Genotypic variation in calling song and female preferences of the field cricket Teleogryllus oceanicus

ANIMAL BEHAVIOUR, 2004, 68, 313e322 doi:10.1016/j.anbehav.2003.12.004

Genotypic variation in calling song and female preferences of the field cricket Teleogryllus oceanicus LEIG H W. SI MMON S

Evolutionary Biology Research Group, School of Animal Biology, University of Western Australia (Received 15 May 2003; initial acceptance 1 August 2003; final acceptance 2 December 2003; MS. number: 7726R)

Geographically isolated populations often show phenotypic divergence in traits important in reproduction. A large proportion of the phenotypic variation in temporal parameters of the calling song of the field cricket Teleogryllus oceanicus is related to geographical location. Similarity between the songs recorded in different populations reflects geographical proximity. I used a common-garden breeding experiment to determine whether differences between the songs of two populations from the extremes of the geographical and phenotypic distribution (Oahu, Hawaii and Cairns, Australia) have a genetic basis. Differences in the total song duration and the proportion of the long-chirp element in the song remained after five generations of common-garden breeding, indicating that the populations had diverged genetically for these traits. Differences in a third song trait, the intervals between sound pulses and chirps, disappeared after common-garden breeding, suggesting that either the difference between populations in these traits represents phenotypic plasticity or the populations converged as a result of adaptation to the laboratory environment. A prospective analysis of the patterns of genetic variation within populations is presented. Full-sib analyses suggested high levels of genetic variability in song traits. Family mean covariance matrices suggested that populations differ in the genetic architecture of their songs. Females from both populations preferred songs with a high proportion of the long-chirp element, and preferences appeared to have high genetic and residual variability, although the sampling variances on these parameters were high. There was little evidence of a correlation between female preference for the long-chirp element and the amount of the long-chirp element produced by their brothers. Ó 2004 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Geographical isolation between populations of organisms can result in the divergence of traits important in reproduction. Population divergence can arise from founder effects and genetic drift (Barton 1989), adaptive responses to local environments (Schluter 1998), or the processes of sexual selection and sexual conflict (Parker & Partridge 1998; Gray & Cade 2000; Panhuis et al. 2001; Blows 2002), and can ultimately lead to speciation (see Howard & Berlocher 1998 and references therein). Phenotypic divergence in sexual signals produced by males and in female preferences for male signals is best illustrated by studies that have compared these traits across geographically isolated populations (Tilley et al. 1990; Hill 1994; Endler & Houde 1995; Gray & Cade 2000; Correspondence: L. W. Simmons, Evolutionary Biology Research Group, Zoology Building, School of Animal Biology (M092), University of Western Australia, Nedlands, WA 6009, Australia (email: lsimmons@ cyllene.uwa.edu.au). 0003e3472/03/$30.00/0

Tregenza et al. 2000). Studies such as these not only contribute to our understanding of the speciation process, but can also prove extremely useful for testing theoretical models of female preference evolution (Houde 1993). Field crickets (Orthoptera: Gryllidae) have proved remarkably good models for studying sexual selection and speciation (Hill et al. 1972; Wagner et al. 1995; Shaw 1996, 2000; Mousseau & Howard 1998; Gray & Cade 1999a, 2000). The field cricket Teleogryllus oceanicus has a particularly wide geographical range, occurring throughout the northern coastal regions of Australia and on many of the south Pacific islands, and was introduced into the Hawaiian archipelago as early as 1877 (Otte & Alexander 1983; Kevan 1990). Geography explains much of the phenotypic variation in male calling song across populations, and similarity between population songs strongly reflects geographical proximity of these populations (Rotenberry et al. 1996; Zuk et al. 2001). Although there are currently no data on genetic distances between

313 Ó 2004 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

314

ANIMAL BEHAVIOUR, 68, 2

populations, these phenotypic patterns are characteristic of gradual differentiation in allopatry (Tilley et al. 1990; Ryan et al. 1996). Previous work suggests that selection pressures acting on calling songs may vary between populations and thus explain some of the observed song divergence. Unlike that of most crickets, the song of T. oceanicus is complex, consisting of two elements, the long chirp (three to eight sound pulses) followed by a series of short chirps (each with two sound pulses). Populations differ along a gradient of increasing total song length, pulse durations and the intervals between pulses, and along a second gradient of an increasing proportion of the long-chirp element relative to short chirps (Zuk et al. 2001). Simmons et al. (2001) found a consistent preference among females for songs containing a greater proportion of the long-chirp element. By using song models that extended well beyond the natural variation in the proportion of the long-chirp element, they also found considerable variation between populations in the shape of female preference functions. Some populations showed little sign of a preference for an increasing proportion of the long-chirp element, others had open-ended preference functions, and still others showed indications of stabilizing selection (Simmons et al. 2001). Population differences in female preference functions are likely to be important for calling song divergence (Ritchie & Phillips 1998). Furthermore, in the Hawaiian archipelago, T. oceanicus is subject to natural selection from an acoustically orienting parasitoid fly, the tachinid Ormia ochracea (Zuk et al. 1995). The probability of becoming infested is a positive function of the duration of the longchirp element, and to a lesser extent a negative function of the interval between pulses within the long-chirp (Zuk et al. 1998). Natural selection against the long-chirp element in Hawaiian populations might influence divergence of the calling song in these populations. Indeed, within the Hawaiian archipelago about 7% of the observed variation in song structure across populations, and in particular variation in the long-chirp element, is associated with variation in the prevalence of the parasitoid; a further 13% is associated with geographical distance (Zuk et al. 1993; Rotenberry et al. 1996). Previous studies of the geographical divergence in calling song and female preferences in T. oceanicus have focused on phenotypic variation in these traits. Phenotypic divergence may or may not reflect genotypic divergence. Where gene flow prevents genotypic divergence, phenotypic plasticity may account for the observed phenotypic differences between populations (Thompson 1999). Although some studies have examined covariation between measures of genetic divergence using marker loci and phenotypic divergence in quantitative traits, the data suggest that these can change independently and the covariation is at best weak (Butlin & Tregenza 1998). Studies that examine the quantitative genetics of divergent sexual traits are rare (reviewed in Ritchie & Phillips 1998). In this study I examined patterns of genotypic variation in calling song and female preferences between and within two populations of T. oceanicus from the extremes of their geographical range and calling song phenotype: Oahu in the Hawaiian archipelago and Cairns

in northeast Australia. I adopted a combination of two approaches (common-garden rearing and full-sib analysis) to ask two questions. First, I asked whether there is evidence of genetic divergence in quantitative traits. If phenotypic differences between populations persist after common-garden rearing then the divergence in traits must be associated with genetic divergence, rather than phenotypic plasticity. Second, I conducted a prospective analysis of the relative magnitudes of genetic variation in male calling song and female preferences, and asked whether there is any evidence for covariation between these traits. Sexual selection can drive divergence and speciation when there are genetic correlations between signals produced by males and female preferences for them. Patterns of genetic variance and covariation can generally be informative of the processes of preference evolution.

METHODS

Common-garden Breeding About 30 adult female crickets were collected from the grounds of the University of Hawaii, Manoa, on the island of Oahu, and about 30 from the edge of a mangrove, 5 km outside of Cairns, Queensland, Australia. Females were allowed to oviposit into moist cottonwool pads, and these eggs were returned to the laboratory. On hatching, nymphs were reared in cages (30 ! 22 ! 20 cm) supplied with cat chow, a constant supply of water (a jar of water upturned on to a cottonwool pad), and misted with water twice weekly. Cardboard egg cartons were provided as shelter and to increase the surface area in the cages. Population densities were maintained at around 100 mixed-sex individuals per cage, with between 10 and 15 cages for each source population. All cages were maintained in a constant-temperature (29(C) room with a 12:12 h light:dark cycle. Egg to adult generation time is around 90 days (Simmons 2001). Once crickets had reached adulthood, the cottonwool pads were collected from their cages weekly and incubated. On hatching, nymphs from all cages within a population (Oahu or Cairns) were mixed and used to establish new cages. Common-garden breeding was thus continued for five generations.

Full-sib Breeding Design When nymphs of the fifth generation were in the penultimate instar, single-sex cages were established for both populations. Once adult, a single male and single female were placed into a family cage measuring 12! 13 cm and 35 cm high, but otherwise provisioned as above. Egg pads were collected from each family cage and incubated. On hatching, nymphs were reared in two family cages in full-sibling groups, at a starting density of 100 individuals per family cage. Sexes were separated at the final instar. Once adult, male and female crickets were housed in individual cages measuring 7!7 cm and 5 cm high. Originally, 20 families from each of the Oahu and

SIMMONS: GENOTYPIC VARIATION IN A SIGNALLING SYSTEM

Cairns populations were established. However, owing to breeding failures of some pairs and mortality of offspring within families, 12 full-sibling families with enough individuals for experiments were available for each population.

Song Recording and Analysis Two weeks after reaching adulthood, males were transferred from the constant-temperature room to a greenhouse subject to ambient temperature and light conditions (isolated males would not call in the constanttemperature room). Individual male cages were positioned about 1 m apart. Twenty minutes after sunset, I recorded a 2-min sequence of each male’s song using a Sony Professional Walkman. The ambient temperature during each recording was noted. For song analysis I used the software package Canary 1.2 (Laboratory of Ornithology, Cornell University, Ithaca, NY, U.S.A.). From the recording of each male an unbroken sequence of 10 songs was identified. From these songs I measured nine song parameters: the total duration of the song, the duration of the long chirp, the duration of pulses contained in the long chirp, the interval between pulses in the long chirp, the total duration of the short-chirp sequence, the duration of each short chirp, the interval between short chirps, the duration of pulses in short chirps, and the interval between pulses within the short chirps (Fig. 1; terminology after Otte 1992). The mean value across 10 songs was used as a measure of each song parameter for each individual male. The temperature during recordings varied from 21 to 27(C. I corrected the song parameters to 25(C using the empirically derived regression equations for temperature on each song parameter.

Phonotaxis Trials Field data suggest that, predominantly, it is the long chirp that is subject to selection by the parasitoid fly (Zuk et al. 1998), and previous phonotaxis trials suggest that the long chirp is the characteristic of the call that is most attractive to females (Pollack & Hoy 1981; Hennig & Weber 1997). Therefore, for simplicity, and for comparison with studies of phonotaxis by females from source PD

IPI

CD

ICI

PD

IPI

populations (Simmons et al. 2001), I restricted my analysis to female preferences for quantitative variation in the relative proportions of the long-chirp element to shortchirp sequences. I used the artificially constructed songs that Simmons et al. (2001) used in their study of geographical variation in female preference functions. Briefly, song models consisted of either 20(80)% or 40(60)% long-chirp (short-chirp) sequences. In nature, the proportion of the long-chirp element ranges from 0.1 to 0.4 (Simmons et al. 2001). The total duration of the song was held constant at 1.36 s, the duration of pulses within long chirps was 41 ms, the interval between pulses within the long chirp was 23 ms, the duration of short chirps 73 ms, the intervals between short chirps 64 ms, the duration of pulses within short chirps 32 ms, and the interval between short-chirp pulses 9 ms. The carrier frequency was 4.68 kHz. Songs were broadcast continuously using the loopback command in SoundEdit 16 (Macromedia Inc., San Francisco, CA, U.S.A.), via an amplifier through two tweeter horns (Radio Shack 40-1228A). These song models elicit positive phonotaxis by female crickets (Simmons et al. 2001). I used a two-speaker paradigm in preference trials. Two song models were broadcast simultaneously from either end of an arena using the two sound channels available in SoundEdit. The arena measured 100!25 cm and was lined with black fan-folded fly screen as cover for crickets during phonotaxis. A small upturned section of an egg carton was placed at the centre of the arena, equidistant from both speakers, and covered by a glass beaker. The intensities of both song models at the release point were set to 70 dB (72 dB peak) using a Radio Shack, Realistic 33-2050 sound meter. A female was placed under the glass beaker 2 min before her trial. Females invariably entered the upturned section of the egg carton. The glass beaker was removed and the calls broadcast. Females were allowed 5 min to respond. Typically, females would exit the egg carton and walk around or over it for a few seconds and pause, sometimes for some minutes, before moving away towards one of the broadcast calls. Once a female had begun walking towards one of the song models she was never seen to orient to the alternative model. Indeed, females invariably entered and remained within the tweeter horn to which they had oriented. The song broadcast from the entered horn was noted. The speakers broadcasting the alternative song models were reversed on each trial and females were used only once. All trials were performed in an anechoic room at 25(C.

RESULTS Long-chirp duration

Short-chirp duration

Song Structure Common-garden analysis

Song duration Figure 1. Stylized sonagram of the calling song of Teleogryllus oceanicus, showing the song parameters measured (PD: pulse duration; IPI: interpulse interval; CD: chirp duration; ICI: interchirp interval; note that PD and IPI were measured for the long-chirp sequence and the short-chirp sequence).

Principal components analysis (PCA) was used to summarize variation in the nine song parameters across all individuals from the Oahu and Cairns populations. PCA yielded three PCs with eigenvalues greater than 1.0, and collectively these PCs explained 74.6% of the variation in song structure (Table 1). Table 1 also gives the

315

ANIMAL BEHAVIOUR, 68, 2

Table 1. Principal components analysis of song parameters of Teleogryllus oceanicus from Oahu and Cairns after five generations of common-garden rearing

4 3

Principal component

2

Eigenvalue % Variance explained Song duration

1

2

3

2.916 32.40

2.269 25.21

1.528 16.98

0.543

0.661*


0.476*

Long chirp Duration Pulse duration Pulse interval

0.608
0.371 0.871*

0.056 0.710*
0.206

0.520* 0.397 0.160

Short chirps Total duration Chirp duration Chirp interval Pulse duration Pulse interval

0.405
0.041 0.785*
0.131 0.746*

0.673* 0.627* 0.062 0.620*
0.221


0.594* 0.237 0.044 0.576* 0.323

Principal component 3

316

1 0 –1 –2 –3 –4 –6

–4

–2

0

2

4

6

Principal component 2

The body of the table contains correlations between principal components and the original song variables. *Correlations greater than 70% of the largest correlation were considered to contribute significantly to the PC (Mardia et al. 1979).

Figure 2. Separation by principal components analysis of songs produced by male Teleogryllus oceanicus from Oahu (B) and Cairns (C) populations after five generations of common-garden rearing.

correlations between each PC and the original variables. No weight is placed on the significance of these individual correlations. Rather, to examine which of the original variables contributed to each PC, I adopted the criterion of Mardia et al. (1979); those traits with correlations greater than 70% of the largest correlation were considered to have contributed significantly to the PC. PC1 was a measure of the intervals between pulses in the long chirp, of the intervals between chirps in the short-chirp sequence, and of the intervals between pulses in the short chirps. PC2 was a measure of increasing song duration, associated with increases in the duration of pulses within the long chirp, the total duration of the short-chirp sequence, and the duration of chirps and pulses within the short-chirp sequence. In contrast, PC3 was a measure of decreasing song duration, associated with increasing duration of the long chirp and decreasing duration of the short-chirp sequence (and thus increasing proportion of the long-chirp element), and increasing duration of short-chirp pulses. Strictly speaking not all of the songs were independent because some came from full siblings. I therefore calculated family mean PC scores for further analysis. A MANOVA using the three PCs entered as dependent variables and source population entered as the main effect revealed significant separation of populations based on song structure (exact F3;20 ¼ 46:23, P ! 0:0001). Univariate tests showed that the Oahu and Cairns populations differed on PC2 and PC3 (PC1: t22 ¼ 1:583, P ¼ 0:128; PC2: t22 ¼ 6:512, P ! 0:0001; PC3: t22 ¼ 7:189, P ! 0:0001; Bonferroni critical P ¼ 0:017). Songs from males reared from Oahu populations had more negative scores on PC2, being shorter, with shorter pulse and short-chirp durations, and a shorter sequence of short chirps, and they had more positive scores on PC3, having a relatively longer long-chirp element than short-chirp sequences (a greater proportion of the long-chirp element; Fig. 2).

Genetic analyses Within populations, quantitative genetic analyses were conducted using the original values of song parameters, following the methods of Becker (1984) and Roff (1997). A MANOVA entering all nine song parameters as dependent variables revealed a significant family effect on offspring song structure within both populations (Cairns: F99;759 ¼ 1:323, P ¼ 0:025; Oahu: F99;787 ¼ 1:574, P ¼ 0:0006). Within the Cairns population only the interval between pulses of the long chirp and the duration of the short chirps showed significant family effects and heritabilities (Table 2). Within the Oahu population the total duration of songs, the duration of the long chirp and the interval between pulses in the long chirp all showed significant family effects and heritabilities (Table 3). Coefficients of genetic variation (CV) were calculated as pffiffiffiffi 100 V =X, where V refers to either additive genetic (A) or residual (R) variance (Houle 1992; Table 4). Sampling variances for CVs were estimated using the method of Lynch & Walsh (1998). Because this was a full-sibling analysis, the estimates of the genetic parameters in Tables 2e4 include additive genetic variance, dominance variance, all maternal effects and common environment effects (Becker 1984; Roff 1997). As such, they overestimate the true genetic variances within populations and represent ‘broad-sense’ estimates (Falconer & MacKay 1996). Nevertheless, the estimates are adequate for the comparisons made here, where interest lies in the relative measures of parameters between rather than within populations. Since both populations were reared in the same common environment, any differences in estimates between populations should reflect genetic differences between them. To test for significant differences between population CVs, the original data were log transformed and used to perform one-tailed variance ratio tests (Zar 1984) contrasting between-family or between-progeny (residual) variances

SIMMONS: GENOTYPIC VARIATION IN A SIGNALLING SYSTEM

Table 2. Results of full-sib analysis of sire effects on the song elements of Teleogryllus oceanicus from Cairns, together with estimates of heritability and their standard errors Variable Song duration Long chirp Duration Pulse duration Pulse interval Short chirps Total duration Chirp duration Chirp interval Pulse duration Pulse interval

Variance among Families Progeny

Mean square 195 299 201 546

Families Progeny Families Progeny Families Progeny

7563 7593 54.08 32.86 106.28 55.63

Families Progeny Families Progeny Families Progeny Families Progeny Families Progeny

155 757 173 860 1397 648 632.11 451.45 33.42 21.51 21.73 16.38

F11,115

P

h2

SE (h2)

0.969

0.478


0.006

0.084

0.996

0.454


0.001

0.086

1.646

0.095

0.116

0.126

1.910

0.048

0.159

0.139

0.896

0.547


0.019

0.082

2.158

0.021

0.227

0.159

1.400

0.182

0.073

0.112

1.554

0.122

0.100

0.121

1.327

0.219

0.060

0.107

Heritabilities and SEs were calculated for unequal family sizes after Roff (1997). SEs are estimates only so that the sib analysis provides a more reliable estimate of the variances from families. Family size varied from 7 to 15 with the weighted family size k ¼ 10:53. Significant family effects are denoted by bold heritabilities.

between populations. Considering only parameters with significant family effects, CVAs for song duration (F11;11 ¼ 1:173, P ¼ 0:193), the duration of the long chirp (F11;11 ¼ 1:618, P ¼ 0:219) and the interval between pulses in the long chirp (F11;11 ¼ 1:046, P ¼ 0:471) were similar for the Oahu and Cairns populations, as were the CVRs (song duration: F115;118 ¼ 1:299, P ¼ 0:078; duration of the

long chirp: F115;118 ¼ 1:16, P ¼ 0:212; interval between pulses in the long chirp: F115;118 ¼ 1:266, P ¼ 0:102). In contrast, for the duration of short chirps, the CVA was higher in Cairns (F11;11 ¼ 3:118, P ¼ 0:036) and the CVR lower in Oahu (F115;118 ¼ 2:258, P ! 0:0001). Only this latter contrast is robust to Bonferroni correction for the eight contrasts made (Pcrit ¼ 0:006).

Table 3. Results of full-sib analysis of sire effects on the song elements of Teleogryllus oceanicus from Oahu, together with estimates of heritability and their standard errors Variable Song duration Long chirp Duration Pulse duration Pulse interval Short chirps Total duration Chirp duration Chirp interval Pulse duration Pulse interval

Variance among Families Progeny

Mean square 200 524 96 341

Families Progeny Families Progeny Families Progeny

19 197 9519 43.76 33.91 180.48 69.83

Families Progeny Families Progeny Families Progeny Families Progeny Families Progeny

125 574 75 734 331.63 185.39 878.35 512.43 29.48 46.45 36.16 26.87

h2

SE (h2)

F11,118

P

2.081

0.027

0.183

0.144

2.017

0.033

0.173

0.141

1.290

0.238

0.053

0.103

2.585

0.006

0.257

0.165

1.658

0.092

0.115

0.124

1.789

0.063

0.137

0.130

1.714

0.078

0.125

0.127

0.635

0.796


0.070

0.057

1.346

0.208

0.062

0.106

Heritabilities and SEs were calculated for unequal family sizes after Roff (1997). SEs are estimates only so that the sib analysis provides a more reliable estimate of the variances from families. Family size varied from 7 to 20 with the weighted family size k ¼ 10:75. Significant family effects are denoted by bold heritabilities.

317

318

ANIMAL BEHAVIOUR, 68, 2

Table 4. Observational coefficients of variation (CV) in the song parameters of Teleogryllus oceanicus from Oahu and Cairns Variable (ms)

Population

MeanGSD

CVA (SE)*

CVR (SE)*

Song duration

Oahu Cairns

1370.65G324.38 1747.81G448.33

32.67 (7.67) 25.28 (5.72)

70.68 (6.51) 79.31 (7.86)

Oahu Cairns Oahu Cairns Oahu Cairns

391.86G101.71 327.88G87.12 38.29G5.90 38.60G5.89 36.08G8.90 31.43G7.75

48.99 26.52 17.28 19.05 37.24 32.80

(12.71) (6.04) (3.79) (4.21) (8.97) (7.71)

69.21 82.06 47.47 45.85 72.29 73.27

(6.30) (8.29) (3.72) (3.60) (6.73) (6.96)

Oahu Cairns Oahu Cairns Oahu Cairns Oahu Cairns Oahu Cairns

978.70G282.81 1426.69G415.07 79.22G14.07 94.84G26.70 94.57G23.24 85.18G21.26 32.94G6.71 32.66G4.75 14.27G5.26 11.89G4.10

36.21 27.66 22.99 39.41 31.34 29.52 16.48 17.70 42.14 39.20

(8.67) (6.33) (5.15) (9.62) (7.31) (6.82) (3.61) (3.89) (10.46) (9.56)

79.95 90.24 53.65 82.84 74.72 77.02 64.58 43.85 113.39 105.09

(7.86) (9.65) (4.38) (8.41) (7.08) (7.51) (5.69) (3.40) (13.95) (12.41)

Long chirp Duration Pulse duration Pulse interval Short chirps Total duration Chirp duration Chirp interval Pulse duration Pulse interval

CVA: additive genetic variance; CVR: residual variance. *Sampling variance of CVs estimated using the method of Lynch & Walsh (1998).

Finally, I estimated the genetic covariance structure for song parameters from family mean correlations (Table 5). Family mean correlations provide conservative estimates of the underlying genetic correlations (Roff 1997; Lynch & Walsh 1998). The genetic architecture of songs in the two populations appeared to differ. For Oahu, increased song duration was correlated with an increase in the duration of the long chirp and the total duration of the short-chirp sequence. In contrast, for Cairns, increased song duration was correlated only with an increase in the total duration of the short-chirp sequence (Table 5). In general, the Oahu songs were characterized by more positive covariances between song parameters (30 of 36 correlations) than

Cairns songs (19 of 36 correlations; chi-square test: c21 ¼ 7:73, P ¼ 0:008).

Female Preference Common-garden analysis Again, not all the data on female preferences are strictly independent because some of the females were full siblings. I therefore calculated the proportion of females within a family choosing the 40(60)% over the 20(80)% long-chirp (short-chirp) song model. The mean proportion of full-sibling females choosing the 40% long-chirp

Table 5. Genetic correlation matrix for song parameters estimated from the family mean correlations for Oahu (above the diagonal) and Cairns (below the diagonal) Long chirp Song duration Song duration

Duration 0.758**

Short chirp

Pulse duration

Pulse interval

Total duration

0.433

0.505

Long chirp Duration Pulse duration Pulse interval

0.321
0.576* 0.610*


0.218 0.704**


0.702*

Short chirp Total duration Chirp duration Chirp interval Pulse duration Pulse interval

0.915***y
0.642* 0.585*
0.576* 0.197

0.197
0.084 0.489 0.108 0.679*


0.444 0.699*
0.295 0.833***y
0.213

0.036

0.870***y
0.271

0.377
0.604* 0.729
0.399 0.576*

Chirp duration

Chirp interval

Pulse duration

Pulse interval

0.980***y

0.421

0.674

0.138

0.235

0.656* 0.474 0.389

0.483 0.456 0.288

0.449 0.689* 0.279


0.039 0.696*
0.067

0.484
0.431 0.709**

0.402

0.657* 0.415

0.195 0.502 0.427


0.342 0.327
0.444 0.079


0.632* 0.593*
0.079


0.295 0.623*

N ¼ 12 sires for both populations. *P%0:05; **P%0:01; ***P%0:001; ysignificant after Bonferoni correction for 36 pairwise correlations in a population.

0.012

0.177 0.168 0.182
0.170

SIMMONS: GENOTYPIC VARIATION IN A SIGNALLING SYSTEM

0.34 0.32

Proportion of long chirp

0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 30

40

50

60

70

80

90

Preference for 40% long chirp Figure 3. Covariation between family mean preference for 40% long chirp over 20% long chirp, and the proportion of the long-chirp element contained in male songs (B: Oahu; C: Cairns).

song model differed significantly from 50% for both populations (Oahu: 0:59 G 0:04; one-sample t test: t11 ¼ 2:468, P ¼ 0:031; Cairns: 0:63 G 0:02; t11 ¼ 6:306, P ! 0:0001) but the difference between populations in this proportion was not significant (t23 ¼ 0:791, P ¼ 0:437).

Genetic analyses A logistic general linear model revealed no significant family effects on female preference for the 40% long-chirp song model for either the Oahu (Wald c211 ¼ 9:724, P ¼ 0:555) or Cairns (Wald c211 ¼ 4:422, P ¼ 0:956) population. Patterns of genetic variance were examined by treating female preference as a threshold trait ( females choosing the 20% long chirp were coded as 0 and females choosing the 40% long chirp were coded as 1) using the ANOVA method (Roff 1997). Calculation of the intraclass correlations yielded negative values for both populations so that h2ð0;1Þ z0. The CVs on the underlying scale were very large for both populations, as were the sampling variances (Oahu: CVAð0;1Þ G SE ¼ 372:6 G 93:9; CVRð0;1Þ ¼ 380:8 G 115:7; Cairns: CVAð0;1Þ ¼ 178:4 G 106:6; CVRð0;1Þ ¼ 284:2 G 37:05). Finally, there was no significant correlation between the family mean female preference for the 40% long-chirp song model and the proportion of the long-chirp element produced by males (see Fig. 3; across populations: r21 ¼
0:228 (lower, upper 95% CI:
0.590, 0.430), P ¼ 0:285; within populations: Oahu: r11 ¼
0:254ð
0:706; 0:344Þ, P ¼ 0:425, Cairns r11 ¼ 0:056ð
0:510; 0:588Þ, P ¼ 0:863). The confidence intervals for these correlations are wide. Family mean correlations can provide conservative estimates of the true genetic correlations between traits (Lynch & Walsh 1998). DISCUSSION The patterns of phenotypic variation revealed by principal components analysis of nine song parameters in this

study were similar to those revealed in the broader study of 18 song parameters from recordings made in the field of 15 geographically separated populations (Zuk et al. 2001). Like Zuk et al.’s (2001) analysis, mine returned PCs that explained variation in intervals (PC1 here, PC3 in Zuk et al.), song duration (PC2 here, PC1 in Zuk et al.) and the duration of long chirps relative to that of short chirps (PC3 here, PC4 in Zuk et al.). Calling songs recorded directly from field populations in Oahu and Cairns differed in song duration (Oahu scored negatively and Cairns positively on Zuk et al.’s PC1), the duration of the long chirp relative to that of the short-chirp sequence (although both populations scored negatively for Zuk et al.’s PC4, in contrast to Cairns, Oahu was close to zero), and the intervals between pulses (Oahu scored positively on Zuk et al.’s PC3 and Cairns negatively; see figure 3 in Zuk et al. 2001). Thus, in contrast to those from Cairns, calling songs recorded in Oahu were shorter, had a greater proportion of the long-chirp element and longer intervals between pulses. After five generations of common-garden rearing, some of these differences between populations persisted. In the current study I found that males bred from the Oahu population had shorter songs (Oahu scored negatively and Cairns positively on PC2), and their songs contained more of the long-chirp element and less of the short-chirp sequence (Oahu scored positively and Cairns negatively on PC3) than those bred from the Cairns population. Thus, population divergence in song duration and the proportion of the long-chirp element in the song must reflect underlying genetic divergence in these song parameters. In contrast, unlike songs recorded from field populations, there were no differences in interpulse or interchirp intervals between commongarden-reared Oahu and Cairns males (PC1 did not differ significantly between populations). The higher values for these song parameters noted from field recordings from Oahu relative to Cairns (Zuk et al. 2001) may reflect phenotypic plasticity in this trait. Alternatively, the convergence of populations after common-garden rearing could have arisen because of adaptation to the common laboratory environment. Phenotypic plasticity seems more likely, given the strong dependence of these variables on temperature (Walker 1962; Bennet-Clark 1989). Within populations there were significant family effects on song structure, and the coefficients of genetic variation were generally high for all song parameters. Consistent with the phenotypic differences in song structure between populations, the family mean correlation matrices suggest that the genetic architecture of song structure differs greatly between the Oahu and Cairns populations. Estimates of genetic variation may be inflated because of the full-sib breeding design; they represent broad-sense estimates which include common environment and dominance variance. Nevertheless, broad-sense estimates can be useful in providing an upper limit to genetic variation in the narrow sense (Falconer & Mackay 1996; Roff 1997). There are two forms of common environment effects that could have contributed to the resemblance between full siblings in this study. First, full siblings may resemble each other simply because they share the same environment during their development to the final instar.

319

320

ANIMAL BEHAVIOUR, 68, 2

However, I randomly drew animals from two family cages so that not all would have shared the same common environment. Furthermore, the common-garden design specifically controlled rearing environments between as well as within families, so that rearing environments experienced by full siblings were the same as those experienced by unrelated individuals. Thus, there is little reason to suspect that full siblings should resemble each other any more than unrelated individuals based on preadult rearing environment alone. Cricket calling song does appear to be sensitive to the environment experienced by adults; thus, temporal parameters of male calls vary with the availability of nutrient resources to the adult male (Wagner & Hoback 1999; Scheuber et al. 2003). In the experiments reported here, all animals were housed individually during adult development and for final recording so that common environment effects that are known to influence male song can be ruled out as contributing to the observed resemblance between relatives. Second, maternal effects can also generate resemblance between relatives. Maternal effects are a special form of common environment effects and can influence morphological traits of offspring. However, maternal effects rarely persist to adulthood (Mousseau & Dingle 1991; Mousseau & Fox 1998) and it is difficult to see how they could influence a behavioural trait such as calling song. Bearing in mind that the estimates for song parameters reported here are upper estimates of the narrow-sense genetic variances, their magnitudes are consistent with the general finding that traits subject to sexual selection have high CVAs (Pomiankowski & Møller 1995). A previous study of the genetics of song parameters in T. oceanicus adopted a hybridization method using T. commodus (Bentley 1971; Bentley & Hoy 1972). These studies concluded that the patterns of inheritance were characteristic of a polygenetic system. More importantly, they suggested that the inheritance of the short-chip and long-chirp sequences may differ. For short-chirp sequence parameters, Bentley & Hoy (1972) found that hybrid male crickets produced song closer to that of their maternal species. Since crickets have an XX/XO sex determination system (Hewitt 1979), song parameters associated with the short-chirp sequence must be the result of genes located on the X chromosome (nongenetic maternal effects seem unlikely given the common laboratory rearing of both species used in Bentley & Hoy’s studies). In contrast, hybrid song parameters associated with the long chirp were intermediate between maternal and paternal species, suggesting they are influenced by genes located on the autosomes. If males inherit their short-chirp song parameters from their mother, female preferences for short-chirp components of the song are unlikely to evolve under indirect models of preference evolution. Consistent with this conclusion, the long chirp appears to be the only important parameter of the song for female phonotaxis (Pollack & Hoy 1981; Hennig & Weber 1997), and females derived directly from field populations from Oahu and Cairns prefer songs with a greater proportion of the longchirp element (Simmons et al. 2001). However, parasitoid predation in Oahu exerts selection against the duration

of the long chirp and the interval between pulses in the long chirp (Zuk et al. 1998). All else being equal, we might predict that the patterns of genetic variation in these traits would differ between populations, given the known differences in selection pressures. Indeed, selection pressures via parasitoid predation have been argued to maintain genetic variation in the nightly calling duration of Gryllus texensis (Gray & Cade 1999b). Although the broad-sense heritabilities for long-chirp parameters tended to be higher, and the heritability of long-chirp duration was significant only for the Oahu population, there were no significant between-population differences in the CVAs or CVRs for any long-chirp element. More parasitoidinfested and uninfested populations need to be analysed before any conclusions can be made regarding the role of parasitoids in explaining levels of genetic variation between populations of T. oceanicus. Preference functions derived from females in Oahu suggest that selection on the proportion of the long-chirp element may be stabilizing, in contrast with Cairns where it is clearly open ended for an increased long-chirp element (Simmons et al. 2001). Stabilizing female preference and opposing selection via parasitoid infestation both have the potential to result in the greater canalization of long-chirp song elements in Oahu than in Cairns. There appears to be stabilizing selection via female preference for pulse rate in G. texensis, and as expected where stabilizing selection is strong and persistent (Houle 1992; Pomiankowski & Møller 1995), CVA for pulse rate is low at just 3.2% (Gray & Cade 1999a, 2000). Again, an examination of the CVAs for long-chirp song elements of T. oceanicus gives no indication of lower genetic variability in Oahu than in Cairns. On the contrary, there appears to be generally high genetic variability for these traits, consistent with persistent directional selection in both populations (Pomiankowski & Møller 1995). The female preferences for increased proportion of the long-chirp element persisted after five generations of common-garden rearing, suggesting that they have a genetic basis. The strength of this preference did not differ between populations. As for song parameters, the full-sib breeding design means that the genetic parameters for female preference are likely to be overestimates of the underlying genetic variability. The high CVRs and their sampling variances are consistent with the generally high phenotypic variability in female preference functions found across populations (Simmons et al. 2001). Nevertheless, the equally high CVAs suggest that these preferences are likely to have some underlying genetic variation. Indeed, hybridization studies of Teleogryllus have shown female preferences to have a genetic basis (Hoy 1974; Hoy et al. 1977), and quantitative genetic studies of female preferences in G. texensis have revealed significant heritability (Gray & Cade 1999a, 2000). Female hybrids between T. oceanicus and T. commodus respond more to songs of their male siblings than to males of the reciprocal cross (Hoy et al. 1977). Coupled with the observation that male hybrids produce songs with parameters of the long chirp intermediate between those of the parental species, these observations led Hoy et al. (1977) to conclude that male songs and female preferences

SIMMONS: GENOTYPIC VARIATION IN A SIGNALLING SYSTEM

in Teleogryllus were genetically coupled (that is, one or more genes exert pleiotropic effects on male songs and female preferences). Pleiotropy would be expected to drive the coevolution of male songs and female preferences so that population divergence in both of these traits would potentially facilitate speciation (Butlin & Ritchie 1989). However, across 11 populations there was no covariation between female preference for the long chirp and the proportion of the long-chirp element in male songs, suggesting that these traits may not be coevolving (Simmons et al. 2001). The intermediate songs and preferences of Teleogryllus hybrids could arise with independent genetic control of these traits. Similarly, apparent genetic coupling may reflect linkage disequilibrium between genes that independently influence male song and female preferences (Butlin & Ritchie 1989), as has been found in G. texensis (Gray & Cade 1999a, 2000). Linkage disequilibrium will decline by 50% per generation when normal mate choice is disrupted (Bakker & Pomiankowski 1995; Gray & Cade 1999a). Thus, the laboratory populations studied here can tell us little regarding linkage disequilibrium. In contrast, covariation resulting from pleiotropy should be robust to common-garden rearing and is expected to generate a strong correlation between male song traits and female preferences. Indeed, there was strong and significant evidence for pleiotropy between song parameters. The correlations between family mean proportion of the long-chirp element and preference for the long chirp for T. oceanicus were small and not significant for either population, or across populations. The statistical power of the correlations was undoubtedly weak, and the experiment cannot detect linkage disequilibrium in natural populations. Nevertheless, the experiment was powerful enough to detect pleiotropy between song parameters. Given the general lack of phenotypic covariation between the proportion of the long-chirp element in male song and female preferences for long chirps across populations (Simmons et al. 2001), the combined evidence lends little support for the hypothesis that male songs and female preferences have coevolved in this species. Acknowledgments I thank Marlene Zuk for collecting cricket eggs on Oahu, Kylie Shau-Gaull, Oliver Berry and Julie Wernham for assistance with song analysis, and John Hunt for assistance with song recordings and phonotaxis trials. Mike Ritchie, Mike Johnson and Marlene Zuk provided valuable comments on the manuscript. This research was supported by the Australian Research Council. References Bakker, T. C. M. & Pomiankowski, A. 1995. The genetic basis of female mate preferences. Journal of Evolutionary Biology, 8, 129e171. Barton, N. H. 1989. Founder effect speciation. In: Speciation and its Consequences (Ed. by D. Otte & J. A. Endler), pp. 229e256. Sunderland, Massachusetts: Sinauer.

Becker, W. A. 1984. Manual of Quantitative Genetics. Washington: Pullman. Bennet-Clark, H. C. 1989. Songs and the physics of sound production. In: Cricket Behavior and Neurobiology (Ed. by F. Huber, T. E. Moore & W. Loher), pp. 227e261. Ithaca: Cornell University Press. Bentley, D. R. 1971. Genetic control of an insect neuronal network. Science, 174, 1139e1141. Bentley, D. R. & Hoy, R. R. 1972. Genetic control of the neuronal network generating cricket (Teleogryllus gryllus) song patterns. Animal Behaviour, 20, 478e492. Blows, M. W. 2002. Interaction between natural and sexual selection during the evolution of mate recognition. Proceedings of the Royal Society of London, Series B, 269, 1113e1118. Butlin, R. K. & Ritchie, M. G. 1989. Genetic coupling in mate recognition systems: what is the evidence? Biological Journal of the Linnean Society, 37, 237e246. Butlin, R. K. & Tregenza, T. 1998. Levels of genetic polymorphism: marker loci versus quantitative traits. Philosophical Transactions of the Royal Society of London, Series B, 353, 187e198. Endler, J. A. & Houde, A. E. 1995. Geographic variation in female preferences for male traits in Poecilia reticulata. Evolution, 49, 456e468. Falconer, D. S. & Mackay, T. F. C. 1996. Introduction to Quantitative Genetics. Harlow: Longman. Gray, D. A. & Cade, W. H. 1999a. Quantitative genetics of sexual selection in the field cricket, Gryllus integer. Evolution, 53, 848e854. Gray, D. A. & Cade, W. H. 1999b. Sex, death and genetic variation: natural and sexual selection on cricket song. Proceedings of the Royal Society of London, Series B, 266, 707e709. Gray, D. A. & Cade, W. H. 2000. Sexual selection and speciation in field crickets. Proceedings of the National Academy of Sciences, U.S.A., 97, 14449e14454. Hennig, R. M. & Weber, T. 1997. Filtering of temporal parameters of the calling song by females of two closely related species: a behavioral analysis. Journal of Comparative Physiology A, 180, 621e630. Hewitt, G. M. 1979. Insecta 1. Orthoptera: Grasshoppers and Crickets. 1st edn. Berlin: Gebru ¨ der Borntraeger. Hill, G. E. 1994. Geographic variation in male ornamentation and female mate preference in the house finch: a comparative test of models of sexual selection. Behavioral Ecology, 5, 64e73. Hill, K. G., Loftus-Hills, J. J. & Gartside, D. F. 1972. Pre-mating isolation between the Australian field crickets Teleogryllus commodus and T. oceanicus (Orthoptera: Gryllidae). Australian Journal of Zoology, 20, 153e163. Houde, A. E. 1993. Evolution by sexual selection: what can population comparisons tell us?. American Naturalist, 141, 796e803. Houle, D. 1992. Comparing evolvability and variability of quantitative traits. Genetics, 130, 195e204. Howard, D. J. & Berlocher, S. H. 1998. Endless Forms: Species and Speciation. Oxford: Oxford University Press. Hoy, R. R. 1974. Genetic control of acoustic behavior in crickets. American Zoologist, 14, 1067e1080. Hoy, R. R., Hahn, J. & Paul, R. C. 1977. Hybrid cricket auditory behavior: evidence for genetic coupling in animal communication. Science, 195, 82e84. Kevan, D. K. M. 1990. Introduced grasshoppers and crickets in Micronesia. Boletin de Sanidad Vegetal, 20, 105e123. Lynch, M. & Walsh, B. 1998. Genetics and Analysis of Quantitative Traits. Sunderland, Massachusetts: Sinauer. Mardia, K. V., Kent, J. T. & Bibby, J. M. 1979. Multivariate Analysis. London: Academic Press.

321

322

ANIMAL BEHAVIOUR, 68, 2

Mousseau, T. A. & Dingle, H. 1991. Maternal effects in insect life histories. Annual Review of Entomology, 36, 511e534. Mousseau, T. A. & Fox, C. W. 1998. The adaptive significance of maternal effects. Trends in Ecology and Evolution, 13, 403e407. Mousseau, T. A. & Howard, D. J. 1998. Genetic variation in cricket calling song across a hybrid zone between two sibling species. Evolution, 52, 1104e1110. Otte, D. 1992. Evolution of cricket songs. Journal of Orthopteran Research, 1, 25e49. Otte, D. & Alexander, R. D. 1983. The Australian Crickets (Orthoptera: Gryllidae). Philadelphia: Academy of Natural Sciences of Philadelphia. Panhuis, T. M., Butlin, R., Zuk, M. & Tregenza, T. 2001. Sexual selection and speciation. Trends in Ecology and Evolution, 16, 364e371. Parker, G. A. & Partridge, L. 1998. Sexual conflict and speciation. Philosophical Transactions of the Royal Society of London, Series B, 353, 261e274. Pollack, G. S. & Hoy, R. R. 1981. Phonotaxis to individual rhythmic components of a complex cricket calling song. Journal of Comparative Physiology A, 144, 367e373. Pomiankowski, A. & Møller, A. P. 1995. A resolution of the lek paradox. Proceedings of the Royal Society of London, Series B, 260, 21e29. Ritchie, M. G. & Phillips, S. D. F. 1998. The genetics of sexual isolation. In: Endless Forms: Species and Speciation (Ed. by D. J. Howard & S. H. Berlocher), pp. 291e308. Oxford: Oxford University Press. Roff, D. A. 1997. Evolutionary Quantitative Genetics. New York: Chapman & Hall. Rotenberry, J. T., Zuk, M., Simmons, L. W. & Hayes, C. 1996. Phonotactic parasitoids and cricket song structure: an evaluation of alternative hypotheses. Evolutionary Ecology, 10, 233e243. Ryan, M. J., Rand, A. S. & Weigt, L. A. 1996. Allozyme and advertisement call variation in the tu´ngara frog, Physalaemus pustulosus. Evolution, 50, 2435e2453. Scheuber, H., Jacot, A. & Brinkhof, M. W. G. 2003. Condition dependence of a multicomponent sexual signal in the field cricket Gryllus campestris. Animal Behaviour, 65, 721e727. Schluter, D. 1998. Ecological causes of speciation. In: Endless Forms: Species and Speciation (Ed. by D. J. Howard & S. H. Berlocher), pp. 114e129. Oxford: Oxford University Press. Shaw, K. L. 1996. Polygenic inheritance of a behavioral phenotype: interspecific genetics of song in the Hawaiian cricket genus Laupala. Evolution, 50, 256e266.

Shaw, K. L. 2000. Interspecific genetics of mate recognition: inheritance of female acoustic preferences in Hawaiian crickets. Evolution, 54, 1303e1312. Simmons, L. W. 2001. The evolution of polyandry: an examination of the genetic incompatibility and good-sperm hypotheses. Journal of Evolutionary Biology, 14, 585e594. Simmons, L. W., Zuk, M. & Rotenberry, J. T. 2001. Geographic variation in female preference functions and male songs of the field cricket Teleogryllus oceanicus. Evolution, 55, 1386e1394. Thompson, D. B. 1999. Different spatial scales of natural selection and gene flow: the evolution of behavioural geographic variation and phenotypic plasticity. In: Geographic Variation in Behaviour (Ed. by S. A. Foster & J. A. Endler), pp. 33e51. Oxford: Oxford University Press. Tilley, S. G., Verrell, P. A. & Arnold, S. J. 1990. Correspondence between sexual isolation and allozyme differentiation: a test in the salamander Desmognathus ochrophaeus. Proceedings of the National Academy of Sciences, U.S.A., 87, 2715e2719. Tregenza, T., Pritchard, V. L. & Butlin, R. K. 2000. The origins of premating isolation: testing hypotheses in the grasshopper Chorthippus parallelus. Evolution, 54, 1687e1698. Wagner, W. E. & Hoback, W. W. 1999. Nutritional effects on male calling behaviour in the variable cricket. Animal Behaviour, 57, 89e95. Wagner, W. E., Murray, A.-M. & Cade, W. H. 1995. Phenotypic variation in the mating preferences of female crickets, Gryllus integer. Animal Behaviour, 49, 1269e1281. Walker, T. J. 1962. Factors responsible for intraspecific variation in the calling songs of crickets. Evolution, 16, 407e428. Zar, J. H. 1984. Biostatistical Analysis. Englewood Cliffs, New Jersey: Prentice-Hall. Zuk, M., Simmons, L. W. & Cupp, L. 1993. Calling characteristics of parasitized and unparasitized populations of the field cricket Teleogryllus oceanicus. Behavioural Ecology and Sociobiology, 33, 339e343. Zuk, M., Simmons, L. W. & Rotenberry, J. T. 1995. Acoustically-orienting parasitoids in calling and silent males of the field cricket Teleogryllus oceanicus. Ecological Entomology, 20, 380e383. Zuk, M., Rotenberry, J. T. & Simmons, L. W. 1998. Calling songs of field crickets (Teleogryllus oceanicus) with and without phonotactic parasitoid infection. Evolution, 52, 166e171. Zuk, M., Rotenberry, J. T. & Simons, L. W. 2001. Geographic variation in calling song of the field cricket Teleogryllus oceanicus: the importance of spatial scale. Journal of Evolutionary Biology, 14, 731e741.