Eastern Brook trout: Joint venture

1
Eastern Brook trout Joint venture

• Editors
• M. Hudy, USDA Forest Service
• T.M. Thieling, USDA Forest Service, James Madison
  University
• N. Gnat Gillespie, Trout Unlimited
• Eric P. Smith , Virginia Tech

2
and a cast of thousands

• New York
• D. Bishop
• J. Robins
• B. Hammers
• F. Angold
• W. Pearsall
• C. Guthrie
• D. Zielinski
• F. Linhart
• D. Cornwell
• W. Elliot
• L. Suprenant
• B. Angyal
• R. Pierce
• M. Flaherty
• F. Flack
• R. Preall
• J. Daley

• Virginia
• Larry Mohn
• Paul Bugas
• Steve Reeser
• Maine
• Merry Gallagher
• Paul Johnson
• Gregory Burr
• Rick Jordan
• Ron Brokaw
• Forrest Bonney
• David Howard
• James Pellern
• Francis Brautigan
• Timothy Obrey
• Nels Kramer
• David Basley
• North Carolina
• Doug Bestler

3
The Big Picture Through the eyes of a brook
trout!
4
Lack of large scale evaluation

• Large scale assessments useful for
• Identifying problems and information gaps
• Setting priorities for restoration and funding

Population Data
Habitat Information
5
Assessment goals

• Assess the loss of reproducing brook trout
  habitat as it relates to historic
  (pre-settlement) levels.
• Assess watershed perturbations by expert opinion
  
• Assess watershed level metrics using GIS
• Make an interactive database on the web (ArcIMS)

6
Objectives

• Calculate landscape metrics as a measure of
  habitat characteristics
• Develop a model that uses landscape metrics and
  known brook trout status to predict areas where
  brook trout data is missing
• Indicate landscape thresholds or warning flags

7
What the assessment is not

• Classification of wild trout
• Classification of recreational fishing quality or
  potential
• A value judgment on past or current management
  practices
• A viability assessment

8
What scale?
9
(No Transcript)
10
Appropriate Scale

• 5th level too big
• Stream reach too small (too many)
• 6th level just right

• How big is a 6th level watershed?
• 41-163 km2
• 1 quadrangle map
• 3 x Harrisonburg

11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
Final Classifications

• Extirpated
• Predicted Extirpated from Unknown and Present
  qualitative
• Reduced gt 50
• Predicted Reduced gt 50 from Unknown and Present
  qualitative
• Intact gt 50
• Predicted Intact gt 50 from Unknown and Present
  qualitative

17
Results
18
GIS Analysis
19
Limiting factors by watershed(expert opinion)
20
Top Ten Threats Streams Cumulative Categories 1
2n 4,484

• Agriculture 36
• High Water Temperature 35
• Sediment-Roads 27
• All Exotics 26
• Urbanization 25
• Riparian Habitat 22.
• Brown trout 19
• Stream Fragmentation - Roads Culverts 17
• Dams 15
• Forestry 15

21
GIS dataSubwatershed and Water Corridor Metrics

• Over 60 Metrics
• Road Density
• Dams/area
• Road/Stream Crossings
• Population Density
• NO3 and SO4 Deposition
• Soil pH
• Elevation
• Land Use (21 land use classes)

22
National Land Cover Data (30m)
Human Uses
Natural Cover

• Low Intensity Residential
• High Intensity Residential
• Commercial/Industrial/ Transportation
• Quarries/Strip Mines/Gravel Pits
• Transitional
• Orchards/Vineyards
• Pasture/Hay
• Row Crops
• Small Grains
• Fallow
• Urban/Recreational Grasses

• Woody Wetlands
• Emergent Herbaceous Wetlands
• Open Water
• Perennial Ice/Snow
• Bare Rock/Sand/Clay
• Deciduous Forest
• Evergreen Forest
• Mixed Forest
• Shrubland
• Grasslands/Herbaceous

Derived Cover

• Total Forested
• Agriculture
• Residential
• Human Use

23
Screening

• Completeness

• Range

• Redundancy

• Responsiveness

24
Final Core Metrics for Analysis

• Forested land
• Agricultural land
• Combined N03 SO4 deposition (kg/ha)
• Road density (km/km2)
• Mixed forested land in water corridor
• Latitude (decimal degrees)

25
(No Transcript)
26
(No Transcript)
27
Model Development
28
Objectives

• Create a model to predict the status
  (Extirpated/Reduced/Intact) of the Unknown and
  Present subwatersheds
• Indicate thresholds metric cutoff points for
  land use managers

29
Model Development

• We developed Binomial and Trinomial models
• Single variable logistic regression
• Multi variable logistic regression
• CART Classification Trees
• Discriminate
• Linear
• Quadratic
• Nearest neighbor

Resubstitution Ten-fold Cross-Validation
30
Correct Classification Rates

• Binomial
• Presence/Extirpated CCR 66-79 (cv)
• Multivariable Logistic Regression (79)
• Nearest neighbor Discriminant Analysis (77)
• Classification trees (77)
• Trinomial
• Extirpated/Reduced/Intact CCR 51-65 (cv)
• Classification trees (65)
• Nearest neighbor (64)
• Multivariable Logistic (64)

31
We picked CART classification trees
Because

• Higher of correct predictions
• (better balance)
• Easier interpretation than logistic (especially
  trinomial) multi metric
• Helps in development of thresholds for land
  managers

32
What CART classification trees do

• Look at all possible combination of metrics and
  metric values to most efficiently divide the
  dataset
• Sets up a decision tree using different metric
  values as splitting criteria (20 80 couplets)
• Predicts the probability of correct
  classifications at terminal nodes

33
M3 Extirpated Reduced gt 50 Intact gt 5071
correct overall76 Extirpated64 Reduced79
Intact

• Forest lt68
• Deposition lt 28 kg/ha
• Deposition lt 19 kg/ha
• Agriculture lt 27
• Road Density lt 1.67 km/km2
• Deposition lt 18 kg/ha

34
(No Transcript)
35
(No Transcript)
36
Study Area 6th level watersheds

• Extirpated 21
• Predicated extirpated 8
• Reduced gt50
• 28
• Predicated Reduced gt50 7
• Intact gt50 14
• Predicated Intact gt50 17
• Absent Unknown History 5

37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
Model Analysis
41
Areas of misclassification
1. Extirpated subwatersheds misclassified as
presentExotic species?
42
Areas of misclassification
2. Reduced and Intact subwatersheds predicted as
Extirpated Low Total Forest and High
Deposition. Watershed size??
43
Key findings
44
Trouts there be good store in every brook,
ordinarily two and twenty inches

• John Josselyn New England 1674

45
Brook trout are extirpated from 29 of the
subwatersheds and reduced gt 50 in another 35

• The majority of large riverine habitats are gone

46
Presence does not equal persistence
47
Even with no further degradation many of the
Reduced gt 50 populations could become Extirpated.

• No connectivity or redundancy to reestablish
  populations after stochastic events
• Exotics fill in
• 330 subwatersheds highly vulnerable to extirpation

48
I have given the matter considerable thought,
and frankly I can think of not one stream that I
would classify as predominately brook trout. This
state and neighboring states have spent most of
their time and money stocking brown trout in what
were good brook trout waters

• All about Brook Trout from Maine to California
  Bob Elliot 1950

49
Exotics, Exotics, Exotics!!!

• Biggest non land use threat
• Rainbow trout in south east
• Brown trout in New York, New England
• Smallmouth bass in lakes
• Metric ??

50
If you dont know where you are going any road
will get you there !
51
Important quantitative data gaps exist for many
stream habitats (33 ) in large portions of
Maine, New Hampshire, New York with smaller gaps
in portions of Vermont, Massachusetts and West
Virginia.

• Need to validate the predictive models
• Quantitative needed for monitoring land use
  changes and exotics

52
very large and nice trout were formerly caught
here but since the introduction of pickerel about
the year 1820 but very few trout have been
taken
53
Lake populations have all but been eliminated
except for a few strong holds in Maine

• Vulnerable to exotics
• Vulnerable to land ownership changes

54
While many extirpations and losses occurred at
the turn of the century, many documented losses
have occurred in the last ten years.

• 75,000 dams
• 2 million miles road
• 90 million people

55
Land use metrics at the subwatershed level are
useful predictors of brook trout for land managers
56
Core Metric Forest

• Subwatershed threshold
• 68 forested land
• Only 6 of Intact gt 50 subwatersheds have less
  than 68 Total Forest.
• 85 of Extirpated subwatersheds lt 68 Total Forest

68
68
57
Core Metric Agriculture

• Subwatershed threshold
• 12-19 range or greater
• Only 17 of Intact gt 50 subwatersheds have
  greater than 19 Agriculture
• 74 of Extirpated subwatersheds have greater than
  12 Agriculture

12-19
12-19
58
Core Metric NO3 SO4 Deposition (kg/ha)

• Subwatershed threshold
• 24 - 33 kg/ha
• Only 23 of Intact gt 50 subwatersheds have a
  Deposition greater than 33 kg/ha
• 94 of Extirpated subwatersheds have a Deposition
  greater than 24 kg/ha

24 kg/ha
24 kg/ha
59
Core Metric Road Density (km/km2)

• Subwatershed threshold
• 1.8-2.0 km/km2
• Only 17 of Intact gt 50 subwatersheds have a
  Road Density greater than 1.8 km/km2
• 72 of Extirpated subwatersheds have a Road
  Density greater than 2.0 km/km2

1.8 km/km2
2.0
60
Next steps Brook trout populations 2015 ?
61
(No Transcript)
62
(No Transcript)