Spatial%20Sampling%20Designs%20Why%20do%20we%20Need%20Them?

1
Long-term monitoring of large, remote areas with
minimal funding hope and encouragement for
natural area managers
Steven Fancy National Monitoring Program
Leader National Park Service
2
State of the Parks Report (circa 2000)

• 80 (1/3) of the natural resource parks had no
  professional natural resource manager
• Another 84 parks had only 1 or 2 natural resource
  professionals.
• Almost all projects/studies were short-term
  staff mostly deals with the crisis of the day.
• Sampling designs mostly done by expert opinion
  representative sites or convenience.

No Time No Money No Clue
3
NPS Natural Resource Challenge
Revitalize and expand the natural resource
program within the NPS improve park management
through greater reliance on scientific knowledge

• Accelerate Inventories
• Design/Implement Vital Signs Monitoring
• Collaboration with scientists and others
• Improve Resource Planning
• Enhance Parks for Science
• Assure Fully Professional Staff
• Control Non-native Species
• Protect Native and Endangered Species
• Enhance Environmental Stewardship
• Expand Air Quality efforts
• Protect and restore Water Resources
• Establish Research Learning Centers

4
Economics 101
Funding level (avg. 100,000 per park) would
allow each park to hire one professional position
(GS-9 or 11) plus about 30-40 K operating
Conclusions/Strategy

• Without integration and cost-sharing, parks can
  only monitor a few things too few to adequately
  track condition of resources
• Park buy-in and cost-leveraging through
  partnerships are critical must be relevant to
  park managers and flexible to allow integration
  and partnerships
• Establish 32 monitoring networks that share
  funding and staffing among parks to gain
  efficiencies and consistency.

5
The Wedding Cake An alternative to One Size Fits
All
National
Network/Ecosystem
Park

• Primary use of data is at local level for park
  management
• Indicators and protocols most relevant to each
  system are very different e.g., Northwest
  Forests vs. South Florida vs. Colorado Plateau
• Cost-leveraging through Opportunity very
  important for early success and efficiency

Servicewide Core Variables
Network/Ecosystem Core Variables
6
Spatial Sampling DesignsWhy do we Need Them?

• We can only sample a very small proportion (often
  lt1) of most natural areas but,
• Our job is to protect, restore, understand, and
  inform others about the entire area, not just
  some convenient portion of it.
• Need to make scientifically-defensible inferences
  to areas beyond the actual areas we sample.
• Statistical, design based inferences can only be
  made to areas that have a chance of being
  included in the sample.

Especially important because we are designing
LONG-TERM MONITORING
7
Reality Check Sometimes you just cant get there
8
Take Home Messages
A few Lessons Learned about Sampling Design for
Long-term Monitoring

• Design should allow inferences to entire park,
  not just some accessible portion of it
• Some sort of probability sample should always be
  taken, even if you can only start with very small
  sample sizes
• Judgement sampling, using representative sites
  selected by experts, should be avoided
• Do not stratify on biological criteria (e.g.,
  vegetation maps). Use stratification sparingly,
  if at all
• Permanent plots that are revisited over time are
  recommended to increase precision
• It is not necessary to visit plots every year -
  rotating panels increase sample size
• Co-locate sample sites for various components of
  the monitoring program

9
Three Examples

• 2-stage mini-grid design for Central Alaska
  Network 3 parks gt21 Million acres (size of
  Indiana)
• GRTS design large river parks in Missouri and
  Arkansas
• Unequal probability sampling of upland sites in
  Northern Colorado Plateau parks (site selection
  based on accessibility, climate, geology and
  soils)

10
First-stage systematic grid with 20 km spacing
11
Second-stage mini-grid sample (25 points)
12
(No Transcript)
13
Buffalo River Streams 100 sites
Sampling of stream segments with co-location and
co-visitation among monitoring indicators

• Geomorphology
• Aquatic Macroinvertebrates
• Fish Communities
• Ozark Hellbender (giant salamander, TE)
• Mercury
• Lead

14
(No Transcript)
15
(No Transcript)
16
Northern Colorado Plateau Network (NCPN)
Sampling Design
17
Integrated Upland Sampling Design - NCPN

• Collocated Vital Signs vegetative structure,
  composition,
• condition of focal communities soil/site
  stability
• upland-hydrologic function soil-crust
  structure
• nutrient cycling

• Systematic grid of points (50-m spacing)
  overlaid on park
• used to define anchor points of sampling plots
• Unequal probability of selection of sampling
  plots
• provides ability to add sites in the future (if
  needed)
• probability defined by accessibility (high, low)
• probability defined by climate, soils, and
  geology
• (rel. static surrogate for vegetative
  communities)

18
Integrated Upland Sampling Design NCPN
Temporal Design

• A split-panel design will be employed that
  features
• spatial interpenetration
• the ability to evaluate inter-annual variability
• a balance between trend and status assessments
• And that meets Vital Sign sampling objectives

19



Split panel (2-7)9, (1-8)9
10
9
8
7
6
5
4
3
2
1
Year/ Panel
X
X
X
1
X
X
2
X
X
3
X
X

4
X
X
5
X
X
6
X
X
7
X
X

8
X
X
X
9
X
X
1

X
2
X
3
X
4
X

5
X
6
X
7
X
8
X
9
X 10 plots 30 plots per year spatially
distributed ca. equally among 4 Ecosites
20
Weight of Evidence
21
http//science.nature.nps.gov/im/monitor
22
http//science.nature.nps.gov/im/monitor
23
Take Home Messages
A few Lessons Learned about Sampling Design for
Long-term Monitoring

• Design should allow inferences to entire park,
  not just some accessible portion of it
• Some sort of probability sample should always be
  taken, even if you can only start with very small
  sample sizes
• Judgement sampling, using representative sites
  selected by experts, should be avoided
• Do not stratify on biological criteria (e.g.,
  vegetation maps). Use stratification sparingly,
  if at all
• Permanent plots that are revisited over time are
  recommended to increase precision
• It is not necessary to visit plots every year -
  rotating panels increase sample size
• Co-locate sample sites for various components of
  the monitoring program

A scientifically-defensible sampling design can
be done, and it should be done. Assistance and
examples are available from others.