Stridulation rate as a thermometer in Tettigonia viridissima: a season-long automated record and a test against archive recordings

Bertrand Mercier des Rochettes 

That an insect's song rate follows its temperature has been known since Dolbear proposed reading a cricket as a thermometer in 1897. The relation is not in doubt; its functional form is. Walker (1975a) found stridulation rate approximately linear in ten species of Orchelimum and, reanalysing fifteen published datasets, found linear fits better than exponential in thirteen — but his data were hand-timed and spanned 12–26 °C.

Here I report a season-long automated record. A phone recorder and an on-site weather station logged the nightly chorus of Tettigonia viridissima — identified acoustically from the recordings by two specialists working independently — in one garden in the Vexin français from 26 June to 24 September 2026. Echeme repetition rate was measured automatically from the amplitude-modulation spectrum of the 9–12 kHz carrier band and paired with air temperature logged every five minutes at the microphone. Thirty-seven nights spanned 14.6–28.3 °C and a 2.89-fold range of rate; rate and temperature correlate at r = 0.986, with residual scatter equivalent to 0.63 °C.

Comparing five functional forms on common residuals, a quadratic fits best, and all seven nights below 17.4 °C fall below an exponential extrapolated from the warmer thirty. Walker's split-half test cannot separate linear from exponential here either, but admitting a quadratic resolves it out of sample: the relation is convex and sub-exponential, refining his finding rather than contradicting it. Decomposing the two-syllable echeme shows its two intervals carrying different apparent activation energies (55.5 and 36.1 kJ/mol), so the measured rate is a composite.

Tested against 178 archive recordings from 22 recordists in 14 countries, the relation reproduces but its coefficient does not: precision is 0.65 °C where calibrated and about 1.6 °C where not, though for four of five recordists the uncalibrated curve is accurate to roughly half a degree. Archives collected for other taxa may therefore carry a usable temperature record, at a precision set by what the recordist wrote down.