Foraging body temperature, sprint speed, morphology and fitness data collection
We collected lizard foraging body temperature data in the field yearly from 1981 to 1991, and then again in 2004, 2005, 2018 and 2019, all between the months of April to October, and during the day (6:30 - 17:00), when lizards are active. When encountering a lizard, we first observed its behaviour and then caught it using a mealworm-baited (Tenebrio molitor ) fishing pole. We identified foraging behaviour in lizards actively searching their surroundings and flicking their tongues, with bodies not pressed against substrates as they would be during basking. Upon capture, we immediately measured body temperature by inserting a ±0.1 °C, K-type thermocouple (TX-10, Yokogawa, Japan) into the cloaca. We then measured the nearest air temperature in the shade, noted time of day, marked the individual with a marker, and then released it at the site of capture. All data are thus known to be from separate individuals. We should also note that our data for Miyake island lizards are limited to 1983 due to the incidental population collapse of P. latiscutatus following the deliberate release of the Japanese weasel (Mustela itatsi ) for the control of rats in the mid 1980’s (Hasegawa 2003). We retrieved observatory air temperature data from the Japan Meteorological Agency (http://www.data.jma.go.jp/) both as daily mean averages for each year, and at the finest time scale available to match that of our collected body temperature (the finest being 10 minute averages, and the coarsest being day averages). Such data however were not available for Kozu from 1981 to 1991, or for Mikura or Toshima at any time. Although data for Hachijo-Kojima were not available, we used the air temperature data for the larger, adjacent Hachijojima as a proxy, which lies ~7 km away.
We collected sprint speed data for lizards in two different ways - directly on open field terrain, for a better comparison with snake crawling speeds, and on a racetrack, to compare lizard sprint speeds between two islands with and without the snake predator in a controlled setting. We measured racetrack sprint speed for lizards on Hachijo-Kojima and Kozu islands from May to June 2018 and from July to August 2019. For this, we brought a 90 cm long wooden racetrack marked at 5 cm intervals, as well as a camera and tripod, to the field. We then caught lizards using the baited fishing pole method and either briefly placed them in a bucket, or immediately coerced them to run along the track, with each individual only running once. The lizards sprinted along the track and into a cloth bag, at which point we immediately measured body temperature via the cloaca, as well as morphological measurements, which included snout-vent length (SVL), total tail length, tail break distance (length between the tail base and the first point at which the tail has broken off in the past) and mass. For the 2019 experiments, we additionally measured hind leg length (joint to tip of the longest toe) and hind foot length (bottom of the heel to tip of the longest toe). We video recorded each run at 60 frames per second to obtain a measurement of the fastest 10 cm run in cm/s.
We measured open field sprint speed for lizards on Kozu from July to September 1990, on concrete and earthen surfaces they regularly occupy. Within 2-3 m from a lizard, and with stopwatch in hand, one person would stomp their foot, prompting the lizard to run. Once the lizard reached a visible landmark approximately 1-3m away, such as a rock or crack, we clocked the time, immediately caught the individual by hand or baited fishing pole, and inserted a K-type thermocouple into the cloaca to measure body temperature. We then measured the starting point to landmark distance to calculate sprint speed in cm/s. For all open field and racetrack sprint speed measurements, we conducted experiments on a variety of days and times of the year to include the range of temperatures lizards can experience in the field, and only included non-juvenile individuals (SVL, >35mm). We also discounted sprint speed data obtained from individuals showing limb deformities or impairment.
We measured crawling speed for the snakes on Kozu in April 1994 and August 1996, as well as on Niijima in May 1996. For this, we caught individuals by hand in the field and briefly kept them in cloth bags to carry over to a flat earthen surface. We conducted experiments on different days and times of the year to include as much as possible the range of body temperatures snakes naturally experience in the field. With stopwatch in hand, we then released each snake with the head positioned at one marked spot on the ground and re-caught it after it had dashed a distance of approximately 2-3m, at which point we marked the exact position of the head at recapture, and immediately measured body temperatures via the cloaca by inserting a K-type thermocouple. We then measured the distance between the two marked positions, and together with the time clocked by the stopwatch, obtained an estimate of crawling speed in cm/s. For both lizard and snake experiments, we released all individuals at respective sites of capture following completion.
To determine fitness in the context of survival against predation by snakes, we collected lizard morphological data including SVL, hind leg length, and tail break distance, from two islands with the snake predator (Kozu and Mikura) and two without (Hachijo-Kojima and Miyake), separately from sprint speed and body temperature measurements. We measured live individuals on Mikura in 1981, as well as on Kozu and Hachijo-Kojima in 2019.
We measured SVL and tail break distance on Miyake individuals in 1979, before the weasel-induced population collapse, and fixed and preserved individuals in 10% formalin and 70% ethanol, respectively. In 2020, we measured the hind leg length of those preserved individuals. We estimated hind leg lengths when alive by major axis regression, with hind leg length when alive as the response and that when preserved as the predictor. We based the model on living and preserved individuals from Oshima (n =6), Mikura (n=14) and Kozu (n=22), for which we had measured hind leg length of the same individuals under both conditions (y = 9.48 + 0.80x ; R 2 = 0.73).