Anim. Behav., 1997, 54, 1509–1515

Antennal asymmetry and sexual selection in a cerambycid

ANDERS PAPE MØLLER* & CARMEN ZAMORA-MUNx OZ† *Laboratoire d’Ecologie, CNRS URS 258, Universite´ Pierre et Marie Curie †Departamento Biologı´a , Facultad de Ciencias, Universidad de Granada

(Received 22 November 1996; initial acceptance 11 February 1997; final acceptance 30 April 1997; MS. number: 5394)

Abstract. Cerambycid have exaggerated antennae that are usually sexually size-dimorphic. We investigated the relationship between antenna morphology and sexual selection in the species melanura (L.) in which males on average have antennae that are 13% longer than those of females. Males and females aggregate at flowers near oviposition sites for feeding during June–August. We sampled both copulating and single individuals at these sites. Fluctuating asymmetry (a measure of developmental instability) in antennae was considerably larger than in tibia and elytra and males had larger degrees of asymmetry in their antennae than females. Mated individuals did not differ from unmated individuals with respect to any of three size variables, but antennal asymmetry was smaller in mated individuals of both sexes. When two males were released with a female on a flower, males with symmetric antennae more often won the fight over the female than expected by chance. When two females and a male were released on a flower, the male more often preferred the female with more symmetric antennae than expected by chance. These results suggest that antennal symmetry, but not length, is currently under sexual selection.  1997 The Association for the Study of Animal Behaviour

There has been a tremendous increase in our many cerambycid species, with males having the knowledge of certain aspects of sexual selection longest antennae. This obviously implies a role for over the last couple of decades (for a review see sexual selection, as already suggested by Darwin Andersson 1994). The general existence and (1871) and Linsley (1959) for this family. Surpris- importance of female choice and male–male com- ingly, there are, to the best of our knowledge, no petition have been repeatedly demonstrated. This previous studies of sexual selection with respect to is particularly the case for many . For this character in any cerambycid beetle. species with elaborate ornamentation, however, Our main aim in the present study was to there are few detailed studies of sexual selection, determine to what extent the morphology of the with horned beetles being a notable exception antennae and other morphological characters are (Brown & Balaton 1986; Conner 1988, 1989). subject to current sexual selection in the species Beetles of the family Cerambycidae have . More specifically, we investi- extremely elongated antennae, which are often gated the relationship between antennal length considerably longer than the body, and it is not and asymmetry and mating success. We did this difficult to imagine that such long antennae may by sampling mated and unmated individuals in be very costly for their bearers. These exaggerated the field, and by staging male–male competition structures can also be beneficial of course, when for a female and male choice of females in the used to detect potential predators. The length of field. the antennae is clearly sexually size-dimorphic in Stenurella melanura is a very common North European cerambycid beetle, occurring in conifer- Correspondence: A. P. Møller, Laboratoire d’Ecologie, ous forests as an adult from June to September. CNRS URS 258, Universite´ Pierre et Marie Curie, Oviposition takes place in dead wood, as is the Baˆt.A,7e`me e´tage, 7 quai St Bernard, F-75252 Paris case in most species of the family. Individuals of Cedex 5, (email: [email protected]). C. Zamora-Mun˜oz is at the Departamento Biologı´a both sexes feed on flowers near oviposition sites, Animal, Facultad de Ciencias, Universidad de Granada, and these flowers also act as sites for mate search- E-2800 Granada, . ing. Almost all the individuals we studied were

0003–3472/97/121509+07 $25.00/0/ar970565  1997 The Association for the Study of Animal Behaviour 1509 1510 Animal Behaviour, 54, 6 observed and captured on the flowers of the Eppendorf vials with absolute alcohol until we composite Knautia arvensis. While feeding on could measure them. Only three individuals flowers, males also search for females. If a female escaped during capture attempts. is approached by a male, however, this does not necessarily imply the male will mate successfully, Experimental Interactions since the female may either leave the flower or resist the male’s approach. Interactions between During July 1993 single beetles were captured males and take-over attempts are frequent. Inter- and stored in Eppendorf vials before being used in actions between males consist of the individuals competition or mate-choice experiments. We grabbing each other’s antennae and limbs while staged male–male competition trials by simul- vigorously wrestling. This wrestling may some- taneously releasing two males on to a flower that times result in the amputation of an antenna or a already held a female. The release distance leg (Linsley 1959). Males of several cerambycid between each of the males and the female was less species have apparently evolved particularly than 1 cm and between the males less than 0.5 cm. powerful mandibles for this purpose (Thornhill When the Eppendorf vials were removed, the & Alcock 1983). Similarly, male–female inter- activities of the males were observed until one actions also involve frequent use of the antennae, male had started copulating with the female. We apparently for determining position and move- performed 25 trials, but no copulation occurred in ment of the other individual. Often antennal tips four and one male escaped in two. Both males appear to be used for determining the position of were re-captured in the remaining 19 trials. the tip of the abdomen of the other individual. In the second experiment two females were Females are able to terminate copulations by released on to a flower next to a male in the same simply flying away (Michelsen 1963). Copulatory way as described for the previous male–male mate guarding is common, as in other cerambycid competition experiment. We performed only 15 species (Parker 1970). Females regularly copulate trials owing to a male-biased operational sex ratio with more than one male (personal observation). in the population. One female escaped re-capture in two of these leaving 13 trials with re-captured females for analyses. MATERIALS AND METHODS Measuring Beetles Study Area Beetles were measured under a stereo micro- A.P.M. studied Stenurella melanura at a co- scope with a magnification of #12 for antenna niferous plantation near Kraghede (57 12 N, ) * and elytra and #25 for tibia, mainly by C.Z.M. 10)00*E), , regularly throughout July who was unaware of the origin of the different 1993. This area consists of 20–75-year-old stands samples that were identifiable only by identifi- of Picea abies with many fallen trunks which the cation numbers. The length of the three characters beetles use as oviposition sites. The beetles forage (antenna, tibia, elytra) was based on pre- on flowers, particularly of Knautia arvensis,in determined landmarks determined by the end open areas in the plantation, which have a very points of the characters. All characters were fully warm and sunny micro-climate. extended (antennae) and in the plane of the micro- scope field when measured. We measured 10 Sampling Beetles individuals on two subsequent days to estimate repeatability of character length and asymmetry A.P.M. walked along 300-m transects in open (Becker 1984; Falconer 1989). The measurements areas of the plantation between 0900 and 1500 from the first trial were covered when making the hours while inspecting all flowers for the presence second measurements. The average lengths of left of beetles. The individuals included in this study and right sides were highly repeatable (antenna: were all those recorded during the study period, in R=0.998, F9,10=849.79, P<0.001; tibia: R=0.985, total 67 males and 31 females. Males guarding F9,10=151.17, P<0.001; elytra: R=0.998, F9,10= females and individuals in copula we considered 1067.77, P<0.001). Similarly, repeatabilities of to be mated. Captured individuals were stored in asymmetries were also statistically significant, Møller & Zamora-Mun˜oz: Beetle antennal asymmetry 1511

Table I. Summary statistics for the size and asymmetry of three morphological characters in Stenurella melanura

Antenna Tibia Elytra

Males (N=67) Length (mm) 8.95 &0.07 1.66&0.01 6.12&0.05 Signed left"right value (mm) "0.001&0.01 "0.002&0.003 0.001&0.005 Kurtosis 0.521 "0.564 0.007 Skewness "0.324 0.172 "1.164 Absolute asymmetry (mm) 0.081&0.017 0.017&0.002 0.021&0.004 Relative asymmetry 0.011&0.002 0.010&0.001 0.003&0.001 Females (N=31) Length (mm) 7.89 &0.11 1.68 &0.02 6.35 &0.10 Signed left"right value (mm) "0.025&0.022 "0.005&0.003 0.007&0.006 Kurtosis 0.553 "0.520 1.000 Skewness "1.383 0.089 0.326 Absolute asymmetry (mm) 0.058&0.007 0.016&0.003 0.020&0.005 Relative asymmetry 0.007&0.001 0.010&0.002 0.003&0.001

Values are means&. Relative asymmetry is absolute asymmetry divided by length. with the exception of tibia, and measurement differences in variances of left-minus-right errors were smaller than the asymmetries for character values (Sokal & Rohlf 1995). Tests for antenna and elytra (antenna: R=0.714, the experiments were paired t-tests.

F9,10=8.41, P<0.01; tibia: R=0.667, F9,10=5.30, When more than a single test was made on a P<0.1; elytra: R=0.712, F9,10=6.45, P<0.01). data set, we corrected the P-values for multiple Hence, the asymmetry estimates for antenna and tests using sequential Bonferroni adjustment elytra were reliable, while tibial asymmetry was (Holm 1979). excluded from the analyses because of its low All values reported are means&. repeatability. Care was taken to exclude all individuals that had damaged characters. For example, damaged RESULTS or missing antennal segments could readily be determined. We excluded seven individuals with Sexual Size Dimorphism damaged antennae and two with damaged tibia from the calculations for this reason. Sexual size dimorphism was statistically highly significant for the length of antennae (Table I;

t96=8.08, P<0.0001), males having antennae that Statistical Methods were on average 13.4% longer than those of We tested whether the three morphological females. There was also a marginally significant ff characters demonstrated directional asymmetry or di erence in the length of elytra (t96=2.34, anti-symmetry by comparing the frequency distri- P<0.05), with females on average 3.7% larger butions of signed left-minus-right character values than males. Tibia were also slightly longer in with normal distributions using Lilliefors’ test. females than in males, although not significantly  Kurtosis and skewness were also tested against the so (t96=0.76, ). The single most sexually di- null expectation values of zero. Finally, we used morphic character was therefore the length of the one-sample t-tests to test whether mean left- antennae. minus-right character values differed significantly from zero. Tests for Fluctuating Asymmetry We used F-tests to investigate differences in variance between characters and sexes (Sokal & The morphological characters demonstrated Rohlf 1995). Morphological differences between fluctuating asymmetry as shown by the frequency mated and unmated individuals were tested using distributions of signed left-minus-right character unpaired t-tests for length values and F-tests for values not deviating significantly from normal 1512 Animal Behaviour, 54, 6

Table II. Morphology of mated and unmated individual Stenurella melanura from field-collected samples in 1993

Test Character Mated Unmated statistic P

Males Antenna length (mm) 8.90&0.15 8.97&0.09 "0.39  Tibia length (mm) 1.68&0.03 1.69&0.02 "0.92  Elytra length (mm) 6.08&0.10 6.13&0.06 "0.47  Antenna asymmetry (mm2) 0.007 0.022 3.14 <0.01* Elytra asymmetry (mm2) 0.002 0.002 1.03  N 18 49 Females Antenna length (mm) 7.98&0.14 7.77&0.18 0.95  Tibia length (mm) 1.68&0.03 1.69&0.03 "0.31  Elytra length (mm) 6.33&0.14 6.38&0.14 "0.25  Antenna asymmetry (mm2) 0.002 0.007 3.93 <0.01* Elytra asymmetry (mm2) 0.001 0.002 1.98 <0.1 N 18 13

Values are means&. Asymmetries are variances in left-minus-right character values. Test statistics are unpaired t-tests for mean values and F-tests for variances. *P<0.05 after Bonferroni adjustment.

distributions (Lilliefors’ tests: P>0.10). Kurtosis F48,17=2.75, P<0.05; females: F12,17=2.33, and skewness were moderate in most cases with P<0.05). None of the other asymmetries differed only one value differing significantly from zero between mating categories. Individuals of both (elytra in males), but this significance disappeared sexes were therefore favoured by sexual selection after Bonferroni adjustment for multiple tests if they had more symmetric antennae. (Table I). There were only weak relationships between Experimental Results asymmetry and mean size of characters (all ). Given these weak relationships we decided not to When pairs of males were left with single correct the asymmetry estimates for size since this females on a flower, the only difference in mor- would result in only marginal changes in values. phology between winning and losing males was Fluctuating asymmetry in the length of antennal asymmetry (Table III). Winners were antennae was significantly larger than in the more symmetric than losers. In the mate-choice tibia and elytra of both sexes (males: elytra: experiment, the only statistically significant differ-

F66,66=7.02, P<0.05; females: tibia: F30,30=3.51, ence between the two groups was again a smaller P<0.05; elytra: F30,30=7.13, P<0.05). There degree of antennal asymmetry among the females was also a statistically significant difference in chosen by the males (Table III). However, the antennal asymmetry between sexes (F66,30=2.22, significance of this relationship disappeared after P<0.05), but not for elytra. Bonferroni correction.

Morphology and Mating Status DISCUSSION None of the size measures of characters differed significantly between mated and unmated indi- Our main findings can be summarized as follows. viduals in either sex (Table II). If anything, mated The antennae of the cerambycid beetle Stenurella individuals were slightly smaller than unmated melanura were sexually size-dimorphic and exag- individuals in five out of six comparisons. Asym- gerated, and their asymmetry exceeded that of two metry measures differed significantly for antennae sexually size-monomorphic traits. Individuals of in males and females, with mated individuals both sexes with symmetric antennae were more being more symmetric (Table II; males: often involved in mating than other individuals. Møller & Zamora-Mun˜oz: Beetle antennal asymmetry 1513

Table III. Morphology in relation to mating staus among individual Stenurella melanura involved in the two sexual selection experiments

Test Character Mated Unmated statistic P

Male–male competition experiment Antenna length (mm) 8.85&0.16 8.89&0.15 "0.07  Tibia length (mm) 1.69&0.03 1.69&0.02 "0.03  Elytra length (mm) 6.01&0.09 6.03&0.08 "0.05  Antenna asymmetry (mm2) 0.006 0.024 4.32 0.001* Elytra asymmetry (mm2) 0.002 0.002 1.06  N 19 19 Male–choice experiment Antenna length (mm) 7.95&0.17 7.82&0.18 0.22  Tibia length (mm) 1.67&0.03 1.70&0.03 0.14  Elytra length (mm) 6.31&0.15 6.41&0.16 "0.18  Antenna asymmetry (mm2) 0.003 0.011 3.88 <0.05 Elytra asymmetry (mm2) 0.002 0.002 1.11  N 13 13

Values are means&. Asymmetries are variances in left-minus-right character values. Test statistics are paired t-tests for mean values and F-tests for variances. *P<0.05 after Bonferroni adjustment.

There was no current directional sexual selec- Thornhill, in press). Since females sometimes tion for increased length of antennae. This result copulate with more than one male, our results suggests that the sexually different elongation of may be subject to bias. This seems unlikely, how- the antennae in this cerambycid beetle was not ever, given that we sampled beetles throughout maintained by current directional selection for the season, and so should have sampled first as increased length. This result is similar to those of well as later copulating males. Our observations a number of studies of birds that have found no or that cerambycids attempted to grab each other’s very little evidence of current sexual selection for antennae and amputate them also suggest that the increased ornamentation of highly exaggerated beetles attempted to exploit this preference to secondary sexual characters (Møller 1993). Bird their own advantage. Since individuals with sym- species with no or little evidence of directional metric antennae are at a selective advantage, while sexual selection mainly have a very skewed mating individuals with long antennae are not, antennal success characteristic of highly polygynous or amputation may have the effect of increasing lekking mating systems. This contrasts with a antennal asymmetry and hence reduce mating number of species with clear directional sexual success of opponents. Such spiteful behaviour selection and less skewed distributions of mating will be effective only at a local scale, but since success (Møller 1993). The cerambycid beetle S. males are competing with local individuals for melanura may fit this pattern by being highly fertilization of particular females, antennal polygynous as demonstrated by a high frequency damage of competitors may directly influence of multiple mating. potential future fitness loss to other males in Both observational and experimental results the same site. Antennal damage was common confirmed that individual beetles with symmetric in our field samples: 8.1% (N=74) of all males antennae enjoyed a mating advantage. This find- and 6.1% (N=33) of all females had broken ing is in agreement with a large number of obser- or otherwise damaged antennae. Similar examples vational and experimental studies demonstrating of amputation appear to be widespread within an effect of asymmetry on sexual selection in the family (Linsley 1959), suggesting that insects in particular (e.g. Markow & Ricker preferences for antennal symmetry may be 1992; Thornhill 1992a, b; Liggett et al. 1993; widespread. McLachlan & Cant 1995; Simmons 1995) and in Our experiments on male mate choice and in general (for a review see Møller & male–male competition revealed a selective 1514 Animal Behaviour, 54, 6 advantage for symmetric individuals. When two from the Swedish Natural Science Research males fought for access to a female, the male with Council to A.P.M. and a post-doctoral grant from the more symmetric antennae won more often Universidad de Granada to C.Z.M. than expected by chance (Table III). Similarly, when a male was left with two females on a flower, the female chosen by the male had more sym- REFERENCES metric antennae more often than predicted by chance (Table III). Although these simple exper- Andersson, M. 1994. Sexual Selection. Princeton: Princeton University Press. iments do not demonstrate a causal relationship Becker, W. A. 1984. Manual of Quantitative Genetics. between antennal symmetry and sexual selection, 4th edn. Pulman: Academic Enterprises. they support the conclusion from the obser- Brown, L. & Balaton, J. 1986. Behavioral correlates of vational study that symmetry or a correlated male morphology in a horned beetle. Am. Nat., 127, character provided individuals with a selective 565–570. advantage. Conner, J. 1988. Field measurements of natural and sexual selection in the fungus beetle, Bolitotherus Fluctuating asymmetry in antenna length was cornutus. Evolution, 42, 735–749. significantly larger than for the sexually size- Conner, J. 1989. Density-dependent sexual selection in monomorphic character. This pattern is in agree- the fungus beetle Bolitotherus cornutus. Evolution, 43, ment with a growing body of evidence suggesting 1378–1386. that secondary sexual characters have increased Darwin, C. 1871. The Descent of Man and Selection in Relation to Sex asymmetry because of their recent history of . London: John Murray. Falconer, D. S. 1989. Introduction to Quantitative directional selection (Møller & Pomiankowski Genetics. 3rd edn. New York: Longman. 1993a, b, c; review in Møller & Swaddle 1997). Holm, S. 1979. A simple sequentially rejective multiple Directional selection generally results in selection test procedure. Scand. J. Stat., 6, 65–70. against genetic mechanisms that control the Liggett, A. C., Harvey, I. F. & Manning, J. T. 1993. expression of a trait and thereby, as a conse- Fluctuating asymmetry in Scatophaga stercoraria L.: Anim. Behav. quence, also results in increased asymmetry successful males are more symmetrical. , 45, 1041–1043. (Møller & Pomiankowski 1993a, b; review in Linsley, E. G. 1959. Ecology of Cerambycidae. A. Rev. Møller & Swaddle 1997). Stabilizing selection has Entomol., 4, 99–138. the opposite effect by increasing the developmen- McLachlan, A. & Cant, M. 1995. Small males are more tal control of characters (Møller & Swaddle 1997). symmetrical: mating success in the midge Chironomus Antennal asymmetry was unrelated to the length plumosus L. (Diptera: Chironomidae). Anim. Behav., 50, 841–846. of antennae. This absence of a relationship Markow, T. A. & Ricker, J. P. 1992. Male size, between asymmetry and size is also consistent developmental stability, and mating success in natural with findings from studies of sexual selection in populations of three Drosophila species. Heredity, 69, birds (Møller 1993; Møller & Pomiankowski 1122–1127. 1993c). In bird species with a directional mate Michelsen, A. 1963. Observations on the sexual preference for increased size of secondary sexual behaviour of some longicorn beetles, subfamily (Coleoptera, Cerambycidae). Behaviour, characters, the most exaggerated secondary sexual 22, 152–166. characters are also the most symmetric ones, while Møller, A. P. 1993. Patterns of fluctuating asymmetry in there is no relationship between size and asym- sexual ornaments predict female choice. J. evol. Biol., metry in species with an absence of a directional 6, 481–491. preference (Møller 1993). A number of additional Møller, A. P. & Pomiankowski, A. 1993a. Fluctuating Genetica examples consistent with this pattern have been asymmetry and sexual selection. , 89, 267–279. published since 1993 (Møller & Swaddle 1997). Møller, A. P. & Pomiankowski, A. 1993b. Punctuated Stenurella melanura apparently belongs to the equilibria or gradual evolution: fluctuating asymmetry second category of species. and variation in the rate of evolution. J. theor. Biol., 161, 359–367. Møller, A. P. & Pomiankowski, A. 1993c. Why have Behav. Ecol. ACKNOWLEDGMENTS birds got multiple sexual ornaments? Sociobiol., 32, 167–176. Møller, A. P. & Swaddle, J. P. 1997. Asymmetry, J. Soler and R. Thornhill provided constructive Developmental Stability and Evolution. Oxford: criticism. This study was supported by a grant Oxford University Press. Møller & Zamora-Mun˜oz: Beetle antennal asymmetry 1515

Møller, A. P. & Thornhill, R. In press. Bilateral sym- Thornhill, R. 1992a. Female preference for the metry and sexual selection: a meta-analysis. Am. Nat. pheromone of males with low fluctuating asymmetry Parker, G. A. 1970. Sperm competition and its in the Japanese scorpionfly (Panorpa japonica, evolutionary consequences in the insects. Biol. Rev., Mecoptera). Behav. Ecol., 3, 277–283. 45, 525–567. Thornhill, R. 1992b. Fluctuating asymmetry and the Simmons, L. W. 1995. Correlates of male quality in mating system of the Japanese scorpionfly, Panorpa the field cricket, Gryllus campestris L.: age, size, japonica. Anim. Behav., 44, 867–879. and symmetry determine pairing success in field Thornhill, R. & Alcock, J. 1983. The Evolution of populations. Behav. Ecol., 6, 376–381. Insect Mating Systems. Cambridge, Massachusetts: Sokal, R. R. & Rohlf, F. J. 1995. Biometry. 3rd edn. San Harvard University Press. Francisco: Freeman.