Landscape Scale Modeling of Trout Habitat Conditions

1
Landscape Scale Modeling of Trout Habitat
Conditions
Beth Gardner CSS 620
2
The BeaverKill
3
The BeaverKill

• Historically prestigious trout fishery
• Human Impacts Tanneries, Over fishing, Highway
  17, Development
• Declining trout populations

4
(No Transcript)
5
Temperature Impacts

• Behavior / Movement
• Growth
• Mortality
• Production
• Community Dynamics

6
Brook Trout Temperature

• Optimal 11 - 16 C
• Sub-optimal 16 - 23 C
• Hazardous gt23 C

7
BeaverKill - Temperature

• The main stem warms to gt23 C
• Fish are seeking thermal refuges

http//www.billcotter.com/onteora/chiefs.htm
8
Objectives

• Predict stream temperatures across the
  BeaverKill Watershed
• Characterize species abundance and mortality in
  relation to temperature

9
Outline

• Data Collection
• Example
• Methods
• Analysis
• Future Analysis

10
Data Collection - Temperature

• 72 temperature loggers
• Hourly temperatures recorded during the summer
  of 2000

11
Data Collection - Temperature

• Sites were chosen to reflect temperature at the
  mouth of a tributary and in the main stem nearby

Trib mouth
Upstream main stem
12
Data Collection Biological Sampling

• Randomly selected
• Fish abundance estimates
• Habitat evaluations

13
Data Collection Mapping

• All sites were geographically referenced using
  a hand-held GPS unit
• Hydrography maps were obtained from the EPA and
  CUGIR websites

14
BeaverKill Study Locations
15
Example

• Two observations of stream temperature across
  the watershed from July 15, 2000
• Main stem heating up to hazardous levels

16
100 AM

17
700 PM
18
Methods

• Use GIS (ArcView) to evaluate

• Stream Network
• Inter-point Distances

19
Stream Network

• Connect disjoint stream sections
• Assign directionality

20
Inter-point Distances

• 1. Calculated distances between each temperature
  logger along the stream network

21
Distance Matrix
22
Inter-point Distances

• 2. Calculated the distance between each
  temperature logger and each point to be predicted
• This returned another distance matrix, used to
  calculate k

23
Analysis

• Inter-point distances used to calculate

• Variogram
• Prediction Model

24
Variogram

• Average variation in difference of observations

25
Variogram
Sill
g(h)
Range
Nugget
26
Kriging

• This spatial variation is quantified by the
  semi-variogram
• Predictions weighted by distance and covariance

27
Variogram Without Network

• Fitted Spherical Model

28
Predicted Model
Results returned from Splus
29
Predicted Model
Transferred into ArcView with Stream System
30
Variogram 2-way Network
31
Predicted Model
32
Predicted Model
33
2-way Model Problems
34
1-way Vs. 2-way
A
B
C
35
Variogram 1-way Network
Sill 2.3 Range 7500 Nugget 2.5
36
Predicted Model

• Technical issues in predicting values based on
  the 1-way distance matrix between the predicted
  points and the temperature loggers

37
Future Analysis

• Calculate kriging estimates along the stream
  network in a systematic manner
• Expand systematic placement of points to be
  estimated every 1000 meters on the stream network

38
Systematic Grid
39
Future Analysis

• Weight the predictions by elevation or stream
  discharge

40
Future Analysis

• Make inferences about species distribution/abunda
  nce related to temperature

41
Brook Trout
Temp F
Number/m2
42
Number of Species Caught
Max 11 Min 1
43
Acknowledgements

• New York State Department of Environmental
  Conservation
• Clifford Kraft
• Arthur Lembo
• Dan Zielinski, Tamara Smith,
• Tim Patronski, Mark Roche, Nat Gillespie

44
Thanks
45