- Email: [email protected]
The quantitative genetics of sound production in Gryllus firmus
Anim. Behav., 1992,44, 823-832
The quantitative genetics of sound production in Gryllusfirmus K A R E N L Y N N W E B B & D E R E K A. R O F F Department of Biology, McGUl University, Montreal, Quebec, H3A 1B1 Canada
(Received 29 August 1991; initial acceptance 11 November 1991; final acceptance 25 January 1992; MS. number: A6144)
Abstract. The species-specific calling songs of male crickets are used by females for species recognition and mate choice. Heritability of variation of morphological structures involved in song production, components of the calling song, and body size were estimated for the cricket Gryllusfirmus. The heritabilities of morphological structures range from 0'42 to 0.73, while those of the song components range from 0.10 to 0-35. Comparison between sire and dam components suggests that there may be non-additive and/or maternal effects. As a result of the low correlation between body size and song components, it is unlikely that female G.firmus use the calling song to assess male body size or wing morph (micropterous or macropterous).
In Orthoptera, male song is a major component of the mate recognition system (Regen 1913; Walker 1957; Popov & Shuvalov 1977; Pollack & Hoy 1981; Weber et al. 1981; Pollack 1982; Thorson et al. 1982; Stout et al. 1983; Doherty 1985a, b). Song may be used to convey information not only about the location of the caller but also about the caller's physical characteristics, such as size: for example, Simmons (1988b) found a significant increase in the pulse-repetition rate and a decrease in chirp duration with increasing male size for the cricket Gryllus bimaculatus. That these song components can be assessed by the female cricket is evidenced by the observation that females preferred the calls of large males in playback experiments (Simmons 1988b). Components of calling song found to be important in many species of crickets for species recognition and female phonotaxis are number of pulses per chirp, pulse length, pulse repetition rate within chirps (l/pulse period), chirp length, chirp repetition rate (1/chirp period) and most intense frequency component or carrier frequency (Doherty & Hoy 1985). Females appear to evaluate several of these components (reviewed by Doherty & Hoy 1985); if females differ in their evaluation of song attractiveness, and song components are heritable, variations in song may persist. Hedrick (1988) found that calling-bout length, in the cricket Gryllus integer, was significantly heritable (h2=0.75, P<0-0005), and virgin females were preferentially attracted to long calling-bout 0003-3472/92/110823 + 10 $08.00/0
lengths (Hedrick 1986). However, little is known, overall, about genetic variation in the components of the calling song within Orthopteran species, and their effects on female mate choice (Crankshaw 1979; Hedrick 1986, 1988; Simmons 1987, 1988b). Butlin & Hewitt (1986) examined the heritability of morphology and song characteristics of the grasshopper Chorthippus brunneus. Heritabilities for morphological features were generally higher than song components (h2= -0.21-0.28 for song components, and h 2 = 0-16-0.89 for morphological features). However, none of the heritability estimates for song components were significant, though, this may be a result of small sample sizes ( 9 0 < N < 9 5 , where N = individuals per sample) with the consequently high standard errors (Butlin & Hewitt 1986). The sand cricket, Gryllus firmus, is a large ground-dwelling cricket native to the south-eastern United States (Alexander 1968), and is well suited as a subject for the analysis of heritability of both morphological and song components. It is easy to rear in large numbers and its generation time is relatively short (approximately 60 days from egg to adult at 28~ Males of G.firmus produce three types of calls: a calling song that attracts females from a distance, a courtship song that is used by the male when in very close proximity to the female, and an aggressive song used primarily in male-male interactions. The courtship song may be an important secondary mechanism by which species recognition may occur, and is required by females
9 1992The Association for the Study of Animal Behaviour 823
Animal Behaviour, 44, 5
824
Chirp length
'1 Pulse period
0
50
IO0
I
I
I
Pulse length
Time (ms) Figure 1. Simulated oscillogram of one chirp of a Gryllusfirmus calling song.
of some species to mate (Crankshaw 1979; Burk 1982). However, the courtship song is more complex than the calling song, and difficult to quantify (see Alexander 1961; also personal observation). In this study our analysis was restricted to the primary mechanism of species discrimination, the calling song. The calling song of G.firmus is simple in structure consisting of distinct chirps of three to five pulses per chirp (Fig. 1). The calling song is produced by lifting the tegmina at a 3 0 4 5 ~ angle from the abdomen, and rubbing the scraper of the lowermost tegmen over the stridulatory file on the lower surface of the uppermost tegmen (usually the right tegmen) upon wing closure (Rakshpal 1960). In this paper we report on an experiment designed to determine narrow sense heritabilities and phenotypic correlations between the following species-specific components of the calling song: number of pulses per chirp; pulse length; pulse rate; chirp length; and carrier frequency. The chirp repetition rate, determined as 1/inter-chirp period, was excluded as it has been found to be highly variable for individuals, coefficients of variation ranging from 19 to 54% (N=78 individuals), making measurements unreliable. This variability may have been due in part to the inability to distinguish
between inter-chirp period within a calling bout and the chirp period between consecutive calling bouts. The close proximity of the female might have contributed to variation in chirp rate (J. A. Doherty, personal communication). However, in the single case where we recorded a male that called in the absence of a female the chirp repetition rate was also highly variable, with a coefficient of variation of 39% ( N = 188 chirps). This highly variable chirp rate is consistent with Alexander's (1961) grouping of G.firmus in the category of crickets with irregularly delivered chirps. On this basis, Alexander (1961, page 194) suggested that 'chirp rate is a character of minor importance'. In addition to the song components we estimated the heritability of the sound producing structures: the number of teeth on the stridulatory file; the length of the stridulatory file covered by stridulatory teeth; and the resonator area. Although the harp is the radiator of the calling song carrier frequency of Gryllus males (Michelsen & Nocke 1974), it was not measured, as no definite upper boundary exists (Fig. 2). The resonator contributes to the overall power output of the sound produced, contributing to the efficiency of the tegmen as a sound producing organ (Lutz & Hicks 1930; Michelsen & Nocke 1974). In addition to heritability estimates, we measured the
Webb & Roff: Quantitative geneties of song
825
calling song. We examined the likelihood that female G.firmus can use components of the calling song to assess size by computing the phenotypic correlation between body size, as measured by femur length, and the song components. Additionally, we estimated the phenotypic correlations between femur length and the other morphological structures examined (number of stridulatory teeth, file length, and resonator area) and the heritability of femur length. Gryllus firmus is dimorphic with respect to flight wing length, comprising macropterous and micropterous forms, only the former being able to fly. Although macropterous females suffer a reproductive cost, in terms of a delay in age at first reproduction and a reduced fecundity, the benefit of producing a few dispersing offspring is likely to offset that cost, leading to the maintenance of the two forms in the population (Roff 1984). Differences in morphology, and variations in song between morphs was examined to determine whether the calling song of G.firmus males could be used to assess wing morph, and thereby enable females to preferentially select males of a particular wing morph by using only information within the calling song. METHODS
0
2
4
Lenglh (mm)
Figure 2. Right tegmen of Gryllusfirmus, ventral view. Most of the lateral part, L, of the tegmen is removed; A: stridulatory vein; the part of vein A having teeth is considered the stridulatory file; R: resonator (hatched region); H: harp (cross-hatched region). phenotypic correlations between song and soundproducing structures to determine the effects of variation in these structures on the sound produced. The study by Simmons (1988b) suggests that female G. bimaculatus may prefer large males, and discriminate size based on characteristics of the
The crickets used in the present study were descended from an original group of approximately 40 individuals (about 20 females) collected, 7 years before the start of this study, from a single location in northern Florida (Roff 1986). We obtained heritability estimates using a halfsibling design (Falconer 1981). The parental generation consisted of offspring from a large control stock (approximately 200 breeding adults) maintained during the course of a selection experiment (see Roff 1990). We raised 60 nymphs per cage at 30~ with a 15:9 h light:dark photoperiod in 29 x 19 x 13 cm polystyrene mouse cages covered with a glass sheet. Air circulation was provided by an approximately 1-cm diameter hole on each side of the cages, covered with wire mesh to prevent nymphs from escaping. Nymphs were fed Purina Rabbit Chow ad libitum, and water was supplied through a shell vial with a cotton plug. Once moult to the adult stage had commenced, we inspected cages daily and removed new adults. Male and female virgins obtained in this manner were chosen randomly as parents for the study. The
826
Animal Behaviour, 44, 5
analysis of Robertson (1959) suggests that for a half-sibling design, four dams per sire, and family sizes of 10 individuals are optimal. In the present experiment, logistic considerations dictated that we use eight sires and four dams per sire, giving a total of 32 families, and a maximum sample size of 320 offspring. Once females had mated (spermataphore accepted from the male and visible), we provided each mated dam with an oviposition box (plastic sandwich box filled with moist sterilized soil). When the eggs started to hatch, the boxes were removed and kept at 28~ Boxes were checked daily, and all nymphs removed. To permit a nested analysis for the estimation of common environment effects, we set up from each dam's offspring two nymph cages, comprising 60 nymphs each. Nymphs were raised at 28~ with a 15:9 light:dark photoperiod in white 4-1itre plastic buckets (Vulcan Industrial Packaging, Ltd). Air circulation was provided by two 1-cm holes covered with wire mesh in the bucket lids. Nymphs were fed Purina Rabbit Chow ad libitum, and supplied water through a shell vial with a cotton stopper. Ten adult males of known age were collected from each dam family (attempts were made to collect five from each cage, otherwise four and six were collected). The adult males were individually caged to minimize the risk of damage in aggressive interactions, and for individual identification. We housed males individually in cages by inserting a loop of wire mesh (approximate height of 3 cm) secured with masking tape between the top and bottom of a 150 x 15 mm plastic disposable petri dish. Water was supplied by a cheesecloth wick which extended through a hole in the floor of the cage to a reservoir of water contained in a second petri dish. Purina Rabbit Chow was supplied ad libitum. Cade & Wyatt (1984) found that calling by the adult male, on average, occurs by age 7 days post-eclosion at 19-26~ for G. integer, G. pennsylvanicus, G. veletis, and Teleogryllus africanus. Therefore, the first recording attempts were made at the minimum age of 7 days. In the laboratory, G.firmus males call at or near dawn (onset of the light phase of the light cycle) and continue to call throughout the day, with little or no calling during the night (personal observation). We began recording sessions each day within an hour of the onset of the light phase of the light cycle. We made recordings using a Uher 4200 Report recorder with a Uher M538 microphone, at a tape
speed of 9.5 cm/s. To minimize extraneous noise, we made all recordings in an anechoic chamber. The temperature was maintained between 30~ and 33~ as Doherty (1985b) found little change in song characteristics with temperature changes above 24~ in the chirping species G. bimaculatus. Heating of the anechoic chamber above 22~ was provided by an infrared 250 Philips heating lamp suspended approximately 50 cm above the cricket cage, and intensity was adjusted with a light dimmer, as needed, to maintain the temperature within the specified range. Attempts were made to acquire a minimum of 100-200 chirps per individual. To increase the likelihood of the male calling at the time of recording, an adult female was placed in the, otherwise isolated, male's cage. When a male and female come into very close contact the male switches to its courtship song, but courting males, interrupted by the departure of the female, may start calling again (Alexander 1962), although this calling may be sporadic (J. A. Doherty, personal communication, personal observation). Therefore, whenever the male switched to his courtship song the pair was disturbed to prevent mating (after which the male remains silent), and to re-initiate calling. The calling songs were analysed at Brock University, St Catharines, Ontario, using the Brock University Digital Signal Processing Network. Of the 320 individuals 170 songs were recorded and analysed (except frequency, where N = 163), which was one to 10 individuals per dam family ( N = 32 dam families, mean = 5.3, median = 5.5, SE= 0'4). The hind-left femur (the right, if the left was missing) of each male was mounted between two slides. We measured femur length by projecting onto a Houston Instrument Hipad@ Digitizer with an Omega Enlarger B22. The uppermost tegmen was removed. A clear replica was made of the lower surface to facilitate the measurement of the stridulatory file length. We measured only the part of the stridulatory vein covered by stridulatory teeth (Fig. 2), which could be seen easily on the enlarged image of a tegmen replica. The original tegmina are too opaque for this purpose. We made replicas by dissolving an acetate sheet in acetone, stirring the mixture until it was smooth and had the consistency of nail varnish and sweeping this solution onto the ventral surface of the tegmen. This solution dried within 1-2min and was easily removed from the original tegmen without any apparent damage to the tegmen or significant
Webb & Roff: Quantitative genetics of song shrinkage of the replica. This method proved better than the use of nail varnish (Ragge 1969), which takes longer to dry and often leads to the damage of the original tegmen, and the replica, when they are separated. We counted the number of teeth on the stridulatory file using a Carl Zeiss light microscope on the F40/65 objective with Kpl wl0 x/18 oculars. We measured the length of the stridulatory file by projecting the image of the tegmen replica on a digitizer pad, using a Ken-A-Vision Microprojector, model x-1000-1. The resonator area was measured using the original tegmen on the same digitizer as the femur lengths. For the morphological features, a total of 320 offspring were measured (except the resonator area, where N = 319 were measured, due to half the tegmen being missing from one individual), comprising 10 male offspring from each of the dam families (for resonator area nine individuals were measured in one of the dam families). Formulas for the calculation of the heritabilities and their standard errors were taken from Becker (1985, pp. 57-65). Statistical analyses were done using Statgraphics version 2.6, SAS and SYSTAT.
827
Table I. Character means (___sn), and coefficient of variation (CV) Character
Mean ( _ sE)
CV (%)
12'8 ___0.04 4.1 + 0.02 5'7 _ 0.04 188.5 +0.5
5 7 12 4
3.7 __+0.02 24'6 + 0.3 20-9__+0.1 153.7 + 1.4 4.0 _ 0.02
8 14 9 12 5
Morphological(N= 320) Femur length (mm) File length (ram) Resonator area (mm2)* Tooth number Song (N= 170) Pulses/chirp Pulse length (ms) Pulse rate (pulses/s) Chirp length (ms) Frequency (kHz)t *N=319. t N = 163. statistically possible, resulting from negative estimates of variance components. Strictly speaking, in such cases h 2 was undefined since no valid estimate could be made. For both the morphological and song components h2dam were considerably larger than hasiresuggesting the presence of maternal and/ or non-additive effects.
RESULTS
Phenotypic Correlations A nested analysis for the estimation of common environment effects revealed no significant cage effects. Therefore, the data from individual cages were pooled for each family. The experimental animals consisted of 176 micropterous (short-winged), 136 macropterous (long-winged, capable of flight) individuals, and eight individuals of intermediate wing length.
Heritability Both morphological features and song components exhibited moderate ranges in phenotypic variation, with coefficients of variation ranging from 4 to 14% (Table I). Heritability estimates for all the morphological features were significant for the dam component and the sire + dam estimates (Table II). However, the heritability estimates of only one of the song components, the intra-chirp pulse rate, was significant (full-sibling estimate; Table II). The heritabilities of two song components (pulses per chirp and length) estimated from the sire component were negative. Negative heritabilities have no biological meaning but were
Correlations between song components The song components, chirp length (el), pulse rate (pr), number of pulses per chirp (ppe), and pulse length (pl) are mathematically related according to the formula cl = [(1000/pr)(ppc- 1)] + pl The examination of a simulated oscillogram of a single chirp helps illustrate how this formula was derived (Fig. 1). Multiplying the intra-chirp pulse period (pulse period (ms)= 1000/pulse rate (number/s)) by the number of pulses minus 1, results in the chirp length minus 1 full pulse length. Simply adding pulse length to the above determines the chirp length. Considering pairwise combinations of song components, the above relationship suggests that these components will be correlated in the manner shown in the upper right-hand corner of Table III. As predicted, chirp length was positively correlated with the number of pulses per chirp and pulse length, and negatively correlated with pulse rate (Table III). However, the remaining correlations
Animal Behaviour, 44, 5
828
Table II. Heritability estimates (_+SE) h2sire
h2dam
h2sire + dam
1.01+0.36" 0.42 + 0.20* 0-85• 0.31* 0.65+0.27*
0.64+0.19" 0.55 __+0.21* 0.73 + 0.24* 0.42+0.14"
0.66+0.39 0.22 • 0.26 0.52+0.33 0.61+0.37 0.18__+0-28
0.19-t-0.15 0.22 __+0.15 0.35+0.17" 0.23___0.15 0.10__+0.13
Morphological
Femur length File length Resonator area Tooth number
0.26+0-29 0"67__+0.40 0.62 • 0.43 0.18•
Song
Pulses/chirp Pulse length Pulse rate Chirp length Frequency
-0.28• 0-23 __+0.22 0.18+0.24 -0-15__0.12 0.20+0.14
*P < 0.05.
Table III. Partial correlation, matrix of song components (adjusted for wing morph), N= 163
Pulses/chirp Pulses/chirp Pulse length Pulse rate Chirp length Frequency
Pulselength
Pulserate
Chirp length
-
+
+
+
+
- 0.66* 0.40*
-0-35*
0.30* -0.10 0.77* -0.11
-0.48* 0,59" -0.33*
Predicted correlations from a pairwise analysis are shown in the upper triangle. *P<0.001 (probability levelsare not Bonferroni-adjusted).
were all opposite in sign to those predicted by a pairwise analysis. This can be explained by the dominating influence of correlations with chirp length (cl). For example, consider the correlation between pulse length (pl) and the number of pulses per chirp (ppc). Rearranging the previous equation gives pl = cl - (1000/pr)(ppc- 1) A simple pairwise analysis suggests that the number of pulses per chirp and pulse length will be negatively correlated. But, chirp length and the number of pulses per chirp are strongly positively correlated. Therefore, an increase in the number of pulses per chirp is accomplished by an increase in chirp length; if the latter increase exceeds the increase in the number of pulses per chirp, then pulse length will increase, not decrease. A significant positive correlation between pulse length and the number of pulses per chirp was, in fact, observed.
Correlations between morphology and song components Phenotypic correlations between morphological structures were all large and highly significant (Table IV). However, correlations between song components and morphological structures were very weak. Only the number of pulses per chirp and frequency were significantly correlated with femur length (Table IV). However, neither correlation remained significant when the significance level was corrected for the total number of tests (Bonferroniadjustment), and therefore it is unlikely that the calling song of the male was used by G. firmus females to assess male body size.
Variations in morphology between wing morphs Intermediate individuals were omitted from the analysis as they comprised too small a sample size for statistically meaningful comparisons.
Webb & Roff: Quantitative genetics of song
829
IV. Partial phenotypic correlation (adjusted for wing morph) of morphological characters on morphological (N= 319) and song components (N= 162) Table
Femur length File length Resonator area Tooth number Pulses/chirp Pulse length Pulse rate Chirp length Frequency
File length
0.64** 0.59** 0.32** -
0.17*
0-03 - 0.02 - 0"08 - 0.19*
Resonator area
0.63** 0.56**
Tooth number
0-33**
- 0.18 *
- 0.19*
0.12 0.09 - 0.15 -- 0.07
0.07 - 0.01 - 0.11 - 0.07
--0.03 0.06 0.10 -0.06 -0.11
*P<0.05; **P<0.001 (probability levels are not Bonferroni-adjusted). V. Character means (+ SE) of morphological and song components by wing morph Table
Character Morphological
Femur length (mm) File length (mm) Resonator area (mm z) Tooth number Song Pulses/chirp Pulse length (ms) Pulse rate (pulses/s) Chirp length (ms) Frequency (kHz)
Micropterous
Macropterous
( N = 176)
( N = 136)
12-8 _+0-06 4.0 + 0.02 5.3 -I-0.05" 188.4 -I-0.6 (N = 89) 3.7 ___0.03 23.3 + 0.3 21.0__+0.2 153-0 + 2.0 4.0 + 0'02t
12.9 + 0.05 4.3 ___0.02 6.1 + 0-05 188.7 __+0.5 (N= 77) 3.7 __+0.03 26.1 _+0.4 20.8 __+0.2 154.8 + 2.2 4.0 + 0.02:~
P
0.210 < 0.001 < 0.001 0.748 0.933 < 0.001 0.503 0.558 0.822
Probability level reflects ANOVA by wing morph. *N= 175. tN=85. :~N=75.
Macropterous individuals had significantly longer files and larger resonator areas than micropterous individuals, but did not differ in either femur length or tooth number (Table V). O f particular interest was the observed difference in overall morphology between micropterous and macropterous individuals. N o t only did the length of the flight wings differ, but the morphology of the tegmina and the correlations between the morphology of the song-producing structures and body size differed significantly in several respects (covariance analysis, no heterogeneity in slopes). For a given femur length, micropterous individuals had smaller file lengths: ( N = 3 1 2 , F-ratios: Fmorph = 139"626, P<0"001; Ff . . . . l e n g t h = 2 1 4 " 3 3 6 , P<0-001) Y=O'21X+O'23D+l'4, where X=
femur length, Y=file length, D = 0 for microptery and D = 1 for macroptery). Similarly, micropterous individuals have smaller resonator areas ( N = 311, F-ratios: Fmorph = 173' 781, P < 0"001 ; F r. . . . length = 161.590, P < 0-001): Y = 0-51 X + 0 . 7 4 D - 1.2, where X = femur length, Y = resonator area, and D is as above). Micropterous males have more teeth than macropterous males for a given file length (N = 312, F-ratios: Fmorph=29"816, P
Animal Behaviour, 44, 5
830
reasons for this difference may be more complex than simply the result of micropterous males having smaller files (see Webb 1991). There is considerable overlap in pulse length between morphs (Micropterous range = 13"5-30"9, X'_+SE= 23'3 + 0"3; Macropterous range=18-1-37'8, ~'-+SE= 26' 1 + 0.4), making pulse length a poor predictor of wing morph.
DISCUSSION Morphological features typically possess higher heritabilities than behavioural traits (Mousseau & Roff 1987; Roff & Mousseau 1987). This general observation is supported by the present analysis, the heritabilities for the morphological features, (hZsire+dam 0"42 -- 0"73), being significantly higher than the song components, (h2slro+d,m=0"10 0"35, Table II). All of the morphological features possessed significant heritability (h2sire+dam), while one of the song components, pulse rate, was significant. The generally larger values of hZaamcompared with h2si,e suggests that maternal and/or nonadditive effects may be important. It would be premature to assume that the heritabilities of the other song components are zero, as the standard errors are relatively large, though approximately equal to those of the morphological features. If we assume the heritability estimates are not zero, and estimate heritability to be 0.20 (approximately the mean h2sjro+aam of the nonsignificant estimates), then the family size required for significance to be detected may be determined. The expected intra-class correlation for halfsiblings is t = h2/4 (Robertson 1959, 1960; Falconer 1981), and the required intra-elass correlation is t =0.05. The optimal number of offspring per dam is determined by N=l/(2t) (Robertson 1959), resulting in N = 10. A mean of 5"3_+0"4 offspring per dam family was measured for the song components. Thus to detect a heritability of 0.20 requires twice the number of offspring per dam as measured in the present experiment. Femur length, an index of body size, has a high heritability (Table II). File length and tooth number, which are closely associated with important features of the calling song, possess lower heritabilities than the resonator area, although the differences were not statistically significant. Resonator area is associated with song intensity (Michelsen & Nocke 1974). However, due to =
attenuation, song intensity may not be of importance in female choice (Simmons 1988b). Trade-offs between traits chosen by females may contribute to the genetic variation in these traits. For example, femur length may be undergoing directional selection, with larger males being favoured by females (Simmons 1987). But the song components important in species recognition and female phonotaxis, and their associated structures, may be under stabilizing selection (Butlin & Hewitt 1986). As song-producing structures and body size are correlated (Table IV), changes in one may affect the other and, in turn, influence calling song. Thus, at some point trade-offs may occur between preferred song components and body size, contributing to the maintenance of genetic variation in the calling song. Female G. bimaculatus prefer larger males, and there exists the possibility that the calling song, used to attract females, could contain information about the size of the advertiser (Simmons 1987, 1988a, b). Simmons (1988b) found a significant increase in pulse rate and a decrease in chirp length with increasing body size (with pronotum width used as the index of body size) for G. bimaculatus. The reduction in chirp length was attributed to a negative correlation in pulse rate. Pulse rate was determined to be unaffected by distance degradation, suggesting females could detect differences in pulse rate between males (Simmons 1988b). In G.firmus there was no significant correlation between pulse rate and femur length (used as the index of body size). Only the number of pulses per chirp and frequency form weak negative correlations (N=162; t = - 0 . 1 7 , P<0'05, and r = - 0-19, P < 0'05, respectively) with femur length, and are, therefore, not likely to be used by females to assess body size. Note that using pronotum width as a measure of body size (Simmons 1988b), instead of femur length, would not alter the results for G.firmus, as there is a highly significant positive correlation between femur length and pronotum width for G.firmus (N= 106, r=0.77, P<0.001, D. A. Roff, unpunished data). Micropterous and macropterous individuals differed not only in the length of their flight wings, but also in other morphological features. Micropterous males possessed significantly shorter files, and smaller resonator areas, but did not differ in either femur length or tooth number, from macropterous males. Micropterous individuals possessed five more teeth than macropterous individuals with the
Webb & Roff." Quantitative genetics o f song same file length. Carrier frequency, determined by the number of teeth struck per unit time, should differ between morphs if the number of teeth struck differs, but the speed of wing closure does not. Howe;eer, no significant difference in frequency was observed for a given file length. It is, therefore, likely that the effects, if any, on the speed of wing closure, or carrier frequency of such a small difference in the number o f teeth, were too small to be measured. Phenotypic correlations between song and morphology suggest that changes in one may result in changes in the other. This is contingent on the genetic correlations between the traits; accurate estimates of these require more accurate estimates of the heritabilities of the song components. Differences in the tegminal morphology of micropterous and macropterous individuals resulted in measurable differences in the calling song. It is not unreasonable to expect morphological changes in the sound-producing structures accompanied by changes in the calling song with selection on wing morph. Further research, however, is required to determine whether such variation in song could influence mate choice by females.
ACKNOWLEDGMENTS We gratefully acknowledge the assistance of the following people: S. David, P. Shannon, D r R. Lemon, M. Bradford and D. Webb. Special thanks to Dr W. Cade and T. M a c D o n a l d for their integral roles in s o n g analysis. The comments of Drs G. Pollack, D. Kramer, A. Hedrick and J. Doherty on earlier versions of this paper are gratefully acknowledged. This work was supported by a grant from national Sciences and Engineering Research Council of Canada to D.A.R.
REFERENCES Alexander, R. D. 1961. Aggressiveness, territoriality, and sexual behavior in field crickets (Orthoptera: Gryllidae). Behaviour, 17, 130-223. Alexander, R. D. 1962. Evolutionary change in cricket acoustical communication. Evolution, 16, 443-467. Alexander, R. D. 1968. Life cycle origins, specialization and related phenomena in crickets. Q. Rev. Biol., 43, 1-41. Becker, W. A. 1985. ManualofQuantitative Genetics. Ann Arbor, Michigan: McNaughton & Gunn. Burk, T. 1982. Male aggression and female choice in a field cricket (Teleogryllus oceanicus): the importance
831
of courtship song. In: Orthopteran Mating Systems: Sexual Competition in a Diverse Group of Insects (Ed. by D. T. Gwynne & G. K. Morris), pp. 97-119. Boulder, Colorado: Westview Press. Butlin, R. K. & Hewitt, G. M. 1986. Heritability estimates for characters under sexual selection in the grasshopper, Chorthippus brunneus. AnOn. Behav., 34, t256-t261. Cade, W. H. & Wyatt, D. 1984. Factors affecting calling behaviour in the field crickets, Teleogryllus and Gryllus (age, weight, density and parasites). Behaviour, 88, 61-75. Crankshaw, O. S. 1979. Female choice in relation to calling and courtship songs in Acheta domestieus. Anim. Behav., 27, 1274-1275. Doherty, J. A. 1985a. Temperature coupling and tradeoff phenomena in the acoustic communication system of the cricket, Gryllus bimaculatus DeGeer. J. exp. Biol., 114, 17-35. Doherty, J. A. 1985b. Trade-off phenomena in calling song recognition and phonotaxis in the cricket, Gryllus bimaeulatus (Orthoptera, Gryllidae). J. eomp. Physiol., 156, 787-801. Doherty, J. A. & Hoy, R. 1985. Communication in insects. III. The auditory behavior of crickets: some views of genetic coupling, song recognition, and predator detection. Q. Rev. Biol., 60(4), 457-472. Falconer, D. S. 1981. Introduction to Quantitative Genetics. 2nd edn. London: Lougman. Hedrick, A. V. 1986. Female preferences for male calling bout duration in a field cricket. Behav. Ecol. Soeiobiol., 19, 73 77. Hedrick, A. V. 1988. Female choice and the heritability of attractive male traits: an empirical study. Am. Nat., 132(2), 267-276. Lutz, F. E. & Hicks, W. P. 1930. An analysis by movietone of a cricket's chirp (Gryllus assimilis). Am. Mus. Novit., 420, 1-14. Michelsen, A. & Nocke, H. 1974. Biophysical aspects of sound communication in insects. Adv. Insect Physiol., 18, 247-296. Mousseau, T. A. & Roff, D. A. 1987. Natural selection and the heritability of fitness components. Heredity, 59, 181-197. Pollack, G. S. 1982. Sexual differences in cricket calling song recognition. J. comp. Physiol., 146, 217-221. Pollack, G. S. & Hoy, R. 1981. Phonotaxis to individual rhythmic components of a complex cricket calling song. J. comp. Physiol., 144, 367-373. Popov, A. V. & Shuvalov, V. F. 1977. Phonotactic behavior of crickets, or. comp. Physiol., 119, 111-126. Ragge, D. R. 1969. A revision of the African species of Pseudorhynchus Serville (Orthoptera: Tettigoniidae). Bull. Brit. Mus. nat. Hist. Entomol., 23, 167-190. Rakshpal, R. 1960. Sound producing organs and mechanism of sound production in the field crickets of the genus Acheta fabricus (Orthoptera, Gryllidae). Can. J. Zool., 38, 499-507. Regen, J. 1913. Uber die anlickung des weibchens von Gryllus campestris L. durch telephonnisch/ibertragene stridulatioslaute des mannchens. Pfliigers Arch. ges. Physiol., 155, 193-200.
832
Animal Behaviour, 44, 5
Robertson, A. 1959. Experimental design in the evaluation of genetic parameters. Biometrics, 15(2), 219-226. Robertson, A. 1960. Experimental design on the measurement of heritabilities and genetic correlations. In:
Biometrical Genetics. Proceedings of an International Symposium Sponsored by the Biometrics Society and the International Union of Biometrical Sciences (Ed. by O. Kempthorn), pp. 101-106. London: Pergamon Press. Roff, D. A. 1984. The cost of being able to fly: a study of wingpolymorphismin two species of crickets. Oecologia (BerL), 63, 30-37. Roff, D. A. 1986. The genetic basis of wing dimorphism in the sand cricket, Gryllusfirmus and its relevance to the evolution of wing dimorphism in insects. Heredity, 57, 221-231. Roff, D. A. 1990. Selection for changes in the incidence of wing dimorphism in Gryllus firmus. Heredity, 65, 163-168. Roff, D. A. & Mousseau, T. A. 1987. Quantitative genetics and fitness: lessons from Drosophila. Heredity, 58, 103-118. Simmons, L. W. 1987. Heritability of a male character chosen by females of the field cricket, Gryllus bimaculatus. Behav. Ecol. Sociobiol., 21, 129-133.
Simmons, L. W. 1988a. Male size, mating potential and lifetime reproductive success in the field cricket, Gryllus bimaculatus (De Geer). Anita. Behav., 36, 372-379. Simmons, L. W. 1988b. The calling song of the field cricket, Gryllus bimaculatus (De Geer): constraints on transmission and its role in inter-male competition and female choice. Anita. Behav., 36, 380-394. Stout, J. F., DeHaan, C. H. & McGhee, R. W. 1983. Attractiveness of the male Acheta domesticus calling song to females. J. comp. PhysioL, 153, 509-521. Thorson, J., Weber, T. & Huber, F. 1982. Auditory behavior of the cricket. II. Simplicity of the calling-song recognition in Gryllus, and anomalous phonotaxis at abnormal carrier frequencies. J. comp. Physiol., 146, 361 378. Walker, T. J. 1957. Specificity in the response of the female tree crickets (Orthoptera, Gryllidae, Oecanthinae) to calling songs of the males. Ann. Entomol. Soc. Am., 50, 626636. Webb, K. L. 1991. The quantitative genetics of sound production in Gryllus firmus. M.Sc. thesis. McGill University. Weber, T., Thorson, J. & Huber, F. 1981. Auditory behavior of the cricket. I. Dynamics of compensated walking and discrimination paradigms on the Kramer treadmill. J. comp. Physiol., 141,215-232.
The quantitative genetics of sound production in Gryllusfirmus K A R E N L Y N N W E B B & D E R E K A. R O F F Department of Biology, McGUl University, Montreal, Quebec, H3A 1B1 Canada
(Received 29 August 1991; initial acceptance 11 November 1991; final acceptance 25 January 1992; MS. number: A6144)
Abstract. The species-specific calling songs of male crickets are used by females for species recognition and mate choice. Heritability of variation of morphological structures involved in song production, components of the calling song, and body size were estimated for the cricket Gryllusfirmus. The heritabilities of morphological structures range from 0'42 to 0.73, while those of the song components range from 0.10 to 0-35. Comparison between sire and dam components suggests that there may be non-additive and/or maternal effects. As a result of the low correlation between body size and song components, it is unlikely that female G.firmus use the calling song to assess male body size or wing morph (micropterous or macropterous).
In Orthoptera, male song is a major component of the mate recognition system (Regen 1913; Walker 1957; Popov & Shuvalov 1977; Pollack & Hoy 1981; Weber et al. 1981; Pollack 1982; Thorson et al. 1982; Stout et al. 1983; Doherty 1985a, b). Song may be used to convey information not only about the location of the caller but also about the caller's physical characteristics, such as size: for example, Simmons (1988b) found a significant increase in the pulse-repetition rate and a decrease in chirp duration with increasing male size for the cricket Gryllus bimaculatus. That these song components can be assessed by the female cricket is evidenced by the observation that females preferred the calls of large males in playback experiments (Simmons 1988b). Components of calling song found to be important in many species of crickets for species recognition and female phonotaxis are number of pulses per chirp, pulse length, pulse repetition rate within chirps (l/pulse period), chirp length, chirp repetition rate (1/chirp period) and most intense frequency component or carrier frequency (Doherty & Hoy 1985). Females appear to evaluate several of these components (reviewed by Doherty & Hoy 1985); if females differ in their evaluation of song attractiveness, and song components are heritable, variations in song may persist. Hedrick (1988) found that calling-bout length, in the cricket Gryllus integer, was significantly heritable (h2=0.75, P<0-0005), and virgin females were preferentially attracted to long calling-bout 0003-3472/92/110823 + 10 $08.00/0
lengths (Hedrick 1986). However, little is known, overall, about genetic variation in the components of the calling song within Orthopteran species, and their effects on female mate choice (Crankshaw 1979; Hedrick 1986, 1988; Simmons 1987, 1988b). Butlin & Hewitt (1986) examined the heritability of morphology and song characteristics of the grasshopper Chorthippus brunneus. Heritabilities for morphological features were generally higher than song components (h2= -0.21-0.28 for song components, and h 2 = 0-16-0.89 for morphological features). However, none of the heritability estimates for song components were significant, though, this may be a result of small sample sizes ( 9 0 < N < 9 5 , where N = individuals per sample) with the consequently high standard errors (Butlin & Hewitt 1986). The sand cricket, Gryllus firmus, is a large ground-dwelling cricket native to the south-eastern United States (Alexander 1968), and is well suited as a subject for the analysis of heritability of both morphological and song components. It is easy to rear in large numbers and its generation time is relatively short (approximately 60 days from egg to adult at 28~ Males of G.firmus produce three types of calls: a calling song that attracts females from a distance, a courtship song that is used by the male when in very close proximity to the female, and an aggressive song used primarily in male-male interactions. The courtship song may be an important secondary mechanism by which species recognition may occur, and is required by females
9 1992The Association for the Study of Animal Behaviour 823
Animal Behaviour, 44, 5
824
Chirp length
'1 Pulse period
0
50
IO0
I
I
I
Pulse length
Time (ms) Figure 1. Simulated oscillogram of one chirp of a Gryllusfirmus calling song.
of some species to mate (Crankshaw 1979; Burk 1982). However, the courtship song is more complex than the calling song, and difficult to quantify (see Alexander 1961; also personal observation). In this study our analysis was restricted to the primary mechanism of species discrimination, the calling song. The calling song of G.firmus is simple in structure consisting of distinct chirps of three to five pulses per chirp (Fig. 1). The calling song is produced by lifting the tegmina at a 3 0 4 5 ~ angle from the abdomen, and rubbing the scraper of the lowermost tegmen over the stridulatory file on the lower surface of the uppermost tegmen (usually the right tegmen) upon wing closure (Rakshpal 1960). In this paper we report on an experiment designed to determine narrow sense heritabilities and phenotypic correlations between the following species-specific components of the calling song: number of pulses per chirp; pulse length; pulse rate; chirp length; and carrier frequency. The chirp repetition rate, determined as 1/inter-chirp period, was excluded as it has been found to be highly variable for individuals, coefficients of variation ranging from 19 to 54% (N=78 individuals), making measurements unreliable. This variability may have been due in part to the inability to distinguish
between inter-chirp period within a calling bout and the chirp period between consecutive calling bouts. The close proximity of the female might have contributed to variation in chirp rate (J. A. Doherty, personal communication). However, in the single case where we recorded a male that called in the absence of a female the chirp repetition rate was also highly variable, with a coefficient of variation of 39% ( N = 188 chirps). This highly variable chirp rate is consistent with Alexander's (1961) grouping of G.firmus in the category of crickets with irregularly delivered chirps. On this basis, Alexander (1961, page 194) suggested that 'chirp rate is a character of minor importance'. In addition to the song components we estimated the heritability of the sound producing structures: the number of teeth on the stridulatory file; the length of the stridulatory file covered by stridulatory teeth; and the resonator area. Although the harp is the radiator of the calling song carrier frequency of Gryllus males (Michelsen & Nocke 1974), it was not measured, as no definite upper boundary exists (Fig. 2). The resonator contributes to the overall power output of the sound produced, contributing to the efficiency of the tegmen as a sound producing organ (Lutz & Hicks 1930; Michelsen & Nocke 1974). In addition to heritability estimates, we measured the
Webb & Roff: Quantitative geneties of song
825
calling song. We examined the likelihood that female G.firmus can use components of the calling song to assess size by computing the phenotypic correlation between body size, as measured by femur length, and the song components. Additionally, we estimated the phenotypic correlations between femur length and the other morphological structures examined (number of stridulatory teeth, file length, and resonator area) and the heritability of femur length. Gryllus firmus is dimorphic with respect to flight wing length, comprising macropterous and micropterous forms, only the former being able to fly. Although macropterous females suffer a reproductive cost, in terms of a delay in age at first reproduction and a reduced fecundity, the benefit of producing a few dispersing offspring is likely to offset that cost, leading to the maintenance of the two forms in the population (Roff 1984). Differences in morphology, and variations in song between morphs was examined to determine whether the calling song of G.firmus males could be used to assess wing morph, and thereby enable females to preferentially select males of a particular wing morph by using only information within the calling song. METHODS
0
2
4
Lenglh (mm)
Figure 2. Right tegmen of Gryllusfirmus, ventral view. Most of the lateral part, L, of the tegmen is removed; A: stridulatory vein; the part of vein A having teeth is considered the stridulatory file; R: resonator (hatched region); H: harp (cross-hatched region). phenotypic correlations between song and soundproducing structures to determine the effects of variation in these structures on the sound produced. The study by Simmons (1988b) suggests that female G. bimaculatus may prefer large males, and discriminate size based on characteristics of the
The crickets used in the present study were descended from an original group of approximately 40 individuals (about 20 females) collected, 7 years before the start of this study, from a single location in northern Florida (Roff 1986). We obtained heritability estimates using a halfsibling design (Falconer 1981). The parental generation consisted of offspring from a large control stock (approximately 200 breeding adults) maintained during the course of a selection experiment (see Roff 1990). We raised 60 nymphs per cage at 30~ with a 15:9 h light:dark photoperiod in 29 x 19 x 13 cm polystyrene mouse cages covered with a glass sheet. Air circulation was provided by an approximately 1-cm diameter hole on each side of the cages, covered with wire mesh to prevent nymphs from escaping. Nymphs were fed Purina Rabbit Chow ad libitum, and water was supplied through a shell vial with a cotton plug. Once moult to the adult stage had commenced, we inspected cages daily and removed new adults. Male and female virgins obtained in this manner were chosen randomly as parents for the study. The
826
Animal Behaviour, 44, 5
analysis of Robertson (1959) suggests that for a half-sibling design, four dams per sire, and family sizes of 10 individuals are optimal. In the present experiment, logistic considerations dictated that we use eight sires and four dams per sire, giving a total of 32 families, and a maximum sample size of 320 offspring. Once females had mated (spermataphore accepted from the male and visible), we provided each mated dam with an oviposition box (plastic sandwich box filled with moist sterilized soil). When the eggs started to hatch, the boxes were removed and kept at 28~ Boxes were checked daily, and all nymphs removed. To permit a nested analysis for the estimation of common environment effects, we set up from each dam's offspring two nymph cages, comprising 60 nymphs each. Nymphs were raised at 28~ with a 15:9 light:dark photoperiod in white 4-1itre plastic buckets (Vulcan Industrial Packaging, Ltd). Air circulation was provided by two 1-cm holes covered with wire mesh in the bucket lids. Nymphs were fed Purina Rabbit Chow ad libitum, and supplied water through a shell vial with a cotton stopper. Ten adult males of known age were collected from each dam family (attempts were made to collect five from each cage, otherwise four and six were collected). The adult males were individually caged to minimize the risk of damage in aggressive interactions, and for individual identification. We housed males individually in cages by inserting a loop of wire mesh (approximate height of 3 cm) secured with masking tape between the top and bottom of a 150 x 15 mm plastic disposable petri dish. Water was supplied by a cheesecloth wick which extended through a hole in the floor of the cage to a reservoir of water contained in a second petri dish. Purina Rabbit Chow was supplied ad libitum. Cade & Wyatt (1984) found that calling by the adult male, on average, occurs by age 7 days post-eclosion at 19-26~ for G. integer, G. pennsylvanicus, G. veletis, and Teleogryllus africanus. Therefore, the first recording attempts were made at the minimum age of 7 days. In the laboratory, G.firmus males call at or near dawn (onset of the light phase of the light cycle) and continue to call throughout the day, with little or no calling during the night (personal observation). We began recording sessions each day within an hour of the onset of the light phase of the light cycle. We made recordings using a Uher 4200 Report recorder with a Uher M538 microphone, at a tape
speed of 9.5 cm/s. To minimize extraneous noise, we made all recordings in an anechoic chamber. The temperature was maintained between 30~ and 33~ as Doherty (1985b) found little change in song characteristics with temperature changes above 24~ in the chirping species G. bimaculatus. Heating of the anechoic chamber above 22~ was provided by an infrared 250 Philips heating lamp suspended approximately 50 cm above the cricket cage, and intensity was adjusted with a light dimmer, as needed, to maintain the temperature within the specified range. Attempts were made to acquire a minimum of 100-200 chirps per individual. To increase the likelihood of the male calling at the time of recording, an adult female was placed in the, otherwise isolated, male's cage. When a male and female come into very close contact the male switches to its courtship song, but courting males, interrupted by the departure of the female, may start calling again (Alexander 1962), although this calling may be sporadic (J. A. Doherty, personal communication, personal observation). Therefore, whenever the male switched to his courtship song the pair was disturbed to prevent mating (after which the male remains silent), and to re-initiate calling. The calling songs were analysed at Brock University, St Catharines, Ontario, using the Brock University Digital Signal Processing Network. Of the 320 individuals 170 songs were recorded and analysed (except frequency, where N = 163), which was one to 10 individuals per dam family ( N = 32 dam families, mean = 5.3, median = 5.5, SE= 0'4). The hind-left femur (the right, if the left was missing) of each male was mounted between two slides. We measured femur length by projecting onto a Houston Instrument Hipad@ Digitizer with an Omega Enlarger B22. The uppermost tegmen was removed. A clear replica was made of the lower surface to facilitate the measurement of the stridulatory file length. We measured only the part of the stridulatory vein covered by stridulatory teeth (Fig. 2), which could be seen easily on the enlarged image of a tegmen replica. The original tegmina are too opaque for this purpose. We made replicas by dissolving an acetate sheet in acetone, stirring the mixture until it was smooth and had the consistency of nail varnish and sweeping this solution onto the ventral surface of the tegmen. This solution dried within 1-2min and was easily removed from the original tegmen without any apparent damage to the tegmen or significant
Webb & Roff: Quantitative genetics of song shrinkage of the replica. This method proved better than the use of nail varnish (Ragge 1969), which takes longer to dry and often leads to the damage of the original tegmen, and the replica, when they are separated. We counted the number of teeth on the stridulatory file using a Carl Zeiss light microscope on the F40/65 objective with Kpl wl0 x/18 oculars. We measured the length of the stridulatory file by projecting the image of the tegmen replica on a digitizer pad, using a Ken-A-Vision Microprojector, model x-1000-1. The resonator area was measured using the original tegmen on the same digitizer as the femur lengths. For the morphological features, a total of 320 offspring were measured (except the resonator area, where N = 319 were measured, due to half the tegmen being missing from one individual), comprising 10 male offspring from each of the dam families (for resonator area nine individuals were measured in one of the dam families). Formulas for the calculation of the heritabilities and their standard errors were taken from Becker (1985, pp. 57-65). Statistical analyses were done using Statgraphics version 2.6, SAS and SYSTAT.
827
Table I. Character means (___sn), and coefficient of variation (CV) Character
Mean ( _ sE)
CV (%)
12'8 ___0.04 4.1 + 0.02 5'7 _ 0.04 188.5 +0.5
5 7 12 4
3.7 __+0.02 24'6 + 0.3 20-9__+0.1 153.7 + 1.4 4.0 _ 0.02
8 14 9 12 5
Morphological(N= 320) Femur length (mm) File length (ram) Resonator area (mm2)* Tooth number Song (N= 170) Pulses/chirp Pulse length (ms) Pulse rate (pulses/s) Chirp length (ms) Frequency (kHz)t *N=319. t N = 163. statistically possible, resulting from negative estimates of variance components. Strictly speaking, in such cases h 2 was undefined since no valid estimate could be made. For both the morphological and song components h2dam were considerably larger than hasiresuggesting the presence of maternal and/ or non-additive effects.
RESULTS
Phenotypic Correlations A nested analysis for the estimation of common environment effects revealed no significant cage effects. Therefore, the data from individual cages were pooled for each family. The experimental animals consisted of 176 micropterous (short-winged), 136 macropterous (long-winged, capable of flight) individuals, and eight individuals of intermediate wing length.
Heritability Both morphological features and song components exhibited moderate ranges in phenotypic variation, with coefficients of variation ranging from 4 to 14% (Table I). Heritability estimates for all the morphological features were significant for the dam component and the sire + dam estimates (Table II). However, the heritability estimates of only one of the song components, the intra-chirp pulse rate, was significant (full-sibling estimate; Table II). The heritabilities of two song components (pulses per chirp and length) estimated from the sire component were negative. Negative heritabilities have no biological meaning but were
Correlations between song components The song components, chirp length (el), pulse rate (pr), number of pulses per chirp (ppe), and pulse length (pl) are mathematically related according to the formula cl = [(1000/pr)(ppc- 1)] + pl The examination of a simulated oscillogram of a single chirp helps illustrate how this formula was derived (Fig. 1). Multiplying the intra-chirp pulse period (pulse period (ms)= 1000/pulse rate (number/s)) by the number of pulses minus 1, results in the chirp length minus 1 full pulse length. Simply adding pulse length to the above determines the chirp length. Considering pairwise combinations of song components, the above relationship suggests that these components will be correlated in the manner shown in the upper right-hand corner of Table III. As predicted, chirp length was positively correlated with the number of pulses per chirp and pulse length, and negatively correlated with pulse rate (Table III). However, the remaining correlations
Animal Behaviour, 44, 5
828
Table II. Heritability estimates (_+SE) h2sire
h2dam
h2sire + dam
1.01+0.36" 0.42 + 0.20* 0-85• 0.31* 0.65+0.27*
0.64+0.19" 0.55 __+0.21* 0.73 + 0.24* 0.42+0.14"
0.66+0.39 0.22 • 0.26 0.52+0.33 0.61+0.37 0.18__+0-28
0.19-t-0.15 0.22 __+0.15 0.35+0.17" 0.23___0.15 0.10__+0.13
Morphological
Femur length File length Resonator area Tooth number
0.26+0-29 0"67__+0.40 0.62 • 0.43 0.18•
Song
Pulses/chirp Pulse length Pulse rate Chirp length Frequency
-0.28• 0-23 __+0.22 0.18+0.24 -0-15__0.12 0.20+0.14
*P < 0.05.
Table III. Partial correlation, matrix of song components (adjusted for wing morph), N= 163
Pulses/chirp Pulses/chirp Pulse length Pulse rate Chirp length Frequency
Pulselength
Pulserate
Chirp length
-
+
+
+
+
- 0.66* 0.40*
-0-35*
0.30* -0.10 0.77* -0.11
-0.48* 0,59" -0.33*
Predicted correlations from a pairwise analysis are shown in the upper triangle. *P<0.001 (probability levelsare not Bonferroni-adjusted).
were all opposite in sign to those predicted by a pairwise analysis. This can be explained by the dominating influence of correlations with chirp length (cl). For example, consider the correlation between pulse length (pl) and the number of pulses per chirp (ppc). Rearranging the previous equation gives pl = cl - (1000/pr)(ppc- 1) A simple pairwise analysis suggests that the number of pulses per chirp and pulse length will be negatively correlated. But, chirp length and the number of pulses per chirp are strongly positively correlated. Therefore, an increase in the number of pulses per chirp is accomplished by an increase in chirp length; if the latter increase exceeds the increase in the number of pulses per chirp, then pulse length will increase, not decrease. A significant positive correlation between pulse length and the number of pulses per chirp was, in fact, observed.
Correlations between morphology and song components Phenotypic correlations between morphological structures were all large and highly significant (Table IV). However, correlations between song components and morphological structures were very weak. Only the number of pulses per chirp and frequency were significantly correlated with femur length (Table IV). However, neither correlation remained significant when the significance level was corrected for the total number of tests (Bonferroniadjustment), and therefore it is unlikely that the calling song of the male was used by G. firmus females to assess male body size.
Variations in morphology between wing morphs Intermediate individuals were omitted from the analysis as they comprised too small a sample size for statistically meaningful comparisons.
Webb & Roff: Quantitative genetics of song
829
IV. Partial phenotypic correlation (adjusted for wing morph) of morphological characters on morphological (N= 319) and song components (N= 162) Table
Femur length File length Resonator area Tooth number Pulses/chirp Pulse length Pulse rate Chirp length Frequency
File length
0.64** 0.59** 0.32** -
0.17*
0-03 - 0.02 - 0"08 - 0.19*
Resonator area
0.63** 0.56**
Tooth number
0-33**
- 0.18 *
- 0.19*
0.12 0.09 - 0.15 -- 0.07
0.07 - 0.01 - 0.11 - 0.07
--0.03 0.06 0.10 -0.06 -0.11
*P<0.05; **P<0.001 (probability levels are not Bonferroni-adjusted). V. Character means (+ SE) of morphological and song components by wing morph Table
Character Morphological
Femur length (mm) File length (mm) Resonator area (mm z) Tooth number Song Pulses/chirp Pulse length (ms) Pulse rate (pulses/s) Chirp length (ms) Frequency (kHz)
Micropterous
Macropterous
( N = 176)
( N = 136)
12-8 _+0-06 4.0 + 0.02 5.3 -I-0.05" 188.4 -I-0.6 (N = 89) 3.7 ___0.03 23.3 + 0.3 21.0__+0.2 153-0 + 2.0 4.0 + 0'02t
12.9 + 0.05 4.3 ___0.02 6.1 + 0-05 188.7 __+0.5 (N= 77) 3.7 __+0.03 26.1 _+0.4 20.8 __+0.2 154.8 + 2.2 4.0 + 0.02:~
P
0.210 < 0.001 < 0.001 0.748 0.933 < 0.001 0.503 0.558 0.822
Probability level reflects ANOVA by wing morph. *N= 175. tN=85. :~N=75.
Macropterous individuals had significantly longer files and larger resonator areas than micropterous individuals, but did not differ in either femur length or tooth number (Table V). O f particular interest was the observed difference in overall morphology between micropterous and macropterous individuals. N o t only did the length of the flight wings differ, but the morphology of the tegmina and the correlations between the morphology of the song-producing structures and body size differed significantly in several respects (covariance analysis, no heterogeneity in slopes). For a given femur length, micropterous individuals had smaller file lengths: ( N = 3 1 2 , F-ratios: Fmorph = 139"626, P<0"001; Ff . . . . l e n g t h = 2 1 4 " 3 3 6 , P<0-001) Y=O'21X+O'23D+l'4, where X=
femur length, Y=file length, D = 0 for microptery and D = 1 for macroptery). Similarly, micropterous individuals have smaller resonator areas ( N = 311, F-ratios: Fmorph = 173' 781, P < 0"001 ; F r. . . . length = 161.590, P < 0-001): Y = 0-51 X + 0 . 7 4 D - 1.2, where X = femur length, Y = resonator area, and D is as above). Micropterous males have more teeth than macropterous males for a given file length (N = 312, F-ratios: Fmorph=29"816, P
Animal Behaviour, 44, 5
830
reasons for this difference may be more complex than simply the result of micropterous males having smaller files (see Webb 1991). There is considerable overlap in pulse length between morphs (Micropterous range = 13"5-30"9, X'_+SE= 23'3 + 0"3; Macropterous range=18-1-37'8, ~'-+SE= 26' 1 + 0.4), making pulse length a poor predictor of wing morph.
DISCUSSION Morphological features typically possess higher heritabilities than behavioural traits (Mousseau & Roff 1987; Roff & Mousseau 1987). This general observation is supported by the present analysis, the heritabilities for the morphological features, (hZsire+dam 0"42 -- 0"73), being significantly higher than the song components, (h2slro+d,m=0"10 0"35, Table II). All of the morphological features possessed significant heritability (h2sire+dam), while one of the song components, pulse rate, was significant. The generally larger values of hZaamcompared with h2si,e suggests that maternal and/or nonadditive effects may be important. It would be premature to assume that the heritabilities of the other song components are zero, as the standard errors are relatively large, though approximately equal to those of the morphological features. If we assume the heritability estimates are not zero, and estimate heritability to be 0.20 (approximately the mean h2sjro+aam of the nonsignificant estimates), then the family size required for significance to be detected may be determined. The expected intra-class correlation for halfsiblings is t = h2/4 (Robertson 1959, 1960; Falconer 1981), and the required intra-elass correlation is t =0.05. The optimal number of offspring per dam is determined by N=l/(2t) (Robertson 1959), resulting in N = 10. A mean of 5"3_+0"4 offspring per dam family was measured for the song components. Thus to detect a heritability of 0.20 requires twice the number of offspring per dam as measured in the present experiment. Femur length, an index of body size, has a high heritability (Table II). File length and tooth number, which are closely associated with important features of the calling song, possess lower heritabilities than the resonator area, although the differences were not statistically significant. Resonator area is associated with song intensity (Michelsen & Nocke 1974). However, due to =
attenuation, song intensity may not be of importance in female choice (Simmons 1988b). Trade-offs between traits chosen by females may contribute to the genetic variation in these traits. For example, femur length may be undergoing directional selection, with larger males being favoured by females (Simmons 1987). But the song components important in species recognition and female phonotaxis, and their associated structures, may be under stabilizing selection (Butlin & Hewitt 1986). As song-producing structures and body size are correlated (Table IV), changes in one may affect the other and, in turn, influence calling song. Thus, at some point trade-offs may occur between preferred song components and body size, contributing to the maintenance of genetic variation in the calling song. Female G. bimaculatus prefer larger males, and there exists the possibility that the calling song, used to attract females, could contain information about the size of the advertiser (Simmons 1987, 1988a, b). Simmons (1988b) found a significant increase in pulse rate and a decrease in chirp length with increasing body size (with pronotum width used as the index of body size) for G. bimaculatus. The reduction in chirp length was attributed to a negative correlation in pulse rate. Pulse rate was determined to be unaffected by distance degradation, suggesting females could detect differences in pulse rate between males (Simmons 1988b). In G.firmus there was no significant correlation between pulse rate and femur length (used as the index of body size). Only the number of pulses per chirp and frequency form weak negative correlations (N=162; t = - 0 . 1 7 , P<0'05, and r = - 0-19, P < 0'05, respectively) with femur length, and are, therefore, not likely to be used by females to assess body size. Note that using pronotum width as a measure of body size (Simmons 1988b), instead of femur length, would not alter the results for G.firmus, as there is a highly significant positive correlation between femur length and pronotum width for G.firmus (N= 106, r=0.77, P<0.001, D. A. Roff, unpunished data). Micropterous and macropterous individuals differed not only in the length of their flight wings, but also in other morphological features. Micropterous males possessed significantly shorter files, and smaller resonator areas, but did not differ in either femur length or tooth number, from macropterous males. Micropterous individuals possessed five more teeth than macropterous individuals with the
Webb & Roff." Quantitative genetics o f song same file length. Carrier frequency, determined by the number of teeth struck per unit time, should differ between morphs if the number of teeth struck differs, but the speed of wing closure does not. Howe;eer, no significant difference in frequency was observed for a given file length. It is, therefore, likely that the effects, if any, on the speed of wing closure, or carrier frequency of such a small difference in the number o f teeth, were too small to be measured. Phenotypic correlations between song and morphology suggest that changes in one may result in changes in the other. This is contingent on the genetic correlations between the traits; accurate estimates of these require more accurate estimates of the heritabilities of the song components. Differences in the tegminal morphology of micropterous and macropterous individuals resulted in measurable differences in the calling song. It is not unreasonable to expect morphological changes in the sound-producing structures accompanied by changes in the calling song with selection on wing morph. Further research, however, is required to determine whether such variation in song could influence mate choice by females.
ACKNOWLEDGMENTS We gratefully acknowledge the assistance of the following people: S. David, P. Shannon, D r R. Lemon, M. Bradford and D. Webb. Special thanks to Dr W. Cade and T. M a c D o n a l d for their integral roles in s o n g analysis. The comments of Drs G. Pollack, D. Kramer, A. Hedrick and J. Doherty on earlier versions of this paper are gratefully acknowledged. This work was supported by a grant from national Sciences and Engineering Research Council of Canada to D.A.R.
REFERENCES Alexander, R. D. 1961. Aggressiveness, territoriality, and sexual behavior in field crickets (Orthoptera: Gryllidae). Behaviour, 17, 130-223. Alexander, R. D. 1962. Evolutionary change in cricket acoustical communication. Evolution, 16, 443-467. Alexander, R. D. 1968. Life cycle origins, specialization and related phenomena in crickets. Q. Rev. Biol., 43, 1-41. Becker, W. A. 1985. ManualofQuantitative Genetics. Ann Arbor, Michigan: McNaughton & Gunn. Burk, T. 1982. Male aggression and female choice in a field cricket (Teleogryllus oceanicus): the importance
831
of courtship song. In: Orthopteran Mating Systems: Sexual Competition in a Diverse Group of Insects (Ed. by D. T. Gwynne & G. K. Morris), pp. 97-119. Boulder, Colorado: Westview Press. Butlin, R. K. & Hewitt, G. M. 1986. Heritability estimates for characters under sexual selection in the grasshopper, Chorthippus brunneus. AnOn. Behav., 34, t256-t261. Cade, W. H. & Wyatt, D. 1984. Factors affecting calling behaviour in the field crickets, Teleogryllus and Gryllus (age, weight, density and parasites). Behaviour, 88, 61-75. Crankshaw, O. S. 1979. Female choice in relation to calling and courtship songs in Acheta domestieus. Anim. Behav., 27, 1274-1275. Doherty, J. A. 1985a. Temperature coupling and tradeoff phenomena in the acoustic communication system of the cricket, Gryllus bimaculatus DeGeer. J. exp. Biol., 114, 17-35. Doherty, J. A. 1985b. Trade-off phenomena in calling song recognition and phonotaxis in the cricket, Gryllus bimaeulatus (Orthoptera, Gryllidae). J. eomp. Physiol., 156, 787-801. Doherty, J. A. & Hoy, R. 1985. Communication in insects. III. The auditory behavior of crickets: some views of genetic coupling, song recognition, and predator detection. Q. Rev. Biol., 60(4), 457-472. Falconer, D. S. 1981. Introduction to Quantitative Genetics. 2nd edn. London: Lougman. Hedrick, A. V. 1986. Female preferences for male calling bout duration in a field cricket. Behav. Ecol. Soeiobiol., 19, 73 77. Hedrick, A. V. 1988. Female choice and the heritability of attractive male traits: an empirical study. Am. Nat., 132(2), 267-276. Lutz, F. E. & Hicks, W. P. 1930. An analysis by movietone of a cricket's chirp (Gryllus assimilis). Am. Mus. Novit., 420, 1-14. Michelsen, A. & Nocke, H. 1974. Biophysical aspects of sound communication in insects. Adv. Insect Physiol., 18, 247-296. Mousseau, T. A. & Roff, D. A. 1987. Natural selection and the heritability of fitness components. Heredity, 59, 181-197. Pollack, G. S. 1982. Sexual differences in cricket calling song recognition. J. comp. Physiol., 146, 217-221. Pollack, G. S. & Hoy, R. 1981. Phonotaxis to individual rhythmic components of a complex cricket calling song. J. comp. Physiol., 144, 367-373. Popov, A. V. & Shuvalov, V. F. 1977. Phonotactic behavior of crickets, or. comp. Physiol., 119, 111-126. Ragge, D. R. 1969. A revision of the African species of Pseudorhynchus Serville (Orthoptera: Tettigoniidae). Bull. Brit. Mus. nat. Hist. Entomol., 23, 167-190. Rakshpal, R. 1960. Sound producing organs and mechanism of sound production in the field crickets of the genus Acheta fabricus (Orthoptera, Gryllidae). Can. J. Zool., 38, 499-507. Regen, J. 1913. Uber die anlickung des weibchens von Gryllus campestris L. durch telephonnisch/ibertragene stridulatioslaute des mannchens. Pfliigers Arch. ges. Physiol., 155, 193-200.
832
Animal Behaviour, 44, 5
Robertson, A. 1959. Experimental design in the evaluation of genetic parameters. Biometrics, 15(2), 219-226. Robertson, A. 1960. Experimental design on the measurement of heritabilities and genetic correlations. In:
Biometrical Genetics. Proceedings of an International Symposium Sponsored by the Biometrics Society and the International Union of Biometrical Sciences (Ed. by O. Kempthorn), pp. 101-106. London: Pergamon Press. Roff, D. A. 1984. The cost of being able to fly: a study of wingpolymorphismin two species of crickets. Oecologia (BerL), 63, 30-37. Roff, D. A. 1986. The genetic basis of wing dimorphism in the sand cricket, Gryllusfirmus and its relevance to the evolution of wing dimorphism in insects. Heredity, 57, 221-231. Roff, D. A. 1990. Selection for changes in the incidence of wing dimorphism in Gryllus firmus. Heredity, 65, 163-168. Roff, D. A. & Mousseau, T. A. 1987. Quantitative genetics and fitness: lessons from Drosophila. Heredity, 58, 103-118. Simmons, L. W. 1987. Heritability of a male character chosen by females of the field cricket, Gryllus bimaculatus. Behav. Ecol. Sociobiol., 21, 129-133.
Simmons, L. W. 1988a. Male size, mating potential and lifetime reproductive success in the field cricket, Gryllus bimaculatus (De Geer). Anita. Behav., 36, 372-379. Simmons, L. W. 1988b. The calling song of the field cricket, Gryllus bimaculatus (De Geer): constraints on transmission and its role in inter-male competition and female choice. Anita. Behav., 36, 380-394. Stout, J. F., DeHaan, C. H. & McGhee, R. W. 1983. Attractiveness of the male Acheta domesticus calling song to females. J. comp. PhysioL, 153, 509-521. Thorson, J., Weber, T. & Huber, F. 1982. Auditory behavior of the cricket. II. Simplicity of the calling-song recognition in Gryllus, and anomalous phonotaxis at abnormal carrier frequencies. J. comp. Physiol., 146, 361 378. Walker, T. J. 1957. Specificity in the response of the female tree crickets (Orthoptera, Gryllidae, Oecanthinae) to calling songs of the males. Ann. Entomol. Soc. Am., 50, 626636. Webb, K. L. 1991. The quantitative genetics of sound production in Gryllus firmus. M.Sc. thesis. McGill University. Weber, T., Thorson, J. & Huber, F. 1981. Auditory behavior of the cricket. I. Dynamics of compensated walking and discrimination paradigms on the Kramer treadmill. J. comp. Physiol., 141,215-232.