A National Framework for Monitoring Multiple Species at a Broad Scale

1
A National Framework for Monitoring Multiple
Species at a Broad Scale
Development and Pilot Pat Manley, USFS
Research pmanley_at_fs.fed.us Presentation Bea Van
Horne, USFS National Wildlife Program Leader for
Research bvanhorne_at_fs.fed.us
2

• We need a national framework for our monitoring
  to assure data are saved and used to the MAX
• The framework should be the most efficient, that
  is, provide the most information per dollar spent
• The framework should integrate with and
  compliment existing monitoring
• The framework should represent our best attempt
  to satisfy legal requirements

3
MSIM Objective

• Nationally consistent protocol to provide
  presence data (status) on an extensive array of
  vertebrate and plant species and their habitats
  across a broad scale in time and space.

4
Ancillary Benefits achieved for some species with
high detection rates

•   Indications of suitable habitat
• Estimates of abundance
• Measures of reproductive status, age ratios, and
  sex ratios for small mammals (based on captures) 
• Detection of spatially explicit change in
  distribution

5
Ancillary Benefits cont.

•    Information to evaluate existing and
  potential indicator species
• Provide a foundation for more intensive sampling
  and research

6
Why sample multiple species?

• Experience humility

Synchronous information
7
Stellers jay
Northern flicker
American robin
song birds
Goshawk
Spotted Owl
Marten
voles
woodrats
hares
ground squirrels
deer mouse
tree squirrels
chipmunks
8
Why use a grid-based approach?

• Stratification by habitat type is not a viable
  long-term strategy
• Consistency in site location
• Problems with road-based sampling

9
Challenges

• Bottom-up vs top-down protocol developmentstandar
  dization
• Adapt existing monitoring to the new realities
  of GIS, spatially explicit landscapes, and a
  broad-scale perspective

10
By standardizing sampling techniques we can
scale up to answer broad questions.

• Are bird populations declining in riparian zones
  of western forests?

• Does thinning have a positive or negative
  influence on small mammals?

11
MSIM Protocol contributes to research
effectiveness by . . .

• Providing baseline information about communities
  and ecosystems

12
Who is working on MSIM?

• Core Team
• Bea Van Horne, Pat Manley, Christina Hargis
• Coordinate Core and TT teams
• Craft National Framework and write FSM Technical
  Guide
• Design Team
• Suite of FSR scientists and NFS ecologists
• Steer overall sampling design
• Set statistical and other standards for protocols
• Inform data storage and management
• Recommend and develop analytic approaches
• Taxon/Technique Teams
• Review and recommend standard sampling protocols

13
MSIM Protocol meets information needs for
populations and habitats .

• ranges and trends for many species

1.0
100
Index of Riparian Condition
Presence in Riparian
50
0.5
0
0
02
10
06
Year
14
(No Transcript)
15
MSIM Protocol contributes to research
effectiveness by . . .

• Allowing researchers to place their studies in a
  larger context by tracking change across space
  and time in
• habitat
• species or taxon groups

Downy Woodpecker 1995
Downy Woodpecker 2002
16
Spatial display of Species-specific Annual
Survey Results
17
Theoretical Relationship between Presence and
Abundance over Time
18
Species occupancy
0 1 2 3 4 5 6 7 8 9 10
Year
Plant species richness
Canopy closure
0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10
Year
Year
19
MSIM Protocol meets information needs for
populations and habitats .

• ranges and trends for many species
• effects of management actions

100
Burned
Presence in Riparian
50
Not Burned
0
02
06
10
Year
20
Integrated Monitoring and Research
21
Burned
Sample Sites
MSIM Long Term
Riparian
22
Integrated Monitoring and Research
MSIM indicate that occupancy and abundance are
associated with prescribed fire
MSIM Data show substantial increase in prairie
dog occupancy and abundance
Research initiated and reveals thresholds of
and effects of prescribed fire
Management adapts to incorporate prescribed fire
and thresholds into management strategy
23
MSIM Protocol meets information needs for
populations and habitats .

• ranges and trends for many species
• effects of management actions
• larger context of management actions

24
MSIM Protocol meets information needs for
populations and habitats .

• ranges and trends for many species
• effects of management actions
• larger context of management actions
• habitat associations

FIA point - Plant and animal species -
Detailed vegetation data
Landscape around point - Remotely-sensed data
(veg, soils, disturbance)
25
Monitoring Data are Integrated
26
In sum, MSIM will work at a large scale to
provide information on

• Species range and distribution
• Co-occurrence among species
• Occupancy of specific habitats
• Trends in occupancy over space and time
• Trends in habitat over space and time

27
Conceptual Model of MSIM
Scale
National
Ecoregional
National Framework
Regional
Forest
Multiple-species add-ons
Single-species add-ons
MSIM monitoring plan
Population and habitat monitoring strategy
28
National Framework

• Ecoregion-scale application and sample site
  locations linked to FIA grid points

29
National Framework

• Ecoregion-scale application and sample site
  locations linked to FIA grid points
• Primary multiple-species detection methods
  identified that cover major vertebrate taxonomic
  group and plants

30
National Framework

• Ecoregion-scale application and sample site
  locations linked to FIA grid points
• Primary multiple-species detection methods
  identified that cover major vertebrate taxonomic
  group and plants
• Core multiple-species detection methods that
  detect the largest numbers of vertebrates and
  commonly selected MIS species and basic habitat
  parameters subset of primary methods required

31
National Framework

• Ecoregion-scale application and sample site
  locations linked to FIA grid points
• Primary multiple-species detection methods
  identified that cover major vertebrate taxonomic
  group and plants
• Core multiple-species detection methods that
  detect the largest numbers of vertebrates and
  commonly selected MIS species and basic habitat
  parameters subset of primary methods
  mandatory
• Sampling design and data management and analysis
  specifications

32
Ecoregion Plan Development

• Quantify the number of FIA points in ecoregion
  and habitat representation

33
Ecoregion Plan Development

• Quantify the number of FIA points in ecoregion
  and habitat representation
• Estimate the number and range of species expected
  to be adequately detected identify short-falls
  (e.g., MIS, SOC, certain habitat associates)

34
Concern Criteria
Exotic, domestic,and native pest
Extirpated/ Potentially extirpated
Potentially imperiled
Potentially vulnerable
Endemic
Threatened, endangered, special concern
Population and range characteristics
Life history traits
Rarity
35
Ecoregion Plan Development

• Quantify the number of FIA points in ecoregion
  and habitat representation
• Estimate the number and range of species expected
  to be adequately detected identify short-falls
  (e.g., MIS, SOC, certain habitat associates)
• Identify annual monitoring budget

36
Ecoregion Plan Development

• Quantify the number of FIA points in ecoregion
  and habitat representation
• Estimate the number and range of species expected
  to be adequately detected identify short-falls
  (e.g., MIS, SOC, certain habitat associates)
• Identify annual monitoring budget
• Develop options for proportion of FIA points to
  be included in sample, location of points to be
  added, primary and detection methods to be added
  to core set, single-species methods to be added

37
Pilot Study Area
Lake Tahoe
Lake Tahoe
Sierra
Sierra
Nevada
Nevada
FIA hexagon
FIA hexagon
clusters
clusters
California
California
38
Species Specs

• 465 vertebrate species and 4000 plant species
• 213 species at risk
• 46 management indicator species
• Forest Service required to monitor populations
  and habitat for all MIS, and needs to be able to
  assess the effects of management on species at
  risk

39
2002 Forest-scale Application
40
Detection Methods2001-2002 pilot tests

• Bird point counts 7-8 stations, 3 v
• Bat surveys (mn and ac) 3 sites, 2-6 v
• Tomahawk traps 54 traps, 4 nights
• Sherman traps 102 traps, 4 nights
• Track plates 5 stations, 5 checks
• Vertebrate area searches (aq and terr) 2 v
• Nocturnal broadcast surveys 3 km2, 2 v
• Pitfall arrays - 2 arrays, 10 visits
• Coverboards 10 boards, 10 visits
• Plant surveys 8 1m2 quadrats, 2v

41
Monitoring point
Track stations
Lake
Small pond
Bat mist nets
Live trapping
Meadow
Bird point counts
Pitfalls
Aq. vert. surveys
Conifer forest
Plant surveys
Habitat measures
Riparian
Note not to scale.
42
Primary Detection Methods Birds
43
Primary Detection Methods Mammals
44
Primary Detection Methods Amphibians and Reptiles
45
Primary Detection Methods Vascular Plants
46
Primary Measurement Methods Habitat
47
Secondary Detection Methods
48
Multiple-Species Approach Dry Run

• What if
• We implemented 10 standardized, commonly employed
  multiple species detection methods for
  vertebrates
• At each FIA grid point on NFS lands in the Sierra
  Nevada, and
• Based on estimates of the number of points in
  each species range and their probability of
  detection with the 10 protocols, then
• Which species would we expect to observe at
  enough points to detect gt 20 change between two
  time periods with 80 confidence and power?

49
MSIM Predicted Effectiveness

• Based on modeling, over 60 of all vertebrate
  species, and 80 of all vertebrate MIS species in
  the Sierra Nevada were predicted to be observed
  frequently enough to detect.
• 20 relative change in observations with
• 80 confidence and power

50
MSIM Predicted Effectiveness

• Based on modeling, over 60 of all vertebrate
  species, and over 80 of all vertebrate MIS
  species in the Sierra Nevada were predicted to be
  observed frequently enough to detect.
• Most species considered potential indicator
  species were also adequately detected

51
MSIM Predicted Effectiveness

• Based on modeling, over 60 of all vertebrate
  species, and over 80 of all vertebrate MIS
  species in the Sierra Nevada were predicted to be
  observed frequently enough to detect.
• Most species considered potential indicator
  species were also adequately detected
• Species predicted to be adequately detected
  represented a balance of life history
  characteristics and habitat associations,
  including species of concern

52
Specific 2002 Pilot ResultsSpecies Detections
53
General Field Test ResultsPredicted
Efficiencies Realized

• Diversity of species detected
• Over 50 of all species expected were detected
• Logistically feasible to implement
• Habitat data needs to be collected at all sample
  sites, not just center point
• Cost estimates refined
• QA/QC measures need to be developed

54
Factors Affecting Optimal Designs
Effort per point
Number of points
Probability of detection
Proportion of points with observations
Proportion of points occupied
Remeasurement frequency

• Habitat relationships
• Cause-effect relationships
• Indicator species dev./eval.

• Trend detection
• - occupancy, richness, comp
• Spatially explicit trend detection

55
Parameters of Interest

• Probability of detection
• Estimates of
• proportion of points occupied
• probability of observation
• habitat relationships and indicator species
• potential cause-effect relationships
• Statistical power to
• detect change
• detect spatially
• explicit change

56
Estimating Probability of Detection and
Proportion of Points Occupied

• Based on presence/absence data binomial
  distribution
• Maximum likelihood estimation to derive
  probability of detection and proportion of points
  occupied estimates
• Program PRESENCE on the Patuxent Wildlife
  Research Center website used for protocols with
  one sample unit per point
• SAS program written by Jim Baldwin (USFS
  statistician) used for protocols with multiple
  sample units per point (e.g., bats)

57
Observed vs. Estimated Proportion of Points
Occupied
58
Tahoe vs. Sierra Nevada Trackplate and Camera
Results
59
Change Detection Validation

• We were able to estimate p and PPO for about half
  the species we detected
• We expect this percentage to improve
  substantially with a larger number of monitoring
  points
• Pilot had 20 to 40 monitoring points NFS lands
  in Sierra Nevada have 1800 FIA points

60
Change Detection Validation cont.

• For the 80 species for which we could estimate
  probability of detection.
• 78 were estimated to be adequately detected based
  on our best guess of p and PPO (dry run)
• 75 are estimated to be adequately detected based
  on empirical data

61
Change Detection Validation cont.

• The 80 species with estimates of p and PPO were
  not always detected at a high percentage of
  points

62
Bird Species Accumulation by Visit
63
Bird Species Accumulation by Survey Duration
64
Bird Species Accumulation by Station
65
Relative Efficiency of Sites vs. Visits for Bats
Assemblage
66
(No Transcript)
67
Budget Core Data Collection
Cost includes equipment, housing, training,
habitat data
68
Budget Additional Methods
Cost includes equipment, housing, training,
habitat data