Data for "Acoustic degradation of bat contact calls" Chaverri & Carter

See Chaverri &amp; Carter methods. Each row is a recording of a playback of a bat contact call recorded at a specific distance in a specific habitat type. For playbacks we used recordings of one contact call from each of 35 individual bats of four species: the common vampire bat (Desmodus rotundus, n = 9), the white-winged vampire bat (Diaemus youngi n = 10), Spix's disc-winged bat (Thyroptera tricolor, n = 10), and Thomas’s fruit-eating bat (Dermanura watsoni, n = 6). To record calls, we used an Avisoft CM16 microphone (frequency range 10–200 kHz, ±3dB frequency response 25-150kHz, Avisoft Bioacoustics, Berlin, Germany), an Avisoft UltraSoundGate 116 or 416 (16-bit resolution, 250 kHz sampling rate for phyllostomids, 384 kHz sampling rate for Thyroptera), and a PC laptop running Avisoft RECORDER. We recorded Desmodus and Diaemus individually in a small mesh cage within 10–30 cm of the microphone, Thyroptera at distances of 1-3 m within a large flight cage, and Dermanura underneath a tent-roost at a distance of 0.5-1 m. We conducted playbacks in the Golfito Wildlife Refuge and Barú Research Station, southwestern Costa Rica (mean annual temperatures of 24-28 °C and ca. 90% relative humidity) during the daytime (9 am-5 pm). To determine how signal degradation of bat social calls is affected by habitat, we performed playback sessions at each of five sites within secondary forests with high-density understory vegetation, five primary forest sites with low-density understory vegetation, and seven open area sites with no trees or shrubs within at least 5 m. In each playback session, we broadcast the 35 social calls four times with varying distances of 2, 4, 6 or 8 m between speaker and microphone. We mounted an Avisoft Ultrasonic Dynamic Speaker Vifa (frequency response curve at Avisoft.com) and an Avisoft CM16 microphone on tripods at 1.5 m above ground, pointing directly towards each other. In open and primary forest habitats there was always a clear line of sight between the microphone and speaker; in secondary forests this was not always the case, but we verified that both instruments were pointing directly towards each other. To digitize calls, we used an Avisoft UltraSoundGate player and recorder (216H and 116Hme, 16-bit resolution, 250 kHz sampling rate) connected to a laptop computer running Avisoft-RECORDER. Calls were always broadcast and received by the same speaker and microphone in all trials. We analyzed the recordings of calls that were played back using automated measurements in Avisoft SASLab Pro. To help match recorded calls to their original file, we used spectrogram cross-correlation in SASLab. After the automatic measurement procedure, we also manually inspected calls to determine if the start and end positions matched the signal’s waveform, and if signals were properly measured. We removed three measures of start frequency where Avisoft was actually measuring background noise. Of the 2380 recorded calls, 13 calls were not detected or analyzed by Avisoft, so we removed these observations from our statistical analysis, because it was not certain whether the missing call was due to degradation by habitat or technical problem. These included four Thyroptera calls in primary forest and nine calls from secondary forests (three Desmodus, one Dermanura, one Diaemus, and four Thyroptera calls). For each call recording, we extracted (FFT length 256, Frame size= 100%, FlatTop window) four signal measurements: call duration, peak frequency at the call start (“start frequency”), peak frequency of the call’s overall spectrum (“peak frequency”), and peak frequency at the call end (“end frequency”). We did not compare recordings of playbacks with the original files; we compared recordings of playbacks degraded over different distances. Recordings and analysis were either automated or blind to minimize observer bias.<br>

详见Chaverri与Carter的研究方法。每一行对应一段回放录音：该录音是在特定生境类型下、以特定距离录制的蝙蝠联络叫声回放信号。本次回放实验使用了4个物种共35只个体蝙蝠的单份联络叫声录音：普通吸血蝠（*Desmodus rotundus*，n=9）、白翼吸血蝠（*Diaemus youngi*，n=10）、斯皮克斯盘翼蝠（*Thyroptera tricolor*，n=10）以及托马斯食果蝠（*Dermanura watsoni*，n=6）。为录制叫声，我们使用了德国柏林Avisoft Bioacoustics公司生产的Avisoft CM16型麦克风（频率范围10~200 kHz，±3dB频响范围25~150 kHz）、Avisoft UltraSoundGate 116或416型采集设备（16位采样精度，叶口蝠科采样率250 kHz，盘翼蝠属采样率384 kHz），以及搭载Avisoft RECORDER软件的便携式PC电脑。普通吸血蝠与白翼吸血蝠的录音是在小型网笼中完成，麦克风与个体间距为10~30 cm；斯皮克斯盘翼蝠的录音在大型飞行笼中完成，间距为1~3 m；托马斯食果蝠的录音则在帐篷形栖息处下方完成，间距为0.5~1 m。本实验的回放环节在哥斯达黎加西南部的戈尔菲托野生动物保护区（Golfito Wildlife Refuge）与巴鲁研究站（Barú Research Station）开展，该区域年均温24~28 ℃，相对湿度约90%，实验均在日间（9:00~17:00）进行。为探究生境对蝙蝠社交叫声信号衰减的影响，我们在三类生境中开展回放实验：林下植被高密度的次生林（设5个采样点）、林下植被低密度的原生林（设5个采样点），以及周边至少5米内无乔灌木的开阔区域（设7个采样点）。每场回放实验中，我们将35份社交叫声各播放4次，扬声器与麦克风的间距设置为2、4、6或8米不等。我们将Avisoft Ultrasonic Dynamic Speaker Vifa型超声动态扬声器（频响曲线详见Avisoft官网）与Avisoft CM16型麦克风固定在离地1.5米的三脚架上，二者正对放置。在开阔区域与原生林生境中，麦克风与扬声器始终保持无遮挡的直视路径；在次生林生境中虽未必能保证无遮挡，但我们已确认两台设备始终正对放置。为完成叫声的数字化处理，我们使用搭载Avisoft-RECORDER软件的便携式电脑，连接Avisoft UltraSoundGate 216H与116Hme型播放采集设备（16位采样精度，采样率250 kHz）。所有实验中，叫声的播放与录制均使用同一套扬声器与麦克风设备。我们采用Avisoft SASLab Pro软件的自动化测量功能，对回放录制的叫声进行分析。为将录制得到的叫声与原始文件对应，我们在SASLab中使用声谱图互相关分析。自动化测量流程完成后，我们还会对每段叫声进行人工检视，确认信号的起止位置与波形匹配度，以及测量结果是否准确。我们剔除了3组被Avisoft软件误判为起始频率的背景噪声测量值。在2380段录制得到的叫声中，有13段未被Avisoft软件检测或分析，因此我们将这些数据从统计分析中剔除——因无法确定该类缺失是由生境信号衰减还是技术故障导致。该13段无效数据包括原生林中的4段盘翼蝠叫声，以及次生林中的9段叫声（3段普通吸血蝠、1段托马斯食果蝠、1段白翼吸血蝠与4段盘翼蝠叫声）。针对每段叫声录音，我们采用FFT长度256、帧长100%、平顶窗的参数，提取了4项信号测量指标：叫声时长、叫声起始峰值频率（即“起始频率”）、叫声全频谱峰值频率（即“峰值频率”），以及叫声结束峰值频率（即“结束频率”）。本研究未将回放录制的叫声与原始文件进行对比，仅对不同传播距离下产生衰减的回放录音进行组间比较。为最小化观察者偏差，所有录制与分析流程均采用自动化或盲法处理。