Dynamic infrared thermography
Overview . We quantified the dynamics of leaf temperature using an infrared camera (ICI 8640P, ICI Inc., Beaumont, TX) after transient radiative warming. The leaf was illuminated for 5-8 sec with a 275 W near-infrared heat lamp (model ZB:IL04-06, Serfory), causing its temperature to increase by approximately 15 – 40 K. The mean temperature in each of two regions of interest (”patches”), one with hairs and one without, was then recorded with the infrared camera at 5 Hz until temperature had ceased to decline noticeably (< 60 sec). Wind speed was measured with a hot-wire anemometer (FMA903R-V1, Omega, Stamford, CT) mounted 3 cm below the leading edge of the leaf. Air temperature was measured with a fine-wire type-T thermocouple (TT-T-36, Omega) mounted on the anemometer. Each such ”cooling curve” experiment was repeated 3-5 times for each of 5-6 leaves in five species (Table 2).
Leaf and patch preparation . Each leaf was excised from the plant shortly before measurements, and vacuum grease (DC 976, Dow-Corning, Midland, MI) was applied to the cut petiole to minimize water loss. The leaf was then placed in a mesh of fishing lines to secure it for measurements (an example, not from an actual experiment but set up for demonstrative purposes, is shown in Fig S1), such that the lamina was parallel to the direction of the wind flow. The mesh had a 45-degree opening in the direction facing incoming airflow; all temperature measurements were performed on regions located in that gap, to ensure that the fishing lines did not influence airflow upstream of the patches. We selected two leaf patches: one with leaf hairs and one without. In three species (OLEU, MAGR and PLOC), we were able to remove hairs from a patch, either by gently scraping the leaf surface with a razor blade oriented nearly perpendicular to the surface (OLEU and MAGR), or by gently rubbing the hairs off using a thumb and forefinger (PLOC). Although we were unable to remove hairs from POTO leaves by hand, this species exhibits extreme variation in trichome density among leaves, even of similar age. Thus, for POTO, we mounted two leaves in the fishing line mesh – one hairy and one hairless. We did the same for SOLY, comparing mutants with excessive hairs or without hairs. For each experiment with POTO and SOLY, we chose two leaves of similar size and shape; in the case of POTO, both leaves were always sampled from the same branch.
A rectangular region of interest (ROI) was located in each patch using the camera software. The ROIs were equal in size (approximately 5 x 5 mm) and were typically positioned with around 3 mm of distance between the ROI and the leading edge of the leaf. The patches could not always be located at precisely the same distance from the leaf edge across experiments, due to the practical difficulty of finding a suitable area in each leaf where hairs could be removed without damaging the underlying tissue. However, within each experiment, the two patches were always located at the same distance from the leaf edge as one another, and were placed in locations such that the adjacent leaf edge was oriented at a similar angle with respect to the direction of airflow for both patches. This ensured that heat transfer properties could be directly compared between the two patches in a given experiment.
Wind . Wind was generated by CPU fans arrayed horizontally in the leaf plane and located at one end of a 2-m long and 25-cm diameter cylindrical wind tunnel made of mylar-covered bubble wrap (Reflectix; Reflectix, Inc., Markleville, IN). The leaf was located approximately 10 cm from the other end of the wind tunnel. Tests with the anemometer showed that wind speed did not vary with position within the range of locations where the leaf patches were located, nor between the leaf plane and the location 3 cm below it where the anemometer was located during experiments. For each species, we adjusted the number of CPU fans and/or their voltage input to produce the highest wind speed that would not cause noticeable flapping or other leaf movement during the experiment. PLOC, SOLY and POTO leaves were too physically unstable at the higher wind speed generated using four CPU fans; most of the results described below were therefore limited to lower wind speeds generated using two CPU fans. Wind speed was steady during each experiment (average SD = 0.029 m s-1), and wind speed varied between 0.76 and 2.13 m s-1 among experiments.
Procedural details . Inference of leaf boundary layer conductance from temperature dynamics (see Theory below) requires the radiative environment of the leaf to be constant during the experiment, because this eliminates incoming radiation terms from the derivative of leaf temperature with respect to time, leaving dependences only on factors that can be more easily measured (namely leaf temperature, air temperature, stomatal conductance and leaf heat capacity). We therefore designed the experimental apparatus (Figure 1) to ensure that incoming radiation was constant during each cooling curve. First, the heat lamp was mounted on a large articulating arm, which allowed an operator to move the lamp away from the leaf after heating, into a position located in the same plane as the leaf but about 1 m away. We also attached a piece of Reflectix to the side of the lamp head that was facing the leaf when in the latter position, to block radiation. Second, we used an A-frame Reflectix shield mounted over the camera to shield the camera and leaf from any fluctuations in incoming infrared radiation caused by people operating the apparatus. Third, the camera was moved out of position using a sliding boom stand (model BBB, Amscope, Irvine, CA) while heating the leaf, so that the camera would not be heated by the lamp. Fourth, during leaf heating, we shielded the air temperature and wind speed sensors (which were located 3 cm below the leading edge of the leaf), as well as other objects below the leaf, from the heat lamp with a plastic plate covered with aluminum foil and mounted on a sliding boom stand. We then moved this shield out of the way after heating the leaf, and simultaneously rotated the lamp out of the way using its articulating arm, and moved the camera back into place using its sliding boom stand. The transition from heating to measuring took approximately one second. Figures 1b and 1c illustrate the positions of each component of the apparatus during leaf heating and measurement, respectively.
Black-body calibration . To ensure that leaf temperatures measured by infrared thermography were commensurable with air temperatures measured with the fine-wire thermocouple, which allowed modeling of the leaf-air temperature difference (see Theory ), we calibrated the infrared camera against a black-body reference before each cooling curve. The black-body consisted of a basketball that had been cut in half, the two halves inverted (which increased their rigidity), coated internally with graphite with the aid of spray adhesive, reassembled using duct tape, and placed into an insulating foam box covered in Reflectix. The black-body contained a fine-wire thermocouple previously calibrated to match the fine-wire thermocouple used to measure air temperature below the leaf. A square 2 x 2 cm window was cut into both the basketball and the foam box. The black-body was located such that the camera’s field of view was focused on the window into the black-body when the camera was moved out of the way for leaf heating. We recorded the mean IR temperature in the black-body window for 2 seconds, computed a correction factor by comparing that measurement with the output of the thermocouple located in the black-body, and applied this correction factor to IR temperature records for the subsequent cooling curve.