Use of acoustic methods for monitoring avian communities in temperate and tropical ecosystems

This research addressed acoustic monitoring for avian populations as a monitoring protocol in three different habitats in Tennessee and Kentucky (grassland at Fort Campbell Military Reserve in 2000; oldfield at Freel\u27s Bend Wildlife Management Area, Oak Ridge in 2000; and mixed hardwood forest at Cherokee National Forest, Tellico District in 2002 and 2003) and two habitats in Thailand in 2002 (hill evergreen forest at Phu Luang Wildlife Sanctuary, Loei province; and grassland at Khao Yai National Park, Na Korn Ratchasima province). Four recording devices, originally built in 2000, were comprised of Sennheiser MKR20 omni-directional microphones with 18-volt phantom power supplies, Jensen videocassette recorders (Hi-Fi VCRs) with 12-volt marine batteries, and microphone amplifiers with 9-volt batteries. In 2002, the recording devices were modified in that VCRs were replaced by computers as recorders. A 9-ha plot (300 m x 300 m) was set up in each habitat and included the four monitoring stations at grid intersections with 150-m spacing between each station. On IO mornings during the breeding season, the sites were acoustically monitored for 2 hours. The acoustic method was tested by conducting two standard census techniques currently used for bird monitoring: a series of I 0-minute, unlimited-distance point counts at each monitoring station and territory mapping. In most habitats, acoustic monitoring detected an equal or greater number of bird species when compared to unlimited-distance point counts or territory mapping when these 3 methods were conducted simultaneously. Some overlooked species at great distances as well as species during the dawn chorus were detected acoustically but not by other methods. On the other hand, secretive species and non-vocal flyovers were missed by the acoustic method. Sampling effort representing different combinations of number of visits, number of monitoring stations, and recording periods were investigated. In general, a greater number of recording periods, visits, and stations may be needed to detect most species in the area when species richness is high. I recorded 45 species in Fort Campbell grasslands and 54 species in Freel\u27 s Bend oldfields based on I 0-day data from the 3 methods; the results suggested using ten 90-minute visits with 4 stations and ten 120-minute visits with 4 stations in those areas, respectively. Similar results were found in the temperate forest habitat. I recorded 33 species in Cherokee National Forest, Tennessee, based on 8 days of monitoring by the 3 methods. The optimal sampling effort was eight SO-minute visits with 4 acoustic monitoring stations to document the maximum number of species detected on Cherokee NF. In tropical ecosystems, I detected 72 species in Khao Y ai based on 5 days of monitoring and 69 species in Phu Luang, Thailand based on 8 days of observation with the 3 methods. The optimal sampling effort for the maximum number of species was five I 00-minute visits with 4 stations and eight 110-minute visits with 4 stations for Khao Y ai and Phu Luang, respectively. The number of species detected within I 0-minute increments during 2 hours of recording was used to estimate the detection probability of individual species by the acoustic method. Most species were detected each day within 2 hours of recording and were detected within 80-100 minutes in I visit for all habitats. Detection probability estimated by acoustic method was similar to aural observations from previous studies indicating that the capacity of acoustic devices to detect individual avian vocalizations was equivalent to the ability of human hearing. Based on the results of this study, acoustic monitoring should be viewed as a suitable monitoring technique under certain conditions: I) when many sites need to be monitored simultaneously and expert observers are limited, 2) when the study sites are in area of restricted access, and 3) when the number and densities of species present are great. Acoustic approaches cannot provide abundance estimates unless the individual vocalization is identified by an array of microphones or by individual voice recognition software. An index to relative abundance can be developed with the acoustic method by using multiple monitoring sites and calculating (the number of sites with a species)/(total number of sites)