marine science institute university of california santa ... · metapopulations and metacommunities...
TRANSCRIPT
Marine Science Institute
University of California Santa Barbara
Who we are
Robert Miller, David Siegel, Craig Carlson, Daniel Reed, BS
Manjunath, Deborah Iglesias-Rodriguez, Doug McCauley, Milton Love
Principal/Associate Investigators
UCSB
Florida State University
USGSKevin Lafferty
UCSD - SIO
John Hildebrand
Andrew Thompson
NOAA – NMFS SWFSC
Andrew Rassweiler
Partners
Plumes and Blooms (NASA)
BOEM Pacific Region
Santa Barbara Coastal LTER (NSF)
Channel Islands National Marine Sanctuary
Southern California Coastal Water Research Project
Southern California Coastal Ocean Observing
System (SCCOOS)
Channel Islands National Park
Gray Whales Count
CalCOFI
1. Provide data to inform managers and society about
patterns of biodiversity across taxa, space, and time
• Integrate existing data
• Develop new methods & products
2. Build a framework to facilitate MBON development under
diverse circumstances
MBON Prototype:
Progress, by Taxonomic Group
Measurement Class:
D: Occurrence (EBV “distribution”)
A: abundance or density (EBV “abundance”)
Focus on time series oldest: 35 yrs youngest: 18 yrs
14 Data packages published
stable, immutablewith DOI
Spanning taxa from microbes to whales
Net Environmental Benefit Analysis of offshore platform decommissioning alternatives
Users:
– US Bureau of Ocean Energy Management
– CA State Lands Commission
– Petroleum industry (e.g. Exxon, Chevron, Venoco)
Leave in
place Partial
Removal
Total
Removal
Meyer-Gutbrod et al., in prephttps://meyer-gutbrod.shinyapps.io/Decommissioning_WebApp/
• Platform structures are dominated by rockfish
• If removed, sites will return to mud inhabited by soft bottom species such as flatfish
• Removal of 24 platforms will result in the loss of 28,700 kg of fish biomass, and an additional loss of 4,700 kg of annual fish production
Comparing platform decommissioning scenarios
Meyer-Gutbrod et al. In prep.
Webinization of Condition Reports
How does biodiversity affect ecosystem function and
stability?
However, most empirical and theoretical studies which have investigated ecological
variability and its relation with biodiversity have focused on local scales
Understanding the factors that dampen variability of biomass production is
a core concept of Ecology
Species diversity begets stability
Asynchronous species
dynamics trumped
asynchronous spatial
dynamics to stabilize
metacommunity
biomass in the kelp
forest understory
10
Real-world ecosystems are complex: from populations of single species to
multi-species communities to metapopulations and metacommunities
Lamy et al. 2019, Ecology, Wang et al. 2019, Ecography
Recommendation: data
integration and management
needs science involvement
Typical machine learning
Engineer required
throughout
Time consuming
(months)
Only works on
specific type of
dataTrain & test
Select classifier
Select features
Images
annotations
BenefitsGeneralizes to your data
Fully automated - no feature selection
High accuracy
LeveragesScalable services
Annotation system
Cluster processing
Fast classification on GPUs
Deep learning
New Products: Deep learning for image analysis
|
Image management & annotation |
Model building & usage |
Identification & segmentation|
Recommendation: machine
learning requires time but
saves work in the long run
New Products: AcousticsAcoustic Detection of Marine Mammals
Blue Whale
D-callT
ota
l D
ete
ctio
ns p
er W
ee
k
Jul09 Jan10 Jul10 Jan11 Jul11 Jan12 Jul12 Jan13 Jul13 Jan14 Jul14 Jan15 Jul15 Jan16 Jul16 Jan170
1.1288
2.2577x 10
4
0
50
100
Pe
rce
nta
ge
of
Eff
ort
pe
r W
ee
k
Tota
l D
ete
ctio
ns p
er
We
ek
Jul09 Jan10 Jul10 Jan11 Jul11 Jan12 Jul12 Jan13 Jul13 Jan14 Jul14 Jan15 Jul15 Jan16 Jul16 Jan170
9937.5
19875
0
50
100
Pe
rce
nta
ge
of
Eff
ort
pe
r W
ee
k
Site H
Site N
Seasonal patterns of blue whale song
New Products: Remote SensingPhytoplankton functional diversity
Phytoplankton pigment community clusters define five clusters
Catlett and Siegel, 2018.
Bio-optically modeled PFTs give us the links to PACE/ESBG
350 400 450 500 550 600 650 700Wavelength [nm]
-0.02
-0.01
0
0.01
0.02
0.03
0.04
a' p
h(6
) coe
ffic
ients
[A
(6)]
fo
r F
uco
(a)Mean95% C.I.
350 400 450 500 550 600 650 700
Wavelength [nm]
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
a'' p
h(6
) co
eff
icie
nts
[B
(6)]
for
Fu
co
(b)
Prediction of fucoxanthin
(diatom marker) from
optical derivative spectra
Works well - r2 = 0.86
Similar performance for
other marker pigments’
Requires info from the
entire spectrum
Bio-Optical Model Extends Biomarker Pigment Time Series
*Catlett et al., in prep.
Model Retrievals R2
TChlb (green algae) 0.815
HexFuco (haptophytes) 0.733
Fuco (diatoms) 0.856
Perid (dinoflagellates) 0.887
Zea (picoplankton) 0.541Pigment EOF Mode 1 (Early
upwelling mixed bloom) 0.884Pigment EOF Mode 2
(Diatoms vs. mixed nano-/pico-plankton) 0.852
Pigment EOF Mode 3 (Pico-plankton vs. haptophytes) 0.454
Pigment EOF Mode 4 (Dinoflagellates vs. mixed
diatoms/haptophytes) 0.809
PFTs and Climate• Investigating PFT associations with climate
1997-2015
• SBC dinoflagellate blooms associated with low or negative NPGO + MEI
• Hypothesis: climate-driven variations in surface-ocean circulation drive variations in PFTs– Low/negative NPGO = weaker equatorward
flows in Cal Current, dino-rich SoCal Bight waters can be transported North into SBC to seed blooms
– Low/negative MEI = weak El Nino/La Nina provides favorable conditions for dinoblooms
*Catlett et al. preliminary
97 99 01 03 05 07 09 11 13 15
Year
-2
-1
0
1
2
3
NP
GO
/ME
I
0
1
2
Peri
din
in
NPGO
MEI
Perid
PFTs and Surface Ocean Circulation
• ROMS particle trajectory modeling used to quantify water mass origins for each PnB cruise
• Particles released back in time from the PnB transect (red), count how many originated in each box (blue)
• ~30% increase in proportion of particles from Box 7 for dino blooms relative to diatom or no bloom
• (Climate-driven) surface ocean circulation may be important in biodiversity monitoring
*Catlett, Siegel, Simons, et al, preliminary1 2 3 4 5 6 7 8
Origin Box Number
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
Diffe
ren
ce
in
Pro
po
rtio
n o
f P
art
icle
s
Dino minus no bloom
Diatom minus no bloom
15 day advection time
Dino-rich SoCal
Bight
Diatom-rich Cal.
Current
Formation, Development, and Propagation of a
Coastal Coccolithophore Bloom
VIIRS 2015-06-05
Matson et al. (in press) JGR Oceans
No
bloom
Growing
Bloom
Senescing
Bloom
Genomic Links with Pigment PFTs + Bio-optics
• Correspondence is found in genomic & pigment PFTs
• Seasonality evident in PFTs
• Work on-going to pull apart the Eukgenomic data
*Catlett et al., preliminary
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Seq. Rel. Abundance (Bacillariophyta)
-2
0
2
Lo
g1
0(F
uco
Co
nce
ntr
atio
n)
0
0.5
1
Dia
tom
% C
HL
(C
HE
MT
AX
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Seq. Rel. Abundance (Dinophyta)
0
0.5
Pe
rid C
oncen
tration
0
0.5
Din
ofla
ge
llate
% C
HL
(C
HE
MT
AX
)
Genomics Pigments
Fall-Winter
Spring
Upwelling
Spring Bloom
Summer-
Fall
Fall-
WinterSpring Upwelling
Spring
BloomSummer-
Fall
Recommendation:
Determine remotely sensible
and validated PFTs
New Products: Remote Sensing
a.
b.
c.
Min
imu
m
Em
erg
en
ce
Ag
e (
d)
Chl:C
(mg
mg
-1)
Bio
ma
ss
(kg
m-2
)
April June August+
500 m
Kelp condition, age, and forest extent
Bell & Siegel in prepBell et al. L&O 2018
PIs: Siegel, Bell, Cavanaugh, Miller, Nidzieko, Nelson, Reed
Aerial extent: color imagery
NIR
Re
flecta
nce
(Bio
ma
ss)
Low
High
Bell et al. in prep
Biomass and condition: multispectral
25
Giant kelp creates a diverse ecosystem
Lamy et al. Ecology in review
Giant kelp increases biodiversity through physical engineering
Giant kelp stability indirectly stabilizes the community via its effect on biodiversity
Miller et al. 2018, Proc. Royal Soc. B.,Castorani et al. 2018 Ecology
Recommendation: focus on
dynamics of foundation
species (best with RS!)
New Products: GenomicsMicrobial diversity & community structure
Eu
rya
rch
ae
ota
: M
ari
ne
Gro
up
II
Th
au
ma
rch
ae
ota
: N
itro
so
pu
milu
s
Ac
tin
ob
acte
ria:
OC
S1
55
Fla
vo
ba
cte
ria
: P
ola
rib
ac
ter
Fla
vo
bac
teri
a:
NS
4
Fla
vo
bac
teri
a:
NS
5
De
ferr
ibac
tere
s:
SA
R4
06
SA
R11 S
urf
ace
1
SA
R11 S
urf
ace
2
SA
R11
De
ep
1
Ro
se
ob
ac
ter
NA
C11
-7
Ro
se
ob
ac
ter
DC
5-8
0-3
Ro
seo
ba
cte
r O
CT
Oth
er
Rh
od
ob
ac
tera
ce
a
OC
S11
6
SA
R11
6
Nit
ros
pin
a
SA
R32
4 (
Ma
rin
e g
rou
p B
)
SA
R9
2
OM
60(N
OR
5)
Ps
eu
do
sp
iril
lum
SA
R8
6
Verr
uc
om
icro
bia
: M
B11C
04
Oth
er,
<0
.1%
-10
-5-5
-4
-3
-2
-1
0
1
2
35
10
Lo
g2-f
old
rati
o t
o e
xp
ecte
d
V4
V4-5 V3-4
V1-2
Alphaproteobacteria Deltaproteobacteria Gammaproteobacteria
Mock community deviation from expected
abundance with four different primer sets:
Ordination plots of the same marine time-series
samples sequenced with four primer sets
Wear et al. 2018
Phytoplankton Genomics Method Validation
• Mock communities revealed unexpected issue with common methods
• Outliers correspond to specific primer indices used for sample multiplexing that dramatically reduce accuracy + precision
*Catlett et al., 2019, submitted, Environmental Microbiology
Rela
tive A
bundance
Recommendation: method
validation is critical for
genomic surveys
Environmental DNA captures the fine scale and hierarchical
spatial structure of kelp forest fish communities
Spatially stratified sampling of 49 water samples on 27
transects across 11 rocky reefs
Sheephead
Senorita
Blacksmith Garibaldi
Visual Survey
(2001-2019 SBC
LTER)
eDNA
(12S)
Kelp bass
Leopard sharkLingcod
Japanese pilchard
Engraulis spp.
White shark
Barred
surfperch
Lizardfish
Lamy et al., in prep.
Lafferty et. al 2018, Frontiers Mar. Sci.
eDNA and Acoustic Telemetry Detection of Great White Sharks
Recommendation: eDNA
useful especially for known
targets
For each method:- Rarify biodiversity to 100
individuals observed - Calculate associated
confidence intervals (CI)- Uncertainty is CI/mean
Uncertainty-effort curves:- Randomly subsample
replicates- Calculate expected CI
and uncertainty
0D = Richness, 1D ≈Shannon Diversity, 1D ≈Simpson Diversity
Information about biodiversity
management actions and outcomes
Sampling method 2
Sampling method 1
Comparing the efficiency of alternative methods of monitoring biodiversity
Biodiversity information:status of rare speciesrichness and evennessTrends in abundanceSpatial patterns of diversity
√
0D(richness)
1D(~Shannon)
2D(~Simpson)
Pow
er t
o d
etec
t 2
5%
ch
ange
in b
iod
iver
sity
Sampling Effort (diver hours)
Compare power to detect change:
0D(richness)
1D(~Shannon)
2D(~Simpson)
Un
cert
ain
ty (
CI/
mea
n)
Sampling Effort (diver hours)
Compare sampling efficiency across methods:
Program Development
Recommendation: MBON
support should be
competitive but predictable
SBC MBON: by the numbers
• Publications: 47
• Undergraduate Students: 67
• Graduate students: 17
• Postdocs: 7