Acoustical Wildlife Presence Monitoring
Status - FY 08
Demonstration Deployments of CLO's ARUs
The following is a summary of the deployments of Cornell Lab of Ornithology's ARUs in Montana associated with the Northern Region's Landbird Monitoring Program and the the Cerulean warbler monitoring on the Allegheny National Forest.
This summary will be provided to the Inventory & Monitoring Steering Committee as a document that could be used to garner support from other federal partners (BLM, USFWS, USGS, USNPS) in further (more fully) promoting the development of bioacoustical monitoring technology.
Demonstration Deployment in the USFS' Northern Region (National Forests in Montana)
Point Count Sites
During the 2007 field season Dr. Dick Hutto, with the University of Montana, Avian Science Center deployed 10 stereo ARUs for periods of two to several days at existing point-count sites that were surveyed by field biologist for the USFS' Northern Regions' Landbird Monitoring Program.
Each ARU was deployed one to two days in advance at the site of the planned human point count, and retrieved one to two days after the point count is conducted. Each unit was then re-deployed at another point count site. Deployments were conducted in three different vegetation types near Missoula, Montana. Each ARU was programmed to record for the hours of each day during which human point counts were conducted. The goal was to conduct deployments at 30 sites in each vegetation type, for a total of 90 samples.
At the conclusion of the field season, all ARUs were shipped back to Cornell Lab of Ornithology for data extraction.
results are updated as they become available
During the winter following the field season, Dr. Hutto and a student assistant (Ryan Stuzman) independently reviewed the ARU recordings. The reviewers opened the entire audio record from that ARU unit using CLO's Raven software, which provides easy navigation, visualization, and playback of the sounds. The reviewers used the software to navigate directly to the logged time of a field-taken point count and listened to the recording of the ten-minute count period on stereo headphones, in four 2.5-minute segments (as per the field point count protocol). A log was kept of all species identified in each 2.5-minute segment.
(Note: in addition to the 10-minute field taken point count, the reviewer also reviewed the ten-minute period immediately before and immediately after the field taken point count.)
Preliminary data from Univ. of Montana's, Avian Science center is shown in the table below. The preliminary results show that Dr. Hutto and the ARU observed the same 272 individuals at all point counts (a 52% overlap). The ARU recorded 63 individuals that Dr. Hutto missed (12% of individuals). Dr. Hutto observed 185 individuals that the ARU missed (36% of all individuals). The results for Dr. Hutto's assistant (Stuzman) was similar.
These preliminary results from this single deployment suggest that the ARU detected half of the aindividuals that might be expected in a 10-minute point count survey.
||Number of individuals (of all species) at Point Count Sites
||ARU & Observer
Further analysis of data on species and individuals that the ARU recorded at point count locations before and after a human observer arrived at the count location is forthcoming.
Demonstration Deployment on the Allegheny National Forest
During the 2007 field season Dr. Scott Stoleson, USFS Northern Research Station (Forestry Sciences Lab - Irvine, PA Field Office), deployed 3 ARUs at 9 known and potential Cerulean territories. The ARUs took autonomous, continuous recordings between 6-11 am for several days at each locations to produce about 250 hours of audio recordings.
Wildlife Acoustics automatic processing of data
SDTDC contracted Wildlife Acoustics, Inc. to use their Song Scope™ software to process the 250 hours of audio data. With about 60 hours of effort, Wildlife Acoustics, Inc was able detect and confirm 1552 Cerulean vocalizations. (See diagram illustrating the detection process.) The Cerulean vocalizations were independently confirmed by Dr. Stoleson, as well.
Its noteworthy that most of the effort in time (53%) was in the development of the "recognizer" that Song Scope used to automatically identify Cerulean vocalizations of a specific locality. This is a one-time investment of time. Once the recognizer is developed, it can be used in subsequent field seasons.
For the entire report by Wildlife Acoustics, Inc. which documents the process used to automatically detect Cerulean vocalizations from the Allegheny deployment - click here.
Cornell Lab of Ornithology automatic processing of data
SDTDC also requested CLO to use their software (XBAT data template detector) to independently process the same 250 hours of audio data to get a feel for the ease-of-use, cost effectiveness, and sensitivity of detection of CLO's XBAT verses Wildlife Acoustics' Song Scope.
To date CLO has provided a Preliminary Report of their automatic detection process.
more details on the comparison of the two automatic dection softwares will be provided as they become available.
Song Meter™ Test Results
Wildlife Acoustic Inc. recently (October 2007) developed an autonomous recording device - Song Meter™. SDTDC purchased a device and will test it and report the results on this webpage.
Biologist with the US Geological Survey's Amphibian Research Monitoring Imitative (ARMI) also purchased several Song Meters for testing and use for amphibian bioacoustical monitoring.
Song Meter™ by Wildlife Acoustics, Inc.