ArticlePDF Available

A comparison of automated and traditional monitoring techniques for marbled murrelets using passive acoustic sensors

  • Conservation Metrics, Inc.

Abstract and Figures

Autonomous sensors and automated analysis have great potential to reduce cost and increaseefficacy ofwildlifemonitoring.Byincreasingsampling effort, autonomoussensors arepowerfulat detecting rareandelusivespeciessuchasthemarbledmurrelet(Brachyramphusmarmoratus).Newapproachesmustbetestedforcomparability to existing methodologies, so we compared the results of inland audio–visual and of automatedacoustic monitoring for marbled murrelets, conducted during the 2010 breeding season, at 7 sites in the SantaCruz Mountains, California, USA. We found automated acoustic surveys and analysis had fewer detections permorning compared with audio–visual surveyors, but the rate of automated acoustic detections per morning waspositively and strongly correlated with the rate of audio–visual detections per mornings (r¼0.96,P<0.01).Furthermore, acoustic monitoring sampled 10 times more mornings per site (x¼48) than were monitored byhuman surveyors (x¼4.4) at a comparable cost. We used resampling to estimate the power to detect murreletpresence with acoustic sensors at>80% within 8 continuous days of recordings, even at low-activity sites. Ourresults suggest that autonomous sensor and automated analysis approaches could greatly increase the scale andefficacy of murrelet monitoring, allowing formore cost-effective surveyingof large and remoteareas ofpotentialhabitat, as well as, improved ability to measure changes in inland activity. Further study of passive acousticrecordingswouldbevaluableto examine foracousticsignsofbreedingphenology,and site occupancy,if acousticsurveys are to replace the utility of audio–visual surveys. A comparison of automated and traditional monitoring techniques for marbled murrelets using passive acoustic sensors. Available from: [accessed Jan 11, 2016].
Content may be subject to copyright.
Tools and Technology
A Comparison of Automated and Traditional
Monitoring Techniques for Marbled Murrelets
Using Passive Acoustic Sensors
Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, Center for Ocean Health, 100
Shaffer Road, Santa Cruz, CA 95060, USA
PORTIA HALBERT, California State Park, 303 Big Trees Park Road, Felton, CA 95018, USA
MATTHEW W. McKOWN, Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, Center for Ocean Health, 100
Shaffer Road, Santa Cruz, CA 95060, USA
BERNIE R. TERSHY, Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, Center for Ocean Health, 100 Shaffer
Road, Santa Cruz, CA 95060, USA
DONALD A. CROLL, Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, Center for Ocean Health, 100
Shaffer Road, Santa Cruz, CA 95060, USA
ABSTRACT Autonomous sensors and automated analysis have great potential to reduce cost and increase
efficacy of wildlife monitoring.By increasing sampling effort, autonomous sensors are powerful at detecting rare
and elusive speciessuch as the marbled murrelet (Brachyramphusmarmoratus). Newapproaches must be tested for
comparability to existing methodologies, so we compared the results of inland audio–visual and of automated
acoustic monitoring for marbled murrelets, conducted during the 2010 breeding season, at 7 sites in the Santa
Cruz Mountains, California, USA. We found automated acoustic surveys and analysis had fewer detections per
morning compared with audio–visual surveyors, but the rate of automated acoustic detections per morning was
positively and strongly correlated with the rate of audio–visual detections per mornings (r¼0.96, P<0.01).
Furthermore, acoustic monitoring sampled 10 times more mornings per site (x¼48) than were monitored by
human surveyors (x¼4.4) at a comparable cost. We used resampling to estimate the power to detect murrelet
presence with acoustic sensors at >80% within 8 continuous days of recordings, even at low-activity sites. Our
results suggest that autonomous sensor and automated analysis approaches could greatly increase the scale and
efficacy of murrelet monitoring, allowing for more cost-effective surveying of large and remote areas of potential
habitat, as well as, improved ability to measure changes in inland activity. Further study of passive acoustic
recordings would be valuable to examine for acoustic signs of breeding phenology, and site occupancy, if acoustic
surveys are to replace the utility of audio–visual surveys. Ó2015 The Wildlife Society.
KEY WORDS bioacoustics, Brachyramphus marmoratus, California, management, marbled murrelet, methods,
monitoring, murrelets, wildlife.
Marbled murrelets (Brachyramphus marmoratus), a small
seabird (188–269 g) that nests in old-growth forest from
central California to coastal Alaska, USA, are recognized as
exceptionally difficult to monitor because of low detection
rates and remote habitats (Nelson 1997). In fact, the first
record of a marbled murrelet nest was not discovered until
1974, making it one of the last bird species in North America
to have its nest site described (Binford et al. 1975). They are
sensitive to habitat disturbance (Malt and Lank 2009), and
have been U.S. Federally listed in California, Oregon, and
Washington as “threatened” since 1992 (Department of
Interior Fish and Wildlife Service 1992). In particular, their
requirement for old-growth forest for breeding has led to
significant conservation conflict with commercial timber
harvest interests (Raphael 2006). Management decisions are
currently based on a combination of coastal surveys,
watershed-scale radar surveys and inland audio–visual
surveys. Audio–visual surveys by human observers measure
site occupancy, presence or probable absence, and map
breeding distribution (Mack et al. 2003).
However, human surveys are logistically challenging and
costly. Recent comparisons of monitoring data from radar
surveys and audio–visual surveys have revealed potential
sources of detection error, including observer skill (Bigger
et al. 2006b) and murrelet visibility (Rodway and Regehr
2000). Despite advances in radar monitoring (Bigger et al.
2006a), radar units are limited to forest stands with road
access and sites with open lines of sight (Mack et al. 2003).
Although radar may be an effective tool for watershed-wide
population estimates in accessible areas (Burger 2001,
Cooper and Blaha 2002), it has not been considered a
Received: 26 November 2013; Accepted: 5 September 2015
Wildlife Society Bulletin; DOI: 10.1002/wsb.608
Borker et al. Acoustic Monitoring of Murrelets 1
substitute for the inland audio–visual surveys (Mack et al.
2003), which can be used to determine critical breeding
Here, we test the use of autonomous passive acoustic
sensors and automated acoustic analysis as an alternative to
traditional inland monitoring techniques. Autonomous
acoustic sensors are increasingly applied as a tool for
monitoring elusive wildlife, and automated acoustic
analysis helps process large data streams to provide
information at large spatial and temporal scales (Van
Parijs et al. 2009, Buxton 2010, Thompson et al. 2010).
Autonomous acoustic sensors are a potentially cost-
effective way to assess presence and relative activity levels
across large spatial scales for marbled murrelets. In
addition, they can help reduce high sampling variability,
observer bias, and costly repetitive visits to remote field sites
We explored correlations between murrelet activity
measured using traditional inland audio–visual surveys
with indices measured using autonomous acoustic sensor
data processed using automated acoustic analysis at 7
murrelet monitoring sites in the Santa Cruz Mountains,
California. In particular, we examined the power of
autonomous acoustic monitoring and automated analysis
(hereafter, referred to as “acoustic monitoring”) to detect
murrelets and measure levels in activity, as compared with
human audio–visual surveys.
We selected 7 historical inland murrelet monitoring sites in
Big Basin and Butano State Parks, Santa Cruz County,
California (D. L. Suddjian, Command Oil Spill Trustee
Council, unpublished data; Supporting Material, Fig. S1 and
Table S1). Sites were selected to encompassa range of murrelet
activity levels based on data from historical counts in the Santa
Cruz Mountains. This population, in Zone 6 of the Northwest
Forest Plan, is the southernmost extent of the marbled
murrelet range (Raphael 2006). Murrelet detection rates
during traditional surveys in Big Basin State Park have
declined 92% from 1995to 2008, from a mean of 54.5 morning
detections/site to 4 morning detections/site (D. L. Suddjian,
Command Oil Spill Trustee Council, unpublished data).
California State Parks and a private contractor conducted 31
standard inland audio–visual surveys at 7 sites between 16
June and 5 August 2010. Surveys by 5 trained observers were
conducted according to inland forest survey protocol (IFSP;
Mack et al. 2003), for 150 min beginning 30 min before
dawn. All surveys were conducted within the IFSP
monitoring window of 15 April–5 August.
At each of the 7 monitoring sites, we also deployed
a SongMeter SM2 passive acoustic recorder (Wildlife
Acoustics, Concord, MA). We secured each SongMeter
to the trunk of a tree 3–4 m off the ground within 10 m of the
human surveyor’s location. We deployed sensors in mid-June
and collected them in September; we programmed sensors to
record for 3 hours, beginning an hour before dawn. For this
study, we only analyzed recordings during the IFSP survey
period from 15 June to 5 August. We attached an omni-
directional microphone (SMX-II; sensitivity: x¼36 4 dB,
frequency response: 20 Hz–20 kHz, signal-to-noise ratio:
>62 dB) directly to the SongMeter, and recorded on a single
channel at a samplingrate of 20 kHz. This sample rate captured
the range of marbledmurrelet vocalization as wellas other birds
present in the study area.
To detect marbled murrelet “keer” calls (Nelson 1997), we
identified sounds of interest using the spectrogram cross-
correlation detection tool in the eXtensible BioAcoustic Tool
(XBAT; Figueroa 2007)—a bioacoustics analysis package for
MATLAB (The MathWorks 2010). We modified the
software to improve performance in this complicated
soundscape. Specifically, we added 1) stationary noise
reduction to remove the noise component that is uniformly
distributed across time; 2) frequency shifting to increase
detection robustness to shifts in absolute frequency; and 3) an
approach that uses the distribution of cross-correlation scores
across templates for more fine-grained detection accuracy.
For search templates, we selected 5 murrelet keer calls of high
signal-to-noise ratio from the field recordings collected for
this study. We carried out processing with spectrograms
calculated at a fast Fourier transform size of 512, Hann
window, and frame advance setting of 0.336.
After automated processing, we manually reviewed all
events identified as potential murrelet calls by the detector.
Thus, a human reviewer confirmed all murrelet vocalizations
and removed all events that were incorrectly classified by the
software by viewing the spectrogram and listening to
the recording. Most misclassifications were generated by
the songs of American robins (Turdus migratorius) and songs
of Swainson’s thrush (Catharus ustulatus), with features
similar to murrelet keer calls. Finally, we grouped all murrelet
calls separated by <5 s into calling bouts to meet IFSP
guidelines (Mack et al. 2003); each calling bout is henceforth
referred to as a murrelet detection.
We compared human surveys and automated acoustic
monitoring at 2 temporal scales. At the seasonal scale, we
calculated mean rate of morning detections for all sites (n¼7)
with each method (inland audio–visual and acoustic) and
compared them using Pearson’s product-moment correlation
coefficient. We calculated a 95% confidence interval around
the correlation coefficient by resampling 7 points with
replacement bootstrapped 1,000 times to examine the
influence of outliers. At the individual morning scale, we
compared automated and human IFSP detections during
simultaneous surveys (n¼29) using the same approach
(Supporting Material, Fig. S2). Two sample comparisons of
detection rates weredone with paired Wilcoxon signed-ranked
tests to address non-normality. All statistical comparisons and
analysis were conducted in the R programming environment
(R Development Core Team 2011). We used a P-value
threshold of 0.05 to assess significance and a P-value of 0.10 for
trends (i.e., marginal support) for all tests.
We used a resampling exercise to estimate the power of
acoustic monitoring to detect the presence of murrelets if not
2 Wildlife Society Bulletin 9999
deploying sensorsfor the entire breeding season. We calculated
the cumulative likelihood of detecting 1 murrelet call given
successive mornings of acoustic monitoring from resampled
continuous setsof mornings with random start datesduring the
IFSP monitoring window.
We estimated costs of a hypothetical 10-year acoustic
monitoring program based on initial equipment investment
and annual staff time needed to collect and analyze
recordings, and produce a final summary report (Supporting
Material, Table S2). We compared our estimates of cost
per site per season with the cost of previous marbled
murrelet monitoring activities conducted at Big Basin State
Park (P. Halbert, personal communication).
Between 15 June and 5 August, autonomous acoustic sensors
recorded for 338 mornings at 7 sites, tallying 2,463 murrelet
detections. Simultaneously, we conducted 29 inland audio–
visual surveys, with a minimum of 3 surveys/site, and tallied
724 detections. Both acoustic monitoring and audio–visual
surveys detected marbled murrelets at all 7 monitoring sites
(Table 1; Fig. 1).
Automated Recording and Analysis
Autonomous sensors performed well, with only one sensor
malfunctioning after 34 mornings. Other sensors recorded
from 45 to 52 continuous mornings spanning peak murrelet
activity. These sensors functioned for as long as 82 mornings,
but we removed these recordings outside the IFSP window
from our analysis. Spectrogram cross-correlation flagged
19,216 sounds as potential murrelet vocalizations all of which
were reviewed by 2 human observers (34 hr of review).
Thirty-eight percent of those sounds were positively
identified as murrelets (7,218 calls). Of those, 2,463 murrelet
calls occurred >5 s after other murrelet calls and were
recorded as independent murrelet detections.
Seasonal Comparisons of Acoustic Activity and Human
Audio–Visual Surveys
At all monitoring sites the mean rate of automated acoustic
detections was less than half of the mean rate of human inland
audio–visual detections (x¼6.8 vs. 19.3 detections/morning;
1-tailed paired Wilcoxon signed-rank test, W¼2, P¼0.02;
Table 1). Because, on average, autonomous acoustic sensors
sampled 10 times more mornings than human surveyors, the
mean total number of automated acoustic detections across
sites throughout the season was >3 times higher than the total
number of human detections per site; however, nonparametric
approaches found only marginal support for a greater central
tendency (x¼351.9 vs. 103.4 detections/season, 1-tailed
paired Wilcoxon signed-rank test, W¼23.5, P¼0.06). The
mean rate (acoustic detections per morning) of acoustic activity
at each site was positively correlated with the mean rate of
human audio–visual detections per morning (Bootstrapped
Pearson’s correlation coeff. ¼0.96, 95% CI ¼0.857–0.999;
Fig. 2).
Table 1. Marbled murrelet activity detected during the 2010 breeding season by acoustic and traditional audio–visual surveys at 7 sites in the Santa Cruz
Mountains, California, USA.
Automated acoustic detections/morning Human detections/morning
Site xSD Mornings sampled xSD Mornings sampled
GMCA 11.9 10.5 51 48.6 32.4 9
GSCR 1.0 2.5 45 6.9 4.2 7
HUCK 1.3 5.6 52 0.3 0.6 3
HUND 0.1 0.3 52 2.0 2.6 3
LBUT 29.4 22.1 52 61.3 53.7 3
RDWD 3.9 6.6 52 15.3 18.9 3
SEMP 0.1 0.5 34 0.7 1.2 3
GMCA, Gazos Camp; GSCR, Girl Scout Creek; HUCK, Huckleberry; HUND, 100 Acre Woods; LBUT, Little Butano Creek; RDWD, Redwood
Meadow; SEMP, Sempevirens.
Figure 1. Seasonal activity of marbled murrelets detected by automated
acoustic sensors and analysis (black lines) and human audio–visual surveys
(gray triangles connected by lines) during the 2010 breeding season at 7
sites in the Santa Cruz Mountains, California, USA. GMCA, Gazos
Camp; GSCR, Girl Scout Creek; HUCK, Huckleberry; HUND, 100 Acre
Woods; LBUT, Little Butano Creek; RDWD, Redwood Meadow;
SEMP, Sempevirens.
Borker et al. Acoustic Monitoring of Murrelets 3
Simultaneous Comparisons of Acoustic Activity and
Human Audio–Visual Surveys
audio–visual surveyors (x¼4.8 vs. 24.9 detections/
morning [1-tailed paired Wilcoxon signed-rank test,
W¼9, P<0.01]), but these indices of activity were
positively correlated (Bootstrapped Pearson’s correlation
coeff. ¼0.82, 95% CI ¼0.588–0.937).
Power Analysis for Presence or Absence
Automated acoustic monitoring exceeded a 90% mean
likelihood of detecting murrelets after 10 mornings of
acoustic monitoring at all 7 sites. In general, sites with lower
activity levels required a greater number of surveys in order to
achieve a 90% likelihood of detecting presence (Fig. 3).
Cost Estimates
The cost of monitoring 7 sites with acoustic sensors,
including staff time for analysis, and long-term data storage,
was estimated at US$7,780 (US$1,111/site) with purchasing
acoustic sensors. Assuming a 10-year monitoring program
(and a 10-year life of equipment), with equipment
investment spread across years, automated acoustic moni-
toring costs US$427/site/year. Previous contracts in
the Santa Cruz Mountains have cost approximately US
$432/survey to conduct human audio–visual surveys (P.
Halbert, personal communication). Given the norm of 3
surveys/year, the cost of site per year is approximately US
$1,296, or US$9,072 for 7 sites.
This study provides more evidence that automated sensors
are a powerful tool for wildlife monitoring, by increasing the
temporal and spatial scale of sampling and reducing biases
(Gauthreaux and Belser 2003, Porter et al. 2005, Rovero and
Figure 2. The relationship between relative activity levels of marbled
murrelets measured with 2 methods, automated acoustic monitoring, and
human audio–visual surveys during the 2010 breeding season, at 7 sites the
Santa Cruz Mountains, California, USA.
Figure 3. Likelihood of detecting marbled murrelets after successive days of automated acoustic monitoring during the 2010 breeding season, at 7 sites in the
Santa Cruz Mountains, California, USA. Shading is ranked from lowest levels of activity (light gray) to highest levels of activity (black) as measured by human
surveyors. Symbols denote different sites. GMCA, Gazos Camp; GSCR, Girl Scout Creek; HUCK, Huckleberry; HUND, 100 Acre Woods; LBUT, Little
Butano Creek; RDWD, Redwood Meadow; SEMP, Sempevirens.
4 Wildlife Society Bulletin 9999
Marshall 2009). Passive acoustic monitoring of vocal wildlife
is a scalable solution for achieving monitoring goals (Grava
et al. 2008, Blumstein et al. 2011, Borker et al. 2014), and has
proven effective with rare and elusive species (Wade et al.
2006, Thompson et al. 2010). Removal of a human observer
comes with some statistical and cost advantages, but
no microphone will match the ecological insights to be
gained from a human observer. Our results indicate that
compared with traditional surveys, automated acoustic
sensors detected fewer murrelets during each morning, but
greatly expand the number of mornings that can be sampled
and the total number of murrelet detections at the seasonal
scale for a comparable cost. At a broad scale, detecting the
presence of marbled murrelets is important to prioritize areas
for occupancy surveys and potential management actions.
Both traditional and acoustic methods succeeded in
detecting murrelets at all sites in the study. Given the
relatively high cost of human surveys, and the ineffectiveness
of a single survey to suggest probable absence, automated
acoustic recording seems a promising technique to survey
large remote areas for murrelets.
Acoustic monitoring had an 80% likelihood of detecting
murrelet presence by 8 mornings, even at sites with low levels
of murrelet activity. These results suggest that sensors could
be moved across sites throughout a season to survey multiple
sites for detecting presence, providing potential cost savings.
The inland forest survey protocol has set a standard of 5
audio–visual surveys to determine murrelet presence,
costing an estimated US$1,400–2,160. Comparatively, near-
continuous acoustic sampling of a site throughout the entire
breeding season could cost US$1,111, or less if part of a
larger or longer term monitoring program. Bigger et al.
(2006a) reported a less costly US$280/survey in northern
California, but did not include costs of data management and
report writing.
Although acoustic monitoring was able to detect murrelets
at all sites, it barely managed to do so at 2 sites with <5
detections. At both those sites audio–visual surveyors only
detected murrelets on one of the 3 surveys. Given limited
budget and unlimited time, acoustic sensors may provide a
more effective tool for detecting presence. However,
ignoring expense, repeated audio–visual surveys would
provide the fastest way to assess murrelet presence at very
low activity sites, because they achieve higher rates of
detection given an equal number of mornings sampled.
Activity levels measured by human surveyors and acoustic
sensors were highly correlated. At a given sampling effort
(mornings), human observers consistently detected more
The potential utility of autonomous sensors and analysis
for detecting threatened species is promising, because staff
and financial resources are a major limitation in all
monitoring programs. Furthermore, costly monitoring can
divert resources from important conservation actions. For
murrelets, acoustic sensors could be widely applied over
large areas of suspected breeding habitat to survey areas for
presence and probable absence. Acoustic monitoring
results could guide intensive surveys for determining
occupancy, which is currently delineated mainly by visual
cues and very rare acoustic events (Mack et al. 2003).
Future directions should include comparing acoustic
activity at occupied and unoccupied breeding sites, and
trying to detect rare acoustic signs of breeding occupancy.
Ongoing studies have used acoustic tools to measure
seabird abundance, phenology, and breeding occupancy
(Buxton and Jones 2012, Borker et al. 2014). Seasonal
patterns of acoustic activity at occupied sites should be
investigated as an indicator of breeding effort where
breeding phenology is independently measured. Finally, a
watershed-scale network of sensors might be used to
identify areas of high calling activity and gradients,
identifying potential nesting areas. Acoustic activity across
a watershed could be paired with radar monitoring to
evaluate acoustic indices of abundance.
Unlike traditional human audio–visual surveys that
produce no permanent record, acoustic sensors created
>1,000 hr of acoustic recordings that can be reanalyzed and
interpreted as new questions arrive. By allowing reanalysis of
archived data sets, these data streams eliminate inter- and
intra-observer biases. Compared with human surveys, an
entire season of acoustic recordings can be collected at
minimal cost, and sensors can be deployed in remote areas
not accessible by trained observers for multiple human or
radar surveys.
Further study is needed to examine the relationship
between acoustic activity and breeding status, but sites with
acoustic detections could be prioritized for human surveys
that can be used to establish occupancy. This could greatly
expand the inland surveying effort for murrelets on a fixed
budget. Monitoring at inland sites is important because
murrelets have sensitive breeding requirements and are
threatened by habitat loss and habitat management (Peery
et al. 2004, Raphael et al. 2013). Actions that affect breeding
habitat (including timber harvest or predator control) should
be evaluated at those sites and compared with inland controls
to identify the most effective actions for conservation and for
mitigation of impacts.
The application of passive acoustic monitoring to marbled
murrelets is promising. Despite lower detection rates during
simultaneous audio–visual surveys and acoustic recordings
(likely to be improved with new analysis tools), automated
sensors consistently detected more murrelets by sampling
>10 times more mornings than human surveyors, at less than
a fifth of the cost.
We acknowledge the assistance of S. Singer, T. Kastner, D.
Suddjian, California State Parks, Klamath Wildlife Resour-
ces, C. Sullivan, R. W. Henry, and an uncertain abundance
of marbled murrelets. Additional thanks to anonymous
reviewers, C. Ribic, L. Webb, and T. Mabee for their
comments on the manuscript. Funding was provided by the
Packard Foundation Marine Birds Program. D. Croll, B.
Tershy own shares in, and M. McKown owns shares in and
works for Conservation Metrics Inc., a company that
provides acoustic wildlife monitoring services.
Borker et al. Acoustic Monitoring of Murrelets 5
Bigger, D., M. Z. Peery, J. Baldwin, S. Chinnici, and S. P. Courtney. 2006a.
Power to detect trends in marbled murrelet breeding populations using
audiovisual and radar surveys. Journal of Wildlife Management
Bigger, D., M. Z. Peery, S. Chinnici, and S. P. Courtney. 2006b. Efficacy of
audiovisual and radar surveys for studying marbled murrelets in inland
habitats. Journal of Wildlife Management 70:505–516.
Binford, L. C., B. G. Elliott, and S. W. Singer. 1975. Discovery of a nest and
the downy young of the marbled murrelet. The Wilson Bulletin
Blumstein, D. T., D. J. Mennill, P. Clemins, L. Girod, K. Yao, G. Patricelli,
J. L. Deppe, A. H. Krakauer, C. W. Clark, and K. A. Cortopassi. 2011.
Acoustic monitoring in terrestrial environments using microphone arrays:
applications, technological considerations and prospectus. Journal of
Applied Ecology 48:758–767.
Borker, A. L., M. W. McKown, J. T. Ackerman, C. A. Eagles-Smith, B. R.
Tershy, and D. A. Croll. 2014. Vocal activity as a low cost and scalable
index of seabird colony size. Conservation Biology 28:1100–1108.
Burger, A. E. 2001. Using radar to estimate populations and assess habitat
associations of marbled murrelets. Journal of Wildlife Management
Buxton, R. T. 2010. Monitoring and managing recovery of nocturnal
burrow-nesting seabird populations on recently predator-eradicated
Aleutian Islands. Thesis, Memorial University of Newfoundland, St.
John’s, Canada.
Buxton, R. T., and I. L. Jones. 2012. Measuring nocturnal seabird activity
and status using acoustic recording devices: applications for island
restoration. Journal of Field Ornithology 83:47–60.
Cooper, B. A., and R. J. Blaha. 2002. Comparisons of radar and audio-visual
counts of marbled murrelets during inland forest surveys. Wildlife Society
Bulletin 30:1182–1194.
Department of Interior Fish and Wildlife Service. 1992. Determination of
threatened status for the Washington, Oregon and California population
of the marbled murrelet. Federal Register 57:45328–45337.
Figueroa, H. 2007. XBAT. v5. Cornell University Bioacoustics Research
Program, Ithaca, New York, USA.
Gauthreaux, S., Jr., and C. Belser. 2003. Radar ornithology and biological
conservation. The Auk 120:266–277.
Grava, T., N. Mathevon, E. Place, and P. Balluet. 2008. Individual acoustic
monitoring of the European eagle owl Bubo bubo. Ibis 150:279–287.
Mack, D., W. Ritchie, S. K. Nelson, E. Kuo-Harrison, P. Harrison, and T.
Hamer. 2003. Methods for surveying for marbled murrelets in forests: a
revised protocol for land management and research. Pacific Seabird
Group, Marbled Murrelet Technical Committee, Arcata, California,
Malt, J. M., and D. B. Lank. 2009. Marbled murrelet nest predation risk in
managed forest landscapes: dynamic fragmentation effects at multiple
scales. Ecological Applications 19:1274–1287.
Nelson, S. K. 1997. Marbled murrelet (Brachyramphus marmoratus). Account
276 in A. Poole, editor. The birds of North America online. Cornell Lab
of Ornithology, Ithaca, New York, USA.
bna/species/276. doi:10.2173/bna 276
Peery, M. Z., S. R. Beissinger, S. H. Newman, E. B. Burkett, and T. D.
Williams. 2004. Applying the declining population paradigm: diagnosing
causes of poor reproduction in the marbled murrelet. Conservation
Biology 18:1088–1098.
Porter, J., P. Arzberger, H.-W.Braun, S. Gage, T. Hansen, P. Hanson, C.-C.
Lin, F. Lin, T. Kratz, W. Michener, S. Shapiro, P.Bryant, and T. Williams.
2005. Wireless sensor networks for ecology. BioScience 55:561.
R Development Core Team. 2011. R: a language and environment for
statistical computing. Volume 1. Tertiary R: a language and environment
for statistical computing. R Foundation for Statistical Computing,
Vienna, Austria.
Raphael, M. G. 2006. Conservation of the marbled murrelet under the
Northwest Forest Plan. Conservation Biology 20:297–305.
Raphael, M. G., G. A. Falxa, and J. O’Callaghan. 2013. From trees to seas—
marbled murrelet numbers are down. Science Findings 157. U.S.
Department of Agriculture, Portland, Oregon, USA.
Rodway, M. S., and H. M. Regehr. 2000. Measuring marbled murrelet
activity in valley bottom habitat: bias due to station placement. Journal of
Field Ornithology 71:415–422.
Rovero, F., and A. R. Marshall. 2009. Camera trapping photographic rate as
an index of density in forest ungulates. Journal of Applied Ecology
The MathWorks. 2010. MATLAB: the language of technical computing v.
R2009a. The MathWorks, Natick, Massachusetts, USA.
Thompson, M. E., S. J. Schwager, K. B. Payne, and A. K. Turkalo. 2010.
Acoustic estimation of wildlife abundance: methodology for vocal
mammals in forested habitats. African Journal of Ecology 48:654–661.
D. Risch, and I. I. C. Van Opzeeland. 2009. Management and research
applications of real-time and archival passive acoustic sensors over varying
temporal and spatial scales. Marine Ecology Progress Series 395:21–36.
Wade, P., M. P. Heide-Jørgensen, K. Shelden, J. Barlow, J. Carretta, J.
Durban, R. LeDuc, L. Munger, S. Rankin, A. Sauter, and C. Stinchcomb.
2006. Acoustic detection and satellite-tracking leads to discovery of rare
concentration of endangered North Pacific right whales. Biology Letters
Associate Editor: Webb.
Additional supporting information (survey site information,
cost estimates, and graphical comparison of methods during
simultaneous surveys) may be found in the online version of
this article at the publisher’s web-site.
Figure S1. Map of marbled murrelet monitoring stations
(both automated acoustic and human surveys) used in this
study within the Santa Cruz Mountains, California, USA.
Figure S2. Relationship of automated acoustic and human
detections of marbledmurrelets on the same 29 mornings across
7 sites the Santa Cruz Mountains, USA. Bootstrapped Pearson’s
correlation coefficient ¼0.82, 95% CI ¼0.588–0.937.
Table S1. Locations and full names of marbled murrelet
monitoring sites used in this study within the Santa Cruz
Mountains, California, USA.
Table S2. Cost estimates for acoustic marbled murrelet
monitoring at 7 sites in the Santa Cruz Mountains, USA, for
10 years.
6 Wildlife Society Bulletin 9999
... Automated recording units (ARUs) may overcome challenges associated with surveying for ruffed grouse by increasing temporal sampling effort (Borker et al. 2015, Colbert et al. 2015, Brauer et al. 2016, Darras et al. 2019, Gibb et al. 2019. Recently, the availability of low-cost ARUs such as the AudioMoth (Hill et al. 2019) has lowered the barrier to performing acoustic surveys at large spatial scales. ...
Ruffed grouse (Bonasa umbellus) populations are declining throughout their range, which has prompted efforts to understand drivers of the decline. Ruffed grouse monitoring efforts often rely on acoustic drumming surveys, in which a surveyor listens for the distinctive drumming sound that male ruffed grouse produce during the breeding season. Field‐based drumming surveys can fail to detect ruffed grouse when the birds drum infrequently or irregularly, making this species an excellent candidate for remote acoustic sensing with automated recording units (ARUs). An accurate automated recognition method for ruffed grouse drumming could enable effective and efficient use of ARU data for monitoring efforts; however, no such tool is currently available. Here we develop an automated method for detecting ruffed grouse drumming in audio recordings. Our detector uses a signal processing pipeline designed to recognize the accelerating pattern of drumming. We show that the automated recognition method accurately and efficiently detects drumming events in a set of labeled ARU field recordings. In a case study with 56 locations in Central Pennsylvania, we compared detections of ruffed grouse from 4 survey methods: field‐based acoustic drumming surveys, surveys conducted by humans listening to ARU recordings, and automated recognition for both a 1‐day and a 28‐day period. Field‐based surveys detected drumming at 9 of 56 locations (16%), while surveys conducted by humans listening to ARU recordings detected drumming at 8 locations (14%). Using automated recognition, the 1‐day recording period produced detections at 17 locations (30%) and the 28‐day recording period produced detections at 34 locations (61%). Our case study supports the idea that automated recognition can unlock the value of ARU datasets by temporally expanding the survey period. We provide an open‐source Python implementation of the recognition method to support further use in ruffed grouse monitoring efforts. A novel automated recognition method uses signal processing to identify ruffed grouse drumming in audio recordings. In our case study, using the automated method on field recordings produces more detections than field‐based human point counts and requires minimal manual review.
... Other explanations are the passive quality of recorders, which negate flushing/avoidance effects created by human observers (Darras et al., 2018) and can generally be in the field at times when observers cannot, such as night or dawn. Other studies comparing point counts to identifications from an audio file found either similar outcomes between methods (Alquezar & Machado, 2015;Castro et al., 2019;Darras et al., 2018;McGuire et al., 2011;Van Wilgenburg et al., 2017;Yip et al., 2017) or that recorders outperformed humans (Borker et al., 2015;Digby et al., 2013;Haselmayer & Quinn, 2000;Hutto & Stutzman, 2009;Klingbeil & Willig, 2015;Sedláček et al., 2015;Shaw, Hedes, et al., 2021;Tegeler et al., 2012;Venier et al., 2012;Zwart et al., 2014). However, most of these studies compared consecutive recordings with consecutive point count minutes, where the visual advantage of point counts is maximized and the temporal distribution advantage of recorders is nullified. ...
Full-text available
Manually annotating audio files for bird species richness estimation or machine learning validation is a time‐intensive task. A premium is placed on the subselection of files that will maximize the efficiency of unique additional species identified, to be used for future analyses. Using acoustic data collected in 17 plots, we created 60 subsetting scenarios across three gradients: intensity (minutes in an hour), day phase (dawn, morning, or both), and duration (number of days) for manual annotation. We analyzed the effect of these variables on observed bird species richness and assemblage composition at both the local and entire study area scale. For reference, results were also compared to richness and composition estimated by the traditional point count method. Intensity, day phase, and duration all affected observed richness in decreasing respective order. These variables also significantly affected observed assemblage composition (in the same order of effect size), but only the day phase produced compositional dissimilarity that was due to phenological traits of individual bird species, rather than differences in species richness. All annotation scenarios requiring equal sampling effort to point counts yielded higher species richness than the point count method. Our results show that a great majority of species can be obtained by annotating files at high sampling intensities (every 3 or 6 min) in the morning period (post‐dawn) over a duration of two days. Depending on a study's aim, different subsetting parameters will produce different assemblage compositions, potentially omitting rare or crepuscular species, species representing additional functional groups and natural history guilds, or species of higher conservation concern. We do not recommend one particular subsetting regime for all research objectives, but rather present multiple scenarios for researchers to understand how intensity, day phase, and duration interact to identify the best subsetting regime for one's particular research interests. This study examines the effect of 60 subsampling scenarios on the resulting observed richness and composition estimates of bird assemblages. Results demonstrate how time of morning, duration, and recording intensity differentially affect resulting metrics. Results also identify the most efficient scenarios for accumulating species richness, all of which outperform the traditional point count method with equal or less time effort.
... Passive acoustic observations, made possible by the ongoing development of long-term hydrophone/ microphone monitoring stations, also hold promise for detecting biological responses to management (Borker et al., 2015, Buston et al., 2018. In addition to capturing acoustic signatures of organisms as a proxy for presence/abundance, such recordings can also capture changes in animal behaviors (e.g., reproductive behaviors, aggression). ...
... Compared to previous methods relying on direct field observations, soundscape-based monitoring is non-invasive and non-observer bias (Machado et al., 2017;Myers et al., 2019;Ng et al., 2018;van La and Nudds, 2016). It does not require an observer while many locations can be surveyed simultaneously using automated recorders (Borker et al., 2015;Burivalova et al., 2019). In addition, even night birds and birds with low vocal activity can be counted effectively (Pijanowski et al., 2011a). ...
Urban foresters are addressing the challenge of urban biodiversity loss through management plans in the context of rapid urbanization. Protecting the integrity of the urban ecosystem requires long-term monitoring and planning for resilience as well as effective management. The soundscape assessment has attracted attention in this field, but applying the soundscape assessment in urban ecological monitoring requires a protocol that links soundscapes to the impact of resource management on biodiversity over time. The effective processing and visualization of large-scale data also remains an important challenge. The aim of this study was to better understand the relationship between soundscape and physical environment, and examine the feasibility of this innovative soundscape approach in highly urbanized areas. Soundscape recordings were collected for 20 urban parks twice on 4 consecutive days in Spring. A total of 691,200 min of sound material were automatically obtained. In order to track the spatio-temporal patterns of a soundscape and determine its potential suitability for ecosystem monitoring, our study characterized soundscape information by adopting 4 widely used acoustic indices: acoustic diversity index (ADI), bioacoustic index (BIO), normalized difference vegetation index (NDSI), and power spectral density (PSD). Daily patterns of PSD have provided a potential connection between soundscapes and bird songs, and 1–2 kHz presented a similar pattern that was linked to human activity. Through further modeling, we tested the relationship of soundscapes to physical environment characteristics. The results showed the importance of habitat vegetation structure for acoustic diversity. More vertical heterogeneity, with an uneven canopy height or multilayered vegetation, was associated with more acoustic diversity. This suggests that clearing ground cover may have a significant negative impact on wildlife. Our results suggest that soundscape approaches provide a way to quickly synthesize large-scale recording data into meaningful patterns that can track changes in bird songs and ecosystem conditions. The proposed approach would enable regular assessment of urban parks and forests to inform adaptive planning and management strategies that can maintain or enhance biodiversity.
... Acoustic monitoring can be used to detect their occurrence by recording their vocalisations to detect and identify them rather than their visual presence. This means smaller elusive species, such as marine life, birds, amphibians and insects can be recorded (Borker et al. 2015. Capturing sound can improve coverage as well as resolution compared to visual techniques. ...
Biodiversity data-gaps must be better understood to inform conservation policy. Scalable technology, such as camera traps and satellite imaging, have been shown to increase coverage. This research explores the field of acoustic monitoring, which although long established, has struggled to scale effectively due to cost, usability and power inefficiency of existing equipment. This research aims to investigate whether creating an advanced, power-efficient and low-cost acoustic hardware solution can expand coverage. User-centred design principles and aspects of the collaborative economy are adopted in order to design a fit-for-purpose solution to scalability. The hardware design takes inspiration from the utilitarian construction of single-board computers and exploits the recent availability of advanced smartphone and Internet of Things technology. The resulting open source device, AudioMoth, is described, in which the baseline levels of performance improve on existing tools, with better power efficiency, miniature overall dimensions, reduced material cost, and the ability to capture audible and ultrasonic sound simultaneously. Open source hardware, however, imposes barriers to entry for non technical users such as conservation practitioners. To improve access, it is necessary to remove the do-it-yourself nature of construction while remaining low-cost and flexible to adapt. Barriers can be overcome using new collaborative methods of consumption, where crowds can accumulate funds to bulk manufacture and automate hardware assembly with an economy of scale. A collaborative management framework is proposed, in which guidelines enable users to acquire fully assembled open source hardware from crowdfunding opportunities. The framework is applied to AudioMoth, permitting individual devices to be acquired ready-to-use for $49.99 with approximately 8,000 delivered to date. This general system has provided conservation practitioners with access to an adaptable hardware solution, thus expanding the coverage of monitored biodiversity. Conservation policy should consider user-centred design in all new technical innovations and further explore the work outlined in this thesis, thereby allowing those communities outside of the pockets of wealth and high opportunity to monitor biodiversity at low cost.
... Other studies comparing point counts to acoustic recordings found either similar outcomes between methods (Alquezar and Machado, 2015;Castro et al., 2019;Celis-Murillo et al., 2012;Darras et al., 2018;Klingbeil and Willig, 2015;McGuire et al., 2011;Van Wilgenburg et al., 2017;Yip et al., 2017) or that recorders outperformed humans (Borker et al., 2015;Digby et al., 2013;Haselmayer and Quinn, 2000;Hutto and Stutzman, 2009;Klingbeil and Willig, 2015;Tegeler et al., 2012;Venier et al., 2012;Zwart et al., 2014).The differences between study outcomes depend on factors such as distance from recorder, recorder type, species of interest, vegetation density, habitat type, climatic zone and if detection probabilities were calculated and standardised. These studies were not conducted in winter, however, and they noted that is it difficult to directly compare sampling methods, all factors considered. ...
Full-text available
Resident birds in boreal forests can serve as indicators of habitat quality and are often species of conservation interest, particularly in multifunctional forests also used for timber production. To make informed forest management decisions, we must first understand which structural features provide habitats useful for resident birds. This is particularly true in winter, an understudied and critical season for their survival. The objective of this study was to establish reliable methods for monitoring bird presence and activity during winter, and to use these methods to evaluate the relative importance of stand structural features to make inferences about which features support and increase winter survival potential. Using a hybrid bioacoustic and ecoacoustic approach, we tested the ability of acoustic recordings to identify links between bird diversity and components of structural complexity, and compared these results to those from the traditional point count method. We conducted a vegetation survey, point count surveys and collected acoustic recordings from December 2019–February 2020 in 19 sites in a Swedish boreal forest. First, we compared species richness values derived from point counts and bioacoustic monitoring methods. Bioacoustic species richness was significantly higher than point count richness, although only when the time spent identifying species from recordings exceeded the time spent conducting point counts in the field. Next, we demonstrated that bioacoustic species identification yields additional metrics of bird activity that point counts cannot. We tested the response of these metrics, and point count metrics, to variables of structural heterogeneity and complexity of our sites. Almost all bioacoustic metrics increased significantly with increasing structural complexity, while point count richness and abundance did not, indicating that automated recording is more effective in identifying forest patches of high quality in winter. Lastly, using an ecoacoustic approach, we calculated six of the most common acoustic indices and tested if any could effectively reflect the bird-structure relationships described above. Two indices showed significant positive relationships to bioacoustic metrics, demonstrating their potential as biodiversity assessment proxies that respond to differences in habitat quality. This is the first winter acoustic study to monitor bird assemblages in detail; it employed both bioacoustic and multi-index ecoacoustic approaches, which provided evidence that automated acoustic recording can be an effective and superior method for monitoring resident forest birds.
Full-text available
We present PNW-Cnet v4, a deep neural net with an associated Shiny-based application designed to facilitate efficient data processing to detect terrestrial wildlife species through passive acoustic monitoring. PNW-Cnet v4 is a deep convolutional neural network that detects audio signatures of 37 focal species of birds and mammals that inhabit forests of the Pacific Northwest, USA, along with other commonly occurring forest sounds. The primary objective of developing PNW-Cnet v4 was to support a long-term northern spotted owl ( Strix occidentalis caurina ) monitoring program. By incorporating additional species classes, PNW-Cnet v4 expands applicability of the program to broadscale biodiversity research and monitoring. Using the Shiny app with PNW-Cnet v4, users can process audio data using a graphical user interface, summarize apparent detections visually, and export results in tabular format.
Full-text available
African Penguins (Spheniscus demersus) are endangered and declining seabirds which make extensive use of vocal signals for intra‐specific vocal communication. Accordingly, passive acoustic monitoring tools could be developed as robust population monitoring methods that cause minimal disturbance to the birds. In this study, we collected soundscape recordings at the Stony Point penguin colony (Betty’s Bay, South Africa) during the breeding season in 2019 (i) to document the circadian rhythms of vocal activity of this species (ii) to investigate whether the magnitude of variation of three different ecoacoustic indices correlates with the number of ecstatic and mutual display songs counted in recordings, which might inform on the breeding activity of the colony. Indeed, while ecstatic display songs are produced by males during intersexual competition and territorial defence, mutual display songs are given by parents returning to the nest after foraging trips. We found that the vast majority of the display songs (>80%) occurred between 04:00‐08:00 and 17:30‐21:30. We also found that the Acoustic Entropy index was a good predictor of the number of penguins’ songs within a recording. Overall, our study shows that African penguins vocalisations have the potential to assist the monitoring of this species while minimising disturbance.
Full-text available
The moon phase affects the ecology and vocal activity of nightjars (Caprimulgidae). However, some studies have found contradictory results regarding the impact of the moon phase on the vocal activity of nightjars. To increase our knowledge on this topic, we monitored the vocal behavior of two Neotropical nigthjars, the Little nightjar (Setopagis parvula) and the Common pauraque (Nyctidromus albicollis), over 5 lunar cycles in the Brazilian Pantanal. We tested the relationships between the moon phase and daily vocal output (number of calls uttered) and the proportions of calling activity at dusk, midnight, and dawn. Our results suggest that moonlight stimulated the vocal output of both species, since it was between 6 and 8 times higher during full moon nights than during new moon nights. Likewise, the proportion of calling activity at midnight was significantly higher during full moon nights. In contrast, the proportion of calling activity of both species was higher at dawn during new moon nights than under the full moon. The calling activity of the Common pauraque was also higher at dusk during new moon nights. These findings might be partly related to the much lower vocal output at midnight during full moon nights and therefore higher proportions of vocal activity at dusk and at dawn under new moon scenarios. This is the first study comparing the vocal behavior of two Neotropical nightjars over different moon phases and shows that the impact of moonlight may differ between species and at a daily scale when analyzing the periods with the highest and lowest illumination. The consequences of the increase in vocal output under moonlight are unknown and should be assessed.
Full-text available
Passive acoustic monitoring is a non‐invasive tool for automated wildlife monitoring. This technique has several advantages and addresses many of the biases related to traditional field surveys. However, locating animal sounds using Autonomous Recording Units (ARUs) can be technically challenging, and therefore ARUs have traditionally been little employed to estimate animal density. Nonetheless, several approaches have been proposed in recent years to carry out acoustic‐based bird density estimations. We conducted a literature review of studies that used ARUs for estimating bird densities or bird abundances, in order to describe the applications and improve future monitoring programs. We detected a growing interest in the use of ARUs for estimating bird density in the last six years (2014‐2019), with a total of 31 articles assessing the topic. The most common approach was to estimate the relationship between the number of vocalizations per recording time with bird density or bird abundance estimated in the field (61%). In 26 studies (79%), bird estimates obtained by human surveyors agreed with those obtained using ARUs. Some approaches have proven able to reduce biases in acoustic surveys, such as considering imperfect detection (Spatially Explicit Capture Recapture, using microphone arrays), applying paired acoustic sampling to control for different sampling radius between humans and ARUs, or including relative sound level measurements that allow researchers to estimate bird distance to recorder. However, several studies did not include any covariates to reduce existing biases and some did not estimate the sampling radius of the recorder, which may hamper future comparisons between human and ARU surveys. Future studies should include a measurement of the sampling radius of the recorder employed to be able to obtain density estimations using ARUs. Finally, we provide some guidelines to improve the applicability of ARUs to infer bird population estimates in future studies.
Full-text available
Although wildlife conservation actions have increased globally in number and complexity, the lack of scalable, cost-effective monitoring methods limits adaptive management and the evaluation of conservation efficacy. Automated sensors and computer-aided analyses provide a scalable and increasingly cost-effective tool for conservation monitoring. A key assumption of automated acoustic monitoring of birds is that measures of acoustic activity at colony sites are correlated with the relative abundance of nesting birds. We tested this assumption for nesting Forster's terns (Sterna forsteri) in San Francisco Bay for 2 breeding seasons. Sensors recorded ambient sound at 7 colonies that had 15-111 nests in 2009 and 2010. Colonies were spaced at least 250 m apart and ranged from 36 to 2,571 m(2) . We used spectrogram cross-correlation to automate the detection of tern calls from recordings. We calculated mean seasonal call rate and compared it with mean active nest count at each colony. Acoustic activity explained 71% of the variation in nest abundance between breeding sites and 88% of the change in colony size between years. These results validate a primary assumption of acoustic indices; that is, for terns, acoustic activity is correlated to relative abundance, a fundamental step toward designing rigorous and scalable acoustic monitoring programs to measure the effectiveness of conservation actions for colonial birds and other acoustically active wildlife. La Actividad Vocal como un Índice Escalable y de Bajo Costo del Tamaño de Colonia de las Aves Marinas.
IN THE APPROXIMATELY 60 years since the discovery that birds were responsible for some of the puzzling radar echoes dubbed "angels" by the British (Lack and Varley 1945, Buss 1946), radar has proven to be a useful tool for the detection, monitoring, and quantification of bird movements in the atmosphere (Eastwood 1967; Richardson 1979; Vaughn 1985; Bruderer 1997a, b). Radar has been a particularly valuable tool for descriptive studies of daily and seasonal patterns of bird migration, but the technique has also been used to answer important questions about how birds orient during migration and the role of atmospheric structure in shaping flight strategies of birds. Within the last two decades, radar ornithology has played an increasingly important role in conservation of species that are migratory, endangered, threatened, or of special concern.
Survey stations to measure Marbled Murrelet (Brachyramphus marmoratus) activity in low-elevation forest are often preferentially placed on stream channels because they provide wide visibility and the greatest chances of visually detecting birds and thus behaviors possibly associated with nesting. Detections of birds flying along stream channels to access nesting habitat farther inland will inflate estimates of activity associated with the adjacent forest. We compared numbers and types of murrelet detections between 12 paired stations within 100 m of each other on streambeds and in similar habitat in adjacent forest during 8 Jun–10 Jul. 1997 in Clayoquot Sound, British Columbia. Circling and below-canopy flight, thought to be indicative of nesting, were observed at all streambed survey stations and at less than half of paired forest stations. Numbers of such “occupied” detections were six times greater at streambed than forest sites. Size of opening at survey stations accounted for much of the difference in detection rates between forest and streambed stations, but numbers of total, visual, and occupied detections, specifically those of circling birds, were lower at forest than streambed stations even after the effect of opening size had been considered. Correlations between opening size and numbers of detections at streambed but not at forest locations, also indicated that differences between streambed and forest stations were not solely a function of opening size. Observations at one pair of stations where murrelets were using a flight corridor over the forest station indicated that a corridor effect may not be confined to streambed locations. Results indicated that placement of survey stations on stream channels is appropriate if the goal is to determine murrelet presence or the occurrence of occupied detections in an area. However, if a comparison of murrelet activity between habitat types is the objective, then forest stations with comparable opening sizes may be needed to provide unbiased results.
Results of inland forest surveys for marbled murrelets (Brachyramphus marmoratus) have been used extensively to develop complex land-management plans and answer specific questions about the inland ecology of this threatened species, yet the metric used for these surveys, the detection, is poorly understood. We compared concurrent radar and audio-visual (AV) observations at inland forest sites in Washington and Oregon during 160 mornings in May-July 1997-2000 to determine the relationship between number of murrelets detected with radar at a nest stand and number of detections observed by AV surveyors. Each morning, we collected radar data on all murrelet targets that flew over inland forest sites, while AV observers transmitted their information on murrelet detections at the site to the radar laboratory in real time. For each observation, we then determined whether radar, the AV observer, or both had detected the murrelet. Radar data indicated that 25% of murrelet movements at inland sites occurred before the standardized Inland Forest Survey Protocol (IFSP) survey starting time. AV observers detected an average of only 10-23% of the murrelets detected by radar during the official survey period, with high among-day and among-station variation in the proportion detected. We did not find annual differences or seasonal trends in the proportion of birds detected by AV observers compared to radar detections. Cloud cover did not influence the proportion of birds detected by AV observers or the timing of murrelets detected by radar, although timing of AV detections averaged 21 minutes later on cloudy days. We observed some over-estimation of site use because 14% of AV detections of murrelets were flying to or from another site and thus were not associated with the site being surveyed. Our results suggest that AV survey data are an inaccurate measure of use of a particular stand of trees by murrelets. Therefore, determining habitat use and setting management priorities for habitat protection based on AV data could be problematic. Furthermore, unless future research were to find temporal, physiographic, or observer-related factors that could be used to reliably correct AV counts, our results suggest that those counts are not well suited for long-term monitoring of murrelets at inland sites.
Nocturnal burrow-nesting seabirds breeding on isolated oceanic islands pose challenges to conventional monitoring techniques, resulting in their frequent exclusion from population studies. These seabirds have been devastated by nonnative predator introductions on islands worldwide. After predators are eradicated, recovery has been poorly quantified, but evidence suggests some nocturnal seabird populations have been slow to return. We evaluated the use of automated acoustic recorders and call-recognition software to investigate nocturnal seabird recovery after removal of introduced Arctic foxes (Alopex lagopus) in the Aleutian Archipelago, Alaska. We compared relative seabird abundance among islands by examining levels of vocal activity. We deployed acoustic recorders on Nizki-Alaid, Amatignak, and Little Sitkin islands that had foxes removed in 1975, 1991, and 2000, respectively, and on Buldir, a predator-free seabird colony. Despite frequent gales, only 2.9% of 2230 recording hours from May to August of 2008 and 2009 were unusable due to wind noise. Recording quality and call recognition model success were highest when recording devices were placed at sites offering some wind shelter. We detected greater vocal activity of Fork-tailed (Oceanodroma furcata) and Leach's (O. leucorhoa) storm-petrels and Ancient Murrelets (Synthliboramphus antiquus) on islands with longer time periods since fox eradication. Also, by detecting chick calls in the automated recordings, we confirmed breeding by Ancient Murrelets on an island thought to be abandoned due to fox predation. Acoustic monitoring allowed us to examine the relative abundance of seabirds at remote sites. If a link between vocalizations and population dynamics can be made, acoustic monitoring could be a powerful census method. © 2012 The Authors. Journal of Field Ornithology
We identified six approaches to diagnosing causes of population declines and illustrate the use of the most general one (“multiple competing hypotheses”) to determine which of three candidate limiting factors—food availability, nesting site availability, and nest predation—were responsible for the exceptionally poor reproduction of Marbled Murrelets ( Brachyramphus marmoratus) in central California. We predicted how six attributes of murrelet demography, behavior, and physiology should be affected by the candidate limiting factors and tested predictions with field data collected over 2 years. The average proportion of breeders, as estimated with radiotelemetry, was low (0.31) and varied significantly between years: 0.11 in 2000 and 0.50 in 2001. Murrelets spent significantly more time foraging in 2000 than in 2001, suggesting that low food availability limited breeding in 2000. In 2001, 50% of radio-marked murrelets nested and 67% of females were in breeding condition, suggesting that enough nest sites existed for much of the population to breed. However, rates of nest failure and nest predation were high (0.84 and 0.67–0.81, respectively) and few young were produced, even when a relatively high proportion of murrelets bred. Thus, we suggest that reproduction of Marbled Murrelets in central California is limited by food availability in some years and by nest predation in others, but apparently is not limited by availability of nesting sites. The multiple-competing-hypotheses approach provides a rigorous framework for identifying causes of population declines because it integrates multiple types of data sets and can incorporate elements of other commonly used approaches.
I used high-frequency marine radar to count marbled murrelets (Brachyramphus marmoratus) entering 20 watersheds in Clayoquot Sound, Vancouver Island, British Columbia, Canada, in 1996-1998. My goal was to develop standard protocols for radar inventory and to explain landscape-level habitat associations of this threatened species. Dawn counts were consistently higher and less variable than dusk counts. but both sampling periods produced similar rankings of watersheds and proportionate numbers of murrelets. Most dawn surveys showed a unimodal pre-sunrise pulse of incoming murrelets but a few dawn surveys showed post-sunrise pulses, likely caused by repeat visits by some birds. These post-sunrise pulses, although rare, inflated estimates of incoming murrelets and were avoided by restricting analyses to pre-sunrise counts. Dawn and dusk counts were higher on cloudy days (greater than or equal to 80% cloud cover) than on clear days, but among cloudy days there was no additional effect on counts Caused by precipitation (thick fog or drizzle). Numbers of murrelets entering watersheds varied seasonally, reflecting the breeding chronology, but counts restricted to the core period covering incubation and chick-rearing (mid-May through mid-Jul) showed no significant seasonal effects. Counts varied among years at some stations, but when all stations were considered together, no significant inter-annual variation occurred. Murrelets sometimes flew over low ridges (200-600 m), taking shortcuts into watersheds or crossing from I watershed into another. I therefore adjusted the boundaries of some inland catchment areas (based on topography and likely flight paths) to match correctly counts made at the watershed mouths With the appropriate inland catchment area. Radar counts at 18 watersheds were significantly correlated with total watershed area, areas of mature (>140 year old) forest, and-most strongly-with areas of mature forest below 600 m. Logging produced negative impacts, Three of the 5 watersheds with extensive logging of low-elevation forest had fewer murrelets per area than unlogged watersheds or those that were < 10% logged, but these differences disappeared once remaining low-elevation mature forests were considered. With the removal of old-growth forests. murrelets evidently moved elsewhere and did not pack into the remaining old-growth patches in higher densities.