Monitoring population effects of an emergent disease in wild birds'

1
Monitoring population effects of an emergent
disease in wild birds.

• Shannon L. LaDeau
• Postdoctoral fellow
• Smithsonian Institution
• National Zoo-Migratory Bird Center
• Advisor Peter Marra

2
Disease emergence in U.S. avifauna
Avian tuberculosis (1986) Newcastle disease
(1992) House Finch conjunctivitis (1994) West
Nile virus (1999) H5N1-Avian flu (??)
STEPHEN JAFFE/AFP US Army specialist Steve
Richards captures mosquitos
Ready or not, here it comes. It is being spread
much faster than first predicted from one wild
flock of birds to another, an airborne delivery
system that no government can stop. from
coverage of M. Leavitt speech. March 2006
3
Objective
Identify impacts of West Nile virus in wild bird
populations.

4
West Nile virus
1999 emergence in Queens, NY. Primary avian
host. Mosquito vector 284 avian species in 48
states
Positive bird surveillance by county
From CDC/USGS
5
North American Breeding Bird Survey (BBS)

• Citizen scientists
• 1966 to current
• Over 4100 survey routes
• 24.5 mile along secondary roads

Sauer, J. R., J. E. Hines, and J. Fallon. 2005.
The North American Breeding Bird Survey, Results
and Analysis 1966 - 2004. Version 2005.2. USGS
Patuxent Wildlife Research Center, Laurel, MD
6
Route selection

• Mid-Atlantic states
• Temporal coverage At least 80 data coverage
  from
• 1980 2005 with observations in 5 of 6 years
  after 1999.

2004 Population (people/per sq Mile lt
3500 3500-8850 8851-20850 20851 - 55775
7
West Nile Footprint

• Crows experience high mortality.
• Komar et al. 2003
• Eidson et al. 2001

8
Data
Mean of OBSERVED counts
Average count
WNV emergence in NY
9
Spread of West Nile virus
10
West Nile Footprint

• Crows experience high mortality.
• Komar et al. 2003
• Eidson et al. 2001

2. Population effects will be patchy and
greater near urban areas. Kilpatrick et al.
unpub Hochachka et al. 2004 Caffrey and
Peterson 2003
11
Data model
For a given species, individual counts are
conditionally Poisson where subscripts i and j
refer to observer and route identity,
respectively, and t denotes year. The expected
value for a given annual count after accounting
for route and observer effects is with
random effects for variation among routes, years
and observers.

12
Data
Mean of OBSERVED counts
WNV exposure?
Average count
13
Data versus Predicted
Mean of PREDICTED counts
Mean of OBSERVED counts
Average count
14
Unusual routes after 2000
15
Unusual routes before 1999
16
Summary

• 1. Monitoring disease in wildlife populations
  demands analyses that can accommodate natural
  stochasticity, census data and unplanned
  experiments without replication.
• We may not be collecting data at scales useful
  for monitoring avian disease. Consistent
  sampling across rural to urban
• Modeling/Analyses future
• other species
• state-space approach
• using human or crow data as prior information
  regarding spatial exposure.

17

• Special thanks to.
• USGS and BBS volunteers
• Wayne Thogmartin, Bill Link, John Sauer, Michael
  Lavine, and Jim Clark for discussion and modeling
  input.

18
Extra slides
19
Monitoring wildlife disease is difficult Cant
see the disease - Follow mortality Often there is
no population data prior to disease How disease
regulates/limits wildlife is largely
unknown. Disease emergence in U.S.
avifauna Avian tuberculosis (1986) Newcastle
disease (1992) House Finch conjunctivitis
(1994) West Nile virus (1999) H5N1-Avian flu (??)
20
Do we have the data we need?
WNV exposure rates WNV-related mortality
rates Population size prior to disease
emergence Monitoring of populations at scale of
disease ecology
21
Data
22
Trend Analysis

Identify routes where trend before WNV emergence
differs from post 2000 trend.
23
Trend analysis
Change in trend from 20 year mean
Increase in trend
Decrease in trend
24
Spread of West Nile virus