2.2 Wind tunnel measurements with epiphytes in the University of Oldenburg

Permission to work in the Barro Colorado Nature Monument was granted by the Smithsonian. Tropical Research Institute. A permit to export plants was granted by the Panamanian authorities (SEX/P-3-19).

We collected 112 individuals of five epiphytic bromeliads of varying size. Those five species represent varying growth forms (Fig 2). Tillandsia flexuosa (n = 24) was collected from trees in public areas in the Veraguas province in Panama. The other four species (Guzmania monostachia (n = 24), Tillandsia fasciculata (n = 24), Vriesea sanguinolenta (n = 20), Tillandsia elongata (n = 20)) were collected from Annona glabra trees on Lake Gatun in the areas around Barro Colorado Natural Monument, Panama. All plant names follow the Plant List [55]. Bromeliads were chosen for this study because they have a relatively predictable clear form when subjected as a whole plant to oncoming wind. In contrast, epiphytes that are growing very close to their host will not provide much of a barrier against the wind and thus measuring them individually without their host is rather irrelevant. Before export to Germany, plants were carefully cleaned and rinsed to remove all organic material trapped in the leaf axils. In Oldenburg, Germany, the plants were kept in the greenhouse under moist tropical conditions when not used in experiments. Wind tunnel measurements were completed within a week after arrival from Panama.

Prior to clamping (Fig 1B), all dead leaf parts were removed. Roots were carefully trimmed in order to ensure that measurement of drag force was only on the leaves, as these parts are more prominently exposed to wind flow in nature. Drag force measurements were done with the same force sensor and computer-amplifier setup as the trial tests with spheres. Before each measurement, the longest leaf (LL) of each individual plant was recorded and the force sensor tared. None of the plants tested had a length and width ≤ 8 cm and, therefore, the dominant shape for determining C_d should not be affected. Force measurements were recorded by exposing the plants to wind velocities from 5 to 22 m s^-1, with stepwise increments of 1 m s^-1. Plants were exposed to each wind speed for 30 seconds. Wind velocity stabilised after 10 seconds and mean drag force was recorded over the next 20 seconds (sampling rate of 50 times s^-1). Due to the mounting of the plants, the setup was more exposed than that of the mounting of the spheres. Thus, the drag on the setup alone was measured separately without the plant. These offset values were subsequently subtracted from the total drag of each sample [S2 Fig in S1 File].

To determine the plant frontal area exposed to wind, a small industrial camera (camera: DFK 27AUJ003, 10.7 MP, The Imaging Source, Bremen, Germany; lens: T0412FICS, 1/3" 4mm f1.2 Monofocal, Computar, CBC America LLC) was mounted in the upstream section of the wind tunnel to photograph the plant at each wind velocity (Fig 1A). A portable flood light helped to keep plants brightly lit to maximise the colour contrast against the dark background of the wind tunnel. A scale bar of 1.5 cm was placed on the screw of the clamp, which was visible in all photographs, to establish an image scale. The plant frontal area was determined using Fiji, a distribution of ImageJ for scientific image analysis (version 1.53c) [56]. Additional editing to increase contrast between the plant and the background was done in GIMP (version 2.10.10), if deemed necessary. An additional camera installed at the side of the setup took side-view photographs of the bending and reconfiguration of each plant in increasing wind velocities.