qticpk¦cvkqp Wpkxgtukv{"qh"Vqtqpvq"cv"Uectdqtqwij

Fkxkukqp Fgrctvogpv"qh"Nkhg"Uekgpegu."Kpvgitcvkxg"Dgjcxkqwt"cpf"Pgwtquekgpeg

Cfftguu 3487"Oknkvct{"Vtckn."O3E"3C6."Uectdqtqwij."Vqtqpvq."QP."Ecpcfc

Gockn

Cwvjqt Hcokn{"Pcog Jq{Rctvkeng

Ikxgp"Pcog Tqpcnf"T0Uwhhkz




Gockn

Uejgfwng

Tgegkxgf :"Qevqdgt"4227

Tgxkugf 9"Hgdwtct{"4228

Ceegrvgf 33"Hgdwtct{"4228

Cduvtcev Cpkocnu"gokv"xkuwcn"ukipcnu"vjcv"kpxqnxg"ukownvcpgqwu."ugswgpvkcn"oqxgogpvu"qh"crrgpfcigu"vjcv"wphqnf"ykvjxct{kpi"f{pcokeu"kp"vkog"cpf"urceg0"Cniqtkvjou"jcxg"dggp"tgegpvn{"tgrqtvgf"*g0i0"Rgvgtu"gv"cn0"kp"Cpko"Dgjcx

86<353Î368."4224+"vjcv"gpcdng"swcpvkvcvkxg"ejctcevgtk¦cvkqp"qh"oqxgogpvu"cu"qrvkecn"hnqy"rcvvgtpu0"Hqtfgecfgu."ceqwuvkecn"ukipcnu"jcxg"dggp"tgpfgtgf"d{"vgejpkswgu"vjcv"fgeqorqug"uqwpf"kpvq"cornkvwfg."vkog.cpf"urgevtcn"eqorqpgpvu0"Wukpi"cp"qrvke/hnqy"cniqtkvjo"yg"gzcokpgf"xkuwcn"eqwtvujkr"dgjcxkqwtu"qh"lworkpiurkfgtu"cpf"fgrkev"vjgkt"eqorngz"xkuwcn"ukipcnu"cu"Ðurggf"ycxghqtoÑ."Ðurggf"uwthcegÑ."cpf"ÐurggfycvgthcnnÑ"rnqvu"cpcnqiqwu"vq"ceqwuvke"ycxghqtou."urgevtqitcou."cpf"ycvgthcnn"rnqvu."tgurgevkxgn{0"Kp"cffkvkqp.vjgug"Ðurggf"rtqhknguÑ"ctg"eqorcvkdng"ykvj"cpcn{vkecn"vgejpkswgu"fgxgnqrgf"hqt"cwfkvqt{"cpcn{uku0"Wukpigzcorngu"htqo"vjg"lworkpi"urkfgt"Jcdtqpcvvwu"rwiknnku"yg"ujqy"vjcv"yg"ecp"uvcvkuvkecnn{"fkhhgtgpvkcvg"fkurnc{uqh"fkhhgtgpv"Ðum{"kuncpfÑ"rqrwncvkqpu"uwrrqtvkpi"rtgxkqwu"yqtm"qp"fkxgtukhkecvkqp0"Yg"cnuq"gzcokpgf"xkuwcnfkurnc{u"htqo"vjg"lworkpi"urkfgt"Jcdtqpcvvwu"fquugpwu"cpf"ujqy"vjcv"fkuvkpev"ugkuoke"eqorqpgpvu"qhxkdtcvqt{"fkurnc{u"ctg"rtqfwegf"eqpewttgpvn{"ykvj"uvcvkuvkecnn{"fkuvkpev"oqvkqp"ukipcnu0"Ikxgp"vjcv"f{pcokexkuwcn"ukipcnu"ctg"eqooqp."htqo"kpugevu"vq"dktfu"vq"ocoocnu."yg"rtqrqug"vjcv"qrvkecn/hnqy"cniqtkvjou"cpfvjg"cpcn{ugu"fguetkdgf"jgtg"yknn"dg"wughwn"hqt"ocp{"tgugctejgtu0

Mg{yqtfu Oqvkqp"fkurnc{u"/"Ownvkoqfcn"eqoowpkecvkqp"/"Eqwtvujkr"dgjcxkqwt"/"Oqvkqp"cpcn{uku

Hqqvpqvg"Kphqtocvkqp Fcokcp"Q0"Gnkcu"cpf"Dtweg"T0"Ncpf"eqpvtkdwvgf"gswcnn{

UNCORRECTEDPROOF

123 ORIGINAL PAPER

4 Damian O. Elias Bruce R. Land Andrew C. Mason5 Ronald R. Hoy

6

Measuring and quantifying dynamic visual signals in jumping spiders

78 Received: 8 October 2005 / Revised: 7 February 2006 / Accepted: 11 February 20069 � Springer-Verlag 20061011121314

1516 Abstract Animals emit visual signals that involve17 simultaneous, sequential movements of appendages18 that unfold with varying dynamics in time and space.19 Algorithms have been recently reported (e.g. Peters20 et al. in Anim Behav 64:131–146, 2002) that enable21 quantitative characterization of movements as optical22 flow patterns. For decades, acoustical signals have been23 rendered by techniques that decompose sound into24 amplitude, time, and spectral components. Using an25 optic-flow algorithm we examined visual courtship26 behaviours of jumping spiders and depict their complex27 visual signals as ‘‘speed waveform’’, ‘‘speed surface’’,28 and ‘‘speed waterfall’’ plots analogous to acoustic29 waveforms, spectrograms, and waterfall plots, respec-30 tively. In addition, these ‘‘speed profiles’’ are compati-31 ble with analytical techniques developed for auditory32 analysis. Using examples from the jumping spider33 Habronattus pugillis we show that we can statistically34 differentiate displays of different ‘‘sky island’’ popula-35 tions supporting previous work on diversification. We36 also examined visual displays from the jumping spider37 Habronattus dossenus and show that distinct seismic38 components of vibratory displays are produced39 concurrently with statistically distinct motion signals.40 Given that dynamic visual signals are common,41 from insects to birds to mammals, we propose that

42optical-flow algorithms and the analyses described here43will be useful for many researchers.

44Keywords Motion displays Æ Multimodal45communication Æ Courtship behaviour Æ

46Motion analysis

47Introduction

48Studying animal behaviour often means the analysis of49movements in time and space. While techniques are50readily available for static visual patterns and ornaments51(Endler 1990), this is less so with dynamic sequences of52visual signals (motion displays) (but see Zanker and Zeil532001).54An extensive literature exists on the study of55motion as it pertains to neural processing, navigation56and the extraction of motion information from visual57scenes (reviewed in Barron et al. 1994; Zanker and58Zeil 2001). In neurobiology in particular, techniques59have been motivated by the need to accurately60describe biologically relevant features of motion as an61animal moves through its environment to identify62coding strategies in the processing of visual informa-63tion (Zanker 1996; Zeil and Zanker 1997, 2001; Eckert64and Zeil 2001; Tammero and Dickinson 2002). While65these studies have been integral to an examination of66visual processing, such techniques have limited appli-67cation in studies of behavioural ecology and commu-68nication as they do not provide simple, intuitive69depictions of motion for quantification and compari-70son. Another extensive body of literature on the71analysis of motion exists in the study of biomechan-72ics, particularly in the kinematics of limb motion73(Tammero and Dickinson 2002; Jindrich and Full742002; Nauen and Lauder 2002; Vogel 2003; Fry et al.752003; Alexander 2003; Hedrick et al. 2004). Such76techniques could, in principle, provide extensive77information on motion signals but these computa-78tionally intensive approaches may not efficiently

Damian O. Elias and Bruce R. Land contributed equally

D. O. Elias Æ B. R. Land Æ R. R. HoyDepartment of Neurobiology and Behavior,Cornell University, Seeley G. Mudd Hall,Ithaca, NY, 14853, USAE-mail: [email protected]

A. C. Mason Æ Present address: D. O. Elias (&)Department of Life Sciences,Integrative Behaviour and Neuroscience,University of Toronto at Scarborough,1265 Military Trail, Scarborough, Toronto,ON, CanadaM1C 1A4,E-mail: [email protected].: +1-416-2877465Fax: +1-416-2877642

J Comp Physiol A (2006)DOI 10.1007/s00359-006-0116-7

3 5 9 1 1 6B

Dispatch: 24.2.06 Journal: 359 No. of pages: 13

Journal ID Article ID Author’s disk received 4h Used 4h Corrupted 4 Mismatch 4 Keyed 4

UNCORRECTEDPROOF

79 capture aspects of visual motion signals that are most80 relevant in the context of communication signals. In81 addition, both techniques present the experimenter82 with large data sets and it is often desirable to reduce83 the data in order to glean relevant information.84 In particular, Peters et al.(2002) (Peters and Evans85 2003a, b) have provided a significant advance in the86 analysis of motion signals in communication. Peters87 et al. (2002) described powerful techniques for the88 analysis of visual motion as optical flow patterns in an89 attempt to demonstrate that signals are conspicuous90 against background motion noise (Peters et al. 2002;91 Peters and Evans 2003a, b). Here we build upon these92 pioneering techniques by making use of these previous93 algorithms to develop another depiction of visual sig-94 nals and use these to analyse courtship displays of95 jumping spiders from the genus Habronattus. In addi-96 tion, we show that optical flow approaches are suitable97 for quantification and classification by methods equiv-98 alent to audio analysis (Cortopassi and Bradbury99 2000).

100 Male Habronattus court females by performing101 an elaborate sequence of temporally complex motions102 of multiple colourful body parts and appendages103 (Peckham and Peckham 1889, 1890; Crane 1949;104 Forster 1982b; Jackson 1982; Maddison and McMa-105 hon 2000; Elias et al. 2003). Habronattus has recently106 been used as a model to study species diversification107 (Masta 2000; Maddison and McMahon 2000; Masta108 and Maddison 2002; Maddison and Hedin 2003;109 Hebets and Maddison 2005) and multicomponent110 signalling (Maddison and Stratton 1988; Elias et al.111 2003, 2004, 2005). In these studies there has been an112 implicit assumption that qualitative differences in113 dynamic visual courtship displays can reliably distin-114 guish among species (Richman 1982), populations115 (Maddison 1996; Maddison and McMahon 2000;116 Masta and Maddison 2002; Maddison and Hedin117 2003; Hebets and Maddison 2005), and seismic sig-118 nalling components (Elias et al. 2003, 2004, 2005). It119 has yet to be determined, however, whether such120 qualitative differences can stand up to rigorous sta-121 tistical comparisons (Walker 1974; Higgins and122 Waugaman 2004). To test hypotheses on signal evo-123 lution and function it is crucial to understand the124 signals in question. Thus it is necessary to test whether125 qualitative signal categories are consistently different.126 Our method reduces the dimensionality of visual127 motion signals by integrating over spatial dimensions128 to derive patterns of motion speed as a function of129 time. This method may not be adequate for some130 classes of signal (e.g. which differ solely in position or131 direction of motion components). Our results demon-132 strate, however, that for many signals this technique133 allows objective quantitative comparisons of complex134 visual motion signals. This will potentially provide a135 wide range of useful behavioural measures to a variety136 of disciplines from systematics and behavioural ecology137 to neurobiology and psychology.

138Methods

139Spiders

140Male and female H. pugillis and Habronattus dossenus141were field collected from different mountain ranges142in Arizona (Atascosa—H. dossenus and H. pugillis;143Santa Catalina—H. pugillis; Santa Rita—H. pugillis;144Galiuro—H. pugillis). Animals were housed individually145and kept in the lab on a 12:12 light:dark cycle. Once a146week, spiders were fed fruit flies (Drosophila melanog-147aster) and juvenile crickets (Acheta domesticus).

148Recording procedures

149Recording procedures were similar to a previous study150(Elias et al. 2003). We anaesthetized female jumping spi-151ders with CO2 and tethered them to a wire from the152dorsum of the cephalothorax with lowmelting point wax.153We held females in place with a micromanipulator on a154substrate of stretched nylon fabric (25·30 cm). This al-155lowed us to videotape male courtship from a predictable156position, asmales approach and court females in their line157of sight. Males were dropped 15 cm from the female and158allowed to court freely. Females were awake during159courtship recordings. Recordings commenced when160males approached females. For H. pugillis, we used161standard video taping of courtship behaviour (30 fps,162Navitar Zoom 7000 lens, Panasonic GP-KR222, Sony163DVCAM DSR-20 digital VCR) and then digitized the164footage to computer using Adobe Premiere (San Jose,165CA, USA) with a Cinepak codec. Video files were stored166as *.avi files. For H. dossenus, we used digital high-speed167video (500 fps, RedLake Motionscope PCI 1000, San168Diego, CA,USA) acquired usingMidas software (Xcitex,169Cambridge, MA, USA). We selected suitable video clips170of courtship behaviour based on camera steadiness and171length of behavioural displays (<350 frames). For theH.172pugillis analysis, courtship segments from several indi-173viduals were used (Santa Catalina, N=5; Galiuro, N=8;174Santa Rita, N=6; Atascosa, N=4). The camera was175positioned approximately 30� from a zero azimuth posi-176tion (‘‘head-on’’) (azimuthal range 10�–70�). For the H.177dossenus analysis, different signal components from dif-178ferent individuals (N=5) were analysed. The camera was179positioned approximately 90� from a zero azimuth posi-180tion (azimuthal range 75�–95�). It was difficult to predict181precisely the final courtship position of the animals since182males sometimes did not court the female ‘‘head-on’’, so183we included a wide range of camera angles in the analysis.184All further analysis was conducted using Matlab (The185Mathworks, Natick, MA, USA).

186Motion analysis

187The mathematical methods used for motion analysis188are explained in the next few paragraphs. Full Matlab

3 5 9 1 1 6B



UNCORRECTEDPROOF

189 programs for each analysis step are available at http://190 www.nbb.cornell.edu/neurobio/land/PROJECTS/Moti-191 onDamian/

192 Cropping/intensity normalization

193 Video sequences were shot at either 30 (H. pugillis) or194 500 fps (H. dossenus). High-speed sequences (500 fps)195 were reduced to 250 fps for analysis and the intensity of196 each frame normalized because the high-speed camera197 automatic gain control tended to oscillate slightly.198 Normalization (PN) was achieved by the following199 equation:

PN = PO

PAvg

PFAvg

� �

0.75,

201201 where PN is the normalized pixel intensity, PO is the202 original individual pixel intensity, PAvg is the mean203 pixel value for the whole video sequence, and PFAvg is204 the mean pixel intensity value in the individual frame.205 Frames were cropped so that the animal was com-206 pletely within and spanned nearly the entirety (>75%)207 of the frame.

208 Optical flow calculation

209 The details of this algorithm are published elsewhere210 (Barron et al. 1994; Zeil and Zanker 1997; Peters et al.211 2002). Briefly, we used a simple gradient optical flow212 scheme to estimate motion. If a 2-dimensional (2D)213 video scene includes edges, intensity gradients, or tex-214 tures, motion in the video scene (as an object sweeps215 past a given pixel location) can be represented as216 changing intensity at that pixel. Intensity changes can217 thus be used to summarize motion from video segments.218 Such motion calculations are widely used in robotics and219 machine vision to analyse video sequences (e.g. http://220 www.ifi.unizh.ch/ groups/ailab/projects/sahabot/).221 Our video data were converted into an N·M by T222 matrix where N is the number of pixels in the horizontal223 direction, M is the number of pixels in the vertical224 direction, and T is the number of video frames. The 3D225 matrix was smoothed with a 5·5·5 Gaussian convolu-226 tion kernel with a standard deviation of one pixel227 (Barron et al. 1994). Derivatives in all three directions228 were computed using a second-order (centred, 3-point)229 algorithm. This motion estimate is based on the230 assumption that pixel intensities only change from231 frame-to-frame because of motion of objects passing by232 the pixels. The local speed estimate (vg) was calculated as233 follows:

vg ¼ �ðrIÞ ðdI=dtÞ

jjðrIÞjj2;

235235 where vg is the local object velocity estimate in the

236direction of the spatial intensity gradient, I is an array of237intensities of pixels, t is the frame number, and || is the238magnitude operator (Barron et al. 1994). The local speed239estimate is defined as the magnitude of vg.

240Speed profile plots

241The speed waveform is a simple average of the local242speed estimates (vg) for objects over all pixels in the243frame (Peters and Evans 2003a). We also defined a speed244surface (analogous to a spectrogram). The speed surface245is a 2D plot with frame number on the x-axis, pixel246speed bins on the y-axis, and the colour in each bin247related to the log of the number of pixels moving at that248speed. In other words, at each frame, we plotted a his-249togram of the number of pixels showing movement at a250particular speed range. Both of these plots represented251the complete speed profiles of each video clip. We also252constructed a speed ‘‘waterfall’’ plot which represents253the speed surface as a 3-dimensional (3D) plot, with the254z-axis showing the log of the number of pixels associated255with a speed bin in any given frame.

256Maximum cross-correlation of 1D and 2D signals

257Similarity between speed profiles was computed by258normalized cross-correlation of pairs of sample plots259(with periodic wrapping of samples). Waveforms being260compared were padded with the mean of the sequence,261so that the shorter one became the same length as the262longer one. Both speed waveforms and speed surfaces263were analysed, using a 1-dimensional (1D) correlation264for the speed waveforms and a 2D correlation (with265shifts only along time) for the speed surfaces. For the266next stage of the analysis, we used a measure of dis-267similarity (1.0 minus the maximum correlation) as a268distance measure to construct a matrix of distances269between all pairs of signals.

270Multidimensional scaling (MDS)

271The distance (dissimilarity) matrix was used as input for272a MDS analysis (Cox and Cox 2001). MDS provides an273unbiased, low dimensional representation of the struc-274ture within a distance matrix. A good fit will preserve the275rank order of distances between data points and give a276low value of stress, a measure of the distance distortion.277MDS analysis normally starts with a 1D fit and increases278the dimensionality until the stress levels plateau at a low279value. A Matlab subroutine was used (Steyvers, M.,280http://www.psiexp.ss.uci. edu/research/ software.htm) to281perform the MDS. We used an information theoretic282analysis on the entropy of clustering (Victor and Pur-283pura 1997) on both the 1D and 2D correlations and284determined that more information was contained in the2851D correlation (data not shown); hence, all further

3 5 9 1 1 6B



UNCORRECTEDPROOF

286 analyses were performed on the 1D correlation matrices.287 We fitted our data from one to five dimensions. Most of288 the stress reduction (S1) occurred at either two or three289 dimensions (H. pugillis: 1st dimension, S1=0.38,290 R=0.68; 2nd dimension, S1=0.23, R=0.78; 3rd291 dimension, S1=0.15, R=0.84; 4th dimension, S1=0.12,292 R=0.88; 5th dimension, S1=0.09, R=0.89; H. dossenus:293 1st dimension, S1=0.32, R=0.69; 2nd dimension,294 S1=0.19, R=0.81; 3rd dimension, S1=0.12, R=0.88;295 4th dimension, S1=0.09, R=0.92; 5th dimension,296 S1=0.07, R=0.93), hence all further analysis was per-297 formed on 3D fits. Plots of the various signals in MDS-298 space showed strong clustering. The axes on the MDS299 analysis reflect the structure of the data; we therefore300 performed a one-way ANOVA along different dimen-301 sions to calculate statistical significance of the clustering.302 A Tukey post hoc test was then applied to compare303 different populations (H. pugillis) and signal compo-304 nents (H. dossenus). All statistical analyses were305 performed using Matlab.

306 Results

307 Motion algorithm calibration

308 In order to test the performance of the motion algorithm309 against a predictable and controllable set of motion310 signals, we simulated rotation of a rectangular bar311 against a uniform background using Matlab. Texture312 was added to the bar in the form of four nonparallel313 stripes (Fig. 1a). We programmed the bar to pivot314 around one end using sinusoidal motion. We then sys-315 tematically varied the width (w) and length (l) of the bar,316 as well as the frequency (F) and amplitude (A) of the317 motion (Fig. 1a).318 The three examples in Fig. 1 show the effect of a step-319 change in frequency (F), amplitude (A) and bar length320 (l), respectively (Fig. 1b–d). In each case, the analysis321 depicts the temporal structure of the simulated move-322 ment very well (Fig. 1). As frequency, amplitude, or bar323 length increase, the computed average speed increases324 predictably (see below). The speed waveform and sur-325 face plots show that more pixels ‘‘move’’ at higher326 speeds after the step increase. This detailed shape is327 depicted particularly well in the surface plot.328 The amplitude of the average motion of the simulated329 bar is related to the amplitude of the input wave and its330 frequency by a square law. [Fig. 1b(iv), c(iv)]. Detailed331 examination of the image sequence suggests that at332 higher speeds, the motion in a video clip ‘‘skips pixels’’333 between frames, hence this square law is a result of the334 product of the speed measured at each pixel (which335 is linear) multiplied by the greater number of pixels336 averaged into the motion at higher speeds. The337 amplitude of the average motion of the simulated bar is338 related to bar length by a cube law. [Fig. 1d(iv)]. This339 results from the aforementioned square law increased by340 another linear factor, i.e. the number of pixels covered

341by the edge of the bar. Both bar width and texture342change average motion amplitude only weakly (data not343shown). This small change in average motion can be344attributed to the increase in the total length of edge345contours.346We also modelled amplitude (AM) and frequency347(FM) modulated movement by animating a rotating bar348at a fixed carrier frequency and either AM or FM349modulating the carrier. The carrier and modulating350frequencies are clearly discernable for both AM351(Fig. 2a) and FM (Fig. 2b) movements in all the speed352profiles. AM and FM movements are also easily dis-353tinguishable from one another. Both simple (sinusoidal)354and complex (modulated) motions are thus faithfully355and accurately reflected in the analysis.

356Habronattus pugillis populations

357Habronattus pugillis is found in the Sonoran desert and358local populations on different mountain ranges (‘‘sky359islands’’) have different ornaments, morphologies,360and courtship displays (Masta 2000; Maddison and361McMahon 2000; Masta and Maddison 2002). Courtship362displays from four different populations of H. pugillis363are plotted in Fig. 3. Several repeating patterns are364apparent, especially in Galiuro (Fig. 3a), Atascosa365(Fig. 3c), and Santa Catalina (Fig. 3d) populations. By366comparing videos of courtship behaviour with their367corresponding speed profiles, we verified that features in368the speed profiles corresponded to qualitatively identi-369fiable components of the motion display. For example,370in the Atascosa speed profiles (Fig. 3c), high amplitude371‘‘pulses’’ (e.g. frames 130–150) correspond to single372leg flicks and lower amplitude ‘‘pulses’’ (e.g. frames373150–200) correspond to pedipalp and abdominal move-374ments. Speed profiles also reveal more subtle features of375motion displays. For example, animals from the Santa376Rita mountains make circular movements with their377pedipalps during courtship (Maddison and McMahon3782000). It is evident from the speed surface (Fig. 3b,379frames 0–200) that this behaviour does not occur in a380smooth motion, but rather as a sequence of brief381punctuated, jerky movements (Fig. 3b).382Different populations of H. pugillis vary in behavio-383ural and morphological characters (Maddison and384McMahon 2000). We evaluated two populations that385include unique movement display characters [Gali-386uro—First leg wavy circle, Santa Rita—Palp motion387(circling)] and two that have similar courtship display388characters [Atascosa and Santa Catalina—Late-display389leg flick (single)] (Maddison and McMahon 2000). Santa390Catalina spiders include the rare Body shake motion391character, but this was not analysed (Maddison and392McMahon 2000). Using MDS, all four groups can be393discriminated from each other by the speed profiles394(Fig. 4). Clustering was strong for all population classes.395In order to evaluate the significance of each cluster396we performed a one-way ANOVA on dimension 1

3 5 9 1 1 6B



UNCORRECTEDPROOF

l

w

A

t = 1/FA)

0 20 40 60 80 1000

0.1

0.2

0.3

0.4

0.5

Frame Number20 40 60 80

10

20

30

40

50

01000

0

Speed B

i n

1.5

log

(# p

ixel

s)

-

lo

g (

mea

n #

pix

els)

0.5

-0.5

-1.5

20

40

0

100

20 40 60

Frame Number80

A=40

A=80

Frame Number

0 10 20 30

0.0

0.04

0.12

0.08

0.1

Bar Length

0 40 60 1200

0.02

0.04

0.06

0.08

0.1

Input Amplitude

0 20 40 60 80 1000

0.1

0.2

0.3

0.4

0.5

Aver

age

Spee

d

t = 1/3t = 1/5

20 40 60 80 100

10

20

30

40

50

00

0S

peed

Bin

1.5

log(#

pix

els)

-

log (

mea

n #

pix

els)

0.5

-0.5

-1.5

20

40

0

100

20 40 60

FrameNumber80

Spee

d B

in

Frame Number

B) C)

l =15

l =20

D)

ii)

iii)

iv)

v)

i)

0 1 2 3 4 5 60

0.1

0.2

0.3

0.4

Pea

k A

mpli

tude

Frequency

0 20 40 60 80 1000

0.1

0.2

0.3

0.4

0.5

10

20

30

40

50

020 40 60 80 1000

0

0

Speed

Bin

1.5

log(#

pix

els)

-

log (

mea

n #

pix

els)

0.5

-0.5

-1.5

20

40

0

100

20 40 60

Frame Number80

-1

-0.6

-0.2

0.2

0.6

1

-1

-0.6

-0.20.2

0.6

1

Fig. 1 Simulated movements. a A bar of fixed length (l), width (w)and starting angle (h) was simulated and rotated sinusoidally at afixed peak-to-peak amplitude of A degrees and frequency (F). Thefrequency (b), amplitude (c), and bar length (d) were thensystematically changed. The time-course of the corresponding

stimulus parameters are shown in panel i. Panels ii–iv show theresulting analysis of the simulated movement and step changes: 2D‘‘speed waveform’’ plots (ii), 3D ‘‘speed surface’’ plots (iii) and 3D‘‘speed waterfall plots’’ (iv); and summary of simulated motionamplitude as different parameters are changed (v)

3 5 9 1 1 6B



UNCORRECTEDPROOF

397 (F3,19=25.61, P=6.9·10�7) and saw that the Santa398 Catalina population was significantly different from the399 Galiuro (P<0.001) and Santa Rita (P<0.05) popula-

400tions but not from the Atascosa population (P>0.05)401[Fig. 4a(ii)]. Also, the Galiuro population was signifi-402cantly different from the Santa Rita (P<0.01) and

A) AM movement B) FM movement

0

20

40

Speed

Bin

1.5

log(#

pix

els)

-

log (

mea

n #

pix

els)

0.5

-0.5

-1.5

0 100 200 300 400

Av

erag

e S

pee

d

Frame Number

i.

ii.

iii.

iv.

0 100 200 300 4000

0.2

0.4

0.6

0.8

1

100 200 300 400

20

40

00

100 200 300 400

20

40

0

Sp

eed

Bin

0

0Speed

Bin

1.5

log(#

pix

els)

-

log (

mea

n #

pix

els)

0.5

-0.5

-1.5

20

40

0 100 200 300 400

Frame NumberFrame Number

Frame Number

-1

-0.6

-0.2

0.2

0.6

1

-1

-0.6

-0.20.2

0.6

1

0 100 200 300 4000

0.2

0.4

0.6

0.8

1

Fig. 2 Simulated AM and FM movements. The sinusoidal move-ment of a bar was simulated with either amplitude (AM) (a) orfrequency (FM) (b) modulation. (i) Modulated sine wave used to

move the simulated bar. (ii) 2D ‘‘speed waveform’’ plots. (iii) 3D‘‘speed surface’’ plots. (iv) 3D ‘‘speed waterfall plots’’ (third row).AM and FM motion is easily distinguishable in the analysis

3 5 9 1 1 6B



UNCORRECTEDPROOF

0

log(#pixels)-

log(mean#pixels)

20

40

0100

200

300

Speed Bin

1.5

-1.50

400

0100

200

300

400

0

0.0

4

0.0

8

0.1

2

AH

.pugil

lis

(Gal

iuro

s)B

H.

pugil

lis

(San

ta

Rit

as)

0100

200

300

400

0

0.1

0.2

0.3

AverageSpeed

0100

200

300

400

10

20

30

40

50

CH

.pugil

lis

(Ata

scosa

s)

0100

200

300

400

Fra

me

Num

ber

10

20

30

40

50

DH

.pugil

lis

(San

ta

Cat

alin

as)

SpeedBin

Fra

me

Num

ber

i. ii.

iii.

0 1.0

-1.0

00

20

40

0100

200

300

400

Fra

me

Num

ber

Fra

me

Num

ber

Fra

me

Num

ber

Fra

me

Num

ber

SpeedBin

0

20

40

0100

200

300

400

Speed Bin0

1.0

-1.0

0100

200

300

400

0

0.1

0.2

0.3

01

00

20

03

00

0

0.0

4

0.0

8

0.1

2

100

200

300

10

20

30

40

50

0

20

40

0100

200

300

Speed Bin

1.0

-1.0

0

00

100

200

300

400

10

30

50 0

0

20

40

iv.

-1-0.6

-0.2

0.2

0.6

1 -1-0.6

-0.2

0.2

0.6

1

Fra

me

Num

ber

Fra

me

Num

ber

Fig. 3 Different populations of Habronattus pugillis. Representa-tive from the Galiuro (a), Santa Rita (b), Atascosa (c), and SantaCatalina (d) mountain ranges are shown. Top panel (i) shows anexample of a single video frame at the resolution used in the

analysis. Second panel (ii) shows the 2D ‘‘speed waveform’’ plots.Third panels (iii) show the 3D ‘‘speed surface’’ plots (second row).Fourth panel (iv) shows the 3D ‘‘speed waterfall plots’’ (third row).Frame rate is 30 fps

3 5 9 1 1 6B



UNCORRECTEDPROOF

403 Atascosa (P<0.001) populations and the Santa Rita404 population was significantly different from the Atascosa405 population (P<0.05) [Fig. 4a(ii)]. Performing the same406 analysis on dimension 3 (F3,19=3.68, P=0.0303), the407 Santa Catalina and Atascosa populations are signifi-408 cantly different (P<0.05) [Fig. 4b(ii)]. No other differ-409 ences can be observed along dimension 3 (P>0.05,410 Fig. 4b). Hence, all population classes are statistically411 distinguishable in at least one of the dimensions used in412 the analysis.

413 Habronattus dossenus signals

414 Courtship displays from five different individuals were415 selected and the visual component of different seismic416 signals recorded (scrape, N=5; thump, N=10; buzz,417 N=5) for each individual spider (Elias et al. 2003). Two418 classes of thumps, distinguishable by their seismic419 component, were selected for each individual but were420 not distinguishable based on their speed profiles, hence421 they were combined into one class (Elias et al. 2003).422 First we plotted the speed profiles for each of the signal423 classes (Fig. 5). Speed surfaces capture relatively subtle424 details of movements. For example, during individual425 scrape signals the forelegs first come down followed426 by abdominal movement upward (Elias et al. 2003)427 (Fig. 5a). Individual scrapes produce a rocking motion428 that can be observed clearly in the speed surface as429 a characteristic double peak (e.g. 1–3 in Fig. 5a).

430Furthermore, individual abdominal oscillations are re-431solved in the buzz speed surface (Fig. 5c).432To test whether this technique could distinguish433among the three qualitative signal classes, we applied the434same analysis described above. Clustering was strong for435all signal classes. A one-way ANOVA on dimension 1436(F2,18=20.66, P=2.2·10�5) showed that scrapes are437significantly different from thumps (P<0.001) and438buzzes (P<0.05) (Fig. 6) and thumps are significantly439different from buzzes (P<0.05) (Fig. 6). Hence, all sig-440nal classes are statistically distinguishable in the full441analysis.

442Spatial information

443Our analysis technique extracts temporal patterns and444integrates over spatial dimensions. For comparison, we445derived summaries of spatial patterns of image motion,446integrated over time, for representative signals (Fig. 7).447Although we did not carry out quantitative analyses on448spatial data, some general features are apparent. There449are clear differences between the signal examples in the450location of motion within the video frame (Fig. 7 left451panels), although some features are obscured in dis-452plays where leg flick motions are superimposed on the453movement of the entire animals (Fig. 7b). Summaries454of the motion orientation are more difficult to interpret455(Fig. 7 centre panels). This is most likely due to the456fact that many of the movements are cyclical (rotation

-2 -1 0 1 2

-2

-1

0

1

Dimension 1

Dim

ensi

on

2

1

23 4

5

6

7 8

9

10

11

12

13

14

1516

17

1819

20

21

22

23

A

-1 0 1 2

-0.5

0.5

1.5

1

2

34

5

6

78

9

10

11

12

13

14

15

16

17

18

19

20

21

2223

Dimension 1

Dim

ensi

on

3

Santa CatalinasGaliurosSanta RitasAtascosas

-1

0

1

2

Dim

ensi

on

1

ns

p <0.05*

p <0.001***

p <0.01**

p <0.05*

p <0.001***

Santa

Catalinas

Galiuros Santa

Ritas

Atascosas

Santa CatalinasGaliurosSanta RitasAtascosas

B

-1.5

-0.5

0.5

1.5

Dim

ensi

on

3

p <0.05*

Santa

Catalinas

Galiuros Santa

Ritas

Atascosas-2-1.5

i.ii.

-2

Fig. 4 Multidimensionalscaling analysis of differentHabronattus pugillispopulations. Similarity of thecourtship displays from theGailuro, Santa Rita, Atascosa,and Santa Catalina mountainranges were computed bycircular cross-correlation andthen input into a MDSprocedure. Dimension 1 versusdimension 2 [a(i)] anddimension 1 versus dimension 3[b(i)] are plotted. MDS showedstrong clustering by populationlocation. A one-way ANOVAwith Tukey’s post hoc andBonferonni corrections on thedifferent populations alongdimension 1 [a(ii)] or dimension3 [b(ii)] indicated thatpopulation clusters detected byMDS were significantlydifferent from one another(P<0.05)

3 5 9 1 1 6B



UNCORRECTEDPROOF

457 or back-and-forth movement). Speed isoform plots are458 3D plots that (Fig. 7 right panels) show location of459 motion within the video frame with time on the z-axis.460 These plots parallel the speed surface plots used in the461 analyses above, but whereas speed surface plots depict462 the distribution of motion speed (amplitude) as a463 function of time, speed isoform plots depict the loca-464 tion of image motion as a function of time. These465 could, in principle, be analysed similarly to the speed466 profiles above.

467Discussion

468Jumping spiders communicate using a complex reper-469toire of visual ornaments and dynamic visual (motion)470signals (Jackson 1982; Forster 1982b). Here we use471optical flow techniques for the depiction and quantifi-472cation of motion signals and use the technique as the473basis of a statistical analysis to assess motion signals in474jumping spiders.

40 80 120 160

10

20

30

40

50

0040 80 120 160

10

20

30

40

50

00

A Scrape

1

22

3

1

23 4

B Thump C Buzz

0 100 200 300 400

Frame Number

10

20

30

40

50

Av

erag

e S

pee

di.

ii.

iii.

iv.

0 100 200 300 4000

0.05

0.1

0.15

0.2

Frame Number

0 40 80 120 1600

0.5

1

1.5

0 40 80 120 1600

0.1

0.2

0.3

Sp

eed

Bin

Frame Number

v.

-1

-0.6

-0.2

0.2

0.6

1

0

20

40

0 40 80 120 160

log

(#p

ixel

s)-

log

(m

ean

#p

ixel

s)

Speed

Bin

1.0

0

-1.0

0

20

40

0 100 200 300 400

Frame NumberFrame NumberFrame Number

log(#

pix

els)

-

log(

mea

n#

pix

els)

Speed

Bin

1.0

0

-1.0

-1

-0.6

-0.2

0.2

0.6

1

log

(#p

ixel

s)-

log

(m

ean

#p

ixel

s)

0

20

40

0 40 80 120 160

Speed

Bin

1.0

0

-1.0

1

2 12

0

Fig. 5 Different signals of Habronattus dossenus. Representativeexamples of scrape (a), thump (b), and buzz (c) signals are shown.Top panel (i) shows an example of a single video frame at theresolution used in the analysis. Second panel (ii) illustrates bodypositions with numbers (1–4) illustrating movements of the forelegsand abdomen. Third panel (iii) shows the 2D ‘‘speed waveform’’

plots. Fourth panel (iv) shows the 3D ‘‘speed surface’’ plots. Fifthpanel (v) shows the 3D ‘‘speed waterfall plots’’. Panels iii–v areshown in the same time scale, with numbers (1–4) corresponding tothe body movements illustrated in panel ii. Frame rate is 250 fps(reduced from 500 fps)

3 5 9 1 1 6B



UNCORRECTEDPROOF

475 Quantitative analysis of courtship signals

476 ‘‘Sky island’’ populations

477 We examined variation in the courtship displays of478 different ‘‘sky island’’ populations of H. pugillis479 (Maddison and McMahon 2000; Masta 2000; Masta480 and Maddison 2002). By calculating the differences481 between the speed profiles of displays from different482 populations, we were able to show that courtship dis-483 plays were different between all of the populations484 studied. We could easily distinguish between popula-485 tions with unique display elements [Galiuro—First leg486 wavy circle, Santa Rita—Palp motion (circling)]487 (Maddison and McMahon 2000). Importantly, we could488 also discriminate the Santa Catalina and Atascosa489 populations that had qualitatively similar late stage490 visual displays (Late-display leg flick) (Fig. 4b). Madd-491 ison and McMahon (2000) in their initial descriptions492 and analysis of courtship coded this display as being493 the same between spiders from the Santa Catalina494 and Atascosa Mountains. Masta and Maddison (2002)495 demonstrated that fixation rates between neutral496 (mitochondrial genes) and male phenotypic traits497 (morphological and behavioural characters) were dif-498 ferent and used this as evidence to suggest that sexual499 selection was driving diversification in H. pugillis. This500 study not only supports those previous studies, but also501 suggests that male courtship phenotypes are fixed to an502 even greater extent than previously demonstrated.

503 Multicomponent signals

504 Elias et al. (2003) showed that males in the jumping505 spider H. dossenus produced at least three different506 seismic signals all coordinated with motion signals.507 Given the strict coordination of seismic and motion508 signals, the authors suggested that the signal compo-509 nents in different modalities are functionally linked

510(Elias et al. 2003). If emergent properties of the multi-511modal signal (seismic and visual) are important, one512would predict that unique seismic components would513have unique motion components. Similar predictions514can be made if unique motion displays serve to focus515attention on corresponding seismic components (Hasson5161991). Distinct motion signals would function to prevent517habituation and ensure attention to seismic components518(Hall and Channel 1985; Dill and Heisenberg 1995; Post519and von der Emde 1999; Busse et al. 2005). We measured520different motion signals and found that distinct seismic521signals occurred with specific motion signals suggesting522inter-signal interactions either to focus attention or to523construct integrative signals (Partan and Marler 1999,5242005; Hebets and Papaj 2005). While this is not a con-525clusive test on whether there exist inter-signal interac-526tions, it is suggestive that selection has worked on the527integrated multicomponent, multimodal signal.

528Overall implications and limitations

529In general, there are many potential applications of this530technique for measuring motion signals. Any aspect of531the repeated motion patterns can be measured (i.e.532intervals between patterns, duration of patterns, maxi-533mum and minimum motion of patterns, etc.) for use in534subsequent analysis, and multiple aspects of the speed535profiles can be treated simultaneously in multivariate536analyses.537Rigorous classification techniques are desirable in538many disciplines particularly in studies of animal com-539munication. For example, at the level of entire courtship540displays, this could be used to identify motion parame-541ters as characters for phylogenetic analyses. At the level542of individual signals, this is potentially useful in evalu-543ating natural variation in signals (Ryan and Rand 2003)544and as a way to measure signal complexity (e.g. how545many categories of visual signals can be objectively

2.5 1.5 0.5 -0.5 -1.5

-1.5

-0.1

0.1

1.5

1

2

3

4

5

6

7

89

10

11

1213

14

15

16

1718

19

20

21

Dimension 1

Dim

ensi

on

2

-1.5

-0.5

0.5

1.5

p <0.001***

p <0.05*

p <0.05*

ScrapeThumpBuzz

Scrape BuzzThump

Dim

ensi

on

1

a b

Fig. 6 Multidimensional scaling analysis of Habronattus dossenussignals. Similarity of scrape, thump, and buzz courtship signalswere computed by circular cross-correlation and then input into anMDS procedure. Plot of dimension 1 versus dimension 2 (a). MDS

showed strong clustering by signal category. A one-way ANOVAwith Tukey’s post hoc and Bonferonni corrections on the differentsignals along dimension 1 (b) indicated that signal classes detectedby MDS were significantly different from one another (P<0.05)

3 5 9 1 1 6B



UNCORRECTEDPROOF

546 discriminated). These techniques could also be valuable547 in comparative studies. For example, closely related548 species that signal in different visual environments could549 be compared to investigate the effect of the visual envi-550 ronment on the design of motion displays (Endler 1991,551 1992; Peters et al. 2002; Peters and Evans 2003a, b).552 This method of constructing speed profiles severely553 reduces the information present in the original video-554 tapes. Optic flow analyses reduce video data to essentially555 five dimensions (speed, speed spatial distribution, ori-556 entation, orientation spatial distribution, and time) (Zeil557 and Zanker 1997; Zanker and Zeil 2001; Peters et al.558 2002). Some of these other motion parameters have been559 used in other systems (Peters et al. 2002; Peters and560 Evans 2003a, b). We chose to concentrate on speed and561 time since jumping spider displays can be very complex562 and often include the movement of various body parts563 (forelegs, third leg patella, pedipalps) superimposed564 upon the movement of the entire spider (Fig. 7). Any of565 these dimensions however can be plotted for jumping566 spider displays (Fig. 7). Speed and time parameters may567 be especially important in jumping spiders due to the568 structure of their visual system. Jumping spiders have569 two categories of eyes: primary eyes which have a small570 field of view and are specialized for fine spatial resolu-571 tion, and secondary eyes which have a large field of view572 are specialized for motion detection (Land 1969, 1985;

573Land and Nilsson 2002). Motion signals would most574likely stimulate secondary eyes and therefore timing and575not spatial location is likely to be important since sec-576ondary eyes integrate over a wide field of view (Forster5771982a, b; Land 1985). This hypothesis remains to be578tested and it is possible that jumping spiders are reducing579the visual field into timing information (input from the580secondary eyes) and spatial information (input from the581primary eyes) independently. In this scenario, our use of582speed and timing parameters would match ‘‘data reduc-583tions’’ performed by secondary eyes. This underlines the584importance of picking the correct data reduction strategy585based on insights from sensory physiology and behav-586iour. Combining both spatial and temporal analyses with587an analysis on the primary and secondary eye fields of588view could give insights into how information is chan-589nelled into the nervous system (Strausfeld et al. 1993).590Complex motion signals in different communication591systems may be specialized for different dimensions. By592combining our technique with alternative analyses that593focus on spatial rather than temporal motion patterns, it594may be possible to develop a battery of analytical ap-595proaches to identify and analyse the salient parameters of596a wide range of complex visual motion displays.597Regardless of this data reduction, distinct signal598categories can still be discriminated using average speed599and time parameters. The sheer complexity of motion

20

40

60

80

100

0

40

60

80

100

120

0

0 20 40 60 80 100 120

20

40

60

80

100

Speed integrated over timey

pix

elnum

ber 0

40

60

80

100yp

ixel

nu

mb

er

0

45 90 135 180

x pixel number x pixel number

Orientation integrated over time

1

0

1

2

A)

B)

1

0

1

2

Speed isoform surfaces

x pixel number

ypixel num

ber

Fra

me

num

ber

400

0

100

200

300

0

120 80 0

x pixel number

y pixelnum

ber

Fra

me

num

ber

400

0

100

200

300

0

100100 75 25 0

0 45 90 135 1800160 40

120

Fig. 7 Further optic flow parameters for courtship displays ofdifferent populations of Habronattus pugillis. Representativeexamples (same examples as in Fig. 3) from the Galiuro (a) andSanta Catalina (b) mountain ranges are shown. First column showsspeed spatial distributions throughout the video integrated throughtime. Second column shows speed orientation spatial distribution

throughout the video integrated through time. Third column showspeed distribution isoform surfaces throughout the video with timeon the z-axis and speed distribution on the x- and y-axis. Timingpatterns are difficult to observe in speed and orientation distribu-tion space. Frame rate is 30 fps

3 5 9 1 1 6B



UNCORRECTEDPROOF

600 displays makes data reduction attractive and we feel that601 this method reduces the data while constructing accu-602 rate, intuitive depictions of the structure and timing of603 motion displays in a way that may be biologically604 meaningful to the organism in question. While other605 parameters are no doubt important, we feel that using606 speed parameters and time allows one to easily observe607 repeating patterns in a way that is difficult in other608 parameter spaces (Fig. 7).609 One potential limitation in this technique is the con-610 founding effect of the number of edges on total motion.611 For example, if a moving appendage differed between612 two species solely in the number of stripe patterns, then613 for equivalent leg movements, our analysis would record614 a higher motion signal for the animal with more stripes.615 Such a difference in recorded motion signals due to616 ornaments could, however, reflect true differences in the617 perceived signal at the receiver. Neural processing of618 motion in animal brains is based on the movement of619 edges defined by luminance contrast and not edges de-620 fined by chromatic contrast. Edges defined by chromatic621 contrast are usually not perceived by animals; thus, total622 edge motion is total motion (Borst and Egelhaaf 1993).623 Therefore our algorithm analyses motion in a biological624 way.625 By expanding the technique developed by Peters et al.626 (2002), we have developed a novel way to visualize627 motion data analogous to spectrogram representations628 of auditory data as well as demonstrating statistical629 techniques for analysing motion data. This study dem-630 onstrates the utility of using optic flow techniques to631 reduce and analyse motion in a variety of contexts (Zeil632 and Zanker 1997; Zanker and Zeil 2001; Peters et al.633 2002; Peters and Evans 2003a,b).

634 Acknowledgements We would like to thank M.C.B. Andrade, C.635 Botero, C. Gilbert, J. Bradbury, B. Brennen, M.E. Arnegard, E.A.636 Hebets, W.P. Maddison, M. Lowder, Cornell’s Neuroethology637 Journal Club, an anonymous reviewer, and members of the Hoy638 lab for helpful comments, suggestions, and assistance. Spider639 illustrations were generously provided by Margy Nelson. Funding640 was provided by NIH and HHMI to RRH (N1DCR01 DC00103),641 NSERC to ACM (238882 241419), NIH to BRL, and a HHMI Pre-642 Doctoral Fellowship to DOE. These experiments complied with643 ‘‘Principles of animal care’’, publication no. 86–23, revised 1985 of644 the National Institute of Health, and also with the current laws of645 the country (USA and Canada) in which the experiments were646 performed.

647 References

648 Alexander RM (2003) Principles of animal locomotion. Princeton649 University Press, Princeton650 Barron JL, Fleet DJ, Beauchemin SS (1994) Performance of optic651 flow techniques. IJCV 12:43–77652 Borst A, Egelhaaf M (1993) Detecting visual motion: theory and653 models. Rev Oculomot Res 5:3–27654 Busse L, Roberts KC, Crist RE, Weissman DH, Woldorff MG655 (2005) The spread of attention across modalities and space in a656 multisensory object. PNAS 102:18751–18756657 Cortopassi KA, Bradbury JW (2000) The comparison of har-658 monically rich sounds using spectrographic cross-correlation659 and principal coordinates analysis. Bioacoustics 11:89–127

660Cox TF, Cox MAA (2001) Multidimensional scaling. Chapman661and Hall, Boca Raton662Crane J (1949) Comparative biology of salticid spiders at Rancho663Grande, Venezuela. Part IV an analysis of display. Zoologica66434:159–214665Dill M, Heisenberg M (1995) Visual pattern memory without shape666recognition. Philos Trans R Soc Lond Ser B Biol Sci 349:143–152667Eckert MP, Zeil J (2001) Towards an ecology of motion vision. In:668Zanker JM, Zeil J (eds) Motion vision: computational, neural,669and ecological constraints. Springer, Berlin Heidelberg New670York, pp 333–369671Elias DO, Mason AC, Maddison WP, Hoy RR (2003) Seismic672signals in a courting male jumping spider (Araneae: Salticidae).673J Exp Biol 206:4029–4039674Elias DO, Mason AC, Hoy RR (2004) The effect of substrate on675the efficacy of seismic courtship signal transmission in the676jumping spider Habronattus dossenus (Araneae: Salticidae).677J Exp Biol 207:4105–4110678Elias DO, Hebets EA, Hoy RR, Mason AC (2005) Seismic signals679are crucial for male mating success in a visual specialist jumping680spider (Araneae:Salticidae). Anim Behav 69:931–938681Endler JA (1990) On the measurement and classification of color in682studies of animal color patterns. Biol J Linnean Soc 41:315–352683Endler JA (1991) Variation in the appearance of guppy color684patterns to guppies and their predators under different visual685conditions. Vision Res 31:587–608686Endler JA (1992) Signals, signal conditions, and the direction of687evolution. Am Nat 139:S125–S153688Forster L (1982a) Vision and prey-catching strategies in jumping689spiders. Am Sci 70:165–175690Forster L (1982b) Visual communication in jumping spiders (Sal-691ticidae). In: Witt PN Rovner JS (eds) Spider communication:692mechanisms and ecological significance. Princeton University693Press, Princeton, pp 161–212694Fry SN, Sayaman R, Dickinson MH (2003) The aerodynamics of695free-flight maneuvers in Drosophila. Science 300:495–498696Hall G, Channel S (1985) Differential effects of contextual change697on latent inhibition and on the habituation of an orientating698response. J Exp Psychol Anim Behav Process 11:470–481699Hasson O (1991) Sexual displays as amplifiers: practical examples700with an emphasis on feather decorations. Behav Ecol 2:189–197701Hebets EA, Maddison WP (2005) Xenophilic mating preferences702among populations of the jumping spider Habronattus pugillis703Griswold. Behav Ecol 16:981–988704Hebets EA, Papaj DR (2005) Complex signal function: developing705a framework of testable hypotheses. Behav Ecol Sociobiol70657:197–214707Hedrick TL, Usherwood JR, Biewener AA (2004) Wing inertia and708whole-body acceleration: an analysis of instantaneous aerody-709namic force production in cockatiels (Nymphicus hollandicus)710flying across a range of speeds. J Exp Biol 207:1689–1702711Higgins LA, Waugaman RD (2004) Sexual selection and variation:712a multivariate approach to species-specific calls and preferences.713Anim Behav 68:1139–1153714Jackson RR (1982) The behavior of communicating in jumping715spiders (Salticidae). In: Witt PN, Rovner JS (eds) Spider com-716munication: mechanisms and ecological significance. Princeton717University Press, Princeton, pp 213–247718Jindrich DL, Full RJ (2002) Dynamic stabilization of rapid hexa-719pedal locomotion. J Exp Biol 205:2803–2823720Land MF (1969) Structure of retinae of principal eyes of jumping721spiders (Salticidae : Dendryphantinae) in relation to visual722optics. J Exp Biol 51:443–470723Land MF (1985) The morphology and optics of spider eyes. In:724Barth FG (ed) Neurobiology of arachnids. Springer, Berlin725Heidelberg New York, pp 53–78726Land MF, Nilsson DE (2002) Animal eyes. Oxford University727Press, Oxford728Maddison WP (1996) Pelegrina franganillo and other jumping729spiders formerly placed in the genus Metaphidippus (Ara-730neae: Salticidae). Bull Mus Comp Zool Harvard Univ731154:215–368

3 5 9 1 1 6B



UNCORRECTEDPROOF

732 Maddison W, Hedin M (2003) Phylogeny of Habronattus jumping733 spiders (Araneae : Salticidae), with consideration of genital and734 courtship evolution. Syst Entomol 28:1–21735 Maddison W, McMahon M (2000) Divergence and reticulation736 among montane populations of a jumping spider (Habronattus737 pugillis Griswold). Syst Biol 49:400–421738 Maddison WP, Stratton GE (1988) Sound production and associ-739 ated morphology in male jumping spiders of the Habronattus740 agilis species group (Araneae, Salticidae). J Arachnol 16:199–741 211742 Masta SE (2000) Phylogeography of the jumping spider Habro-743 nattus pugillis (Araneae: Salticidae): recent variance of sky is-744 land populations? Evolution 54:1699–1711745 Masta SE, Maddison WP (2002) Sexual selection driving diversi-746 fication in jumping spiders. PNAS 99:4442–4447747 Nauen JC, Lauder GV (2002) Quantification of the wake of rain-748 bow trout (Oncorhynchus mykiss) using three-dimensional ste-749 reoscopic digital particle image velocimetry. J Exp Biol750 205:3271–3279751 Partan SR, Marler P (1999) Communication goes multimodal.752 Science 283:1272–1273753 Partan SR, Marler P (2005) Issues in the classification of multi-754 modal communication signals. Am Nat 166:231–245755 Peckham GW, Peckham EG (1889) Observations on sexual selec-756 tion in spiders of the family Attidae. Occas Pap Wisconsin Nat757 Hist Soc 1:3–60758 Peckham GW, Peckham EG (1890) Additional observations on759 sexual selection in spiders of the family Attidae, with some re-760 marks on Mr. Wallace’s theory of sexual ornamentation. Occas761 Pap Wisconsin Nat Hist Soc 1:117–151762 Peters RA, Evans CS (2003a) Design of the Jacky dragon visual763 display: signal and noise characteristics in a complex moving764 environment. J Comp Physiol A 189:447–459

765Peters RA, Evans CS (2003b) Introductory tail-flick of the Jacky766dragon visual display: signal efficacy depends upon duration.767J Exp Biol 206:4293–4307768Peters RA, Clifford CWG, Evans CS (2002) Measuring the struc-769ture of dynamic visual signals. Anim Behav 64:131–146770Post N, von der Emde G (1999) The ‘‘novelty response’’ in an771electric fish: response properties and habituation. Physiol Behav77268:115–128773Richman DB (1982) Epigamic display in jumping spiders (Araneae,774Salticidae) and its use in systematics. J Arachnol 10:47–67775Ryan MJ, Rand AS (2003) Sexual selection in female perceptual776space: how female tungara frogs perceive and respond to777complex population variation in acoustic mating signals. Evo-778lution 57:2608–2618779Strausfeld NJ, Weltzien P, Barth FG (1993) Two visual systems in780one brain: neuropils serving the principle eyes of the spider781Cupiennius salei. J Comp Neurol 328:63–72782Tammero LF, Dickinson MH (2002) The influence of visual783landscape on the free flight behavior of the fruit fly Drosophila784melanogaster. J Exp Biol 205:327–343785Victor JD, Purpura KP (1997) Metric-space analysis of spike786trains: theory, algorithms and application. Netw Comp Neural7878:127–164788Vogel S (2003) Comparative biomechanics: life’s physical world.789Princeton University Press, Princeton790Walker TJ (1974) Character displacement and acoustic insects. Am791Zool 14:1137–1150792Zanker JM (1996) Looking at the output of two-dimensional mo-793tion detector arrays. IOVS 37:743794Zanker JM, Zeil J (2001) Motion vision: computational, neural, and795ecological constraints. Springer, Berlin, Heidelberg, New York796Zeil J, Zanker JM (1997) A glimpse into crabworld. Vision Res79737:3417–3426

798

3 5 9 1 1 6B



Queries_ 359_2006_116

1. Section heading Introduction has been introduced. Please confirm the

change if it is ok?

2. Zeil and Zanker (2001) was cited in the text but is missing in the

reference list. Please provide the full reference or delete the citation

from the text.

dear damian elias, - cornell universitydear damian elias, here are the electronic proofs of your...

Documents