Fig. 1. Data applied to test Hawes and Bale's (Hawes and Bale, 2007) hypotheses. Fifteen arthropod species for which seasonal variation in cold hardiness are available were used (see supplementary material Table S1 for the species list and data). Freeze-tolerant species are indicated with squares, freeze-avoiding species with triangles, and a solitary chill-susceptible species (the house spider Acharaenae tepidariorum) with a circle. Note that two points, for Rhabdophaga strobiloides and Mayetiola rigidae, overlap completely. The arrows indicate the hypotheses erected in figure 1 of Hawes and Bale (Hawes and Bale, 2007). No phylogenetic correction was applied to these data, although they represent seven orders and 11 families. Following Hawes and Bale's description, the extent of `evolutionary derivation' was taken as the extreme lower lethal temperature (LLT, in °C), and the `extent of plasticity' as the summer (or non-acclimated) LLT minus winter (or acclimated) LLT (in °C). Note that this procedure might introduce an element of non-independence of the axes (see Brett, 2004), which would have to be given consideration in further tests of the hypothesis. The greater variation in LLT of the freeze-tolerant than in the freeze-avoiding species is likely because of homogeneous nucleation in the latter (see also Zachariassen et al., 2004): no insect supercooling points below –54°C have been reported in either strategy (Addo-Bediako et al., 2000), although freeze-tolerant species can survive well below their supercooling point.