Anticipatory behaviour is the expectation of a near-future event based on information processed in the past and influences an animal’s tactical decisions, particularly when there are significant fitness consequences. The grass lizard (Takydromus viridipunctatus) perches on blades of grass at night which likely reduces the probability of predation by terrestrial predators such as snakes, rodents, and shrews. During twilight (starting 30 mins before sunrise) they move from above the grass to within grass clumps and this is thought to afford the lizard protection while reducing detection by avian predators. Here, we examined how lizards shift their behaviour as a function of visual detectability to their primary predator, the cattle egret (Bubulcus ibis). We show that the lizards shift from their perch site during twilight at the earliest time at which egrets depart communal roosts. At the same time, visual modelling shows a dramatic increase in detectability of the lizards to the visual system of egrets. Therefore, anticipatory behaviour in response to environmental cues acts to reduce predation risk as lizards become more conspicuous and predators become more active. Grass lizard anticipatory behaviour appears to be finely tuned by natural selection to adjust to temporal changes in predation risk.

01 Behavioural pattern.xlsx - raw data for the lizard behaviour patterns. We scored behaviours as follows: 1) eyes open; 2) head movement: head-turning or any body movements less than 1× snout-vent-length (SVL); and 3) body movement: movements more than 1× SVL from the original position. We standardized the time of a behaviour by relating it to the sunrise time of that date (i.e., how many minutes before or after sunrise), which was available from the Central Weather Bureau, Taiwan.

02 Ambient light.xlsx - raw data of ambient light condition. During the twilight period, down-welling light (surface power density, μWatt cm^-2 s^-1) was measured every minute using a power meter (Newport 1935-C, California USA) with a photodiode detector (Newport 918D-SL-OD1R) at the exact moment that behavioural observations were made. The sample rate was 0.1 ms, with 10,000 samples averaged to produce a mean. The power meter was set to detect wavelengths of 416/478/542/607 nm, which indicated the maximum absorbance of the 4 cones: VS/SWS/MWS/LWS in v-type eyes of birds, including cattle egrets (Hart, 2001; Borges et al., 2015; Endler & Mielke, 2005).

03 Reflectance.xlsx - raw data of spectral reflectance. We measured the spectral reflectance of live lizards and grasses in both the breeding (10 males and 7 females) and non-breeding (10 males and 10 females) seasons using a reflectance probe (Ocean Optics QR600-7-SR-125F) and a portable spectrometer (Ocean Optics Flame UV-VIS model).