1 Supplementary Material for Ashcroft and Gollan: “Fine-resolution (25m) topoclimatic grids of near-surface (5cm) extreme temperatures and humidities across various habitats in a large (200km by 300km) and diverse region” S1 Sensor usage details 1.1 DS1923 iButton sensors Technical and calibration information for the DS1923 iButton sensors can be found at: http://datasheets.maxim-ic.com/en/ds/DS1923.pdf. Of particular note, is that the raw observations from the iButtons should not be used directly. Rather, the iButtons contain calibration information in memory locations 240H to 253L, which are used to provide software correction and increase the accuracy of the observations. We used the high resolution settings (0.0625oC / 0.04% Relative Humidity) with quoted accuracy better than ± 0.5oC. Humidity observations are less accurate because the capacitive humidity sensors saturate when exposed to humidities greater than 70%, and provide observations higher than actual. While the manufacturer provides a formula to cater for this saturation, this is not able to cope with the large diurnal range we observed in natural situations, and the fact that saturation persists for some time even once they are removed from the high humidity environment. In effect, the devices typically saturate overnight and maintain some of that saturation during the day (even if humidities drop below 70%) and often into the next night as well. Effects of this saturation may be higher in winter, where humidities are typically higher. We did not apply any correction to remove the humidity saturation, and therefore the humidities we quote can exceed 100% and should be regarded as a moisture index rather than a strict relative humidity. Because of the saturation, observations reflect current conditions as well as past conditions to some extent. Note also that these DS1923 iButtons cannot be buried, since this would block the opening that is used for recording humidity. Indeed, even a small blockage to the hole in the sensors appears to cause spurious humidity readings. It is also important to be careful when retrieving sensors, as spiders appear to nest frequently inside the plastic containers (notably the Australian red-back spider). 1.2 iButton housing and positioning The iButtons were secured in netting inside an inverted plastic jar (Fig. S1), with wooden stakes used where necessary to protect the devices from cattle and other disturbances (Fig. S2). The sensors were placed in a diverse range of topographic positions and habitats (Fig. S1-3). 2 Fig. S1 A DS1923 iButton sensor being placed in a net inside a plastic container (left) and secured to the ground using tent pegs (right). Fig. S2 Wooden stakes were used to protect the devices from cattle and other disturbances. Fig. S3 A selection of locations where the sensors were located. 3 S2 Relationships between soil and air temperatures Soil temperatures were recorded at 34 sites (Fig .1) for the summer period (3rd December 2009 to 25th February 2010) using DS1921G iButtons placed 1cm beneath the surface, and approximately 30cm north of the 5cm air temperature sensors. We calculated the daily maximum and minimum soil temperatures and examined how these related to the 5cm air temperatures at the respective sites. Results were highly variable between different environments, and we therefore present results in terms of five general categories. Rainforests exhibited very stable soil temperatures, both in terms of low-diurnal range and intra-seasonal variability (Fig. S4). During hotter and drier periods the 5cm air temperatures rose slightly, although there was little effect on soil surface temperatures. Moist forests generally had lower canopy cover than rainforests, and were at lower elevations, further inland, or in more exposed locations. They were still fairly stable in soil temperatures, although the air temperatures reached much higher temperatures than the rainforests at all sites, and at some sites the soil temperatures rose also (Fig. S5). The driest forests reached even higher temperatures, both in terms of soil temperatures and air temperatures (Fig. S6). The soil temperatures even exceeded the air temperatures at some of the more exposed sites. The inland grasslands were characterised by high soil temperatures (sometimes reaching 60oC), and these often exceeded the observed air temperatures, especially during hot and dry periods (Fig. S7). The soil temperatures in coastal grasslands remained lower than the air temperatures (Fig. S8), as the buffering effect of distance from coast apparently had more effect on soil temperatures than air temperatures. In terms of average summer minimum temperatures, the 1cm soil temperatures were consistently warmer than the 5cm air temperatures. The difference was approximately 2oC at 20oC, with a few colder sites suggesting differences could be larger at that extreme (Fig. S9). In contrast, the average maximum 1cm soil temperatures were approximately 3oC cooler than the air temperatures between 25oC and 35oC, although soil temperatures became hotter than air temperatures at the hottest sites (Fig. S9). While the results demonstrated that 1cm soil temperatures were quite different from 5cm air temperatures, this was expected given the variability of near surfaces temperatures (Geiger, 1971). We were mostly concerned that the plastic containers might provide too little shelter from radiation, causing the temperatures to be artificially high at exposed sites on hot days. This was clearly unfounded, as the soil temperatures exceeded the air temperatures at the hottest and most exposed sites (inland grasslands and woodlands). 4 Our results highlight the buffering effect of moisture, with soil temperatures in rainforests and moist forests avoiding many of the hottest and coldest conditions present in 5cm air temperatures. In contrast, the drier soils in inland locations experienced more extreme conditions than the air temperatures, with little buffering present. The study area was generally dry prior to a large rainfall event that started on December 26th (Fig. 6). The air and soil temperatures were similar at many sites during the moister times, with differences more apparent during the warmer and drier times (see also Geiger, 1971). Fig. S4 The daily 5cm air and 1cm soil maximum and minimum temperatures (oC) for five rainforest sites during summer 2009/10. 5 Fig. S5 The daily 5cm air and 1cm soil maximum and minimum temperatures (oC) for nine moist forest sites during summer 2009/10. Fig. S6 The daily 5cm air and 1cm soil maximum and minimum temperatures (oC) for nine dry forest and woodland sites during summer 2009/10. 6 Fig. S7 The daily 5cm air and 1cm soil maximum and minimum temperatures (oC) for five inland grassland sites during summer 2009/10. 7 Fig. S8 The daily 5cm air and 1cm soil maximum and minimum temperatures (oC) for six coastal grassland sites during summer 2009/10. 8 Fig. S9 The relationships between the average summer minimum and maximum temperatures taken 1cm below and 5cm above the soil surface. Lines of best fit are fitted using a 2nd degree polynomial. 9 Fig. S10 Partial response graphs for the spatial model of 95th percentile of minimum temperature. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that elevation had far greater effect than all the other predictors. 10 Fig. S11 Partial response graphs for the spatial model of 95th percentile of maximum temperature. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that distance from coast had the greatest effect and exposure to the south the least. 11 Fig. S12 Partial response graphs for the spatial model of 5th percentile of maximum temperature. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that elevation had the greatest effect and exposure to the northeast the least. 12 Fig. S13 Partial response graphs for the spatial model of 5th percentile of minimum humidity. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that distance from coast had the greatest effect and log of flow accumulation the least. 13 Fig. S14 Partial response graphs for the spatial model of 95th percentile of minimum humidity. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that distance from coast had the greatest effect and exposure to the south the least. 14 Fig. S15 Partial response graphs for the spatial model of 95th percentile of maximum humidity. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that elevation had the greatest effect and remotely sensed canopy cover the least. 15 Fig. S16 Partial response graphs for the spatial model of 5th percentile of maximum humidity. The effect of each predictor can be gauged by the range of the Y-axis that each partial response line covers, such that distance to coast had the greatest effect and log of flow accumulation the least.