evaluating the use of autonomous recording units for monitoring yellow rails and other nocturnal wet...
DESCRIPTION
Evaluating the Use of Autonomous Recording Units for Monitoring Yellow Rails and Other Nocturnal Wet Meadow Birds. Anna M. Sidie-Slettedahl , US Fish and Wildlife Service, South Dakota State University Rex Johnson, US Fish and Wildlife Service Todd Arnold, University of Minnesota - PowerPoint PPT PresentationTRANSCRIPT
Evaluating the Use of Autonomous Recording Units for Monitoring Yellow Rails
and Other Nocturnal Wet Meadow BirdsAnna M. Sidie-Slettedahl, US Fish and Wildlife Service, South Dakota State
UniversityRex Johnson, US Fish and Wildlife Service
Todd Arnold, University of MinnesotaJane Austin, USGS Northern Prairie Wildlife Research Center
Joshua Stafford, USGS South Dakota Cooperative Fish & Wildlife Research Unit KC Jensen, South Dakota State University
Introduction• Limited knowledge of these species:
– Yellow Rail– Nelson’s Sparrow– Le Conte’s Sparrow
Conservation• Loss and degradation of
wetlands due to human activity
• All three = Species of Greatest Conservation Need in MN
• Yellow Rail = species of high concern – North American Waterbird Conservation Plan – Northern Prairie and Parkland Waterbird
Conservation Plan
• Systematic surveys are needed– USFWS & USGS: Standardized North American
Marsh Bird Monitoring Protocols (Conway 2009)
• However…– Tendencies to call at night = often times missed by
surveys• Also…
– Secretive habits– Cryptic coloration– Difficult to access habitat
Dr. Jim Petranka
Autonomous Recording Units (ARUs)• Good candidates for surveying with ARUs that
can be analyzed in a laboratory• Benefits:
– Minimize observer biases– Permanent records of surveys– 24 hr/day data collection– Limited numbers of expert field observers
• Disadvantages:– No visual detections – Estimating distances and numbers of birds?– Time spent going through recordings
Song Scope Bioacoustics Recognition Software
• Ideally, you build “recognizers” to scan recordings
Main Objectives
• Compare the detection probabilities of these three species using ARUs versus the standard marsh bird monitoring protocol.
• Use ARU recordings to determine temporal (daily and seasonal) changes in species calling and environmental factors affecting detection, in order to improve survey efforts
Field Methods• 16 survey routes, 10
stations• 22 ARUs (per season; 1-4
ARUs/route)– SM1 Song Meter, Wildlife
Acoustics– 10 min every 15 min, from
20:00 until 08:00
Field Methods
• Standard Marsh Bird Protocol – Call-broadcast surveys--
Yellow Rail call only – Start 1 hour after sunset– May-June = survey season– 4 times/season
Objective 1: Comparing the Two Survey Methods
Methods:• Isolated all 6 minute recorded standard surveys
(172 in total)• Use “recognizers” to automatically detect and
identify target species calls on recordings
• NO!! Ran into problems: missed detections and too many false positives
Manual Scan Method
• Resorted to the “Manual Scan” Method• Quickly visually and
aurally scan through recording to detect target species
• Robust design occupancy model in Program MARK
1 2 3 40
0.2
0.4
0.6
0.8
1
Yellow Rail
Standard Survey
ARU
Survey Repetition
1 2 3 40
0.2
0.4
0.6
0.8
1Le Conte’s Sparrow
Survey Repetition
1 2 3 40
0.2
0.4
0.6
0.8
1Nelson’s Sparrow
Survey Repetition
Probability of Detection
-Each species-Each survey repetition-Each survey method
Why?
• Most calls not detected on ARU recording, but that were detected during Standard Survey, were too faint or not “strong” enough to be recorded by ARU– Reduced detection by ARUs was likely due to
human observers being able to detect birds at greater distances
How many 6 minute ARU recordings to be at least 95% sure of detection?
6 Minute Surveys
Species 1 2 3 4 5
YERA 72.6% 92.3% 97.8% . .
LCSP 53.2% 80.7% 93.1% 97.8% .
NESP 51.1% 75.9% 88.0% 94.0% 97.0%
However
• Because ARUs are in the field for longer periods than human observers, there are more cumulative opportunities for detection
Objective 2: Factors affecting detection
• Looking at temporal and environmental variables that may affect calling and/or detection of these species
• Generalized linear mixed models in R– Presence/absence from 3035 three minute
recordings, from 43 ARU stations– Hourly weather data
Variables of Interest• Random effect = Survey Site • Fixed effects =
– Year – Julian day– Precipitation (yes or no)– Temperature– Wind speed– Atmospheric pressure– Moonlight– Hours after sunset
Yellow Rail• Precipitation
No precip. = 0.63 (95%CI = 0.55 and 0.71)Precip. = 0.47 (95%CI = 0.36 and 0.59)
Le Conte’s Sparrow
Nelson’s Sparrow• Precipitation
No precip. = 0.22 (95%CI = 0.16 and 0.30)
Precip. = 0.08 (95%CI = 0.03 and 0.16
Management Implications
• Incorporate these factors into existing survey protocols to improve survey efforts– Standard surveys– Use of ARUs
• Improvement of systematic surveys
Acknowledgements
• HAPET – US Fish and Wildlife Service
• Agassiz National Wildlife Refuge
• South Dakota State University
Funding: