Avian abundance estimation

The three pillars of wildlife management:

animal populations, their habitats, and the human stakeholders

Measuring populations

Major determinant of harvest quotas for game species
• Our best integrative measure of habitat management

Index

not a true estimate of pop. size, but allows
tracking of trends

• relevant abundance is increasing/decreasing over time relative to ….

Estimate

attempted estimate of true pop. size, with
measure of precision

• capture/mark recapture

Census

a true, complete count (rare)
• Typically for ultra-rare species (California Condors,
Whooping Cranes)

we've got two general problems that we need

to deal with sort of during the study design phase.

So the first would be:

detection (β)
. we dont see and hear everything

• need to account for detection probability

• if you know your detection probability, than you can estimate the true population

Count = β x N
N = Count / β
Two general problems:

Sampling bias

how you actually design

the study to reasonably cover the territory

that you want to cover and come up with an estimate

of populations that you actually believe.

Need enough replicates to estimate mean and variance

Why count birds?

Distribution and density; where they are and how many we have

• we have a lot of birders in north america, so our map of avian distributions is uneven

Region-specific population trends

• point counts

  Monitor the effect of some change or perturbation

• by monitoring birds through time, we are in a

  position to evaluate the effects of changes on the landscape.

Specific research or monitoring question

why is the bobolinks doing so good

1985 is when we had the North American

Waterfowl Management Plan passed, which helped conserve a lot of grasslands and wetlands.

It's also when we had the 1985 Farm Bill signed,

which implemented federal programs like Sodbuster,

saying, thou shalt not till up thine erodible soil.

And so a bunch of conservation programs were implemented in

the mid-1980s that probably benefited this grassland bird.

What to measure?

Presence/absence
• Total abundance
• Relative abundance or density (corrected for effort or area)
• Spacing, habitat use, behavior
• Demographic parameters (reproduction, survival)

3 ways of estimating avian abundance

• Point counts and transects
• Common modifications: callback surveys, roadside
surveys, aerial surveys
• Area searches and spot mapping
• Most common ways to locate nests
• Capture, mark, recapture
• Lots of ways to catch birds
• Lots of ways to mark birds
• Recapturing is harde

Point counts

• Observer goes to fixed point for a fixed
period of time
• Need to consider randomizing for time of day
• Counts all birds seen or heard and
estimates distance out to some
maximum
• Do you want to count flyovers?
• Double counting always a consideration

Point count assumptions

• All birds are detected
• Distance is estimated correctly
• Each bird is independent (flocks)
• Birds are not double-counted
• No differences between observers
• Observers do not influence birds

Breeding Bird Survey (1965)

USGS Patuxent WRC
• Each route is 25 mi and has 50 stops
• 3-minute point counts out to ¼ mile

Line transect surveys

• Systematically set sampling design
• Count animals to pre-determined distance from
transect
• Extrapolate count per unit to the entire area

May BPOP (1955)

Duck pairs and wetlands
counted from low-flying
aircraft along transects
• Covers 2.0 million mi2
• Ground-truthing and
visibility correction (1970s)
• Longest-running wildlife
survey in the world

• width of transect is important

Narrow transect:

high detectability, but high sampling bias

• where you put your lines matters a lot

Wide transect:

worse detectability, better sampling coverage

• wider area

how to correct for transect surveys

Estimate distance to detection

estimating the distance to

detection under the assumption that

detection falls off as a matter of distance so you're

one way that you can curb these data is to throw out the things where your detection probability is bad

Common point count and line
transect modifications

Roadside surveys—transects may not be
optimal, but at least they are consistent

• calculate a visibility correction factor for what airplanes missed

Callback surveys—certainly for secretive
marshbirds. can only hear them if you play it first

Area searches

• Basically, adjacent transects or randomly wander
• Common for nest searching

Spot mapping

• Observers watch for
territorial behavior
during the breeding
season
• Must be able to identify
individuals or assume
territorial exclusion
• Goal is to map area that
is defended; can follow
birds to the nest

• assume territorial exclusion

• individuality

different types of traps

• mist netting

• coral traps

• Funnel trap (various sizes)

• rocket net

• Bal-chatri trap

• Noose carpets

Marking

Usually standard aluminum bird bands (issued
by USGS); other materials exist

• Lots of other options: readable bands (spotting scope), PIT tags (microchip), web tags (ear of small mammal or fin of fish - put in toe pads of ducklings), patagial tags (larger birds, radiotelemetry and will fall off eventually), telemetry, saddles

Age and sex ratios

• plumage characteristics

• skull ossification

  • baby birds still have squishy skulls

Captures: once you have a bird in hand...

• Age and sex ratios
• Morphometrics
• Body condition (fat
score, mass corrected
for size)
• Breeding condition
(molt, brood patch)
• Transmitters of various
types

Lincoln-Peterson

N = population size
n1 = # animals captured and marked at t1
n2 = # animals captured at t2
m = # of marked animals at t2
𝜷 = 𝒎
𝒏𝟏
Can estimate
detection
probability
𝑵 = 𝒏𝟏𝒏𝟐
𝒎
Can estimate populations

example

Catch 12 ducks, mark and release
Days later, catch 9 ducks
3 are marked
𝑵 = 𝒏𝟏𝒏𝟐
𝒎 𝑵 = (𝟏𝟐)(𝟗)
𝟑 = 𝟑𝟔

Assumptions of banding data

• Representative sample
• We are drawing inferences about the population of
interest
• Closed population
• Trapping does not affect subsequent recapture
• No band loss
• Banding does not affect survival