Impact of Disturbance on Habitat Recovery in Habitat Management Areas on the Northern
Edge of Georges Bank
Ecosystem Perturbation Experiment
Woods Hole Oceanographic InstitutionScott Gallager, Steve Lerner, Cameron Fairclough
Fishing CommunityLund’s Fisheries: Wayne Reichle, Jeff Kaelin,
Captain Brady of the F/V Jersey Cape
Annotators: Jinshi Chen, Phil Alatalo, Mary Carman, Patti Keating
2016 – 2018 Scallop Research Set-AsideNOAA-NMFS-NEFSC-2016-2004548
Objective
To complete a series of high resolution Before-After Control-Impact (BACI) habitat
characterizations at three impact scales (heavy, low, none) to evaluate ecosystem and
habitat resiliency
Survey-HabCamV5
*Stereo machine vision cameras*Side scan sonar (50m either side)*CPICS Plankton Imager*CTD*Roll, pitch yawl*7 kts*8 images/sec*Real-time processing on ship*Deep Learning classifier
Impact-Commercial dredge
Went with one level of impact ~20%
Hypotheses
Northeast Reduced Impact Habitat Management Area (NERIA) Areas surveyed:
NERIA: 349.50 km2
NEHMA: 700.36 km2
EGS: 439.06 km2
# Image pairs taken: 852,145Annotated: 12% manual
100% Automated
Total area imaged: ~ 851,000 m2
All images auto-annotated for habitat
Eastern Georges Shoals (EGS)
Northeast Habitat Management Area (NEHMA)
2017 RSA Survey HabCamV5
Total: 44,977,948 Total: 15,549,936
Total: 851,013
Northeast Reduced Impact Habitat Management Area (NERIA)
Gridded to 30 m squares
Ordinary Krigingwith depth as co-variate
Also comparing withEmpirical Bayesian kriging
Bathymetry with Epifauna Contours
60
7080
90
50
40
30
100
70
Exploitable Scallops with Gravel (white) , Epifauna (black) , Shell Hash (brown) Contours
2017 Survey July 20-24
EGS NEHMA
NERIA
Numbers of Scallops
Exploitable Medium Small All Scallops
All Areas 53,303,486 21,755,609 1,575,510 76,634,605
NERIA 44,977,948 15,549,936 851,013 61,378,897
NEHMA 1,260,245 1,029,840 184,850 2,474,935
EGS 7,065,293 5,175,833 539,647 12,780,773
Biomass of Scallops (MT)
Exploitable Medium Small All Scallops
All Areas 14,731.29 687.75 2.49 15,421.53
NERIA 13,158.13 493.36 1.49 13,652.98
NEHMA 299.06 31.61 0.39 331.06
EGS 1,274.10 162.78 0.60 1,437.48
BACI Experimental Sites2012-2016-2017-2018
Impact
1 nm in length10 m wide
Dredge tracks- 10 passesHabCamV5Control tracks
Impact area under the dredge tracksControl area outside the dredge tracks
HabCamV5Treatment tracks
July 2016 pre-impact broad scale and fine scale surveys
July 2012 Initial broad scale and fine scale surveys, part of RSA project (V2)
October 2016 pre-impact fine scale survey, control surveyImpact with scallop dredge
June 2017 10 mo after impact fine scale survey
July 2018 22 mo after impact fine scale survey
Experimental Design for BACI sitesSite 1
Site 2Site 3
Site 4
Site 5
Site 6
Convolutional Deep Neural NetworkClassification
Two approaches
*Holistic- complete image (substrate)*Targets- segmentation (scallop,
seastar, fish, sand dollar, etc.)
*8 images /sec*15 classes (so far)*90-97% accuracy
(class dependent)*runs on NIVIDIA Jetson TX2
5 classes of substrateScallopSeastarRound fishFlat fish(fish to 31 species manually)
Single variate analyses1. number of species (S)2. number of individuals (A)3. Shannon Weiner diversity (H’)4. Margalef’s Richness (d)5. Pielou’s Eveness index (J’) 6. Simpson’s Dominance (λ)
Multivariate analyses (in progress)1. Square root transformed2. Hierarchical clustering to identify sample groupings based on the Bray-Curtis index of similarity and Euclidean distance3. non-metric Multi-dimensional Scaling plots, nMDS4. ANOSIM (Analysis of Similarity) routine was used to test the null hypothesis that there are no differences between biological communities among controls and different times before and after impacts5. significance testing on selected abiotic and biotic variables using two-way Analysis of Variance (ANOVA)
bu
shel
sb
ush
els
bu
shel
s
Tow #
Tow #
Tow #
Bushels removed = 6.5
Bushels removed = 470
Bushels removed = 422
Bushels removed = 2
Bushels removed = 327
Bushels removed = 93
Scallops removed at each Site
SubstrateEpifaunaSandGravel
1-D Habitat Suitability Model for scallop
Organism54 Seastar122 Fish2 Sculpin108 Cod92 Haddock8 Skate
October 2016 -Before Impact- Site 1
SubstrateGravelSandEpifauna
3340.2/m2
100%epifauna
October 2016 -Immediately After Impact - Site 1
Organism40 Seastar15 Fish1 Sculpin12 Cod21 Haddock2 Skate
SubstrateGravelSandEpifauna
200.004/m2
23%epifauna
Removed 6.5 bushels scallops
July 2017 -10 mo After- Site 1
Organism139 Seastar10 Fish5 Sculpin28 Cod15 Haddock1 Skate
SubstrateGravelSandEpifauna
4340.17/m2
92%epifauna
Site 2
Site 3 Site 1
Site 4
Site 5
Site 631 species of fish
BACI Experimental Sites- HAPC 2016-2018
~1 nm boxesas Control
Treatment1 nm x 10m
Site 6Site 3
Site 1
Site 2
Site 4
Site 5
HabCam Survey Tracks-10 passes inside Treatment rectangles, 0.5 nm grid outside Treatment
Surveyed 2016 (Before), 2016 (After), 2017 (After), 2018 (After)
Haddock(Before)
Atlantic Cod(Before)
Little Skate(Before)
Site 1: Epifauna Site 2: Gravel Site 3: Gravel
Site 4: Gravel/Epifauna Site 5: Sand Site 6: Sand/Gravel
Before After 10 mo 22 mo
Co
ntr
ol
Trea
tmen
t
Site 1 only significant change, after 10 mo > 90% recovery
Mean +/- 1 SD
Before After 10 mo 22 mo Before After 10 mo 22 mo
Before After 10 mo 22 mo
Before After 10 mo 22 mo
Before After 10 mo 22 mo
Impact on Dominant SubstrateRecovery to 90%in 1 year
Site 1: Haddock Site 1: Atlantic Cod
Site 1: Monkfish Site 1: All Fish
Before After 10 mo 22 mo Before After 10 mo 22 mo
Before After 10 mo 22 mo Before After 10 mo 22 mo
# /
m2
# /
m2
# /
m2
# /
m2
Trea
tmen
tC
on
tro
l
Recovery to x4in 1 year
Mean +/- 1 SD
substrate scallop seastar fish sculpin cod haddock skate
Site 1 Before 100% epi 0.2 54 122 2 108 92 8
After 44% epi 0.004 40 15 1 12 21 2
10 mo 92% epi 0.17 139 10 5 28 15 1
22 mo 100% epi 0.07 36 1 10 0 1 1
Site 2 Before 55% grav 0.80 10 31 15 19 48 18
After 52% grav 0.03 6 0 0 0 0 0
10 mo 41% grav 0.93 86 19 3 4 21 10
22 mo 55% grav 0.88 50 11 1 10 10 8
Site 3 Before 96% grav 4.17 89 203 0 4 11 14
After 90% grav 0.47 30 1 1 1 3 1
10 mo 88% grav 5.82 106 20 4 2 113 3
22 mo 91% grav 3.18 329 0 1 4 9 0
Site 4 Before 56% grav 0.001 11 107 0 6 43 27
After 52% grav 0.0004 0 0 0 0 0 2
10 mo
22 mo 58% grav 0.002 9 29 1 3 10 1
Site 5 Before 44% sand 0.29 70 1 1 1 5 10
After 53% sand 0.02 0 3 1 1 0 1
10 mo 46% sand 0.31 0 3 1 0 0 1
22 mo 28% sand 0.29 5 10 0 2 3 4
Site 6 Before 98% grav 0.31 0 6 1 0 0 2
After 97% grav 0.002 0 0 0 0 0 0
10 mo 62% grav 0.70 92 55 2 5 161 22
22 mo 91% sand 0.55 26 87 0 0 50 15
Strong evidence of scallop movement intoImpact sites
No evidence of new scallop recruitment within the two years
Summary and ConclusionsSummary1) A BACI whole ecosystem experiment was started in 2016 in Closed Area II, HAPC for the purpose of
examining what level of impact from scallop dredges would cause irreputable damage in anarea that is considered essential fish habitat for Gadids and other commercially important species.
2) Six sites were established in different habitats, 1 nm in length and 10 m wide, and intensively surveyed multiple times before impact and three times ( ~once per year) following impact. Impact consisted of towing a commercial dredge 10 times down the same transect.
Conclusions1) Only Site 1 in a 100% epifauna/mussel/tunicate community was severely impacted as
observed by stereo optical imaging and side scan sonar.2) Biodiversity was significantly reduced (p < .001) from a H’ score of 12.1 to 1.3.3) Within 10 months biodiversity returned to 93% of original form: H’= 11.64) Some species appeared to be stimulated by the impact (cod ^ by factor of 4) while haddock
rebounded to ~ 0.5x original level.
5) Limited trips may be possible in all areas of the HAPC except Site 1, where the epifaunal cover is extremely sensitive to mechanical damage.
Top Related