Figure Legends

FIgure 1: Sampling schedule and illustrations of bat sampling methods. Bat point counts were compared simultaneously against automated ultrasound recording and mist-netting in three oil palm plantation sites (Plots BP1, BP2, and BP3) for nine nights. Drawings by JABU studio.
Figure 2: Identification workflow of bat taxa sampled in oil palm plantations in Sumatra (Indonesia) with bat point counts, as well as traditional mist-netting and ultrasound recording methods. Bat point counts were used for a quarter of the sampling duration of the other methods. Call interval, start and end frequency were also used for the identification but are not shown here. Representative near-infrared photographs are shown; they belong to sequences of multiple pictures. Putative identification pathways are shown with a lighter gray tone. Asterisks denote near-infrared imagery that was not strictly needed for identification but that was used for identity confirmation. For mist-netting, numbers denote captures, and for bat point counts and ultrasound recordings, detections.
Figure 3: Rarefaction and extrapolation sampling curves for bat point counts, compared to established bat sampling methods. Shaded areas show 83% confidence intervals; differences in species richness are statistically significant when they do not overlap (Krzywinski & Altman, 2013). Extrapolated values are only shown up to double the reference sampling size to avoid large prediction errors.
Figure 4 : Detection ranges and sampling locations for bat point counts, mist nets, and automated ultrasound recorders. The sampling rig and ultrasound recorder were set up at the sampling center. The thermal scope’s field of view was scanning the thermal detection area. The curved ranges for ultrasound were drawn manually between the measured range directions. The ultrasound detection ranges are scaled to a maximum of 50 m as they are only representative of our 40 kHz ultrasound emitter otherwise (SPL 48 dB @ 30 cm).