Collaborative Adaptive Management Team (CAMT...
Transcript of Collaborative Adaptive Management Team (CAMT...
Collaborative Adaptive Management Team (CAMT) Investigations: Using New Modeling Approaches to Understand Delta Smelt State Salvage Patterns at the State Water Project and Central Valley Project
Lenny Grimaldo, PhD (ICF)Matt Nobriga (USFWS)William Smith (USFWS)
2016 Bay-Delta Science Conference
Disclaimer
The viewpoints expressed today are those of the authors and do not necessarily reflect the opinions of the U.S. Department of the Interior or the U.S. Fish and Wildlife Service
The Bigger PictureFactors affecting
Entrainment
FWSRPA 1 Existing Management Framework
L~ OMR
Population consequences
Assessment of Impacts
Proposal I Proposal II Proposal Ill Proposal IV
Scoping Team Review
Elevate to CAMT
The Bigger Picture
Why revisit the salvage data?
2008 USFWS Biological Opinion
Previous research did not consider:
1. Finer-scale variability of salvage dynamics (i.e., what drives salvage during first flush periods)
2. Other potentially important predictor variables (e.g., predators, water temperature, etc)
3. Fish behavior during first flush events
4. Population-level impacts (i.e., salvage scaled to previous FMWT abundance)
New conceptual models
DRIVERSWinter Storms
DRIVERS Freshet EventSeasonal Trigger
PHYSICAL RESPONSES
OMR & SJR Flows
Turbidity Distribution
SPECIES RESPONSES
Smelt Dispersal
SPECIES RESPONSES
Prey Availability
Density of Smelt in the South Delta
MANAGEMENT ACTIONS
Operation of Cross Channel
GatesPumping
CONSEQUENCESSalvage &
Entrainment
Conceptual Model for Factors Affecting Entrainment of Adult Delta Smelt at Water Projects Facilities
via Scott Hamilton
2015 CAMT Progress Report
Delta smelt movements during first flush
San Francisco Bay
SacramentoRiver
Suisun Bay
San Joaquin River
Napa River
Sommer et al. 2011
SWPCVP
3.6 km/day
Migration vs diffuse dispersal?
Delta smelt movements during first flush
San Francisco Bay
SacramentoRiver
Suisun Bay
San Joaquin River
Napa River
Murphy and Hamilton 2013
SWPCVP
3.6 km/day
Migration vs diffuse dispersal?
Delta smelt movements during first flush
San Francisco Bay
SacramentoRiver
Suisun Bay
San Joaquin River
Napa River
Murphy and Hamilton 2013
SWPCVP
3.6 km/day
Migration vs diffuse dispersal?
Delta smelt movements during first flush
San Francisco Bay
SacramentoRiver
Suisun Bay
San Joaquin River
Napa River
Murphy and Hamilton 2013
SWPCVP
3.6 km/day
Migration vs diffuse dispersal?
Both models support turbidity bridge CM
Delta Smelt likely employ a combination of unidirectional and diffuse movements during winter storms
2006
80000
60000
40000
20000
400000
300000
200000
100000
0
120967248241
5000
0
-5000
-10000
120967248241
40
30
20
10
0
SAC
Days since Decemeber 1 st
OUT
OMR Total Salvagecfs
2006
80000
60000
40000
20000
400000
300000
200000
100000
0
120967248241
5000
0
-5000
-10000
120967248241
40
30
20
10
0
SAC
Days since Decemeber 1 st
OUT
OMR Total Salvagecfs
Behaviors are complicated! Stay for Ed’s talk
Delta Smelt likely employ a combination of unidirectional and diffuse movements during winter storms
Figure 6, Grimaldo et al. 2009
First flush storms historically led to high Delta Smelt salvage250 ........,l9"'!)~3-------------~ 3 _.._ Salvuge
200
150
100
50
J211f U
IJJ7 mJ/s I
-. ,' ,..""
o ~----~----~~~-4oo ~------------~~--------------~
1996 /• 300
1 \ Y IJIII
J
~ :wo I ' 611mVs \ ~
:w:l ..... .., 100 ;:...
:"l bJ)
.s 0 >
0 5 120
-:;I
00 100 -~ t':l
XO > -;:; 60 rn
"a i5
40 !- 20
0
500
400
300
200
100
0
2(100 I 2 mu 70 7 nd/.s
11
*'
61 () I 12 1
2 -- Outnow •••••• Turbidily
0
- I
-2 J
2 0 ~
:a 0
Ul OJ)
t: 11.1 > t':l
~ ~
0 ::I 0.
-i -I = .....
~
.5 > 0
E ~
X ....... ~
0"
-2 0: -<
4 ,... ;a
) "" ::1 ~ i»
2 2.. t':l > -;;
~· c.. ....... rn
0 r:i 0
- I !-
-2
3
2
u
- I
-2
250
200
150
100
50
0
- I
o ~ .. ~~~----~~~-----~ -2 500 4
400
300
200
100
I')
600
50[)
401)
300
2:00
IOU
0 250
200
150
100
50
j}
l l 6 .1 91 121
3 0 = 2 5 9 < :>.>
0 i5.. -1:: - I &
c.: ? -· - ---...:
) ~ "' 2 ;::; 0. ~ a r::; · 0
0 = -1
-2
4
3
2
0
- I
-l -3
Salvage can be over or nearly over before population estimates are confidently assessed from the Spring Kodiak Trawl
Figure 11, Kimmerer 2008
First flush storms historically led to high Delta Smelt salvage
Too few fish caught in fall nowadays to provide reliable adult abundance estimates, especially using the FMWT net (SKT gear better net)
Figure 11, Kimmerer 2008
0
200
400
600
800
1000
1200
1400
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
711
First flush storms historically led to high Delta Smelt salvage
Too few fish caught in fall nowadays to provide reliable adult abundance estimates, especially using the FMWT net (SKT gear better net)
Figure 11, Kimmerer 2008
0
200
400
600
800
1000
1200
1400
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
711
Why not use the SKT starting in September?
First flush storms historically led to high Delta Smelt salvage
Figure 11, Kimmerer 2008 Source: Matsuishi 2014
Could potentially allow for salvage (losses) to be evaluated in context of recruit-per-spawner models
First flush storms historically led to high Delta Smelt salvage
Revised analyses-Focus on conditions that explain salvage during first flush
12010896847260483624121
5000
4000
3000
2000
1000
0
Days since December 1st
Cum
ulat
ive
salv
age
1992-1993
12010896847260483624121
5000
4000
3000
2000
1000
0
Days since December 1st
Cum
ulat
ive
salv
age 50th percentile
25th percentile
• GLM negative binomial models
• GAM
• Boosted Regression Trees -describe response surface-rather than one best model, average over MANY poorly fitting models-method to sort through MANY potential variables-interactions are automatically included-stepped predictions identify thresholds between high and low risk conditions
Revised analyses-Focus on conditions that explain salvage during first flush
Variables considered- Old and New
4000
020
000
0 40000
0
2000
00
0 10000
5000
0 10000
5000
0 30000
1500
00 806040 16
0800 16128 100
050
00 50
02500
160000800000100000500000400000200000010000500001600080000400002000008060401608001612810005000
SAC
SJR
TOT
SWP
EXPORT
OMR
X2
CCF NTU
Mallard Wat. Temp
Vernalis SSC
Freeport SSC
Also, predators at SWP and CVP
GLM Negative Binomial Model Summary
Model Run
Exports OMR Water Temp
VernalisSSC
Freeport SSC
CCF NTU Outflow PFMWT Bestmodel r-square
25th X X X 0.52
50th X X X 0.54
Annual X X X X 0.65
OMR flow (cfs)
Clift
on C
ourt
Tur
bidi
ty (N
TU)
1 000050000-5000-1 0000
1 40
1 20
1 00
80
60
40
20
> – – – – < 25
25 5050 7575 1 00
1 00 1 251 25
CVPSWP and
Adult Delta Smelt Salvage (1993-2014)
SWP and CVP Exports (cfs)
Clift
on C
ourt
Tur
bidi
ty (c
fs)
1 40001 20001 00008000600040002000
1 40
1 20
1 00
80
60
40
20
> – – – – < 1 0
1 0 2020 3030 4040 50
50
CVSWP and
CombinedSmeltDelta
Adult Delta Smelt Salvage (1993-2014)- 25 Percentile
Boosted Regression Tree Results
Physical PC rotations Relative influence (Rank) full 25th 50th
PCl PC2 PC3 PC4 Variable dataset percentile percentile interpretation Outflow PRJEXP.OMR.SSC CCC.NBAQ.X2 MISDIV
lambda 4.59 1.98 1.31 1.00 Phy.PC.PRJEXP.OMR.SSC 0.28 (1) 0.24 (1) 0.29 ( 1)
l:umulative varian~.:e explained 0.38 0.55 0.66 0.74 FMWT 0.20 (2) 0. 19 (3) 0.23 (2)
Bio.PC.Bull.Pminn 0.13 (3) 0.19 (2) 0.12 (4) Freeport. SSC Phy.PC. CCC.NBAQJC2 0. 10 (4) 0.05 (6) 0.14 (3) Vernalis. SSC
Spawn. day 0.05 (5) 0.08 (4) 0.04 (7) CCC
PRJEXP -o1Q8 -Dl'l Bio.PC .. SWPS:MB.CVPcrappie 0.05 (6) 0.05 (7) 0.05 (5)
NBAQ 4UO Phy.PC.MISDIV 0.03 (7) 0.03 ( 10) 0.04 (6) PREC Bio.PC.CVPpminnS:MB. SWPcrappie 0.03 (8) 0.01 (13) 0.02 (10) MISDV Bio.PC .. CVPL:MBcrappie 0.03 (9) 0.03 (8) 0.01 (11) WEST
RIO ..QJJD Phy.PC.Outflow 0.03 (10) 0.02 (11) 0.02 (9)
OUT ~ Mallard. Temp 0.03 ( 11) 0.07 (5) 0.03 (8)
O::viR ~02 Bio.PC.Blackbass.Pminn 0.03 (12) 0.03 (9) 0.01 (13) X2 ~ Bio.PC.CVPbullCHN.SWPbass 0.02 (13) 0.02 (12) 0.01 (12)
Small differences between facilities but same story
AIC 5 units lower than base model
AIC 2 units lower than base model
DRIVERSWinter Storms
DRIVERS Freshet EventSeasonal Trigger
PHYSICAL RESPONSES
OMR & SJR Flows
Turbidity Distribution
SPECIES RESPONSES
Smelt Dispersal
SPECIES RESPONSES
Prey Availability
Density of Smelt in the South Delta
MANAGEMENT ACTIONS
Operation of Cross Channel
GatesPumping
CONSEQUENCESSalvage &
Entrainment
Conceptual Model for Factors Affecting Entrainment of Adult Delta Smelt at Water Projects Facilities
ExportsOutflow Suspended sediment/Turbidity
Conceptual Model Review
Predation
Conceptual Model Review
Some questions best answered by other approaches-Hydrodynamic models-Tagged fish releases-Predator studies
DRIVERSWinter Storms
DRIVERS Freshet EventSeasonal Trigger
PHYSICAL RESPONSES
OMR & SJR Flows
Turbidity Distribution
SPECIES RESPONSES
Smelt Dispersal
SPECIES RESPONSES
Prey Availability
Density of Smelt in the South Delta
MANAGEMENT ACTIONS
Operation of Cross Channel
GatesPumping
CONSEQUENCESSalvage &
Entrainment
Conceptual Model for Factors Affecting Entrainment of Adult Delta Smelt at Water Projects Facilities
ExportsOutflow Suspended sediment/Turbidity
Predation
CCF NTU
Delta more complex than can be gleamed from single stations
Seriously, stay for the next talk
Acknowledgements
Funding SourcesSFCWA (Proposal Development)DWR and Reclamation
Delta Smelt Scoping Team (led by Steve Culberson and Scott Hamilton)Investigator Team (Ed Gross, Pete Smith, Rob Latour, Michael MacWilliams, Josh Korman) Bruce DiGennaroJason HassrickLeAnne RojasJillian BurnsAndrew KalmbachTara Morgan KingJennifer SibillaJames GleimDean Messer
Broad recognition that ecosystem-based management is preferred to single-species management
Gable FJ. 2005. A large marine ecosystem voluntary environmental management system approach to fisheries practices.
Ecosystem management focuses on:
-Functions of the environment
-Species communities rather than individual species
-Responses in terms of growth and survival rates, not numbers of fish per se