The National Stream Internet Project
An analytical framework for creating new
information from old stream data
Dan Isaak, Erin Peterson, Dave Nagel, Jay Ver Hoef, Jeff Kershner
BIG DATA =
BIG POSSIBILITIES
More Pressure, Fewer Resources
Climate Change
Urbanization &
Population Growth
Shrinking
Budgets
Need to do more
with less
21st Century Will be a Transitional One
Invest
Here
Not
here
Sorry Charlie
Good
Information
Critical for
Decision
Making
Strategically Consistent Information
Across Broad Areas for Efficient Planning
Tactically Precise Information for Local
Decisions & Project Implementation
Regional models
are too coarse
I’m going to invest here…
…instead of here
What’s a Stream Internet?
A network of people, data, digital information systems &
analytical techniques that interact synergistically to create &
transmit massive amounts of “information”
Data (what’s being
“clicked” on?)
People on landscapes
collecting data using
standard protocols
Google Servers & Databases
Corporate databases &
institutional memory
Search algorithms find signal in noise
(information from data)
New stream analyses &
research activities
Key Ingredient #1: NHDPlus Streams
Nationally consistent geospatial stream database
Reach
Descriptors:
All
• Elevation
2,500,000 • Slope
• %Landuse
stream • Precipitation
kilometers100’s more…
Cooter et al. 2010. A nationally consistent NHDPlus framework for identifying interstate
waters: Implications for integrated assessments and interjurisdictional TMDLs.
Environmental Management 46:510-524.
Wang et al. 2011. A Hierarchical Spatial Framework and Database for the National River
Fish Habitat Condition Assessment. Fisheries 36:436-449.
Key Ingredient #2: Statistical Models for
Data on Stream Networks…FINALLY!
Key Innovation of Stream
Models is Covariance Structure
Based On Network Structure
Models “understand”
how information moves among
locations based on network topology
Peterson et al. 2007. Freshwater Biology 52:267-279;
Peterson & Ver Hoef. 2010. Ecology 91:644-651.
Stream Models are Generalizable…
Distribution
& abundance
Response
Metrics
•Gaussian
•Poissan
•Binomial
Stream
Temperature
Statistical stream models
Genetic
Attributes
Water Quality
Parameters
Development Costs & Lineage
Spatial Network Tools: $560,000
CSIRO
$120,000
$17,000
EPA STAR
Grant $150,000
NOAA/NCEAS
$100,000
GNLCC
$40,000
USFS
$140,000
FLoWS
(Functional
Linkage of
Waterbasins
and Streams)
Spatial Stream Statistics
Working Group
Isaak, D.J., E. Peterson, J. V. Hoef, S. Wenger, J. Falke, C. Torgersen, C. Sowder, A. Steel,
M.J. Fortin, C. Jordan, A. Reusch, N. Som, P. Monestiez. 2014. Applications of spatial
statistical network models to stream data. WIREs - Water 1:xxx.
Peterson E.E. & Ver Hoef J.M. 2014. STARS: An ArcGIS toolset used to calculate the
spatial information needed to fit spatial statistical models to stream network data.
Journal of Statistical Software 56(2):1-17.
Peterson E.E., Ver Hoef J.M., Isaak D.J., Falke J.A., Fortin M.J., Jordan C., McNyset K.,
Monestiez P., Ruesch A.S., Sengupta A., Som N., Steel A., Theobald D.M., Torgersen
C.T. & Wenger S.J. 2013. Modeling dendritic ecological networks in space: an
integrated network perspective. Ecology Letters 16:707-719.
Som N.A., Monestiez P., Zimmerman D.L., Ver Hoef J.M. & Peterson E.E. In Press. Spatial
sampling on streams: Principles for inference on aquatic networks. Environmetrics
x:xxx.
Ver Hoef J.M., Peterson E.E., Clifford D. & Shah R. 2014.
SSN: An R package for spatial statistical modeling on
stream networks. Journal of Statistical Software 56(3):1-45.
Key Ingredient #3: Mountains of Data Could
be Mined to Create Valuable “Information”
Free
millions!
Temperature
Discharge – USGS NWIS
Free
millions!
Water
Quality
Free
millions!
Species
distributions
Genetic
Samples
Free
millions!
>45,000,000 hourly records
>15,000 unique stream sites
A BIG DATA Example
= 50% of Data
Stream Internet
>70 agencies
Technologies
in
Action
$10,000,000 data value
Regional Temperature Model
+
Accurate temperature
models
Cross-jurisdictional “maps” of
stream temperatures
Consistent planning
datum for 500,000
kilometers of stream
50 National Forests
& All Lands Between
Data from
New Models are More Accurate
than200
Old Models…
Training on left
2007 validation on right
Mean Summer
Temp
Summer Stream
Mean
Summer Mean
19
Predicted (C°)
14
9
4
4
30
25
20
15
10
5
9
14
Spatial Stream
MWMT Temp =
– 0.0045*Elevation (m)
y = 0.86x + 2.43
+ 0.0085*Radiation
+ 0.48*AirTemp (°C)
– 0.11*Flow (m3/s)
19
cted (C°) Predicted (°C)
Non-spatial Stream Temp =
– 0.0064*Elevation
(m)
y = 0.93x
+ 0.830
+ 0.0104*Radiation
+ 0.39*AirTemp (°C)
– 0.17*Flow (m3/s)
0.68x +RMSE
3.82
r2 y==0.68;
= 1.54°C
19
14
Training on left
2007 v
9
Non-spatial Model
4
Summer Mean
4
19
30
9
14
2
ry ==0.93x
0.93;
RMSE
= 0.74°C
+ 0.830
MWMT
19
y = 0.55x + 7.79
25
14
20
9
15
Spatial Model
4
10
4
5
Isaak et al. 2010. Ecol. Apps. 20:1350-1371
9
14
19
MWMT
Observed
(°C)
Model Interpolations Provide
High-Resolution Network Status Maps
Time 1
Time 2
Trend
Which then facilitate trend assessments…
Stream Thermalscape so Far…
The BLOB…it just keeps growing…
 29,593 summers of data swallowed
 296,000 stream kilometers of thermal ooze
BLOB Space, but BLOB time too…
Where
will the
Coldwater
Refuges
Exist?
1993-2011
2080’s
2040’s A1B
Composite
Here
they
are
The BLOB…it just keeps growing…
 29,593 summers of data swallowed
 296,000 stream kilometers of thermal ooze
Strategic & Tactical Prioritization of
Conservation & Restoration…
•Maintaining/restoring flow…
•Maintaining/restoring riparian…
•Restoring channel form/function…
•Prescribed burns limit wildfire risks…
•Non-native species control…
•Improve/impede fish passage…
Low
High
Priority Priority
The BLOB had a Dream…
“What if I could eat
data everywhere?”
2,500,000 stream kilometers nationally
Stream Internet Project Objectives
1) Develop compatibility between spatial stream analysis tools
and national hydrography layer (NHDPlus, v2)
2) Update STARS stream analysis tools to ArcMap 10.2
3) Host national workshop in 2015 to engage key researchers &
leaders from aquatic programs (i.e., power-users)
Projects like
NorWeST done
routinely &
incentives exist
for database
aggregation
Using the Stream Internet is Easy…
Distribution & abundance
Stream
Temperature
Develop a Database & DO IT!
Genetic
Attributes
Water Quality
Parameters
Stream Statistical Software is Free
SSN/STARS Website
Spatial Stream
Networks (SSN)
Package for R
• Software
• Example Datasets
• Documentation
User Community is Growing Rapidly…
>14,000 Visits to SSN/STARS
website in first 1.5 years
>500 software downloads
Locations of visits to SSN/STARS website in last month
2nd Annual Stream Statistics Training
Workshop in Boise
May 15 – 17
>100 Participants from U.S. & 1 from Egypt!
Ida ho Wa ter Center
3 day workshop
1st day: overview of spatial stream
models (webinar)
2nd/3rd days: work 1-on-1 with
Jay/Erin to model your data
Attendees (15 people); 1st day
webinar viewers (unlimited)
If Interested, contact Dan Isaak
(disaak@fs.fed.us) or go to the
SSN/STARS website for
registration details
More With Less, but What If…
It was Massively More?
Climate Change
Population
growth
Shrinking
Budgets
X
C u on …
stream
Step 2. Link to Covariate Predictors
100’s are Available (NHDPlus, NLCD, DEMs…)
Each Reach:
•
•
•
•
Elevation
Slope
%Landuse
Precipitation
100’s more…
Cooter et al. 2010. A nationally consistent NHDPlus framework for identifying interstate
waters: Implications for integrated assessments and interjurisdictional TMDLs.
Environmental Management 46:510-524.
Wang et al. 2011. A Hierarchical Spatial Framework and Database for the National River
Fish Habitat Condition Assessment. Fisheries 36:436-449.
Crowd-Sourcing Uses Everyone’s Data,
Everyone is Engaged in the Process
Landscape/
Network
GCM
Coordinated
Management
Responses
Management
Decisions
Data Collected by
Local Bios & Hydros
Closing Argument for A Stream InterNet
Technology exists. Statistical theory, computing
horsepower, & geospatial techniques can be used to
develop massive amounts of accurate information (e.g.,
NorWeST temperature project).
Needed. Doing more with less is the “new normal” and all
agencies have mandates to protect, conserve, & manage
aquatic resources.
Scalable. NHDPlus, BIG DATA aquatic
databases, & large user-base of motivated
aquatic resource stewards from hundreds
of organizations across the country.
Costs. Minimal, can we afford not to?
Value. Priceless. How do you value better information?
Integrate with…
Watershed Condition Indicators
Forest Plan
Revisions
Advantages…
1) Huge cost:benefit ratio ($10:$1 – conservatively)
2) Good Information empowers everyone to
manage/conserve more effectively
3) New data are collected more efficiently
All pieces are there – just need institutional &
individual commitments.
Aquatic Surveys Module in NRM
Temperature Surveys Tool in AqS
Data Entry, Uploading, Maintenance Interface
18
14
10
6
Stream
Temperature
(˚C)
7 Survey
Types
in AqS
Temperature
Reach
Passage
Characterization
TimeLake-Pond
Spring
Amphibian
Valley Segment
Callie McConnell’s development team is superb
Surveys/database structure can be evolved
Things We Can do to Help Ourselves
1. Develop/endorse a standard set of protocols
for measuring aquatic things (stream temp,
flow, water chemistry, biological
distributions/genetics)
2. Organize our army’s ammo (all data has to
go into NRM!) USFS has ~600 fish
bios/hydros. (That’s an aquatics army!)
3. Adopt/endorse the NHDPlus 1:100,000 scale
national hydrography layer for use by
national forest systems.
Lots of data
ose data…assuming we get it into a useful format and it’s archived in NRIS so
n and future generations can use it…
w motto should instead of “only you
ires…” “DON’T MAKE SMOKEY
ANDARDIZED DATA PROTOCOLS AND
A Digital Information Revolution is
Better Status & Trend Monitoring
Happening
Better Understanding & Prediction
New relationships
described
Response
Old relationships tested
Predictor
Refined
Rejected
Better Information Enables Better
Decisions, Efficiency, & Resource Stewardship
The 21st-Century will Be
a Transitional One
Pick
these
Not
these
Sorry Charlie
Research/Management Synergy
A Large Land-base
190 Million Acres
“Boots-on-the-Ground”
+
The Green Team has Unique
Advantages Relative to
Other Resource Agencies…
+
Lots of data
being collected
+
USFS has ~600 fish bios/hydros.
(That’s an aquatics army!)
+
Research stations develop
information & connect people
Where do Fish Fit in a Terrestrial World?
This is a Tree not a Fish
“Smart” Maps Developed from Data
Good Maps Significantly
Reduce Uncertainty
We need to make this generation’s maps showing
aquatic resource status
Where do Fish Fit in a Terrestrial World?
This is a Tree not a Fish
Data Needs to be Accessible
Data In
Information Out
#1
Spatially referenced,
corporate database
20
Status
Aquatic
Surveys
Module
Summer Mean Air (C)
22
#4
18
16
14
1970
Status & Trend
Assessments
1975
1980
1985
1990
1995
2000
2005
Time
#3 Spatially Continuous
Resource Maps
#2
#2a
More data,
monitoring
design
Analysis
2010
Key Points (quotes from LCC proposals)
GNLCC NorWeST proposal…objective is not to make recommendations
regarding specific management activities…only to provide tools that enable
development of accurate information that is fundamental to an informed
discussion about prioritizing those activities.
LCC Stream Internet proposal…over time, strong communities,
empowered with good information, will manage and conserve aquatic
resources more effectively.
…our experience in application of the Stream Internet technologies within
the regional NorWeST project is that they are transformative for developing
and solidifying the social networks and interagency collaborations that are
needed for the most effective stewardship.
Also address reviewer comments on proposal
Another “app” fish genome mapping
Show graphic of temp map, but pretend now that’s
a species genetic diversity mapped to the landscape.
Entirely possible—do overlay of Cheap protocols –
collect tissues, process for $5/fish once SNPs
developed, model diversity, # alleles, etc as a
function of landscape characteristics
“Stocking” the national stream (internet):
DNA barcoding the riverscape to identify, delineate, and monitor
units of conservation of aquatic species
Michael Young, Kevin McKelvey, Michael Schwartz, Daniel Isaak
What are conservation units?
Recognizable, substantive
components of the evolutionary
legacy of a taxon
●
●
●
●
lineage
DPS
ESU
GMU
●
●
●
●
MOTU
Stock
Subspecies
Species
Ardren et al. 2011
bull trout
What builds them?
●
●
●
●
Continental dynamics
Climate
Evolution
Interact over deep time to:





Build a river basin
Colonize it
Re-route the basin
Recolonize
Repeat
Why know about them: ectotherms are in trouble
How do we manage?





Translocation
Supplementation
Duplication
Refounding
Broodstocks
Williams et al. 2011
Identifying biodiversity:
use genes
 2003: universal DNA barcode
 ~95% species, 100% accurate
 Library of barcodes and sequences
 GenBank, BOLD
 Results:
 Conservation units
 Cryptic diversity
 Hybridization
 Adaptive traits
Delineating biodiversity:
address space
 Sample
 Broadly
 Intensively
 Randomly
 Map
 Conservation units
 Native species
conservation areas
Kershner et al. 2005
Al-Chokhachy et al. 2010
Westslope cutthroat trout
● State fish of Idaho, Montana
● Widely distributed across the Northern Rockies
● ~50% decline in distribution
● Repeatedly petitioned for federal listing
● “While conducting the new status review for WCT, we found no compelling evidence for
recognizing DPSs of WCT. Instead, for purposes of the new status review, we recognize
WCT as a single taxon in the contiguous United States.” –USFWS 2003
Phylogeography of
westslope cutthroat
trout
● Across five states and two
Canadian provinces
● WSCT are one taxon with:
● Marked “north-south”
split
● Lineages within each
● Terminal divergence =
post-Ice-Age
diversification
Phylogeography
of westslope
cutthroat trout
● Ancestral
● John Day, OR
● Hotspot
● Clearwater, ID
● Stocking
● Detectable
The trouble with sculpin (Cottus)
●
●
●
●
●
Extraordinarily abundant
Salmon/trout stream = sculpin stream
Most challenging group to identify
Hybridization common
Taxonomy unsatisfying
Young et al. 2013
Conservation units
for sculpins in
the Northern Rockies
Range of the new
species, cedar sculpin
Divergent torrent sculpins
are in the Palouse River
In 1932, these fish were described as
C. tubulatus, are confined to the Clearwater,
and could be a distinct species
Shorthead sculpins in the Sinks basins
are morphologically and genetically
distinct from those in the Salmon
Building a biodiversity
portfolio for the U.S.
● Use molecular methods to assess
samples across space & species
● Conservation unit atlas
● Proactive
● Internationally consistent
● Locally – globally relevant
● Data universally available
● Reasonable cost
● Waiting to be done . . .
● Once conservation units are
identified, how do we monitor them?
Monitoring the
biodiversity portfolio:
eDNA
● Pioneering this approach for
identification and detection
● Optimizing for local to rangewide assessments
● Developing user-driven
applications
● Costs: pennies on the dollar
Home across the range:
range-wide eDNA
assessment of bull trout
●
●
●
●
Sample 1000s of sites
Determine occupancy
Refine the model
Set the benchmark
Is anybody else home:
range-wide eDNA
assessment of aquatic
species
● All other species detectable!
● . . . after cons. units known