search
Contact Us
HOME

  • Crossref logo
  • Crossref Similarity Check logo
Journal Search Engine

to

Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1229-3857(Print)
ISSN : 2288-131X(Online)
Korean Journal of Environment and Ecology Vol.29 No.2 pp.174-183
DOI : https://doi.org/10.13047/KJEE.2015.29.2.174

Evaluation of the Forest Bird Community by using a Sound Recording System1a
-Verification for the Avifauna evaluation in the non-breeding season -

Seung-Hwa Yoo2, Hyun-Jin Han3, Dong-Won Kim4, Woo-Young Joo2*
2National Institute of Ecology, Seocheon (325-813), Korea (ecogrus@nie.re.kr)
3Nakdong Estuary Eco Center, Busan (604-020), Korea (gentilis@daum.net)
4National Institute of Biological Resources, Incheon (404-708), Korea (foowl@korea.kr)

a This study was supported by the Korea Forest Service (Subject No. 2012S1200054). This paper contains the reanalysis results which had presented at the proceeding of Fall Meeting of the Korean Society of Environment and Ecology in 2013

Corresponding Author : Tel: +82-41-950-5403, Fax: +82-41-950-5940, joowooye@gmail.com
January 20, 2015 February 27, 2015 February 28, 2015

Abstract

초록

We have conducted a verification test of correlation between bird community indices and bird sound recording results in order to use the equipment which deals with the automatic sound collecting system, the song meter. The study areas include four sites which are Gombaeryoung and Zochimryoung in Jeombongsan Mountain and Jookryoung and Gochiryoung in Sobaeksan National Park. We collected the bird sound data five times using a sound recorder and field survey results in the same place of the study site. As a result of the sound recording from the field survey, the species recorded by three researchers of sound analyst include common resident species or species which have a relatively conspicuous call and song. On the contrary, the species recorded by only one researcher are a relatively rare or inconspicuous species or was familiar with the personal experiences of each researcher. The number of species recorded by only one researcher totaled fourteen species (36.8 %), and that of two researchers totaled twelve species (31.6 %), and that of three researchers totaled twelve species (31.6 %). The correlations of the number of species among three researchers was not significant in some results, but the sum or maximum count of the number of species was useful to estimate a significant correlation between the result of researchers and the indices of field surveys. As a result of correlation analysis by using sound recordings, the maximum number of species among the three researchers significantly correlated with the number of species, the species diversity index and the species abundance index in the field survey at the same site, however it did not correlate with the number of individuals and species evenness index. As a result of sound analysis collected from the stationary sound recording system; song meter, the number of frequencies in bird songs and calls correlated with the number of species in the field survey at the same site. The number of calls and songs decreased as time went by. Finally, we could test the active time range and change the activity strength by using a sound recording system. In particular, that sound recording system is able to collect data in same time and site, so it is expected so that the equipment not only replenish a shortage of researchers and survey periods in field research in a short term monitoring survey, but also be able to acquire statistical objectivity.

ANALYSIS THE BIRD COMMUNITY , AUTOMATIC SOUND MONITORING EQUIPMENT , DIFFERENCE AMONG RESEARCHER , SONG METER , BIOPHONY
    Korea Forest Service

    INTRODUCTION

    The field survey is essential to study a bird community in a forested area, and the line census method and point census method were mainly used (Bibbly et al., 1997). Also, some of the most commonly used community indices include the abundance and diversity of species, and the number of species per unit area (Margalef, 1958; MacArthur, 1965; Whittaker, 1972; May, 1975; Magurran, 2004; Buckland et al., 2005). Field surveys are usually conducted where a researcher makes records of all species identified during a certain period at a selected survey area. At this time, a researcher can identify bird species not only visually, but also through bird sounds. The field survey has been playing a key role in understanding the characteristics of bird communities in various habitats and the correlation between various environmental factors and disturbance. However, bird identification results at the survey site often differed among researchers and, in consequence, lowered the credibility of bird survey data. This is due to not only the researchers' experience, knowledge and age, but also due to a difference in the number of bird species that can be identified according to season and different habitats in a survey area (O'Connor et al., 2000, Thompson 2001, Rosenstock et al., 2002). Due to such limitations in this survey method, a new survey method is required that enables a wide area/long-term ecosystem monitoring and can solve difficulties in simultaneous survey and limited manpower (Lawton et al., 1998; Basset et al., 2000). In particular, compared to waterbirds that can be observed from far off, observing forest birds from various points at the same time is almost impossible due to limited manpower. Also, if the purpose of the survey on a forest bird community is to assess the species diversity of a survey site, the number of times the survey is conducted and input manpower is irrelevant. Rather, it will be necessary to find a method that can estimate habitats with high species diversity (Lawton et al., 1998; Basset et al., 2000).

    It was suggested in various studies and preceding papers that using a sound recording system to survey a bird community is very effective in narrowing the gap between identification results among researchers (Parker III 1991, Hobson et al., 2002, Conway and Gibbs, 2005, Acevedo and Villanueva-Rivera, 2006). In particular, researches substantially depend on hearing to identify bird species and their locations because of the complicated vegetation structures and obstacles (leaves, branches) at habitats in forests (Hutto et al., 1986, Ralph et al., 1995). It was possible to identify more bird species in forests or meadows with a high vegetation density with a sound recording system (Parker III 1991, Allen et al., 2004, Conway and Gibbs, 2005) and it was helpful in finding new species in places such as the Amazon forest where the terrain is difficult for people to survey (Haselmayer and Quinn, 2000). Hobson et al. (2002) used a sound recording system attached with an omnidirectional microphone in the survey of bird community in a mature broadleaf forest in Canada, and analyzed the difference between the field survey and the use of a sound recording system. Allen et al. (2004) and Conway and Gibbs (2005) conducted a survey on the bird community by playing the sounds of birds that were found in the wetlands in South Dakota, the United States through speakers and recorded the sounds of birds that responded to those sounds through speakers. Acevedo and Villanueva-Rivera (2006) used an automatic sound recording system at the southern beach area in Puerto Rico and analyzed the efficiency of surveys on the communities of birds and amphibians. Farina et al. (2011) suggested the possibility of using a sound recording system in monitoring the ecology of birds long-term through the soundscape measurement of central Europe. Accordingly, the U.S. Geological Survey has been conducting a study on monitoring wild animals with a sound sensor. Also, the U.S. National Park Service has been carrying out a study to develop a method which uses a sound sensor in surveying the species diversity and population of frogs (USGS 2011). Similar to this, Ontario State in Canada suggested guidelines for monitoring wild animals by using the sound characteristics of the bird community for effective forest management. It also enacted a law on the mandatory use of sound sensors (Center for Northern Forest Ecosystem Research of Ontario, http://www.cnfer.on.ca/SEP/). Also, in the United States, NEON(National Ecological Observatory Network, http://www.neoninc.org/) was established, and cyber-based facilities are being installed at 62 ecological observatories to enable ecological predictions at a continental scale. At those observatories, changes in the ecological elements and biological resources are monitored with various state-of-the-art sensors and computers and network equipment, and a forecasting model is current under development. Also, in Australia, an environmental sound observatory is planned to be established and currently test operations are in progress to monitor changes in the species diversity of wild animals, population and acoustic behaviour due to climate change.

    In this study, validity was assessed through a comparison of field surveys and record surveys to introduce sound record sensors to Korea. The sensors are used to collect bird ecological data and are used in many developed countries. For an objective verification, blind test results among three analysts were compared. Also, a factor, which is highly correlated with species diversity, was deduced to select indices that express the superiority of a community by using sound recording survey results.

    MATERIALS AND METHODS

    1.Survey period and scope

    The survey areas are situated in Baekdudaegan and they include Gombaeryoung and Zochimryoung in Jeombongsan Mountain and Jookryoung and Gochiryoung in Sobaeksan National Park. 8 places in Gombaeryoung, 11 in Zochimryoung, 14 in Jookryoung and 15 in Gochiryoung were selected as the basic survey points. Since the sound recording survey utilizes the number of appearance ratio to show community indices, 2~3 additional survey points were selected, which were 100~150 m away from the basic survey points (Figure 1). The total number of survey points include 186, and field surveys and recordings were carried out once at each point from August to November 2012 (Figure 2). After installing a stationary sound recorder in the fir colony, which is 1,000 m from Jeombongsan Mountain, bird sounds were collected from November 5th to 15th (Figure 2).

    2.Survey and analysis method

    1)Obtaining sound data and field survey

    An on-the-spot verification was held to check how many species were discovered at the recording spots and if the records of species recorded coincide and whether there is a difference among observers. Birds were observed within 50 m of the survey spot and it was done with the naked eye, through binoculars or by listening for bird sounds. Recordings and field surveys were carried out for 5 minutes at each point. SONY PCM-D50 was used to obtain bird sounds and the sampling was recorded in the 22kHz wave format. In case of sound analysis, the Cooledit2000 program was used to play sounds and were classified by using the spectrogram (Figure 1-left). Bird sounds were classified into sounds to attract a mate (song) and other sounds (call) and the number of times birds made sounds was recorded by species. Because it is very difficult to differentiate calls among species during the non-breeding season, they were classified at a family level. Consecutive songs were regarded as one song, and as for a call, it was deemed one call in case a same syllable did not continue for more than three seconds.

    2)Cross validation

    The field survey on the bird community is conducted by judging data that is collected by observing birds with the naked eye or by listening for bird sounds. Therefore, it is necessary to confirm whether differences among individuals affect the final result because there may be human error occurring in the survey based on only sound recording. For such a reason, three researchers were selected and asked to perform an analysis using a blind test in order to compare sound recorded data and field survey data. The number of species recorded by the three researchers was compared and the number of common species was identified to determine whether there was a significant correlation between them.

    3)Correlation between sound recorded data and field survey data

    After the analysis of the three researchers, the species with the maximum population was selected as a variable and its correlation was examined with the field survey results of each area. Because recorded results could not reflect populations, the recorded results of the detailed areas that were divided into three were combined and the number of appearance ratios were equally compared. As for field survey results, the indices of the number of species, population, species diversity, species abundance and species evenness were used to analyze the correlation with recorded results. A Spearman correlation was used to assess relationships and the SPSS21.0 program was used for analysis.

    4)Changes in the frequency of bird sounds by time slot

    One stationary sound recorder was installed at one recording site in Gombaeryoung (the fir colony in the valley 1,000m elevation, Figure 2) from November 11~15, 2012. A correlation according to time was looked into by keeping the records of species and the number of sounds based on sounds recorded during 5 minutes every hour for a day. SM2 from Wildlife Acoustics (http://www. wildlifeacoustics.com) is a bioacoustics recorder that collects sounds and can record sounds at specific times of day and a set the time duration. Data that was collected during the rain was excluded due to the inactivity of birds and also those gathered between 6pm and 6am due to the low frequency of bird sounds.

    RESULTS

    1.Intercomparison of recorded data

    The number of species recorded by three voice analysts were 28, 26 and 20 respectively, a total of 38 species in field survey (Table 1). As the analysis results, the common species recorded by them include resident birds that can be commonly observed or birds that make distinctive sounds: Parus varius, Emberiza elegans, Sitta europaea, Parus major, Phylloscopus coronatus, Dendrocopus kizuki, Parus palustris, Aegithalos caudatus, Microcelis amaurotis, Parus ater and Corvus macrorhynchos (Table 1). On the other hand, Carduelis spinus, Certhia familiaris, Phylloscopus borealis, Emberiza rustica, Butastur indicus and Coccothraustes coccothraustes, which were recorded by only one analyst, are birds that are hard to find or are those make sounds hard to distinguish or that have voices the analyst is familiar with (Table 1). Among 38 species, 14 (36.8 %) of them were observed by a single analyst and 12 (31.6 %) by two and three analysts respectively.

    As we can see from the results, the analysts concluded that there were inconsistencies among their results. Also, there were cases where a correlation did not exist in comparison among individuals (Figure 3: A-B: Spearman's correlation r=0.294, n=41, p=0.062; A-C: r=0.477, n=43, p<0.001; B-C: r=0.567, n=44, p<0.001). However, the species recorded by each analyst showed a correlation with the total number of species and maximum value (Figure 4: A-Total: r=0.750, n=48, p<0.001; A-Max: r=0.807, n=48, p<0.001; B-Total: r=0.732, n=54, p<0.01, B-Max: r=0.757, n=52, p<0.001; C-Total: r=0.701, n=52, p<0.001, C-Max: r=0.668, n=52, p<0.001).

    2.Relationship between the bird community survey using a recorder and an aggregation indices

    As the result of inquiring into a correlation between the community index surveyed at the field and the analysis results of recorded data, they showed a significant correlation (Figure 5). Also, the maximum value of record files analyzed by the three analysts showed a significant correlation with species diversity and species abundance (the number of species, which was identified through the recording, with the maximum count-the number of species detected through field survey: r=0.612, n=50, p<0.01;s species diversity: r=0.507, n=50, p<0.01; species abundance:

    r=0.572, n=50, p<0.01, Figure 5, 6). However, it did not show a correlation with population and species evenness (the number of species, which was identified through recording, with the biggest population: r=0.275, n=50, p=0.53; species evenness: r=0.205, n=50, p=0.154).

    3.Relationship between bird activity detected with a stationary recorder and an community indices

    Recording bird sounds at specific times of day enabled the researchers to identify changes in the activity frequency of birds by time slot (Figure 7). Birds most actively made sounds in the early morning and the calls gradually decreased over time (r=-0.762, n=11, p<0.01). Also, the number of sounds records showed a high correlation with the number of species observed during the field survey (r=-0.743, n=104, p<0.01, Figure 8).

    DISCUSSION

    Bird identification in the field does not depend merely on sounds, but also visual evidence including morphological and behavioral characteristics (Sibley, 2002). However, because there are various situation depending on species and situations, less bird species will be detected by identifying them through recording only compared to the field survey. The number of species identified through data recorded by the analysts showed differences (Table 1). This is thought to be because the survey was carried out during the non-breeding season (August ~ November), the time when birds make similar alert calls, making it difficult to differentiate their voices. However, we thought we could reduce error in case we combine the analysis results obtained by researchers and use the maximum value of the number of species by point. It was expected that if a survey is to be carried out during the breeding season, the time when it is easy to differentiate sounds made by birds, sounds recorded through a recorder will show a more apparent correlation with results obtained through the field survey. Also, we couldn't verify which of the three analysts was correct. In this study, there was a correlation between Researchers A and B, and Researchers A and C, but Researchers C and B displayed a low correlation, showing that results can differ greatly by individual. The number of samples differed by researcher in the analysis results of the correlation among researchers. This is because researchers could not identify bird sounds when they were distracted by other noises such as sounds from the valley and those caused by the wind. However, it is thought to be important to discover that the analysis results based on the field survey conducted with the naked eye and listening and analysis results based only on recordings showed a correlation with community indices such as the number of species and species diversity (Study Result 2).

    As with the results of this study, the survey results of the recording showed a significant correlation with bird community indices. Bird community indices, the number of species, species diversity and species abundance, have been used as important ecological indices from the ecological perspective. And the correlation between the results of recording and community indices implies that ecosystems can be indirectly assessed through sound recordings (MacArthur and MacArthur, 1961; Bradbury et al., 2005). In particular, because forest birds are often identified through sounds (80% in case of this study), it is expected to use such a method for bird survey and habitat analysis. Also, the fact that the results of measuring the number of bird sounds showed a high correlation with the number of species detected through the field survey implies that a survey based on recordings can be a useful assessment indicator when assessing a bird community (Farina et al., 2011).

    As the results of analyzing data from the stationary recorder, birds were not active at sunset and sunrise, and they were more active in the morning than in the afternoon: it coincided with Kim et al. (2013)'s study result, which stated that the appearance ratio was highest in the morning and gradually decreased over time. It may be possible to identify the activity times, relationship with getting up in the morning and long-term changes in trends through sound recordings. In particular, surveying by time slot is expected to supplement the lack of researchers on hand and field survey times in case of conducting a short-term ecological survey.

    The assessment of the bird community based on direct survey in the field has been performed steadily. However, differences in survey points and time can bring about confusion in assessing the bird community. In particular, in case of comparing the bird communities in the highlands and lowlands, the time taken in moving from one position to another may result in a difference of results. Therefore, an effort to eliminate variations was made by surveying from the lowlands to the highlands and the other way around in case that way makes it easier to approach the summit (Yu et al., 2010). However, the results may still show a difference in results from a simultaneous survey. Therefore, data obtained by recording sounds in a fixed place at set time duration can enable an objective and homogeneous comparison of bird communities. In addition, in case of studies that use sound recorders to conduct long-term ecological surveys and study changes in a biota due to climate change, they may be able to produce meaningful results (Depraetere et al., 2012). However, if all recorded data needs to be analyzed manually, it will be impossible to assess ecosystems using sound recorders. Therefore, it is necessary to build a database on songs and calls and develop a system that can automatically analyze voice data and assess bird communities. However, if it takes time to build up a voice DB and if the accuracy is low, using the number of sounds as the indirect index of the ecosystem assessment needs to be considered.

    Figure

    KJEE-29-174_F1.gif

    Site map of the study area (Site A : Gombaeryoung, B: Zochimryoung, C : Jookryoung, D: Gochiryoung). The arrow mark in Site A means the position of stationary recording

    KJEE-29-174_F2.gif

    Spectrogramatic image of the Varied Tit Parus varius by the Cooledit 2000 program (left) and a picture of mounted song meter in Jeombongsan Mountain (right)

    KJEE-29-174_F3.gif

    Correlation of the number of species among each researcher. The numbers in each point indicate sample size

    KJEE-29-174_F4.gif

    Correlation of the number of species between each researcher and results (left : total, right : maximum). The numbers in each point indicate sample size

    KJEE-29-174_F5.gif

    Correlation between recording methods and field surveys in the same site. The numbers in point indicate the sample size

    KJEE-29-174_F6.gif

    Correlation of species diversity and abundance between sound recording and field survey in the same site

    KJEE-29-174_F7.gif

    Change of average total number of calls by recording times

    KJEE-29-174_F8.gif

    Correlation between the number of calls in the recording and the number of species in the field survey

    Table

    The Number of common species which is recorded by each song analysis researcher

    *SDFS: Species detected in Field Survey

    Reference

    1. Allen T , Finkbeiner SL , Johnson DH (2004) Comparison of Detection Rates of Breeding Marsh Birds in Passive and Playback Surveys at Lacreek National Wildlife Refuge, SouthDakota , Waterbirds, Vol.27; pp.277-281
    2. Acevedo MA , Villanueva-Rivera LJ (2006) Using Automated Digital Recording Systems as Effective Tools for the Monitoring of Birds and Amphibians , Wildlife Society Bulletin, Vol.34; pp.211-214
    3. Basset Y , Novotny V , Miller SE , Pyle R (2000) Quantifying biodiversity: experience with parataxonomists and digital photography in Papua New Guinea and Guyana , BioScience, Vol.50; pp.899-908
    4. Bibbly CJ , Burgess ND , Hill DA (1997) Bird census technique, Academic Press, pp.-257
    5. Bradbury R , Hill RA , Mason DC , Hinsley SA , Wilson JD , Baltzer H , Anderson GQA , Whittingham MJ , Davenport IJ , Bellamy PE (2005) Modelling relationships between birds and vegetation structure using airborne LiDAR data: a review with case studies from agricultural and woodland nvironments , Ibis, Vol.147; pp.443-452
    6. Buckland ST , Magurran AE , Green RE , Fewster RM (2005) Monitoring change in biodiversity through composite indices. Philosophical Transactions of the Royal Society B , Biological Sciences, Vol.360; pp.243-254
    7. Conway CJ , Gibbs JP (2005) Effectiveness of call-broadcast surveys for monitoring marsh birds , The Auk, Vol.122; pp.26-35
    8. Depraetere M , Pavoine S , Jiguet F , Gasc A , Duvail S , Sueur J (2012) Monitoring animal diversity using acoustic indices: Implementation in a temperate woodland , Ecological Indicator, Vol.13 (1) ; pp.46-54
    9. Farina A , Pieretti N , Piccioli L (2011) The soundscape methodology for long-term bird monitoring: A Mediterranean Europe case-study , Ecological Informatics, Vol.6 (6) ; pp.354-363
    10. Haselmayer J , Quinn JS (2000) A comparison of point counts and sound recording as bird survey methods in Amazonian southeast Peru , The Condor, Vol.102; pp.887-893
    11. Hobson KA , Rempel RS , Greenwood H , Turnbull B , ilgenburg SLV (2002) Acoustic surveys of birds using electronic recordings: new potential from an omnidirectional microphone system , Wildlife Society Bulletin , Vol.30; pp.709-720
    12. Hutto RL , Pletschet SM , Hendricks P (1986) A fixed-radius point count method for nonbreeding and breeding season use , Auk, Vol.103; pp.593-602
    13. Kim MJ , Lee SD , Kim JS (2013) Selecting the optimal time for the forest birds census in each season , Korean Journal of Environment and Ecology, Vol.27 (2) ; pp.219-229
    14. Lawton JH , Bignell DE , Bolton B , Bloemers GF , Eggleton P , Hammond PM , Hodda M , Holt RD , Larsen TB , Mawdsley NA , Stork NE , Srivastava TB , Watt AD (1998) Biodiversity inventories, indicator taxa and effects of habitat modification in tropical forest , Nature, Vol.391; pp.72-75
    15. MacArthur RH , MacArthur Z (1961) On bird species diversity , Ecology, Vol.42; pp.594-598
    16. MacArthur RH (1965) Patterns of species diversity , Biological Reviews, Vol.40; pp.510-533
    17. Magurran A (2004) Measuring Biological Diversity, Blackwell,
    18. Margalef R (1958) Information theory in ecology , General Systematics, Vol.3; pp.36-71
    19. Cody ML , Diamond JM (1975) Ecology and evolution of communities, Harvard University Press,
    20. O'Connor RJ , Dunn E , Johnson DH , Jones SL , Petit D , Smith CR , Welling E (2000) A programmatic review of the north american breeding bird survey: Report of a Peer Review Panel , [Online] Available at http://www.mp2-wrc.usgs.gov/bbs/index.html,
    21. Parker III TA (1991) On the Use of Tape Recorders in avifaunal surveys , Auk, Vol.108; pp.443-444
    22. Ralph CJ , Sauer JR , Droege S (1995) Monitoring bird populations by points counts, USDA,
    23. Rosenstock SS , Anderson DR , Giesen KM , Leukering T , Carter MF (2002) Landbird counting techniques: current practices and an alternative , The Auk, Vol.119; pp.46-53
    24. Sibley DA (2002) Sibley's birding basics: How to identify birds, using the clues in feathers, habitats, behaviors, and sounds, Knopf Books, pp.-154
    25. USGS (2011) Advanced Tools for Acoustic Sampling of Calling Amphibians. http://www.armi.usgs.gov/story/story.php?contentid=1626,
    26. Whittaker RH (1972) Evolution and measurement of species diversity , Taxon, Vol.21; pp.213-25
    27. Yu JP , Jin SD , Kim HS , Paek WK , Song HK (2010) Characteristics of birds community in relation to altitude, direction of slope and season in Deogyusan National Park , Korean Journal of Ornithology, Vol.17 (4) ; pp.359-385
    Copyright © Korean Society of Environment & Ecology (KSEE). All right reserved.