Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Lessons learned from sonar BRS: methodological implications for studies with other anthropogenic sound sources
Brandon L. Southall
With thanks and acknowledgments to:
SOCAL-BRS Team
Sonar BRS Sponsors (notably U.S. Navy LMR, ONR)
ESR Co-authors: (P. Tyack, P. Miller, D. Nowacek)
Workshop Organizers (notably C. Harris)
Overview & Talk Outline
• Many talks thus far have illustrated progress in
various areas regarding sonar BRSs.
• Goal: integrate observations across studies
(emphasis on experimental) to identify key
considerations for BRSs with other sources
1. Operational and Logistical Issues
2. Emergent Research Considerations
3. Analytical Approaches
2
Operational and Logistical Issues
Factors driving increased operational size of BRSs
• Most studies (esp. CEEs) inherently require multidisciplinary tools
and skills – and thus field/analysis/support teams
• Increasingly clear need to match context with real operations
Factors driving reduced operational size of BRSs
• Simple funding realities as approaches have matured
• Demonstrated successes of leaner field teams with ability to
adapt to changes in both schedules, field conditions
3
Optimize approach to balance these, given site- and species-specific considerations
Emergent Research Considerations:
Importance of Baseline Data, Controls
Need a representative understanding of baseline
(undisturbed) behavior to interpret response
• How much? Depends on species-typical traits
• Increasingly capable longer-term tags can provide key
baseline information on selected behavioral parameters
(informed interaction of methods on different scales)
Experimental controls (different approaches) within CEEs
critical in publishing defensible results
• Observational studies lack real controls, but can address some
of the same issues with information about periods of
exposure/no exposure
4
Emergent Research Considerations:
Species-Specific Factors Affecting Design
• Life History Factors (e.g., broad distribution, seasonality)
• Group Size
• Natural Response to Threat (e.g., predators)
• Variability Within and Between Behavioral States
• Relative Novelty/Familiarity with Sound Source
5
Emergent Research Considerations:
Context-Dependent Factors (1)
Behavioral State, Prey Distribution, Source RangeSOCAL-BRS blue whale examples
6
Goldbogen et al., (2013)
MFA PRN CON
Playback type
Div
e M
etr
ics r
esp
on
se
02
46
81
012 before
after
Response
MFA PRN CON
Playback type
Pre
y P
atc
h D
ep
ths
-200
-150
-10
0-5
00
before
after
Covariates
MFA PRN CON
Playback type
Ra
tio o
f P
rey P
atc
h D
ep
th / D
ive
Me
tric
01
02
03
04
05
06
0
before
after
Comparison Ratio
Friedlaender et al., (in press)
Emergent Research Considerations:
Context-Dependent Factors (2)
Spatial Orientation of Source-Receiver
7
Stationary Source
Courtesy Buck and Tyack; published in Ellison et al., (2012)
Moving Source
Courtesy 3S (P. Miller)
Analytical Methods
Multivariate data require multivariate analyses
Stimpert et al., (2014)8
Analytical Methods
Complimentary Analyses –
Comprehensive Insight Harris et al., (2016)
9
Group (Across-Individuals) Response Analyses
• Principal component analyses (PCAs) and generalized additive
mixed models (GAMMs)
• Hidden Markov Models (HMM) – Behavioral State Switching
Within-Individual Response Analyses
• General Estimating Equations (GEEs)
• Mahalanobis Distance
• Expert Severity Scoring – Event Survival Analysis
Southall et al: MONDAY 1330 (Imperial A)
Multiple methods in SOCAL-BRS blue whales (5!) where large N (46 CEEs)
Analytical MethodsComplimentary Methods – Comprehensive Insight
10
0
10
20
30
40
50
60
70
80
11-May
12-May
13-May
14-May
15-May
16-May
17-May
18-May
19-May
20-May
21-May
22-May
23-May
24-May
Distance (km)
Day
Before
During
After (0-72 hrs)
After (72-144 hrs)
D. 72-144 hrs afterC. 0-72 hrs afterB. DuringA. Before
E
Before During After
Tyack et al., (2011)
Moretti et al., (2014)
Taking BRSs to Scale
11
• Increasing use of full scale (actual) sources
• Test potential response over larger source-animal range
• Range-RL interaction in terms of p(resp)
• Extend temporal sampling
o CEE results can inform testable hypotheses to be evaluated
with targeted monitoring with longer-term tags/hypotheses
o Multi-scale methods for both experimental and observational
monitoring
o Importance of medium-duration, high-resolution tags
(Calambokidis – SOCAL blue whale example shown)
Synthesis – Key Considerations from
Lessons Learned (Field Studies)
12
• Optimize Field Configurations – Maximize Adaptability
• Understand Baseline Behavior
• Study (or at least measure) Contextual Factors & RL
o Behavioral State
o Source-Animal Spatial Relationships
o Don’t Have to Measure/Understand Every Combination
o Risk Assessment and Mitigation: Different Processes
• Use Complimentary (multi-scale) Methods and Analyses
o Dynamic interaction and overlap between shorter-term, high-resolution and longer-term low resolution methods