O

10 100 Body Length (mm)

FIGURE 6.4 The relationship of size to dominant (a) and low (b) frequency in the insect vibratory signals shown in Figure 6.3, with the analysis restricted to the signals used in mating.

grouped together signals with different functions. Considering all signals together, including those used in mating, alarm communication and attracting ant mutualists, there was no relationship of size and frequency. However, if we restricted our analysis to mate advertisement signals, as we did for the signals of membracid treehoppers, there was a strong negative relationship between body size and both the dominant and the lowest frequency of the signal. Examination of Figures 6.2 to 6.4 reveals a lack of data points in the lower left-hand corner of the figure; although there is considerable variation in frequency for species of a similar size, there were no very small species which produced very low frequency signals.

The question is how should we interpret these results in terms of the reliability of spectral measurements of vibratory signals and in terms of the implications for vibratory signal production and evolution. For the within-species comparisons, the potential challenges for signal measurement were minimal or absent, maximising the probability of detecting a size-frequency relationship. However, we found no suggestion of such a relationship. For the two populations examined (one each for U. crassicornis and E. binotata from Ptelea), variation in signal frequency appears to be decoupled from variation in body size. Within-population variation in signal frequency can be important for female choice, where female preference may exert stabilising or directional selection on male signal frequency (reviewed for insects and anurans in Gerhardt and Huber, 2002). For E. binotata from P. trifoliata, females prefer frequencies near the mean frequency of males in the population (Cocroft and Rodriguez, 2005). Signal frequency differs substantially among species in the E. binotata complex (Rodriguez et al., 2004), and females of each species examined have preferences centred on the mean frequency of males in their population (Rodriguez, Ramaswamy and Cocroft, unpublished data). If E. binotata from Ptelea is representative, however, female preference based on signal frequency is unlikely to lead to correlated changes in male size.

For the broad between-species comparisons reported here, the potential to mask any relationship between size and frequency might exist for the following reasons:

• The sources of variation in measurements of vibratory signal spectra discussed above

• The differences among substrates and transducers

• The limited availability of accurate size measurements

However, for the membracid data set and the overall insect data set, there was a significant negative correlation between body size and dominant frequency and the lowest frequency in of male mating signals. The correlation was higher between size and the lowest frequency in the signal, possibly because the overall bandwidth of the signals was less substrate-dependent than the relative amplitude of different frequencies within that frequency band. The r2 values (reflecting the proportion of variation explained) were higher at the broader level of comparison, suggesting that the relationship between size and frequency is relatively loose, and unlikely to be detected unless there is a large range of values for both variables. The lack of a relationship for the analysis which included signals other than those used in mating, is primarily accounted for by the lycaenid and riodinid caterpillars, which produce signals to attract ant mutualists (DeVries, 1991). These signals are typically broadband and relatively high in frequency, perhaps as a result of selection to produce vibratory signals similar to those of the ants with which they are communicating (DeVries et al, 1993).

Two factors might explain why there is an inverse relationship between body size and frequency in insect vibratory signals. First, an insect resting on six legs can be modelled as a mass on a set of springs (Tieu, 1996; Cocroft et al., 2000; also see Aicher et al., 1983). Other things being equal, the greater the mass, the lower the resonant frequency of a mass-and-spring system. It is not known whether insects use this resonance in signal production (or signal reception; see Aicher et al., 1983), but if they do this could explain why larger insects produce lower frequency vibratory signals. Second, it is likely that at least some species use the thoracic muscles to generate vibratory signals (Gogala, 1985a; Cocroft and McNett, Chapter 23). Wingbeat frequency is inversely correlated with mass in insects (Dudley, 2000), and if the wing muscles are used to produce vibratory signals, this could generate a negative correlation between signal frequency and mass.

In this study, we did not examine the relationship of size with temporal variables or overall signal amplitude. For vibratorily communicating katydids, size was tightly correlated with tremulation rate, larger males signalled at a faster rate (De Luca and Morris, 1998). In communication systems using airborne sound, size has been found to correlate with a variety of temporal signal traits (Gerhardt and Huber, 2002). In general, larger animals can produce higher amplitude signals (see Markl (1983) for vibratory signals). At a finer level (e.g. among individuals within a population or between closely related species), amplitude comparisons will be difficult to make for insects communicating with plant-borne vibrations because the amplitude of plant-borne signals is highly substrate dependent ((Cokl and Virant-Doberlet, 2003). The same insect will produce a higher amplitude signal on a thin stem than on a thick stem, and the amplitude of a signal recorded at different distances from a signaller does not decrease monotonically (Michelsen et al., 1982).

For the membracid data set, there was evidence that close relatives tend to be similar in dominant frequency, though not in size. Ecology and behaviour are also similar within many membracid clades (Wood, 1979, 1984; Dietrich and Deitz, 1991; McKamey and Deitz, 1996), and it would be worthwhile to investigate whether phylogenetic aspects of signal variation reflect adaptation to similar ecological conditions such as low population densities or use of herbaceous vs. woody plants.

Was this article helpful?

0 0

Post a comment