Insectes Sociaux, Paris. 1977, Tome 24, n ~ 2, pp. 191-212.

MATING BEHAVIOR AND SOUND PRODUCTION IN HARVESTER (, FORMICIDAE)

By H. MARKL, B. HOLLDOBLER and T. HOLLDOBLER Faehbereich Biologie, Universil~it Konstanz, Fed. Rep. Germany, and Department o[ Biology, Harvard University, Cambridge, Mass., U.S.A. (*). Re~;u le 25 novembre 1976. Aeceptd le 2 janvier 1977.

SUMMARY

1. Stridulation sound production was investigated in sexuals of the harvester ants Pofonomyrmex barbatus, P. rufosus and P. maricopa of the SW United States during preparation and take-off for nuptial flight, as well as during mate selection, copulation and postcopulatory hch:zvior on the mating leks, and competition in the process of colony foundation. 2. The structures of tile stridulatory organs and the characteristics of air-borne and of substrate-borne signals of the three species are described (Tahles I and II). There are no prominent interspecies differences in the distress signals. 3. Stridulatory communication is not involved in the control of preflight warm-up anti take-off for the nuptial flight. Further, it does not play a role in the selection of species and sex of the mates, or in the competition among males for females. 4. Females of all species stridulate regularly when prevented hy males from leaving the lek after several copulations. It is assumed that this << female liheration signal >> communicates the female's non-receptivity to approaching males and makes them let the female go. 5. Founding queens stridulate regularly when competing for founding burrows, especially if physical fights are involved. Nevertheless,'our experimental results do not suhstantiate a eommunieatory function for this behavior.

ZUSAMMENFASSUNG

Paarungsverhalten und Schallerzeugung bei Ernteameisen (Pogonomyrmex, Formicidae).

1. Die Schallerzengung durch Stridulation wurde an Geschleehtstieren der Ernte- ameisen Poffonomyrmex barbatus, P. ruffosus und P. maricopa aus dem SW der USA sowohl w/ihrend der Vorbereitung zum Hoehzeitsflug und helm Abflug, als auch bei der Geschleehtspartnerauswahl, der Kopulation und dem postkopulatorischen Verhalten auf

(*) Dedicated to Prof. Dr. K. von FRISCH on occasion of his 90th birthday. 192 H. MARKL, B. HOLLDOBLER AND T. HOLLDOBLER den Paarungsarenen und wiihrend Konkurrenzaggression bei der Koloniegriindung unter- sueht. 2. Der Bau des Stridulationsorgans und die Eigensehaften der in Luft bzw im Substrat fortgeleiteten Signale der drei Arten werden beschrieben (Tab. Iund II). Es gibt keine hervorsteehenden interspezifisehen Untersehiede zwisehen den Abwehrsignalen. 3. Stridulationskommunikation spielt keine Rolle bet der Kontrolle des Aufwiirmens vor dem Flug oder des Abflugs zum Hochzeitsflug und genausowenig bet der Erkennung von Art oder Gesehlecht eines Partners oder in der Konkurrenz der M~innehen um die Weibchen. 4. Weibehen aller untersuehten Arten stridulieren regelmiiBig, wenn M~innchen sie daran hindern, naeh mehreren Kopulationen die Balzarena zu verlassen. Es wird an- genommen, dab dieses << weibliche Befreiungssignal >> Mhnnehen bei der Anniiherung die mangelnde Kopulationsbereitsehaft der Weibehen anzeigt und sie dazu bringt, yon den Weibehen ahzulassen. 5. Griinderweibchen stridulieren regelmiissig, wenn sie um Grfindungskammern konkurrieren, besonders wenn es dabei zum Kampf kommt. Niehtsdestoweniger erlauben unsere Experimentalbefunde nicht den SehluB, dab diesem Verhalten dabei eine Ver- sfiindigungsfunkti6n zukommt.

INTRODUCTION

Although it is well known that thousands of species can produce sound with a stridulatory organ located on the dorsal surface of the postpetiolar-gastral joint, the functions of the signals produced have rarely been investigated. For a summary of the literature see MARKL (1967, 1968, 1970, 1973) and SPANGI.ER (1967 a, b, 1974). In leaf-cutting ants (Atta cephalotes) Mark1 found that workers stridulate persistently whenever prevented from moving freely. When the con- ditions are favorable for the conduction of the stridulatory signals as substrate- borne vibrations, they can be perceived by nestmates over some distance. Wor- kers are attracted to the source of these vibrations, where they immediately start with rescue activities, such as freeing nestmates which have been trapped after a cave-in of the nest. Also, sexual castes were found to emit stridulatory vibra.. tions when under physical distress, and the number of responding workers is proportional to the intensity of the signal (MARKL, unpublished). It is of particular interest to learn whether stridulation signals are also involved during reproductive activities. Although no species- or caste-specificity of the widely variable distress stridulation has yet been detected in the ant species studied, it is still possible that distinctive stridulatory signals are emitted during sexual activities. Ant stridulation signals might conceivably be involved in species and mate discrimination, especially in view of the fact that the pheno- menon is known from many other that employ stridulation sounds for sexual communication. Up to now, however, this question could not be answered because most ant species, which produce sound signals loud enough to be reliably monitored, mate during flights, or on inaccessible tree tops or other elevated points. The recent discovery that several sympatric species of harvester ants (Pogonomyrmex) in the mesquite-acacia desert of mate at terrestrial MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 193

mating leks (HOLLDOBLER, 1976 b) provided a unique opportunity to investigate the following questions : I. Are stridulation signals produced by females and/or males in any phase of their sexual behavior, and if so, which are their characteristics .9 II. Are these signals different in closely related, sexually isolated sympatric species .9 III. What are the functions of such signals .9

Stridulation signals of the worker'caste of the species investigated and their biolo- gical significance will be described elsewhere (MARKL, in preparation).

MATER~L AND METHODS

Field observations and sound recordings were made in July 1975, in a flat open mesquite-acacia-desert between Portal, Arizona, and Rodeo, , where Bert H6r~LDOBLER and Turid HSLLDOBr-ER had previously studied the behavior and ecology of the species used in this study (HSLLDOBLER, 1976 a) : Pogonomyrmex barbatus, P. rugosus and P. maricopa (a few observations were also made with P. desertorum). The details of the spatial and temporal organization of the mating leks of these species have been reported elsewhere (H6LLDOnLEn, 1976 b). In brief, males and females of a species assemble for mating after the onset of the summer rains at well defined places either on bushes or on the ground at times of the day that are different for each species. In our study area a partial temporal and spatial overlap in lek usage occurred occa- sionally between P. barbatus and P. rugosus and between P. maricopa and P. desertorum. Take-off flights and landing at the leks of both sexes, mate selection, copulation, post-copulatory behavior, the females' departure from the leks (males stay till death) and the excavation of founding chambers by dealated females were all studied and monitored in the field. The air-borne components of stridulation signals were conve- niently recorded with a Holgate Ultrasonic Receiver (Holgates of Totton Ltd., England; sensitivity--60 dB re 1V/izbar; response flat ___2 dB between 5 and 40 kHz; --10 dB between 1 and 110 kHz), the BFO circuit of which allows a rough estimation of the signal's intensity spectrum. The receiver is sensitive enough to pick up stridulation signals of the reproductive castes over distances of up to 10 era. Mating behavior of individuals was followed in plastic containers of different sizes (10-35 cm). For measur- ing signal parameters, the air-borne stridulation signffls were recorded by means of Bruel and Kjaer calibrated condensor microphones, types 4133 (20 Hz to 40 kHz • dB) and 4145 (20 Hz to 19kHz) on a Nagra SJS magnetic tape recorder (25 Hz to 35 kHz at 15 ips recording speed). During the recording of air-bone signals the ants were held with a thin pair of forceps in 0.5 to 1 cm distance from the microphone in front of a nonrefleeting background, in order to avoid signal distortions. Solid-borne stridulatory vibrations were recorded through calibrated aceelerometers (BnusL and KJAEn types 4335, 4338, 8306, 8307), whereby the types 4335 (0.1 to 15 kHz) and 8307 (0.1 to ca. 30 kHz) proved to be most suitable. Tape recordings were filmed from a Tektronix 5113 storage oscilloscope with a T6nnies Recordine oscilloscope camera; spectrograms and spectra were produced by a Kay Elemetrie Corp. Sonograph type 7029 A, when necessary at reduced tape speeds. The SEM mierographs of the stridulatory files were taken on the AMR 1000 A Scanning Electron Microscope of the Harvard University MCZ Laboratories. For statistical compa- risons t- and ~-~-tests were used. Differences were regarded as significant if error proba- bility for rejection of the null hypothesis was p ~---0.01. 194 H. MARKL, B. HOLLDOBLER AND T, HOLLDOBLER

RESULTS

I. Characteristics of the stridulation organ and stridulatory signals in Pogonomyrmex males and females.

1. The stridulatory organ..As typical in the stridulatory organ of Pogonomyrmex is situated dorsally at the joint between postpetiole

FIG. 1. MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 195 and gaster. For sound production the sharp posterior rim of the postpetiolar tergite (scraper, plectrum) scratches over the actively moved file (pars stridens), a field of horizontally arrayed, parallel ridges in the middle of the anterior end of the first gastral tergite. Sound is typically produced during an upward motion of the gaster, often--usually with less intensity--also during its downward motion (see SPANGLEn, 1967 a, b, for P. occidentalis). During motion, the file can have continuous contact with the scraper, producing an uninterrupted train of sound-pulses, or the contact can be variable, producing amplitude-modulated or even several times interrupted pulse-trains during one upward or downward motion phase. Figure 1 provides SEM micrographs of the stridulatory files of males and females of P. barbatus, P. rugosus and P. maricopu; in Table I the basic measu- rements are assembled from nail-polish replicas. Although there are significant differences between the overall size of the files in different sexes and species, the more important parameters affecting signal production, namely number and distance of ridges in the files, differ very little between P. barbatus and P. rugosus within the same sex. This is also true for the P. maricopa male, hut in the female the ridges are significantly more numerous and narrowly spaced. For the male of P. occidentalis Spangler (1967 b) gives on the average 167 and for the female 173 ridges.

TABLE I. -- Morphology of the stridulatory files of Pogonomyrmex sexuals. Average ----- standard deviation. In each ease n----10.

TABELLE I. -- Morphologie tier Rippenfehler yon Pogonom;lrmex Geschleehtstiercn. Mittclwcrtc und Standardabweichungen. In jedem Fall ist n = 10.

P. bttl'btltlts P. rutjo.gll,~ l P. Inaricopa

cr

l Greatest length 366 ~ 43 393~49 370W_30 396~__27 ]381 ~_ 41 398 -T- 46 (~m) I I J Greatest hreadth 371 ~ 36 418~54 394~46 4'2'2~48 J 2I)~] -- 26 414 ~ 55 (~m) i +

Numher of ridges 191 ~ 9 212~ 8 199~ 6 __217~ 7 ~5220 __252 ~ 10

Distance between ridges in the center 1,9 _~ 0,3 1,8~0,2 1,8~0,2 1,8 -7- 0,2 1,7 ~ 0,2 1,6 ~ 0,1 of the file (Ixm)

FI6. 1. -- Scanning electron mierographs of the stridulatory files of Pogonomgrmex sexuals, night: female; Left : male. Ann. 1.- Rasterelektronenmikroskopisehe Aufnahmen der Rippenfelder der Stridulations- organe von Pogonomyrmex Gesehleehtstieren. Rechts : Weibchen; Links : Mfinnchen. A : Pogonomyrmex barbatus; B : P. rugostls; C : P. marieopa. 19B H. MARKL, B. HOLLDOBLER AND T. HOLLDOBL~R

F /,,

,~.....m..a~t.c.o.p..a..~ ..... 7..m..s ...... PO.G.ON.g..M..V..R..M..E.X..

B 20

C A ...... z ......

D 100 FIGFIT ...... :...... d- ...... #.~l~. ~i -~',-= ~ ..... ~.~k,~i ii ~ii,,,~lll ..... "',,~' ,~i~ i,~,~lt ...... ~ _.~ ...... ~i .....

. . , . i . ~q~,lllll.~ :iinigl,:

E

.. ill lll=.~ . ~le >ill, , lllli< lilllli ...... mlll._. Ill. _

Fro. 2. -- Oscillograms of stridulation signals from Pogonomgrmex sexuals recorded in air or as substrate vibrations. Numbers refer to time-marker spacings in msec. Recording conditions : Above : A, C, D 9; Below : A, D, E : Air-borne signals recoL'ded through Holgate Ultrasonic receiver. Above: D e~: Recorded through Bruel and Kjaer microphone 4133; Abovc : B, E, F; Below : B, C : Vibration signals recorded through Bruel and Kjaer accelerometer 4335. MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 197

2. Stridulation signals. As in all ants capable of stridulation, sound pro- duction is easily released in all castes of harvester ants by holding an by any means and at any part of the body. Stridulation can be in short bursts or in long continuous trains; in females it is often interrupted by vigorous attempts to sting. The stridulation is made up of chirps, caused by the upward or downward movement of the gaster. Each chirp consists of a train of discrete pulses, which are the result of the impact of the scraper on the ridges of the file (for details of the physics of sound production in ants see MA:RKL, 1968). Though the pattern of stridulation is extremely variable, depending on the circumstances (e.g., freedom of movement for the gaster), it can basically be characterized by the parameters given in Table II. Figure 2 shows oscillograms of signals recorded through air or substrate. Before going into the details of Table II it should be emphasized that (1) the data are taken from recordings of experimentally <~ trapped >> (either held by forceps or fastened to an accelerometer; both treatments yielded identical results), so that it is r distress stridulation >> that was analyzed. In only one case (a single female of P. barbatus producing <~ liberation signals >>, see p. 208) was spontaneous stridulation studied : this did not differ significantly from induced <~ distress stridulation >> (Table II : lib.). (2) The variability of the stridulatory pattern is grossly under-represented by Table II, since the only data used were from those animals that produced chirp series of more than 3 chirps in a regular sequence. This was regarded as being the optimal condition to detect sex- or species-differences. In addition, all experimental animals from allspecies produce irregular, odd chirps of extre- mely variable structure ranging from single pulses to full-blown trains without or with more or less pronounced amplitude modulations, sometimes breaking the groups in 2 to 12 <~ pieces >> (similar amplitude modulation can also occur in regular chirp sequences, irrespective of sex" or species). The evaluation of these irregular signals yielded no consistent sex or species characteristic, so that they can be disregarded in the following context. As Table II shows there are several significant differences in some of the parameters between species and between the sexes within a species. For example the female of P. maricopa produces a significantly higher average pulse repetition rate than the females of the other species. However, there also exists considerable variation between different series of chirps of the same animal and of different animals of the same species and sex. Therefore no clearcut species or sex specific signal characteristics can be recognized. This is true even in those species where character displacement should be expected if the signal were involved in repro-

ABB. 2. -- Oszillogramme der Stridulationssignale yon Pogonomyrmex Geschlechtstieren. Die Zahlcn geben die Zeitmarkerabsthnde in msec an. Aufnahmebedingungen : Oben : A, C, D 9; Unten : A, D, E : Aufrrahme der Luftschallsignale mit dem Holgate Ultra- sonic Receiver. Oben : D ~ : Aufnahme mit Bruel und Kjaer Mikrophon 4133. Oben : B, E, F; Unten : B, C : Aufnahme der Vibrationssignale mit Bruel und Kjaer Accelero- meter 4335. INSECTES SOCIAUX, 1977, T. 25, N~ 2. 13 198 H. MARKL, B. H(~LLDOBLER AND T. HOLLDOBLER

., +1"7 II § II +1~' II § II

O+ § II +I'T~ +1~, ]1 Z " ~Y.~ ~.~z .... ~'-~ i r y +1 ~, II

04 ~'~ .~,~ +1"7 II +1"7 § II +1~, II .... ~ ~.~

.~ ~" ~,1 ~'b~.~ ,,-* t... r I *.o +1"~, II +1~. II § II r @,! § II

:ll ""

= ~ .. .~ ~ e~

r162 ~'-~ O+ +1'7 I1"~ II "1"1, I1= -t-I ,

~

*o § § II "H, II +1~' II o ~= . =~.~.~

~ ~ ~ ..~.~

~.,~176

,,: :..~ ~'~ ~z:~ =~ MATING BEHAVIOR AND SOUND PRODUCTION IN PoGONOMYRMEX 199

§ II § 11 r..~=

~ c'-Icq -II~, It § 11 § il~ 3=g

~Lm § II § 114

§ II § II~

§ II "t'u: ~ +[ I1~ r162

~.~.. +I.7 II = +i c"q~ .. ~:~ ". 2.=g.. =

oo co § II +lg, II §

1NS~CTES SOCIAUX, 1977, r. 25, N ~ 2. 13" 200 H. MARKL, B. HOLLDOBLER AND T, HOLLDOBLER

,;,~i~i;;,Z ;::.-: : ,;L...... - ...... -:7111 y-;.v :.--~-~.- : -.: ...... ~- ......

. z u d ,~,,~i ~ ~.i~.~;~-%

a

! ii,:i! !;iii'il-ii/

lihl!i{~iltllli~$ii!l',ltlj~',i',',j;!lillr,!iij!Llild'!~Higlhlii~t~.:~:i,~,..

~:.;~F:.,: -"! :~- .'~! ~ i:~'." : . "~

C 100 2OO

' " q, i

20 ...... i

i

e 20-

1O- f O- 0 100 200 300 msec FIG, 3. -- Spectrograms (left) and spectral sections (right) of stridulation signals from Pogonomgrmex sexuals recorded in air or as substrate vibrations. Recording condi- tions: a) P. barbatus @; b) P. barbatus ~; e) P. maricopa 9: airborne signals recorded through Bruel and Kjaer microphone 4133; d) P. barbatus ~; e) P. rugo- sus ~; f) P. maricopa ~ : vibration signals recorded through Bruel and Kjaer acce- lerometer 4335. In a and c there are spectral sections of the background noise level given in the far right. The noise bands around 9 and 36 kHz (a, b, c) are due to unfavorable environmental recording conditions (background noise source). The minor signal distortions due to them can be easily recognized by comparison of signal and noise sections in a and c. MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 201 ductive isolation (e.g., P. barbatus and P. rugosus). These findings are in agreement with earlier studies on worker stridulation in related myrmicine spe- cies (MARKL, 1968, and unpublished data), among which species specificity could not be found. The maximum peak-to-peak intensity of the stridulation signals in air as measured in 0.5 cm distance from the producer through the Bruel and Kjaer 4133 microphone is 46 to 56 dB (re 2.10 -5 N/m 2) almost regardless of the ~pecies or sex studied, with females of P. maricopa tending to be the loudest in air. Solid-borne stridulatory vibrations were measured between 0.1 and 10 kHz through the Bruel and Kjaer 4335 accelerometer, to which the ant's ventral side was firmly pressed by a strip of masking tape leaving the gaster's motion unim- peded. This arrangement yields only a first approximation of the vibrations transmitted into a substrate (see detailed discussion elsewhere, MARKL, in prep.); however, for interspecific ,comparisons the measurements characterize the energy contents Of stridulatory body-vibrations sufficiently well. Peak-to-peak values ranged between 18 and 35 cm/sec", with no significant sex- Or species-dependent differences apparent, although signals by females of both P. barbatus and P. rugosus are generally of somewhat higher amplitudes--in accordance with the larger body size of these females. The spectral energy distribution of the air- and solid-borne vibrations can be read from spectrograms in fig-are 3. The broad-band signal characteristic is prominent in both cases. In the air-borne signals the energy spreads between ca. 1 and 40 kHz (and beyond, with traces up to 80 kHz, as evidenced with the BFO feature of the Holgate Ultrasonic Receiver) with a broad maximum between 10 and 40 kHz. This seems at first glance at variance with results of Spangler (1967 b) on P. occidentalis (see also KV.RMARREC et al., 1976). The reasons for these inconsistencies lie in the techniques of acoustical measurements and their limitations. These problems will be discussed more thoroughly elsewhere (MARKL, in prep.). It is, however, important to emphasize that all authors agree that ants do not react to the air-borne component of the stridulation signals. Thus differing results concerning the spectral composition do not affect conclusions concerning the functions of stridulatory substrate vibrations in ants. The aecelerometered vibration contains energy between 0.1 and 15 kHz, with minor components up to 25 kHz as measured with type 8307. The broad

ABB. 3. -- Spektrogramme (links) und Spektren (reehts) der Stridulationssignale yon Pogonomyrmex Gesehlechtstieren, die als Luftsehall oder als Substratvibrationen aufgenommen wurden. Aufnahrnebedingungen : a) P. barbatas 9; b) P. barbatus c~; e) P. maricopa ~ : Aufnahme des Luftschallsignals durch Bruel und Kjaer Mikrophon 4133; d) P. barbatus f~; e) P. rugosus 9; f) P. maricopa <2 : Aufnahme des Vibrations- signals durch Bruel und Kjaer Aecelerometer 4335. In a und e sind reehts aufSen Spektren des Hintergrundrausehens wiedergegeben. Die Rauschbanden bet 9 und 36 kHz (a, b, e) riihren yon den ungiinstigen ~iufleren Aufnahmebedingungen her (HintergrundstSrquelle). Die geringfiigige Signalverzerrung, die dureh sie bewirkt wird, kann durch Vergleieh der Signal- und Hintergrundrausehspektren in a und e leieht erkannt werden. 202 H. MARKL, B. HOLLDOBLER AND T. HOLLDOBLER maximum--sometimes multiply banded, but without clearly harmonic struc- ture is between 1 and 8 kHz. Sound and vibration reception have not been investigated in Pogonomyrmex alates. Substrate vibration receptors in ants generally have been found to be most sensitive between a few hundred Hz and several kHz, with lowest accelera- tion threshold around 1 cm/sec 2 (see MAaKL, 1970; Fucns, 1976 a, b). It is pre- sumed that Pogonomyrmex sexuals have similar characteristics of vibration sensitivity enabling them to react to substrate-borne stridulatory vibrations.

II. Nuptial flights and sexual behavior in Pogonomyrmex.

1. Take-off. The nuptial flights and courtship procedures of the species studied have been described in detail by H~LLDOBLER (1976 b). On sunny days following the heavy summer rain showers colonies release alates in large numbers within a period of less than 1/2 hour, at a time of the day different for each species (e.g., forenoon in P. maricopa, afternoon in P. barbatus and P. rugosus). Before the sexuals take flight, they usually crowd together while sunning in front of the nest entrance. During this period and during the actual take off procedure, the workers are extremely agitated and aggressive. They hinder sexuals from leaving the nest too early or prevent them from taking flight when the weather turns unsuitable. We observed the nuptial flight preparations and the take-off procedure in several colonies of P. maricopa, P. barbatus and P. rttgostls, while continuously surveilling the sound production. There were thousands of rake-offs as well as many cases in which the sexuals returned to the nest without having taken flight, where not a singledefinite stridulatory signal could be detected. In other cases, short chirps, never more than 2 or 3 in a row, were heard; in the few cases when the source could be determined with some confidence, these signals proved to be the consequence of males grasping females or workers in futile attempts at courtship. In one case each it was clear that stridulation came from a male or a female which was dragged across the nest surface by workers. In summary, stridulation is neither an integral part of preparation for the nuptial flight nor for the take-off in the harvester ants studied, although it occurs sporadically during forceful interactions between nestmates.

2. Courtship and mating on the mating leks. Males and females reach the mating leks on the wing and mate there on vegetation or on the ground in an agitated melee of wing-whirring males and females. When a female alights she is almost instantaneously surrounded by males, ready to mount her though the females are often reluctant to acquiesce in the beginning (fig. 4). A male grasps the female's thorax with his mandibles, riding on her back, and extends his copulatory apparatus in search of the female's genitalia (see H~LLDOBLER, 1976 b, fig. 6). If she is ready to mate, she turns her abdomen slightly to the MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 203 side, so that the male can couple. After some time, the male relaxes his mandibular grip on the fe- male's thorax. Then while still hanging on her, he rolls over to her ventral side, so that the back of hi~ gaster is now exposed to mandibular kneading by the fe- male, and he simultaneously mas- sages her gaster (fig. 5). In the meantime a second male may catch hold of the female's thorax and wait in line for the first male to uncouple, That occurs, often actively aided by bites from the female, rather swiftly and some- times with such force that the male's copulatory apparatus is detached and left clinging to the female's genitalia. Half a dozen or more additional males some- times surround the pair and try to

Fro. 4. -- Courtship in Pogonomyr- mex barbatus. Above: A female, which had just arrived at the mating lek, is being grasped by a male at the thorax. Middle : A second male takes hold of the fe- male; Below : A few seconds later the female is surrounded by at least 5 males. Ann. 4. -- Paarungsverhalten bet Pogonomyrmex barbatus. Oben : Ein eben in der Balzarena ange- kommenes Weibehen wird yon einem M[innchen am Thorax er- griffen. Mitte : Ein zweites Mtinn- chert fasst das Weibehen. Unten : Eiuige Sekunden spiiter ist das "Weibchen yon m~ndestens 5 MSnn- chen umringt.

get in position for mating themselves (fig. 4). Typically, a female copulates several times. After the last copulation she tries to run away and fly off. But the number of males searching for a mate is often still high, in which case the female is frequently hindered in take-off and has to shake off males again and 204 H. MARKL, B. HOLLDOBLER AND T. HOLLDOBLER again. When a female has entered this departure phase, males cannot succeed in coupling even if they grasp her thorax, because she bends the end of her gaster firmly downwards. The whole courtship and mating process, from arrival at the lek until departure, was followed both in free animals on the lek and in individuals confined inside plastic containers. Most observations were of P. maricopa and P. barbatus, fewer of P. rugosus, and only a very few of P. desertorum. By bringing together barbatus females with rugosus males and vice versa, it was

Fro. 5. -- Copulation of Pogonomyrmex rugosus. The male" massages the female's gaster, simultaneously, the female begins to gnaw -~t the male's abdomen. ABB. 5. Kopulation bei Pogonomyrmex rugosus. Das M~innehen massiert den Gaster des ~Veibchens, w~ihrend gleichzeitig das We[bchen das Abdomen des M~innchens mit ihren Mandibeln zu bearbeiten beginnt. also possible to observe cross-courtship and mating. Since the whole mating activity is accompanied by all but continuous intensive wing-vibration (which is identical in frequencies and intensities in all species studied) the low cutoff characteristic at ca. 1 kHz of the Holgate Receiver is indispensable for detecting low-intensity, high frequency air-borne stridulation signals. Control experi- ments with hand-held animals showed that stridulation in the mating cluster could be clearly detected over several, sometimes up to 10, centimeters. Mating behavior continued entirely undisturbed under close-range visual and acoustical inspection (fig. 6), so that the following results on the occurrence of stridulation can be considered reliable : (1) In all four species no stridulation of the sexuals was reliably witnessed before or during the first copulation of a newly arrived and presumably virgin female, regardless of whether only conspecific sexuals or also heterospecific ones MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 205 came into contact and tried or succeeded in coupling. Even when as many as 10 males were engaged around a single female, there was no regular stridulation involved. Odd, single bursts which were heard in a few instances could not be precisely localized as to the source; however, it is suspected that males some- times stridulate shortly when shaking off homosexual grasps. (2) Sexual behavior in all three species was observed both at the lek under undisturbed conditions, and after animals were placed in containers. In some cases the observations began immediately after the females landed at the lek, in

FIG. 6. -- Monitoring with a Holgate Ultrasonic Receiver the air-borne ultrasounds produced during mating at a lek of Pogonomyrmex maricopa (left) and by an isolated mating cluster (right). Ann. 6. -- Ueberwaehung der Ultraschallsignalerzeugung der Gesehleehtstiere von Pogono- myrmex maricopa wiihrend des Paarungsverhaltens in der natiirliehen Balzarena (links) oder in einem isolierten Paarungskniiuel (reehts) mit Hilfe des Holgate Ultra- sonic Receiver. others females were already engaged in sexual activities. All recordings were carried through until no more copulations occurred, although males still con- tinued to try to mount the females. The data show (Table III) that no female stridulated clearly and persistently as long as she continued to accept successive copulations. However, in 71% of the barbatus-females, in 84 % of the rugosus- females and in 78 % of the maricopa-females, prolonged series of stridulation chirps could be recorded following the last copulation as determined post festum from the departure from the lek or when the females persistently repelled males in the container. There was no evident difference in a female's readiness to stridulate when the courting males were of the same or as opposed to another species (tested between barbatus and rugosus). 206 H. MARKL, B. HOLLDOBLER AND T. HOLLDOBLER

.~ ~ ~,.~

F.

o,,.~~ ~

g~

v

o

t,., ~ ~~

,~ea

o

o ,,~ r.,~ ~ ~ o . t~

eJ ~ ~

~ ~,~= o

I MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 207

o at =

.~-'_ ~ ~.~ "~ =

--~o I d = 208 H. MARKL, B. HOLLDOBLER AND T. HOLLDOBLER

The female stridulation signals can go from a few chirps to several seconds of continuous stridulation, shorter sequences being more frequent (no precise measurements could be taken). No differences in the frequency of the signal production were evident between the three species. That the stridulation signals really originate from the females is concluded from observations in which the abdominal motion could clearly be correlated with the sound heard. A Similar observation was never made in males involved in ~ stridulating >> pairs. Males left alone were never found to produce longer series of chirps, and only one out of 23 barbatus-females, confined without males after they had repelled all males for several minutes, produced a detectable short stridulation (Table III). In one case it was possible to record the stridulation signals produced by a female under continuous attack by males in the laboratory nest. Although the sequence was short, no significant difference from normal distress stridulation could be found (Table II : barbatus ~ lib >>). The fact that most of the females start with prolonged stridulation only after they have ceased to accept further copulations and when they are engaged in freeing themselves from the surroun- ding males led us to call this stridulation signal a <~ female liberation signal >>. Of course the question remains, whether the stridulatory signals emitted by unreceptive females have any function, for example the discouragement of further copulatory attempts by males. Although this hypothesis cannot yet be decided from our data, there is some supportive evidence. Females confined after their last copulation but before their first stridulation were nearly conti- nuously sexually molested by males who were present in a large surplus during these experiments. After the females began to stridulate more fre- quently as a response to these attacks, the intervals of time when the females ran alone in the container became visibly longer until finally the female was often no longer attacked by males at all. Further experiments will have to be conducted in order to decide the full importance of this presumed repellent communication of females after courtship. -

3. Colony-foundation and interfemale competition. Stridulation also occurs during the process of challenge and competition between conspecific or hete- rospecific founding queens for the incipient burrows dug into the soil by foun- dresses. It is not uncommon at all that a mated female searching for a place to dig a chamber runs into the opening of a burrow dug by another female (H6LLDOBLER, 1976 a). In many cases the searching female enters this opening and either leaves immediately after contact with the owner or engages in a more or less extended fight. The fight can end either way, although burrow-owners usually prevail. During these encounters between females stridulation can be registered quite regularly. The following experiments were performed in order to learn whether stridulation is of communicative importance during the encounters. An area with many open founding burrows of P. barbatus and P. rugosus was selected. In different combinations, founding queens of both species caught on the previous or on the same day from their burrows were released from glass vials close to MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX 209

TABLE IV. --- Outcome of interaction between Pogonomyrmex females owning a founding- tunnel and intruding females. It is unclear whether owner or intruder is responsible or begins with stridulation in all but a few cases. TABELLE IV. -- Ergebnis der Auseinandersetzung zwischen Pogonomyrmex Weibehen, die eine Griindungskammer besitzen und Eindringlingen. Ausser in wenigen F~illen ist es unbestimmt, ob die Besitzerin oder ein Eindringling ftir die Stridulation verantwortlich ist oder damit begounen hat.

Intruder Burrow Number Intruder )f encoun- Number of fights Fight won by ru~s away owner ters without fight

with stridulation owner 10 10 intruder 0 1 P, barbatus. 16 owner 2 without stridulation 3 2 intruder 0 P. barbatus. with stridulation owner 6 0 (*) 7 intruder 1 P. rzzgoszzs. owner 0 without stridulation 2 0 intruder 0 I

with stridulation owner 5 intruder 0 P. rugosus. without stridulation owner ! a 1 intruder P. rllgOSllS. owner 3 with stridulation 1 3 intruder 0 P. barbatus owner 2 without stridulation 2 2 intruder

with ] with strid. strid. ] 25 owner Totals 30 14 intruder without I without strid. strid. I 5. / 12

Percentage of encounters with stridulation : 62 % Percentage of fights with stridulation : 84 % Percentage of fights without stridulation : 16 % Percentage of fights in encounters : 68 % Percentage of flight of intruder with stridulation : 14 % Percentage of flight of intruder without stridulation : 86 %

(*) The victorious intruding rugosus-female succeeded afterwards in driving away 3 rugosus and 2 barbatus contenders from her newly occupied burrow; stridulation occurred in 3 of these fights. 210 H. MARKL, B. HOLLDOBLER AND T. HOLLDOBLER the open entrances of the occupied burrows, into which they readily entered. Table IV shows the results; in figure 2 oscillograms of stridulation during a fight are given. There seems to be no significant interspecific difference in these encounters : the percentage of run-aways without fighting is 33 % in homospe- eific contacts, 29 % in heterospecific contacts, and 32 % in total. In 86 % of the cases when ants ran away, no stridulation could be recorded before the intruder left; in the remaining 14 % short stridulation was heard after the intruder entered the burrow and before she left without visible fight. There- fore we cannot regard stridulation of either the owner or the intruder as a decisive communicatory mechanism for avoiding fights. Quite the contrary: in 84 % of those cases in which intrusion led to a more or less prolonged fight (i.e., in 68 % of all encounters) the fights were preceded or accompanied--since fights began in the burrows it could not be decided which was true--by stridu- lation on the part of one or both partners. In most eases it was impossible to decide whether the owner or the intruder stridulated. In a few cases both were found by visual inspection to contribute, although, as figure 2 demonstrates, they rarely did at the same instant. Whether or not stridulation occurred, in almost all fights (97 %) the burrow-owner was victorious. In only a single case a r~zgosus female expelled a barbatus owner and then held the newly won burrow against 3 more ragostts and 2 more barbatus females. From these results it can only be concluded that interfemale encounters in contest for burrows are often accompanied by stridulation and quite regularly so when it comes to a fight. However, no functional significance is evident from the outcome of the encounters observed during the present study.

DISCUSSION

The results of these observations and experinaents on the occurrence and functions of stridulation during mating behavior and colony founding in three sympatric harvester ant species lead to the following conclusions : 1. Sound production is not involved as a signal in the behavior of the three castes before or during take-off of the sexuals for their nuptial flight, nor are sound signals important for occupying or finding the mating lek. 2. Sound Or vibrational signals are used neither for species recognition nor for sex recognition during the mate-selecting process, since almost no stridulation and certainly no prolonged species-specific stridulation occurs during this phase. Stridulation therefore does not function as a reproductive isolating signal in sympatric closely related Pogonomyrmex species. 3. Stridulatory signals are quite regularly produced by females which are no longer ready to mate, whenever they are assaulted by additional sexually excited males (specifically, by 77 % of 26 females which were monitored to the end of their copulatory receptivity). There are indications that males depart MATING BEHAVIOR AND SOUND PRODUCTION IN POGONOMYRMEX ,911 more readily from stridulating females than from silent ones. This would clearly be of advantage to both sexes: males would stop wasting efforts on females already fully charged with sperm; females would no longer run the risk of being mutilated by the males' mandibular grasp (see HOLLDOBLER, 1976 b) and could leave the lek as early as possible, avoiding further danger from predators. One might thus call stridulation in this context a r female liberation signal >> in ants. Still it remains to be proven whether sound production is not merely a behavioral corollary of female repellent behavior without decisive communi- cative significance. 4. Though females fighting with each other for burrows stridulate quite regularly, as do fighting workers of these species, no clear communicative func- tion could be attributed to this sound production. Nevertheless, it has been shown in other eases that vibration signals in ants very often influence (<< modu- late >>) the receivers' behavior in subtle ways detectable only by statistical ana- lysis of large numbers of communicative interactions (Camponotus rapping com- munication : MARKL and FUCHS, 1972; Fucas, 1976 a, b; Novomessor worker stri- dulation: HSLLDOnLER, STANTON and MARKL, in prep.; Pogonomgrmex worker stridulation : MARKL, in prep.). Consequently one should be prepared to revise the negative conclusion concerning Pogonomgrmex if further and more detailed investigations provide new evidence.

Thus stridulation is definitely not a means of communication in court- ship--it is employed neither in mate-finding nor in mate-selection. Since this negative finding has been made in exactly those species of myrmicine ants which can be assumed to be under particularly strong selection pressure for evolving mechanisms of reproductive species discrimination, it seems highly probable that stridulation in ants is not generally involved in reproductive isolation. This conclusion is in agreement with the result of other careful studies of ant stridulation {MARKL, 1967; SPANGt:En, 1967 a). The functions of sound communi- cation in ants seem surprisingly diverse and heterogeneous: they range from cave-in distress alarm (MARKL, 1965, 1967; SPANGLER, 1967 b, among other sug- gested functions), to the presumed (< liberation signal >> of the mated queens (this study) to a special prey-alarm system (H6LLDOBL~R, STANTON and MARKL, in prep.; MAR~L, in prep.). However, the diversity contains a common denominator. Stridulatory vibration apparently functions as a device that modulates reaction thresholds to other stimuli in the receiving partner, the function of which is determined by both the behavioral context (especially the social-behavioral context) and the state of behavioral activity of the receiver. This will be dealt with in more detail elsewhere (MARKL, in prep.).

ACKNOWLEDGEMENTS. -- We thank R. STANTON, n. SELINO, M. BELLEMANN, P. BAUER, H. LANG and J. TAUTZfor technical assistance. This work was supported by the Deutsche Forsehungsgemeinschaft (grants 741, 29-3; Ma 374/2, 4, 9 to H. M.) and by the National Science Foundation (grants BMS 75-06447 and BMS-7412494 to B.H.). 212 H. MARKL, B. H('JLLDOBLER AND T. HOLLDOBLER

REFERENCES

Fucus (S.), 1976 a. -- The response to vibrations of the substrate and reactions to the specific drumming in colonies of carpenter ants (Camponotus, Formieidae, Hymenop- tera). Behav. Ecol. Sociobiol,, 1, 155-184. -- 1976 b. An informational analysis of the alarm communication by drumming behavior in nests of carpenter ants (Campo- notus, Formicidae, ). Behao. Ecol. Sociobiol., 1, 315-336. HSLLDOBLEa (B.), 1976 a. -- Recruitment behavior, home range orientation and territo- riality in harvester ants, Pogonomyrmex. Behao. Ecol. Sociobiol., 1, 3-44. -- 1976 b. The behavioral ecology of mating in harvester ants (Hymenoptera : Formieidae : Pogonomyrmex). Behav. Ecol. Sociobiol., 1, 405-423. HSLLDOBLER (B.), STAr,TON (R.) and MARKL (H.). -- Signals regulating reeruitrrient and food retrieving behavior in Novomessor (Formieidae, Hymenoptera) (in preparation). KEnMAaU~-C (A.), MAULEON (H.) and ABUD ANTUN (A.), 1976. -- La stridulation de Acro- myrmex octospinosus Reich. (Formicidae, Attini) : Biom~trie de l'appareil stridu- lateur et analyse du signal produit. Ins. Soe., 23, 29-48. MAnKZ (H.), 1965. -- Stridulation in leaf-cutting ants. Science, 149, 1392-1393. -- 1967. Die Verstiindigung dureh Stridulationssignale bei Blattschneiderameisen I. Die bio- logische Bedeutung der Stridulation. Z. vergleich. Physiol., 57, 299-330. -- 1968. Die VerstSndigung durch Stridulationssignale bei Blattschneiderameisen II. Erzeugung und Eigenschaften der Signale. Z. vergleich. Physiol., 60, 103-150. -- 1970. Die VerstSndigung dureh Stridulationssignale bei Blattschneiderameisen III. Die Emp- findiichkeit fiir Substratvibrationen. Z. vergleich. Physiol., 69, 6-37. -- 1973. The evolution of stridulatory communication in ants. Proc. 7th Congr. IUSS1, London, pp. 258-265. -- Communication by vibration signals in harvester ants (Pogono- myrmex)--a case of modulatory communication (In preparation). MAnKI. (H.) and FucHs (S.), 1972. -- Klopfsignale mit Alarmfunktion bci RoBameisen (Camponotz~s, Formicidae, Hymenoptera). Z. vergleich. Physiol., 76, 204-225. SPANGLER (H. G.), 1967 a. -- Ant stridulations and their synchronization with abdominal movement. Science, 155, 1687-1689. -- 1967 b. Stridulation and related behavior in harvester ants. Ph.D. Thesis Kansas State University. University Microfilms Ltd., High Wycomb, England, Nr. 67-13, 964. -- 1974. The transmission of ant stridulations through soil. Ann. Ent. Soc. Amer., 67, 458-460.