Bald Eagle Post-Delisting Monitoring Plan

U.S. Fish & Wildlife Service

Post-delisting Monitoring Plan for the
Bald Eagle (Haliaeetus leucocephalus) in
the Contiguous 48 States

OMB Control No. 1018-0143
Expires: 11/30/2012

Paperwork Reduction Act Statement: The Office of Management and Budget has reviewed and approved the
collection of information associated with the Post-Delisting Monitoring Plan for the Bald Eagle. We collect this
information in order to determine the status of the bald eagle. Your response is voluntary. We estimate the
completion time to be 30.8 hours per response, including time for reviewing instructions, gathering and maintaining
data, and providing the information.
You may send comments regarding the burden or any other aspect of the reporting requirements to the Service
Information Collection Clearance Officer, U.S. Fish and Wildlife Service, 1849 C Street, NW., M/S 2042-PDM,
Washington, DC 20240. We may not conduct or sponsor and you are not required to respond to a collection of
information unless it displays a currently valid OMB control number.

Cover Photo: Bald eagle adult and two chicks. Photo by Brad Tedrick, Illinois Department of
Natural Resources.
2

U.S. Fish & Wildlife Service

Post-delisting Monitoring Plan for the
Bald Eagle

March 2009

Recommended Citation
U.S. Fish and Wildlife Service. 2009. Post-delisting Monitoring Plan for the Bald Eagle (Haliaeetus
leucocephalus) in the Contiguous 48 States. U.S. Fish and Wildlife Service, Divisions of Endangered Species
and Migratory Birds and State Programs, Midwest Regional Office, Twin Cities, Minnesota. 75 pp.

3

Table of Contents
Acknowledgements ....................................................................................................................... 6
Background ................................................................................................................................... 6
Purpose and Goal.......................................................................................................................... 9
Implementation ............................................................................................................................. 9
Methods........................................................................................................................................ 12
Protocols....................................................................................................................................... 15
Data Quality ................................................................................................................................ 16
Habitat ......................................................................................................................................... 16
Contaminants .............................................................................................................................. 17
Ongoing and Potential Sources of Mortality ............................................................................ 18
Response Trigger ........................................................................................................................ 18
Reporting ..................................................................................................................................... 18
References.................................................................................................................................... 20
Appendix 1................................................................................................................................... 22
Tables ........................................................................................................................................... 36
Figures.......................................................................................................................................... 45
Appendix 2................................................................................................................................... 55
Appendix 3................................................................................................................................... 62
Appendix 4................................................................................................................................... 70
Appendix 5................................................................................................................................... 74

4

Summary
The Post-delisting Monitoring Plan (Plan) will monitor the status of the bald eagle by
collecting data on occupied nests over a 20-year period with sampling events held once
every 5 years starting in early 2009. The Plan will continue the nest check monitoring
activities conducted by the States over the past years and add census of area sample plots.
The area sample plots will be selected from eagle habitat across the contiguous 48 States
based on known nesting density. The set of known occupied nests (list frame) will be
combined with the numbers of newly identified occupied nests from the area plot samples
(area frame) to provide a dual frame estimate (Appendix 1). Statistically combining the
results of these two data sets will provide a single estimate for the bald eagle population
of the contiguous 48 States that more closely represents the actual nesting population of
bald eagles than either the traditional nest check for occupancy or area plot sampling
alone, based on our pilot studies in Maine, Minnesota, Florida, Washington and Missouri
(Appendix 1). Reduction in future nest check monitoring (list frame) by the States will
be compensated by sampling the list frame during the area plot surveys. In addition, dual
observer sampling protocols are recommended to reduce bias (Appendix 2). Some
States, particularly those with sparse numbers of nesting pairs, are currently collecting
data in a highly accurate manner and may not need to employ the dual frame
methodology. Data from these States will be included as a complete census.
The Plan recommends that the State agencies continue the occupied nest survey data
collection and submission and assist with the area surveys while the U.S. Fish and
Wildlife Service (Service) coordinates the area survey, manages the database, provides
expertise including dual frame sampling design and data analysis, and initially funds the
area sampling. This Plan is not intended to replace specific plans to manage eagles or
monitor them in a different manner for specific management purposes.
The sample design is based on an 80 percent chance of detecting a 25 percent or greater
change in occupied bald eagle nests over any period, measured at five-year intervals. We
believe this is a goal that will both ensure recovery and be cost-effective. Were this
degree of decline to occur with no further increase, the bald eagle population would still
be at a level recognized as recovered (from 9,789 occupied nests when the bald eagle was
delisted in 2007 to 7,342 occupied nests, a 25 percent reduction) based on a population
estimate of 6,471 when the initial proposal to delist was published in 2000. If such
declines are detected, the Service’s Bald Eagle Monitoring Team in conjunction with the
States will investigate causes of these declines, including consideration of natural
population cycles, weather, productivity, contaminants, other mortality factors, habitat
changes or any other significant evidence. The result of the investigation will be to
determine if the population of bald eagles in the contiguous 48 States warrants expanded
monitoring, additional research, and/or resumption of Federal protection under the
Endangered Species Act (ESA). At the end of the 20 year monitoring program, we will
conduct a final review. It is the intention of the Service to work with all our partners
toward maintaining continued species population expansion and management.

5

Acknowledgements
This monitoring plan was written by the U.S. Fish and Wildlife Service Bald Eagle
Monitoring Team: Jody Millar, National Coordinator (Region 3), Suzanne Audet
(Region 1), Greg Beatty (Region 2), Allan Mueller, Candace Martino, Alfredo Begazo
(Region 4), Craig Koppie (Region 5), Dan Mulhern (Region 6), Phil Schempf (Region 7),
and Mary Klee (Region 9); Mark Otto, Migratory Birds program, and John Sauer, US
Geological Survey-Biological Resources. The States of Maine, Minnesota, Florida,
Washington, and Missouri provided critical support in participating in the pilot program.
Background
Between 1952 and 1957 Charles Broley, an avid eagle watcher, reported that about 80
percent of the bald eagle (Haliaeetus leucocephalus) nests in Florida he had been
watching failed to produce any young. By 1958, nesting adult eagles were so scarce in
his study area that he only found 10 nesting pairs where he had found 47 the previous
year, and had found 125 nesting pairs 15 years earlier (Carson 1962). This monitoring
information was ultimately linked to a deadly insecticide in widespread use at that time:
DDT (Carson 1962).
Subsequent bald eagle surveys conducted in the 1960s by the National Audubon Society
and others documented poor nesting success and low numbers of nesting pairs, prompting
the Secretary of the Interior to publish a Federal Register notice (32 FR 4001) on March
11, 1967 listing bald eagles south of 40o N. latitude as endangered under the Endangered
Species Preservation Act of 1966 (Pub. L. No. 89-699, 80 Stat. 926). Bald eagles north
of this line were not included because northern populations were not considered
endangered at that time.
In the 1970s, bald eagle surveys conducted by the Service, other cooperating agencies,
and conservation organizations revealed that the bald eagle population was declining
throughout the contiguous 48 States. On December 31, 1972, DDT was banned from use
in the United States by the Environmental Protection Agency. The following year, the
Endangered Species Act of 1973 (16 U.S.C. 1531-1544) (ESA) was passed. In 1978, the
bald eagle was listed throughout the contiguous 48 States as endangered except in
Michigan, Minnesota, Wisconsin, Washington, and Oregon, where it was listed as
threatened (43 FR 6233, February 14, 1978).
Listing under the ESA and banning of DDT and other harmful organochlorine chemicals
resulted in significant increases in the breeding population of bald eagles throughout the
contiguous 48 States. On February 7, 1990, the Service published an advance notice of a
proposed rule to reclassify the bald eagle from endangered to threatened in 43 States
where it was classified endangered and to retain threatened status for the remaining five
States (55 FR 4209). On July 12, 1994, the Service published the proposed rule for this
reclassification (59 FR 35584), and the final rule was published on July 12, 1995 (60 FR
6

36000). After reclassification, bald eagles continued to improve to the point where the
Service believed the species no longer meets the definition of a threatened species. On
July 6, 1999, the Service published a proposed rule (64 FR 36454) to delist the bald eagle
in the contiguous 48 States, and requested public comments. The comment period on the
proposal to delist was reopened on February 16, 2006 (71 FR 8238). The final rule on
delisting and the Notice of Availability for the draft monitoring plan were published
simultaneously in the Federal Register (72 FR 37346) on July 9, 2007. That July notice
opened a public comment period on the draft post-delisting monitoring plan. After the
comment period closed on October 9, 2007, the Service reviewed each comment received
and edited the Plan appropriately.
On March 6, 2008, the U.S. District Court for the District of Arizona ordered bald eagles
in the Sonoran Desert of central Arizona to again be protected as threatened under the
ESA, pending further review by the Service. To comply with the court order, on May 1,
2008, the Service published a final rule in the Federal Register (73 FR 23966) listing the
potential Sonoran Desert bald eagle distinct population segment in central Arizona as
threatened. On May 20, 2008, the Service published a Federal Register notice (73 FR
29096) initiating a status review of bald eagles in the Sonoran Desert area of central
Arizona and northwestern Mexico. Under the court order, the Service will issue a 12month petition finding on whether listing the Sonoran Desert bald eagle under the ESA is
warranted, and if listing is warranted, then whether those bald eagles should be listed as
threatened or endangered. For the purposes of this monitoring plan, our approach to
include these eagles into our population estimate will not be altered by its listing status.
In the years since Charles Broley’s discovery of declining eagle numbers in Florida, the
States, Tribes, Service, and our non-governmental partners have engaged in the difficult
and costly task of monitoring nesting bald eagles. In the ensuing 25 years since listing,
many States have monitored nesting bald eagles for their entire State annually. With the
recovery of the bald eagle and removal from the Federal List of Threatened and
Endangered Species, many States have since reduced their monitoring efforts.
Post-Delisting Monitoring Requirement of the Endangered Species Act
Post-delisting monitoring is a requirement of the ESA. Section 4(g)(1) requires the
Service to…
implement a system in cooperation with the States to monitor effectively for not
less than five years the status of all species which have recovered to the point at
which the measures provided pursuant to this Act are not longer necessary.
The Plan described herein exceeds the minimum requirement set forth by the ESA by
effectively monitoring the status of the bald eagle over a 20-year period using five
sampling events.

7

History of Plan Development and Pilot Studies
A draft monitoring plan was provided in the proposed rule to delist bald eagles on July 6,
1999 (64 FR 36454). Slightly more than ten percent of all comments we received on that
proposal were concerned with post-delisting monitoring and the draft monitoring plan.
Since then, the monitoring plan has been revised in such a way that it is responsive to the
comments we received.
In September 2000, a bald eagle monitoring workshop was held at the Service’s Patuxent
Wildlife Research Center in Maryland, attended primarily by State biologists involved
with bald eagle monitoring. As a result of that workshop, the Service in cooperation with
the U.S. Geological Survey’s Biological Resources Division (USGS-BRD), proposed a
pilot study. The pilot study, funded by the USGS for 2004 and 2005, incorporated
methods traditionally used by some States to monitor occupied bald eagle nests while
adding an area sample plot census.
The first pilot study was conducted in cooperation with the Maine Department of Inland
Fisheries and Wildlife in spring 2004. In addition to Maine’s yearly aerial survey of bald
eagle nests (list survey), forty-one 10 kilometer (km) x 10 km area plots were surveyed
from the air (area survey) using a dual observer method. Estimates from the area survey,
from Maine’s list of bald eagle occupied nests, and a combination of those data were
compared and analyzed. Those results were presented at a second workshop held at the
Patuxent Wildlife Research Center in October 2004. The purpose of this workshop was
to review results from the first pilot study and to discuss approaches and possible changes
for a broader pilot study to be conducted in winter/spring 2005. Biologists from State
natural resource agencies were invited to this workshop, but emphasis was placed on
representatives from the States proposed for pilot studies in 2005: Florida, Minnesota,
and Washington.
As a result of that workshop, a second pilot study was implemented in three States
(Florida, Minnesota, and Washington) during the 2005 nesting season. The 2004 and the
2005 pilot studies detected 18 to 40 percent more occupied nests with greater precision
using the dual frame approach than calculated for either area or list frame sampling alone
(Appendix 1, p.22). Thus, the dual frame approach to sampling was selected as superior
for the bald eagle post-delisting monitoring plan.
To implement this monitoring plan, we propose to cooperate with and provide technical
assistance to our State, Tribal, Federal, and non-governmental partners in all aspects of
planning and implementing a dual frame approach to bald eagle monitoring. This plan
does not monitor causal factors such as habitat modification or disturbance as defined
under the Bald and Golden Eagle Protection Act. For additional information on
protections for bald eagles under the Bald and Golden Eagle Protection Act (BGEPA),
please refer to the Service's National Bald Eagle Management Guidelines (72 FR 31156)
and regulatory definition of the term "disturb" (72 FR 31132) that were published in the
Federal Register on June 5, 2007. Existing take authorizations for bald eagles issued

8

under the ESA became covered under the BEGPA via a final rule published in the
Federal Register on May 20, 2008 (73 FR 29075).
Additionally, this Plan does not replace specific plans to monitor bald eagles more
regularly or in a different manner for specific management purposes. We encourage
partners with existing plans that meet or exceed this Plan's monitoring standards to work
with us to continue using their own monitoring and conservation efforts, especially where
continuation of those plans will ensure consistency and comparability with existing data
sets.
Purpose and Goal
The purpose of post-delisting monitoring is to determine if at any time the population of
bald eagles in the contiguous 48 States warrants expanded monitoring, additional
research, and/or resumption of Federal protection under the ESA. The population of bald
eagles in the contiguous 48 States will be estimated by monitoring changes in the number
of occupied bald eagle nests in the contiguous 48 States. The goal of the bald eagle postdelisting monitoring plan is to detect a 25 percent or greater change in occupied bald
eagle nests over any period, measured at five-year intervals based on an 80 percent
chance of detecting such a change. We believe this is a goal that both ensures recovery
and is cost-effective. If a 25 percent decline is measured, it means a reduction to a level
still recognized as recovered (from 9,789 occupied nests in 2007 to 7,342 occupied nests,
a 25 percent reduction) based on a population estimate of 6,471 when the initial proposal
to delist was published in 2000, and assuming no further population increase preceding
the decline. If such declines are detected, the Service in conjunction with the States will
investigate causes of those declines. At the end of the 20-year monitoring program, we
will coordinate with States and our other partners to conduct a final review and provide
recommendations to insure a properly managed population of the recovered bald eagle.
Implementation
Bald eagle monitoring will require a well coordinated national effort, involving the
States, Tribes, Federal agencies, and other cooperators. The following describes the roles
and responsibilities of the parties involved in bald eagle monitoring under this Plan.
Service Bald Eagle Monitoring Team
The Service’s national bald eagle monitoring team (Team) comprised of a national
coordinator, regional coordinators from each of the Service’s seven Regions, and a
biometrician (a biological statistician) has been formed to develop and implement the
Plan. The Midwest Region of the Service is the lead Region for this effort (Appendix 4).
This Team will work closely with the States and all other interested parties, particularly
during each 5th-year monitoring effort, to ensure that Plan implementation is wellcoordinated and efficient. States will be directly involved in collecting, reviewing, and
analyzing survey data, as well as identifying significant issues and developing
recommendations to address these issues.
9

The role of the Service’s national coordinator is to coordinate within the Service and with
States, Tribes, other Federal agencies, and non-governmental organizations. Together
with the Team members, the coordinator recommended draft and final Plans to the
Service’s Directorate. The Team will communicate with State resource agencies and
cooperators; coordinate the planning, implementation, and analysis of the surveys and
summarize the monitoring results in cooperation with States and other cooperators; and
prepare interim and final reports on the monitoring results. The team is tasked with
making recommendations based on survey results, seeking partnerships to implement the
Plan, advocating for resources to carry out the monitoring program, and developing
partnerships for any needed studies.
The role of the Service biometrician is to develop and maintain a national database on the
States’ known bald eagle nest list data (spatial and non-spatial); design the surveys based
on State boundaries and Bird Conservation Regions as a means of stratifying high, low,
and trace nesting density; coordinate and maintain a national database of the survey data
from the various States; and conduct the data analysis, interpretation, and summary for
the national surveys.
The Service will fund the area frame surveys for the initial baseline survey in 2009,
including the use of aircraft and pilots from the Service’s Migratory Birds program to
complete the surveys. We will continue to work with the States, Tribes, and our other
partners to secure funding for future surveys.
Coordination with States
Section 4(g)(1) of the ESA states that “The Secretary shall implement a system in
cooperation with the States (emphasis added) to monitor effectively for not less than five
years the status of all species which have recovered to the point at which the measures
provided pursuant to this Act are no longer necessary…” The Service worked with the
States to develop this Plan.
As described in the Background, this Plan is the product of comments from the initial
proposal to delist the bald eagle, two workshops, and two seasons of pilot studies
involving four States. This Plan was framed based on the results of those efforts. Early
in the Plan development process, Service Team members contacted the States within their
respective regions to determine a coordinator for each State with whom we could
coordinate our efforts. States were asked to summarize their bald eagle monitoring
protocol and to provide their most recent survey data. The Service Team also solicited
suggestions regarding Plan content, methods, and format from the State biologists. Once
a draft plan was developed, the Non-game Technical Committees of the Pacific, Central,
Mississippi, and Atlantic Flyway Councils were asked to provide peer review. The Nongame Technical Committees of the various flyway councils are composed primarily of
State biologists. Service Team members presented updates and answered questions on
the draft Bald Eagle Post-delisting Monitoring Plan at the flyway councils’ biannual
10

meeting in March 2007. States were formally requested to provide review and comments
during the public review period. In addition to the technical committee peer review
comments from the flyway councils, 12 States provided comments on the draft Plan.
The Team’s national and regional coordinators will work closely with the State
coordinators and other cooperators to provide technical assistance on implementing the
Plan and submitting the data after each monitoring period. In coordination with the
States, the Service will propose adjustments to the sampling design, if necessary. This
effort will also require the Service biometrician to work closely with the State
coordinators and the Team to select sample areas and maximize sampling efficiency. For
those years that fall between the monitoring years outlined in this Plan, the Service will
have an ongoing request for occupied nest monitoring data that can be entered into a
secure, web-based database (Appendix 3). Data on productivity and any information
regarding major habitat changes, contaminants, or mortality events collected by States or
other partners may also be submitted to the Team at any time.
For the baseline monitoring/first year of sampling in 2009, the States will be asked to:
 Update their nest lists and enter this data on the secure web site.
 Provide experienced bald eagle observers for the aerial part of the survey and
participate in training;
 Provide input on refining stratum boundaries, definitions of bald eagle habitat
specific to the State, and identification of breeding season dates; and
 Reconcile list nests found in the area samples and enter that data into the
secure web site.
Coordination with Tribes
The Service values the cooperation and participation of the Tribes in our bald eagle
recovery efforts, including post-delisting monitoring. Bald eagles are important to the
Tribes for cultural, religious, and ceremonial purposes, giving the Tribes a vested interest
in the continued conservation of the species. While the geographic boundaries of the
States will be the basis of the national monitoring effort, information on nest occupancy
on tribal lands has been and will continue to be important information. We encourage the
Tribes to maintain updated lists of nests on their lands and share that information with the
States and the Service. Prior to initiation of the nest surveys, the Service will coordinate
with Tribes regarding any randomly selected area plots that may fall on tribal lands.
Coordination with Other Partners
Bald eagle monitoring in most States has been carried out by a combination of Federal
agencies, Tribes, private organizations, and individuals. While the Service, in
cooperation with the States, is responsible for post-delisting monitoring of bald eagles,
continued participation and cooperation of all our partners is important for monitoring
success. We anticipate that the combined efforts of all of our partners working together
will provide the necessary resources to implement this Plan.

11

Other Monitoring Efforts
While the dual frame methodology is the formal strategy being recommended to monitor
the breeding population of bald eagles, there are other local and national efforts that will
continue to assist in evaluating the status of the bald eagle population. Many States have
annual flight surveys to count nesting bald eagles. Some States will conduct a second
flight to assess productivity. Wintering bald eagle surveys have been conducted since
1979 and the National Bald Eagle Winter Survey has become institutionalized in many
States across the country (Steenhof et al. 2002). This effort provides an opportunity to
identify and manage for important eagle wintering areas, and provides information on
wintering distribution, abundance, and age class. Biologists at important bird migration
points have been tracking migrating bald eagles for many years. Service National
Wildlife Refuges and other land managers track productivity as well as nesting and
migrating bald eagles. Continuing these efforts post-delisting and providing that
information to the Team members will improve our ability to evaluate the status of the
bald eagle.
Methods
Sampling Design: The Dual Frame Method
These methods are described in more detail in Appendix 1, Contiguous 48 States Bald
Eagle Breeding Pair Survey Design. A generalized description of the methodology
follows.
The Service will work with States, Tribes, Federal agencies, and other partners to conduct
a survey of occupied bald eagle nests that incorporates information from two sampling
strategies. The first strategy uses the list of all currently known occupied bald eagle nest
locations. This is called the list frame and is the method employed for the last 25 to 30
years by the States. The second strategy uses aerial surveys to count occupied bald eagle
nest locations from randomly selected plots. This is called the area frame (Haines and
Pollock 1998). A frame is a set of all possible elements from which we can sample. Data
gathered in these two sampling frames allows aggregation of numbers from occupied
nests found in the list and occupied nests found in the area frames, resulting in an
estimate of the total number of occupied nests that is more accurate than the use of either
frame alone based on results from the pilot studies (Appendix 1).
List Frame
The list frame for this Plan is a summary of all currently known occupied bald eagle nests
for the contiguous 48 States. This includes the current State occupied nest lists and will
include all future updates to those lists. The list will be updated by States and partners
conducting bald eagle nesting surveys in the manner that those surveys were conducted
while the bald eagle was still listed under the ESA. Each State will enter updates to the
list into a secure, web-based database. The updates will include confirmation of activity
12

at known nest sites and the addition of new occupied nests found either through the nest
surveys, or by nest locations found during the area frame survey that were previously
unknown. Historically, States conduct bald eagle nest surveys at varying levels of effort
and at varying time intervals. Thus, the list frame will be composed of data collected in
different years and of varying intensity. Should confidence in the list frame erode over
time due to reduced survey effort, the list may need to be sampled concurrent with the
area frame to provide an estimate of nests that are in the list frame.
Area Frame
The area frame is composed of randomly selected plots within which bald eagle habitat is
carefully surveyed for all occupied bald eagle nests. The plots are 10 km x 10 km each
(or equivalent area), selected from a grid matrix that has been overlaid on a habitat
stratum. The strata are used to increase sampling efficiency, and are based on State
boundaries and Bird Conservation Regions (BCRs). The BCRs group regions with
similar environmental features (Sauer et al. 2003). To accommodate State-specific needs,
we divided BCRs into States, and use these State-BCR units as our initial strata
(Appendix 1, Figure 1). The State-BCR strata are further divided into High, Low, and
Trace categories based on bald eagle habitat quality and nesting density (Appendix 1,
Figure 2). Appendix 2 outlines the protocol for collecting area frame data for this Plan.
Observations of nest occupancy collected in this area-based sample will contain both new
nests and nests that also occur in the list frame.
Dual Frame
The dual frame method of analysis combines sample information from both the list frame
and the area frame to arrive at a more precise estimate of occupied nest density across the
entire study area (Haines and Pollock 1998). To conduct the analysis, occupied nests
identified in the area frame sampling are separated into the two categories: the overlap
(nests in the plots that also occur in the list) and nonoverlap (nests that are newly found in
the plots). The nonoverlap nests are identified, and are used to estimate the total number
of nests not in the list. The sum of the estimates from the area frame and the list frame
are used to determine a total number of occupied eagle nests within the study area.
Monitoring Study Area
The goal of this Plan is to detect changes in the number of occupied bald eagle nests in
the contiguous 48 States. The study area is the entire contiguous 48 States because one
population was identified as the listed entity at the time of removal from the Federal List
of Threatened and Endangered Species. Sampling at this scale will be more costeffective than sampling at a regional or smaller scale.
A GIS-based map has been developed depicting bald eagle nesting density in the
contiguous 48 States (see Appendix 1, Figure 3), which is the study area. The map is
based on the most recent nesting data collected at different times in different ways from
each State and compiled into one list for the contiguous 48 States (list frame).
13

Frequency and Duration of Sampling
If we used a narrow interpretation of the ESA’s requirement for monitoring a minimum
of five years after delisting, this would result in monitoring for one breeding cycle. The
bald eagle’s distinctive white head and tail are not fully visible until four to five years of
age when eagles are considered to have reached sexual maturity. In some areas of the
country, birds in sub-adult plumage have been known to form pair bonds, defend
territories, and construct nests. In other regions with dense populations and competition
for nest sites, eagles may not breed until six or seven years old (Buehler 2000). Thus,
though the exact breeding cycle varies, the majority of bald eagles reach maturity at 5.5
years of age (Buehler 2000). In order to assess several generations of bald eagles after
delisting, this Plan recommends monitoring bald eagle nesting populations at five-year
intervals (which would follow the development cycle to maturity for one generation), for
four generations or a total of 20 years.
Many States monitor bald eagle nests on an annual basis because the surveys provide
valuable resource data. Comments we have received from a number of States, however,
indicates that their future bald eagle monitoring will be greatly reduced due to its
recovery and the need to allocate funding to other areas. Thus, five-year survey intervals
will provide more data for states where surveys are not otherwise planned. It may also
provide a cost savings for other States if they can use these data at five-year intervals to
satisfy their needs.
Sampling the List Frame
The Plan recommends States maintain their lists of known nest sites. It is expected that
this will be done with varying degrees of effort. Ideally, the number of occupied nests in
the list would be determined through a periodic census of all nests on the list. In States
where lists are not maintained, known nest sites can be sampled as part of the area
survey.
Sampling the Area Frame
The area frame must be sampled to obtain unbiased estimates of the total number of
occupied nests. To do this, the Service’s biometrician will determine area frame plot
numbers for each stratum in coordination with the States. Some plots initially selected
for the area sample may have characteristics that make them unreasonable to sample. For
example, the plot may be too far from an airport to be cost effective or allow for safe
reserves of fuel. Plots near urban areas could contain too many obstructions such as
transmission lines or cell towers to permit safe survey conditions. Therefore, we will
initially select an additional 10 percent of plots to sample to ensure alternate plots are
available if different plots are necessary. For example, if 10 plots will be sufficient to
meet the stated goals for precision and accuracy, we will plan to sample 11 plots,
assuming that logistical or safety issues may preclude sampling at one plot. Should all

14

plots be feasible, sampling will cease when the minimum number of plots first selected
has been sampled.
Protocols
The area frames will be sampled using protocols consistent with those developed during
the pilot studies. A detailed discussion of standard operating procedures is included as
Appendix 2. A double observer protocol will be implemented for the area frame
sampling whenever possible to estimate the proportion of nests missed during that area
sampling event (Nichols et al. 2000). Protocols for double observer sampling are also
presented in Appendix 2. Aircraft observers should be familiar with the terrain and
nesting habitats of eagles in their area. The front seat observer should be the primary
data recorder. All occupied nests and the number of visible young should be recorded.
The aircraft should be flown at 200 to 700 feet above ground level (agl) at about 100 mph
or 87 knots. Only the parts of plots that are composed of potential eagle habitat will be
flown. Flight paths will be defined on maps prior to conducting the surveys. We note
that protocols for assessing occupancy status of a nest may differ regionally, and timing
of surveys will also differ regionally. Consequently, protocols for sampling must be
reviewed regionally for consistency as part of survey implementation.
Reproductive Terminology
Standard terminology for describing the status of bald eagle nests and territories is
essential, especially if a meaningful comparison is to be made of the data collected by
different workers over many years and throughout the nation. The following definitions
are derived from Postapulsky (1974), Fraser (1978), Steenhof and Kochert (1982), and
Steenhof (1987). They are entirely separate from, and should not be substituted for,
definitions in other bald eagle documents developed by the Service.
Active nest (breeding): A nest where eggs have been laid. Activity patterns are
diagnostic of breeding eagles (or those with an “active” nest). This category excludes
non-nesting territorial pairs or eagles that may go through the early motions of nest
building and mating, but without laying eggs. From egg-laying to hatching, incubation
typically lasts 35 days (Stalmaster 1987).
Alternate nest: One of several nest structures within a breeding area of one pair of eagles.
Alternate nests may be found on adjacent trees, snags, man-made towers, or on the same
or adjacent cliffs. Depending on the size of the breeding territory, some alternate nests
can be up to a few miles away.
Bald eagle nesting habitat: For this study, bald eagle nesting habitat will need to be
defined for each region to assure sampling efficiency. In general, bald eagle nesting
habitat will include a description of typical nesting structure for the region and proximity
to a food source, usually a larger sized water body. The pilot States generally defined
bald eagle nesting habitat as supercanopy or sturdy-structured trees within one mile of
water bodies greater than 35 acres and rivers greater than 330 feet in width and all coastal
15

waters with suitable nesting substructures nearby. However, local or regional habitat
parameters will have to be modified with input from local biologists to fit local
conditions, including in much of the arid Southwest.
Breeding area (nesting/breeding territories): An area that contains or that was
previously known to contain one or more nests within the territorial range of a mated pair
of eagles.
Nest: A structure, composed largely of sticks, built by bald eagles for breeding.
Unoccupied breeding area/territory/nest: A nest or group of alternate nests at which
none of the activity patterns diagnostic of an occupied nest were observed in a given
breeding season. Breeding areas must be previously determined to be occupied before
they can be recognized and classified as unoccupied.
Occupied nest: Any nest where at least one of the following activity patterns was
observed during the breeding season:
 a recently repaired nest with fresh sticks or fresh boughs on top;
 one adult sitting low in the nest, apparently incubating;
 one or two adults present on or near the nest;
 one adult and one bird in immature plumage at or near a nest, if mating behavior
(display flights, nest repair, coition) was observed;
 eggs were laid (detection of eggs or eggshell fragments);
 any field sign that indicate eggs were laid or nestlings hatched; or
 young were raised
The total number of occupied nests is the index of abundance for this survey. Efforts will
be made to survey during the period when adults are most likely to be incubating and the
identification of occupancy is most often confirmed with the presence of one or more
adults. Observers will make note of nests that are found but where bald eagles are not
present. Distinctions between repaired and not repaired empty nests will be made in
addition to any uncertainties such as questioning whether the empty nest is that of an
osprey. In case of doubt, the default determination will be negative.
Data Quality
The survey data sets used in the Bald Eagle Post-delisting Monitoring Plan will adhere to
the Service’s data quality standards in naming variables and in choosing values those
variables can take (www.fws.gov/stand). The metadata for our spatial data will be
written according to the Federal Government Data Committee standards
http://www.fgdc.gov/metadata/geospatial-metadata-standards.
Habitat
The Service will not monitor changes in bald eagle habitat directly. However, the Team,
in conjunction with the States and other partners, will accept and review data indicating
16

significant changes in bald eagle habitat in the contiguous 48 States. Should a 25 percent
decline in the bald eagle population occur, the Service will consider habitat data when
determining potential causal factors for the decline.
Contaminants
The Service worked with the USGS-BRD to develop a searchable database/library
dedicated to contaminants investigations of bald eagle, osprey (Pandion haliaetus), and
peregrine falcons (Falco peregrinus). The objective was to create a readily available
source of information to consider should the bald eagle (or peregrine) population decline.
Osprey contaminants data are relevant to bald eagles as they occupy a similar niche.
The USGS identified, acquired, and assigned keywords for published and unpublished
literature about contaminants in bald eagles, osprey, and peregrine falcons. The USGS’s
Richard R. Olendorff Memorial Library in Boise, Idaho currently maintains several
hundred references relevant to this topic as part of the Raptor Information System. New
and existing references were assigned contaminant-related keywords, established by the
Service’s contaminants biologists. These keywords are listed on the contaminants
database page at the following website: http://ris.wr.usgs.gov/Contaminants.asp .
Citations for all new references were incorporated into the existing Raptor Information
System database and are served from the existing website (http://ris.wr.usgs.gov/). Many
of the citations include links to the full text of articles that are being served on the World
Wide Web. We will also seek resources from the National Biological Information
Infrastructure (NBII) to serve the PDF files and abstracts as well as the citations from a
separate web site. Keeping this database relevant will require periodic updating. A
Service environmental contaminants biologist will coordinate this effort (see Appendix 4,
Region 5 contacts).
References from the contaminants database can be retrieved by entering the first keyword
in the keyword search box using the autocomplete function. Enter additional keywords
from the keyword popup list, then type in FWSEC as the final keyword in the keyword
box. Not entering FWSEC will bring up references about other species as well as
abstracts and popular articles about the subject species.
By creating this database, biologists in the position of recommending regulatory actions
based on post-delisting monitoring trends will have a clear overview of the most recent
findings of contaminant effects on these three species. Deleterious effects resulting from
contaminant exposure was a major reason the bald eagle and peregrine falcon were listed
under the ESA. Data demonstrating reduction in contaminant exposure supported the
proposal to delist the bald eagle and peregrine falcon. Should additional studies be
needed during post-delisting monitoring, the database will clarify what has been studied
and what has not.

17

Ongoing and Potential Sources of Mortality
In species with a long life span and a relatively low reproductive rate like the bald eagle,
adult mortality can be a very important factor in determining the stability of a population
(Stalmaster 1987). Bald eagles (and many other raptors) are killed as a result of trauma
from collisions with power lines, vehicles, and other obstacles; electrocution; disease;
poisoning; shooting; and other factors.
As part of the Plan, bald eagle mortality will be tracked to alert the Team to new and
potentially significant sources of mortality. We will request information on bald eagle
deaths from sources that are known to encounter dead eagles most frequently: State
wildlife conservation agencies; Service law enforcement officers; wildlife rehabilitators;
the National Wildlife Health Center in Madison, Wisconsin; and the National Fish and
Wildlife Forensic Laboratory in Ashland, Oregon. If an unusually large number of
mortalities occur, the Service and its partners will consider the information in regards to
specific causes and/or locations for investigation of patterns and possible corrective
action, if necessary.
Response Trigger
The Service has established the following Plan goal as the response trigger for additional
investigation:
A 25 percent or greater decline in occupied bald eagle nests between any two periods
measured at 5-year intervals detected with a power of 80 percent and an error rate of
10 percent.
If such declines are detected, the Service’s Bald Eagle Monitoring Team will coordinate
with the States to investigate causes of these declines, including consideration of natural
population cycles, weather, productivity, contaminants, mortality factors, habitat changes,
data from other monitoring efforts, or any other significant evidence. At the end of the
20-year monitoring program, we will coordinate with States and our partners to conduct a
final review and provide recommendations to insure a properly managed bald eagle
population. Any relisting decision by the Service will be made by evaluating the status of
the bald eagle relative to the ESA’s five listing factors (ESA § 4(a)(1)).
Reporting
The Service will issue a report detailing the results of the first breeding population
survey, which will serve as our baseline for future comparison. This report will be
available to the public in printed form and on the internet at
http://www.fws.gov/migratorybirds/BaldEagle.htm within one year of survey completion.
The report will include a description of the geographic areas surveyed, the survey

18

protocol, and an estimate of the breeding population of bald eagles in the contiguous 48
States.
Every five years, the Service intends to issue a report following completion of the
updated continental breeding population sampling. This report will contain information
similar to the baseline report, including an updated breeding population estimate and
comparison with previous data, and will be available to the public. Reports will also
suggest ways to improve sampling protocols or other aspects of the Plan design, if
necessary, and will provide updates to the Plan.
Each report will include comments on the need for any investigative action and the
relationship between the survey results and the response trigger. This Plan is designed to
detect substantial declines in occupied nests with reasonable certainty and precision. If
the response trigger is met or exceeded, the Service will consult with the States and other
partners and make recommendations for future actions. If necessary, an evaluation of the
threats to bald eagles will be made using the five factors required under the ESA to list a
species on the Federal List of Threatened and Endangered Species.
The Service intends to also provide a summary report on bald eagle mortality every five
years. Bald eagle mortality reports will describe the number and causes of reported eagle
deaths during the five-year period, cumulative deaths reported since the completion of
baseline monitoring, and the geographic distribution of the reported deaths. Emphasis
will be on concentrations of multiple deaths. In this way, specific causes and/or locations
of high eagle mortality may be identified for investigation of patterns and possible
corrective action, if necessary.
At the end of the 20-year monitoring period, the Service will review all available
information to determine if continuation of monitoring is appropriate. The decision to
continue or end the monitoring program will be explained in the final monitoring report,
which will be made available to the public as described above.

19

References
Buehler, D.A. 2000. Bald eagle Haliaeetus leucocephalus. No. 506. In A. Poole and F.
Gill [eds.]. The Birds of North America. The Birds of North America,
Inc., Philadelphia, Pennsylvania. 40 pp.
Carson, R.L. 1962. Silent spring. Houghton Mifflin Co., New York, New York. 368
pp.
Fraser, J.D. 1978. Bald eagle reproductive surveys: accuracy, precision, and timing. M.S.
Thesis, Univ. Minnesota. St. Paul, Minnesota. 82 pp.
Grier, J.W. and J. E. Guinn. 2003. Bald eagle habitats and responses to human
disturbance in Minnesota: Final report. Unpublished report. Final report to the
Minnesota Department of Natural Resources, Natural Heritage and Nongame Wildlife
Program, Division of Ecological Sciences. St. Paul, Minnesota. 44 pp.
http://files.dnr.state.mn.us/eco/nongame/projects/consgrant_reports/2003/2003_grier_
guinn.pdf.
Haines, D.E. and K.H. Pollock. 1998. Estimating the number of active and successful
bald eagle nests: an application of the dual frame method. Environmental and
Ecological Statistics 5, 245-256.
Nichols, J. D. J. E. Hines, J. R. Sauer, F. W. Fallon, J. E. Fallon, and H. J. Heglund.
2000. A double-observer approach for estimating detection probability and
abundance from point counts. Auk 117(2):393–408.
Postapulsky, S. 1974. Raptor reproductive success: some problems with methods,
criteria, and terminology. Pages 21-31 in F.N. Hammerstrom, Jr., B.E. Harrell,
and R.R. Olendorff, eds. Management of raptors. Raptor Res. Found., Vermillion,
S.D.
Sauer J.R., J.E. Fallon, and R. Johnson. 2003. Use of North American Breeding Bird
Survey data to estimate population change for bird conservation regions. Journal
of Wildlife Management. 67:372–389.
Stalmaster, M.V. 1987. The bald eagle. Universe Books. New York, New York.
Steenhof, K.L. and M.N Kochert. 1982. An evaluation of methods used to assess raptor
nesting success. Journal of Wildlife Management. 46: 885-893
Steenhof, K. L. 1987. Assessing raptor reproductive success and productivity. Pages
157-170 in B.G. Pendleton, B.A. Milsap, K.W. Cline, D.M. Bird eds. Raptor
Techniques Manual. National Wildlife Federation, Institute for Wildlife Research,
Scientific and Technical Series. No. 10, Washington, D.C.

20

Steenhof, K., L. Bond, K. Bates, and L. Leppert. 2002. Trends in Mid-Winter Counts of
Bald Eagles Across the Contiguous United States, 1986-2000. Bird Populations
6:21-32.

21

Appendix 1
Contiguous 48 States Bald Eagle Breeding Pair Survey Design
by Mark C. Otto and John R. Sauer
Introduction
Bald eagle biologists have focused on site-specific monitoring of eagle nest sites both to
monitor population change and to catalog areas for management of the species (Bartish
1994). These site-specific efforts form a critical resource for eagle monitoring, and the
Plan seeks to maintain these lists of eagle nest sites and to assess their occupancy status.
However, particularly in areas where eagles are increasing, sampling these lists is not an
adequate monitoring program, as many possibly occupied nests exist that are not in the
lists. Consequently, while the delisting criteria are based on the existing lists, and they
form a critical component of the monitoring program, additional sampling is
recommended in conjunction with the list sampling to obtain unbiased estimates of
occupied eagle nests. This dual-frame approach provides estimates of occupied eagle
nest abundances that are not limited to the list of nests, and provides a flexible strategy
for estimating abundance that is not limited to regions where the lists are maintained.
Pilot studies were conducted to assess the feasibility of the dual-frame sampling design.
Generally, eagle pairs choose one nest of a possible number of nests in their territories to
occupy and use for brood-rearing in a given season. Many States collect information on
territories along with information on the nests. Because information on territories is not
collected consistently across all States and because we cannot assign territories to new
nests found on the area survey, we use occupied nests as our sampling unit. We assume
that determining the number of occupied nests is equivalent to determining number of
territories. This imposes a dependence among local nests in occupancy, as only a single
nest in a territory will be occupied. Practically, this lack of independence does not affect
the estimation when we sample nests instead of territories.
Geographic information about suitable bald eagle breeding habitat and existing nests can
be used to develop an efficient sample design for the species. We recommend that the
sample design: (1) stratify by regions (physiographic strata within States) to permit more
intensive samples in regions with a higher density of bald eagle nests; (2) use a dualframe design to sample known nest sites but also conduct an area-based sample to
estimate the total number of nests; (3) use habitat information within regions to further
focus sampling efforts; and (4) account for nests not observed by estimating detection
probabilities using double observer survey methods (Thompson 2004).
Dual-frame sampling uses a sample from the list of known nests, in combination with
additional sampling, to estimate the total number of occupied nests. For the additional
sampling, the study area is divided into plots, and a sample of these plots is randomly
selected and censused for occupied eagle nests. This set of plots is known as a sample
frame, and the selected plots are a random sample from this frame. In accordance with

22

statistical sampling theory (Lohr 1999), results from the sample plots allow us to
calculate an estimate for the entire study area. The additional information from the
sample from the list of nests, however, can be incorporated into the estimation. If nests
that are on the list but are also known to be in the sample plots are identified and removed
from the sample (a process known as unduplication), the remaining occupied nests
observed in the plots can be used to estimate the total number of occupied nests not in the
list. The occupied nests in the list and the occupied nests seen only on the plots can be
added to provide an estimate of the total number of occupied nests.
Additional effort is required during the area surveys to ensure that we account for
occupied nests that are in the sampled area but are not seen during sampling. Use of a
protocol involving two independent observers (double-observer sampling, Nichols et al.
2000) permits estimation of the number of nests that are not observed. For more details
on the double-observer survey method, refer to Appendix 2.
To assess the feasibility of using a dual-frame sample design for bald eagle post-delisting
monitoring, the Biological Resources Division of the U.S. Geological Survey (USGS)
and the Service conducted pilot surveys over three years (2004–2006) in five States
(Maine, Florida, Minnesota, Washington, and Missouri). Based on the results of these
pilot studies, the overall sampling approach described here was developed. The pilot data
were also used to predict the effort needed in a national delisting monitoring program.
In this appendix, we discuss the pilot study results, the national design, and the effects of
list coverage of the nest list on the cost-variance functions. The proposed design expands
on the approach used in the pilot surveys to a national monitoring program for bald
eagles. The discussion includes: (1) stratification and how it can be simplified; (2) list
frame (all known nests from State nest lists in the contiguous 48 States) and how they can
be clustered within plots to reduce flying time; (3) area frame (all 10 km x 10 km plots
covering the contiguous 48 States); and (4) estimation of detection probability using
double observer techniques with the area survey.
Pilot Studies
Pilot studies were conducted in Maine in 2004, and Florida, Minnesota, and Washington
in 2005. Additionally, biologists in Missouri volunteered to test the methods in 2006.
The pilot studies were designed to test the effectiveness of the Haines and Pollock (1998)
dual-frame design in a variety of geographic areas. The States involved in the pilot
studies differed both in eagle abundance and in the completeness of their list frames,
providing a variety of situations for evaluating the dual-frame approach. The eagle nest
surveys for the pilot studies were collaborations among USGS, Service, and State
biologists experienced in bald eagle surveys. The State biologists were consulted on
design issues, conducted surveys of their list frames, and were observers for the area
frame components of the surveys.
For the list frame, State biologists censused or sampled the known nests from the ground,
helicopter, or plane as had been done in previous surveys. The number of occupied nests

23

was determined from the product of the number of nests in the list frame and the
proportion of occupied nests estimated during the survey. If the list was censused (all
nests checked), the variance was zero; if the list was sampled, the variances were
determined according to the methods described by Thompson (1992, p. 35).
To implement the dual-frame protocol, an aerial survey of 10 km x 10 km plots was
conducted over the same strata as the list frame survey. To select plots: (1) each State
was divided into a grid of 10 km x 10 km plots; (2) the plots were assigned to strata based
on the predominant habitat type in each plot; (3) nest densities and their standard
deviations for each stratum were obtained from the previous list frame; and (4) optimal
allocations for the area frame were determined according to survey sample design theory
(Lohr 1999, p. 104). Consequently, higher density, more variable, and less costly strata
were sampled more intensively. Random samples of plots were drawn in proportion to
the optimal allocation in (4).
All eagle habitat (as defined by the State biologists) in each sample plot was examined
during the aerial survey. A double-observer protocol was developed and implemented to
estimate the number of nests missed during the survey. Observers in the front and rear
right side seats of the aircraft made independent observations of bald eagles and eagle
nests. The observations were reconciled immediately after the aircraft had passed the
nest. The “capture history” (i.e. seen-seen, seen-not seen, etc.) of each observation was
recorded (see Appendix 2, Standard Operating Procedures). Detection probabilities for
individual observers and both observers together were obtained using the software
program DOBSERV (Nichols et al. 2000, http://www.mbr-pwrc.usgs.gov/software.html).
Including observer detection in the sample allocation specifies that lower detection
probability and more variable detection probability strata should be sampled more
intensively.
The dual-frame estimate was obtained from combining the list frame with the area frame
surveys by unduplication. Unduplication removes all the nest observations that were on
both the list and the area frames and leaves only the number of new nests. After the
unduplication, list and area estimates become independent from each other because they
have no common observations. The list and area totals can then be added to estimate the
total number of nests. List and area variances can also be summed to estimate a total
variance.
Pilot Study Results
Results indicate that the dual-frame approach with detectability estimation is useful in
providing both (1) an estimate of the number of occupied nests that are not included in
the list of nests and (2) an estimate of the detection rate of nests when sampling plots.
The variances of the dual-frame estimates were smaller than both the mean squared error
of the list total and the sample variance of the area survey. Dual-frame estimates of the
total number of nests were 421 in Maine in 2004, 1,481 in Florida (using the 2003 nest
list), 1,327 in Minnesota and 1,939 in Washington in 2005, and 123 for Missouri in 2006.

24

Detection rates varied among States and due to differences in survey techniques, but
generally were > 95 percent for both observers combined.
The dual-frame sampling design can be applied throughout the contiguous United States
in a manner similar to that conducted at the State level during the pilot studies. Using the
dual-frame method to estimate population size throughout the contiguous U.S. will
require close coordination with the States. Cooperation is needed to continually update
the recently compiled nest list. Assistance from State biologists will be needed to ensure
that bald eagle habitat is properly defined for each area and to confirm or modify the
stratification. Experienced bald eagle observers will be needed to conduct the surveys. If
experienced State pilots will be used for surveying, coordination and effort will be
needed to set up the recording hardware and software and implement the double observer
protocol. Finally, biologists familiar with the nest lists will be especially helpful in
surveying nests on the list and in reconciling the observations in the area survey with
those in the list.
Goals of Sampling
The goal of post-delisting monitoring is to estimate changes in the number of occupied
bald eagle nests in the contiguous 48 States. Because bald eagle populations have been
increasing over most of their range, changes in rate of increase may be an important first
indicator, and can be estimated by comparing the result from any two sample periods.
This survey methodology is designed to detect a 25 percent relative population decline in
the total number of occupied nests in the contiguous 48 States between any two sampling
periods 80 percent of the time with a 10 percent chance of getting a significant decline
just by chance. Statistical methods and pilot data described here provide a rigorous
framework for predicting the number of areas that need to be sampled to meet this goal.
Stratification
Bald eagle distribution and nesting density varies widely both within and among States.
The Plan uses physiographic regions developed for bird conservation (Bird Conservation
Regions, or BCRs; U.S. North American Bird Conservation Initiative Monitoring
Subcommittee 2007) as strata for developing eagle survey plots throughout the
contiguous United States. The BCRs group regions with similar habitats and other
environmental features, and allow for a more consistent regional grouping of habitats
than State boundaries.
To accommodate State-specific needs, we divided BCRs into States (e.g., we considered
DE, MD Coastal Plain as a separate stratum from VA Coastal Plain), and used these
State-BCR units as initial strata (Figure 1). Washington, Oregon, Florida and Maine have
tailored their State strata based on their biologists’ knowledge. We then aggregated these
State-BCR units to larger strata using a clustering procedure that assessed similarity in
the State-BCR units from information on eagle abundance (Table 1). For each StateBCR, the eagle nest list data were overlain on a 10 km x 10 km grid (corresponding to the
proposed plot size) and used to estimate mean and variance of eagle nests in Table 2.

25

After each combining, the overall standard error could be assessed, and large increases in
overall standard error indicate a lack of value in the grouping. In this analysis, we chose
to stop combining when 18 State-BCRs remained as separate strata, and we combined the
aggregated units into a low-abundance stratum. We further separated the 18 contiguous
lowest density strata into a “trace” stratum, a region containing only 14 nests in the
stratum (Figure 2). We view these aggregate strata as reasonable regions for
implementing an initial survey design. Aggregating the similar strata will improve the
estimation by avoiding imprecisely estimating numbers of nests in many small strata.
Variation in plot densities in the aggregated strata is still higher than we would expect for
count data. This suggests that further stratification would be useful. State biologists,
particularly those in States containing the 18 primary strata are encouraged to refine these
strata using their knowledge of eagle populations and habitat use by the species. Also, if
amount of habitat or shoreline are highly correlated with density of occupied nests, we
should weight the sample selection by the correlated variables. Finally, we can obtain
sub-regional or other small area occupied nest estimates (e.g., estimates in portions of
aggregated strata) by post-stratification (Lohr 1999, p. 114) or small-area estimation
(Lohr 1999, p. 397).
List Frame
State biologists have provided lists of nests with their locations, last known status, and
year of the observation. The list frame can be sampled or censused to estimate the
number of known nests that are occupied. Sampling from the list frame is efficient
because locations are known and nests can be observed by flying directly from one nest
to the next. Figure 3 illustrates the distribution of nests collated from State lists.
Because sampling the list is a major cost associated with an eagle monitoring program,
we explored an approach to grouping the list nests for sampling. Efficiency can be
gained by grouping the nests from the lists into “clusters,” then sampling the clusters of
nests in conjunction with the area component of the survey. Clustering increases the
variance relative to a random sample, but this should be compensated by the reduced cost
of sampling (Lohr 1999, pp 154).
To implement this approach, we suggest that the list sampling be directly connected to
the area sampling, by treating the area sample units (the plots) as clusters. List frame
nests within a plot can be defined as a cluster. During the area sample, if a plot is
selected for the survey, all list nests in that plot will also be sampled as the cluster.
Sampling the list where we sample the area allows us best to assess the coverage of the
list and correctly assess nests as occupied or unoccupied. For the rest of this appendix,
sampling of the list frame will be done on clusters of list nests within a plot. Plots with
list nests will be the list frame sampling unit.

26

Area Frame
Within strata, we suggest that the sample unit for area sampling be the 10 km x 10 km
plots used in the pilot studies. For development of a sampling frame for the contiguous
48 States, we used ArcGIS with a Lambert Equal-Area Projection to generate 10 km x 10
km plots. The plots were then categorized by strata. Non-US and water portions of plots
on the coasts or Mexican and Canadian borders were removed. Plots that overlapped two
or more strata were assigned to the stratum that had the majority area. Figure 4 is an
example of the plot grid from the pilot study in Missouri. Filled blocks represent the
selected sample plots; the diamonds indicate location of selected sample list nests.
The 10 km x 10 km grid spanning the contiguous United States was overlain on a map of
the list nest locations, associating each nest with a plot, and means and variances of
number of nests per plot were calculated for each of the strata. These data form the
fundamental information for allocation of samples.
Estimation and Detectability
For estimation and survey design, we follow Haines and Pollock’s (1998) methods and
add additional components for estimating detection probability. We estimate the total
number of occupied nests, N Li , by adding the estimated nests in the list, YˆLi , and the
estimated number of occupied nests in the area frame that were not in the list frame,
.
We use the subscript N instead of A because the nests are new. The stratum estimates are
added up to get the national total,
IL

IA

Yˆ   YˆLi   YˆNi
I

i

i

The I L list or I AL area strata can be different. The list strata effectiveness is determined
by the differences in the percent occupied and the area strata effectiveness by the
differences in the density of the new nests. The latter is determined by unknown
variation in the list coverage along with the occupied nest density itself.
The estimate for occupied nests in the list frame is the sample proportion of occupied
nests, y Li , expanded by the number of nests in the stratum, YˆLi  N Li yˆ Li .
The estimate for new occupied nests expands the density of the new nests seen, y Ni , by
the stratum size,
, and the detection probability, ,
N y
YˆNi  Ai Ni
pˆ i
The A stands for the area frame. The higher the density of occupied nests, the larger the
stratum area, and the lower the detection probability, the larger the stratum total.

27

The variances from the list and area frames are independent, so they can be added to get
the variance of the total1,



IL

 

IA

 

Var Yˆ   Var YˆLi   Var YˆNi
I

i

i

If the list is censused, the variance associated with its estimate is 0. If it is sampled, the
stratum variances for the list frame only depend on the number of occupied nests, N Li ,

and the variability of the estimate, S Li2 n Li y Li 1  y Li  / n Li  1 , and the sample size, nLi ,

 

N S2
Var YˆLi  Li Li  Fixed _ Part 
n Li
So the variance decreases as the number of occupied nests and the sample size increases.
Components of the formula not needed for this analysis are indicted by the term “Fixed
Part,” Fixed _ Part   N Li S Li2 . They represent the parts of the variance that are not
affected by the sample size.
Estimation of the stratum area frame variances is complicated by the detection
probability.
2 2
N Ai S Ni2 N A y Ni n p  1Var  pˆ i 
ˆ
Var Y Ni  2

 Fixed _ Part  .
pˆ i n Ai
pˆ i4 n Ai  1

 

The variance decreases as the area ( N Ai ) density of new nests ( y Ni ) the variance of the

density of new nests ( S Ni2 ) and variance of the detection probability Var  pˆ i  decrease
and as the sample size ( n Ai ) and the detection probability itself ( pˆ i ) increases. Again,
the “Fixed_Part”
2
2
 ˆ 2
Fixed _ Part   N Ai S2 Ni  N A 1  p2 i y Ni
pˆ i
pˆ i
is a part of the variance that is not affected by the sample size. Note that the variance of
the new nest density will drop as fewer new nests are found, i.e., if the list is more
complete, the variability of the new nests drops. The parts of the variance that change
with sample size are used in estimating the required sample size.
Survey Design
To design a survey, we calculate the variances we require given our sampling goals and
look at alternative designs to attain those goals by comparing the sample sizes and
resulting costs needed. The effect size (in terms of relative change), power, significance
level, and variability of the data determine the variance needed from the survey. We use
1

We use variances when we derive the sample sizes we need because the variances of
independent parts of the survey can be added. The standard errors are the square roots of
the variance. We use them in the tables and graphs because they are more understandable
in that they are used to construct confidence intervals. They are on the same scale as the
data.
28

these components to define the least expensive allocation of samples that meets our
variance requirements. We would like to minimize both the costs and the variance,
min  VarYˆ n Li , n Ai     Cost n Li , n Ai 

The variance, VarYˆ , is the variance of the total in the previous section. The variance and
cost are written as functions of the list and area sample sizes nLi and n Ai , as both
parameters depend on sample size.
Costs also vary by stratum. We approximate the cost of sampling a stratum with a linear
function,
L

Cost n Li , n Ai   c 0   (c Li n Li  c Ai n Ai ) ,
i

where the total cost is the sum of c0 , the fixed cost for both the list and area frames, cLi ,
the cost of adding another nest or nest cluster to the list sample and c Ai is the cost of
adding another area sample.
The Lagrangian Multiplier,  , represents the trade off between cost and variance. When
we change the sample sizes we decrease the variance because we are taking a larger
sample and also paying the cost of it. Among all the strata there is some consistent
tradeoff between reducing the variance and increasing the cost,  . The solution to the
minimization of the cost-variance function, requires more sampling in the strata that for
the best price will get the best reduction in variance. The sample is said to be optimally
allocated among the strata.
Since we solve the multiplier for the sample size, the stratum nest densities and their
variances are input, forming the “data” used in the equations. We use the estimates
derived from the nest list and pilot studies (Table 2). Likewise, we also use the nest list
data to construct cost functions for sampling. These calculations yield optimal sampling
procedures, given the pilot data that are input. Following most sampling texts, we
suggest choosing sample sizes 10 percent more than the minimum recommended from
the optimal allocation.
We compare four designs:
1.
2.
3.
4.

List-only,
Area-only,
Dual-frame, and
Combined dual-frame.

1. List-only. The current information comes from the State nest list. A “List-only”
design would involve sampling only from these nest lists without any attempt to estimate
the number of nests missing from the list. The estimates can be derived by just using the
terms in the total and variance equations that have to do with the list. The list estimate is

29

always biased, as occupied nests always exist that are not on the list. Magnitude of the
bias can be expressed as the list coverage, which is the number of nests on the State nest
lists as a percentage of the total nests. We include this bias when comparing the sample
designs by making the estimate of variability the mean squared error,
mse  Var Yˆl  bias 2 .

 

2. Area-only. The area-only design ignores the State nest lists and estimates the number
of occupied nests using just the nests found on a random sample of the plots. The area
estimate uses only the terms for the area survey in the total and variance equations. All
the nests in the plots are used, not just the new nests. This estimate is unbiased.
3. Dual-frame. The dual-frame design includes both the list and the new nests in the area
survey. The total and variance equations are shown above.
4. Combined dual-frame. Finally, we include a special case of the dual-frame called the
combined dual-frame where the list nests are sampled immediately after sample plots.
This could save on flight time. The equations are the same as for the dual-frame, but the
size of the list frame sample is the number plots in the area sample times the portion of
area plots that also are known to contain list nests plus an extra sample of list cluster
plots. We use different cost functions because of the differences in sampling.
Cost Functions for Sampling List-Plot Clusters
The cost functions for sampling the list are determined by simulation, drawing samples
and calculating the shortest distance needed to travel among all the sample clusters plus
the distances to travel within each cluster. Samples of different sizes are drawn, the
distance for the minimum spanning tree is summed for each sample, and a regression is
done with flight miles against sample size. A number of samples of different sizes were
estimated for each stratum using the contiguous 48 State nest list. Cost functions are
derived by a linear regression allowing equations for each stratum to be different. The
totals of the distances traveled among given locations are determined from their minimal
spanning trees (Paradis 2004, the mst function). Minimal spanning trees do not account
for flights starting and returning to one or more airports for each flight.
The distances traveled during the pilot surveys can be converted to costs by assuming the
planes flew 100 miles per hour and the cost of the plane was a rate of $317 per hour (the
Service rate in 2007). Because the conversion factor varies with flight fuel and other
costs that will probably increase before implementation, we use flight miles as our
measure. Actual costs can be applied post hoc using relevant costs. The relative rates do
not change. Only flight miles are accounted for in this analysis. Other costs can be
added on to those found here to make more realistic estimates. As cost estimates per list
cluster or plot are revised, the analysis should be revised. Figure 5 shows the number of
miles needed to fly samples of a given number of list-plot clusters. There is a separate
line for each stratum. Some lines do not extend as far as others because of their varying
number of clusters. This is especially true for the trace stratum which only has 12 nests
in 12 clusters; trace clusters are far apart, hence the line is very steep. The slope affects
30

the proportion of the stratum that is sampled; the steeper the slope, the more expensive
the sampling cost and the less it is likely to be sampled. The intercept affects the initial
cost of sampling the stratum. Since all strata are sampled to some extent, the initial cost
applies to a sample of any size. The other strata have lower, more similar slopes.
Cost Functions for Sampling Plots
The cost functions for the plots in the area frame are similarly determined, see Figure 6.
At a range of sample sizes, the minimum distances among the simulated sample plots are
added to the within-plot costs for each plot. The within plot costs are determined from
the length of the shoreline within each plot. We set a minimum value of 33 miles to scan
a plot with little or no habitat. We also set an upper bound, assuming that complete
coverage by dividing the plot into transects would be more efficient at some point; the
upper bound is never reached.
The linear cost functions are determined as in the list-plot cluster analysis, but here the
samples are plots over the whole stratum. The plots occur more regularly than the list
nests or plot clusters, so the differences in the stratum cost functions are due to the
amount of shoreline in the plots and the size and shape of the stratum. For these cost
functions, the cost of sampling the trace stratum is more expensive than sampling the low
stratum. Costs of the other strata again are lower and generally similar among strata.
The cost functions for the combined dual-frame design are not shown but are slightly
higher than those shown in Figure 6.
Proportion of Occupied Nest and List Coverage Simulations
Since we did not have consistent information from all States on the proportion of
occupied nests or on list coverages in the High-density strata, we simulated values of
occupied nests for each stratum between 0.35 and 0.7, and list coverages from 0.4 to 0.9.
For the Low and Trace strata we assumed a much higher proportion of coverage (0.7 to
0.98). We used these values in 100,000 simulations, in which proportions of occupied
nests and list coverages were simulated for each stratum and the optimal sample
allocation and cost-variance calculations were calculated. The ranges of values provided
as results are based on the simulations.
Optimal Stratum Allocation
As mentioned above, the optimal allocation of sampling effort to strata is independent of
the overall sample size. Consequently, the allocations presented below represent the
relative amount of effort to be allocated to each stratum. The percentages vary because
of simulations due to unknown proportions of occupied nests and list coverage. In
general, sample sizes increase for strata that are larger, more densely populated, more
variable, with a lower detection probability, with a more variable detection estimate, and
are less expensive to sample (Table 3). What is striking is the variation allocated to the
list frame, 23 to 80 percent. If we knew list coverage to be high we would sample more
of the list.
31

The allocations for the area frame are shown in the second column of Table 3. Notice
that trace and low strata are sampled lightly both in relation to proportional sampling and
to the large size of those strata. Although large in area, the strata are very lightly sampled
because they are not densely populated and are costly to sample. Almost two-thirds of
the cost of the survey is due to sampling the Low and Trace strata. (14,000 miles are
needed to sample the High-density strata while 26,000 miles are needed to sample the
Low and Trace strata.) This highlights the need to sample the low and the trace strata
efficiently. During the implementation of the survey, we suggest that the necessity of an
area sample in the Low and Trace strata be evaluated. It is unlikely that such a sample
will provide much useful information on eagle populations due to the low abundances of
nests. In the Low stratum, targeting potential habitat (amount of habitat or shoreline near
a given plot) could improve the sampling and estimation. Non-aerial sampling and
citizen-science based approaches could greatly enhance the efficiency of the survey,
especially in the Low stratum, and we recommend that these approaches be explored.
Survey Sample Design Comparison
With optimally allocated samples for each design, we can compare the three primary
alternative designs (list-only samples, dual-frame samples, and combined sampling
(sampling list clusters after the plots)) in two ways: (1) In terms of cost: what standard
error (SE) can we obtain for a given total cost for each of the sampling designs, and (2) In
terms of precision: what do we have to spend for each design if we require a given
overall standard error. Figure 7 answers the first question. The list-only survey is the
horizontal line on top. We use the mean squared error (MSE), which reflects both bias
and variance. The MSE includes the square of the bias between the list estimate and the
actual number of nests. The bias term dominates at the 60 percent list coverage we
observed in the pilot studies. No amount of sampling will overcome it - even if all the
strata are censused. Dual-frame sampling is more cost effective than area-only sampling
in a survey of over 20,000 flight miles (Figure 7). There is not much difference between
the dual-frame and the combined dual-frame sampling in terms of the standard error
obtained for a given cost.
Table 4 answers the second question: given certain precision requirements (setting
differing levels of effect size (relative change), power, and significance level) for the
estimates, what cost is needed to obtain such a requirement? The first case is the stated
monitoring goals: a 25 percent relative change (effect size) detected 80 percent of the
time (power) at a significance level of 10 percent (significance level). These
specifications determine a required standard error. The cost is shown in the following
columns for the list-only frame (list), plot or area-only frame (plot), dual-frame, and
combined-dual-frame. The list-only flight miles are constant but the estimate obtained
will always be biased with no way to determine the bias. The dual-frame estimates are an
improvement over the area-only frame, with the combined dual-frame design being
slightly more efficient. In practice, the economies associated with the combined
sampling may be greater.

32

These costs represent minimum costs. They do not take into account search time for
nests, flights to and from airports, weather days and per diem for pilots and observers,
differences in flight rates among regions, equipment and preparation costs. Survey
planners should add additional funding to accommodate these incidental costs.
Table 5 shows example sample sizes by State for the proposed survey goals. These will
change as we work with the States to refine the stratification and sampling methods
especially in the Low and Trace strata. Figure 8 shows an example sample selection
including area samples in all strata.
Effects of List Coverage
The dual-frame design provides a way to assess the coverage of the list. The area frame
is used to estimate the number of new (i.e., not in the list) nests, and hence to estimate the
list coverage. The variance of the new nests is the major contribution to the variance of
the total. The list coverage (that is, the proportion of actual nests that are in the list)
affects the relative efficiency of the designs.
Figure 9 shows how the standard error changes for given list coverage when the survey
requirements are the stated survey goals. The dual-frame is better than an area (plotbased) survey only, as list coverage is over 40 percent. The difference improves as the
list coverage improves. The list approaches the dual-frame standard errors as the
coverage approaches 100 percent. Note that an evaluation based only on the list coverage
is problematic without area-based sampling to assess the coverage. That is the value of
dual-frame sampling. It allows an assessment of the list coverage, and permits estimation
of actual numbers of occupied nests.
The optimal allocation treats the same strata in the list and area frames as different strata,
i.e., the Central Florida stratum has a sample of list plots and a sample of area plots.
These are treated as if they are in different strata. It then allocates sampling effort to
each. Because the variability of the list-frame is so much less than the area frame, most
of the sampling effort is allocated to the area-frame. Figure 10 shows the proportion of
flight miles allocated to the list as the list coverage improves. We use cost because it is
the common denominator between the two frames. The cost allocated to the list increases
as the list coverage improved but the costs only vary noticeably when list coverage is
greater than 90 percent. As the cost of the plot frame increases or the plot standard
deviations improve relative to the list frame, more effort (as cost) is allocated to the list
more quickly (red and green lines). The large allocation to the area frame indicates the
importance of reducing the variability in the plot sample.
Misclassification of Occupied Nests
We use the definition of occupied nests from this Post-delisting Monitoring Plan for the
Bald Eagle (p.15–16). Nests in repair were able to be distinguished by each of the State
crews in the pilot study. Fly backs or other methods can be used to verify the nest

33

condition. If an empty nest condition cannot be assessed, the nest is classified as
unoccupied. This provides a conservative estimate of occupied nests.
In those cases where the list survey is conducted separately from the area survey, at least
two independent observations of occupancy can be obtained. Occupancy should be
recorded separately for each visit. The area survey should be run early in the breeding
season when the pairs are most tightly bound to the nest. A Lincoln-Peterson estimate
can assess the number of nests where occupancy is missed on both visits. If the status of
the nest is determined outside the surveys, that better assessment of the status may be
used for that observation, but only observations from the list and area surveys should be
used in Lincoln-Peterson estimate and in the survey total variance calculations.
Conclusions

The goal of the post-delisting monitoring survey is to estimate the change in occupied
bald eagle nests in the contiguous 48 States. To achieve this goal, we use the procedures
outlined above for deciding on strata, clustering nests from the contiguous 48 State nest
list, optimally allocating samples to strata for both the list and plot surveys, and randomly
selecting samples according to that optimal allocation. We estimate sample sizes to
obtain estimates with a given power and precision or within given costs. We use the
dual-frame procedure with a double observer technique in the plot survey to sample more
efficiently, resulting in the reduction of less than half the flight miles. We combine
sampling list clusters with the area plots survey to better assess coverage of the list.
Finally, we use independent estimates of occupancy to assess the number occupied of
nests missed in both the list and area samples.

34

References

Bartish, TM. 1994. Design Considerations for Trust Species Monitoring: Bald Eagles
and Their Habitat in the Great Lakes Region. U.S. Geological Survey, Biomonitoring
of Environmental Status and Trends Program Pilot Project Report, December 1994.
45 pp.
Grier, J.W. and J.E. Guinn. 2003. Bald Eagle Habitats and Responses to Human
Disturbance in Minnesota: Final Report. Unpublished Report. Final report to the
Minnesota Department of Natural Resources, Natural Heritage and Nongame Wildlife
Program, Division of Ecological Sciences. St. Paul, Minnesota. 44 pp.
http://files.dnr.state.mn.us/eco/nongame/projects/consgrant_reports/2003/2003_grier_
guinn.pdf.
Haines, D. E., and K. H. Pollock. 1998. Estimating the number of active and successful
bald eagle nests: an application of the dual-frame method. Environmental and
Ecological Statistics 5:245-256.
Nichols, J. D. J. E. Hines, J. R. Sauer, F. W. Fallon, J. E. Fallon, and H. J. Heglund.
2000. A double-observer approach for estimating detection probability and
abundance from point counts. Auk 117(2):393–408.
Lohr, S.L., 1999. Sampling: Design and Analysis. Brooks/Cole Publishing Co.,
Washington, DC. 494 pp.
Paradis E., Claude J. & Strimmer K. 2004. APE: analyses of phylogenetics and evolution
in R language. Bioinformatics 20: 289–290. PDF, cran.rproject.org/web/packages/ape/ape.pdf
Sauer, J. R., W. A. Link, J. D. Nichols, and J. A. Royle. 2005. Using the North
American Breeding Bird Survey as a tool for conservation: A critique of Bart et al.
(2004). Journal of Wildlife Management 69(4):1321–1326.
Thompson, S. K. 1992. Sampling. John Wiley & Sons, Inc., New York, New York.
343 pp.
Thompson, W. L. 2004. Sampling Rare or Elusive Species. Island Press., Washington,
DC. 429 pp.
U.S. North American Bird Conservation Initiative Monitoring Subcommittee. 2007
Opportunities for Improving Avian Monitoring. U.S. North American Bird
Conservation Initiative Report. 50. pp. Division of Migratory Bird Management,
U.S. Fish and Wildlife Service, Arlington, VA; www.nabci-us.org.

35

Tables
Table 1. Aggregating or merging sequence of adjacent strata. Stratum 1 and 2 are the two

strata to be combined. “Edge” is the name the combination strata are given. “SE” is the
standard error of the survey total of the combination. The start is the individual State
BCRs. Note the SE does not noticeably change until after Edge.156.

36

37

Table 2. Summary statistics of State-BCRs with the number of nests, plot, density per 10 km x

10 km, and standard error of the density. Density is determined from nest list nests. The
larger, higher density strata will be sampled much more intensely. It is also encouraging
that portions of BCRs in different States have similar plot densities. This suggests that
the BCR stratification is effective.
Nests

Plots

Density

Wisconsin

State

Boreal Hardwood Transition

Bird Conservation Region

1635

481

3.40

4.730

Washington

Northern Pacific Rainforest

2002

612

3.27

5.635

Virginia

Southeastern Coastal Plain

473

189

2.50

4.203

Virginia

New England/Mid-Atlantic Coast

159

75

2.12

3.251

Maryland

New England/Mid-Atlantic Coast

363

200

1.82

2.624

Iowa

Prairie Hardwood Transition

129

73

1.77

2.880

Michigan

Boreal Hardwood Transition

1654

1007

1.64

2.806

Illinois

Mississippi Alluvial Valley

6

4

1.50

1.732

Florida

Peninsular Florida

1513

1105

1.37

2.799

Kentucky

Mississippi Alluvial Valley

13

11

1.18

0.982

Minnesota

Boreal Hardwood Transition

1084

937

1.16

2.165

Maine

Atlantic Northern Forest

1055

919

1.15

2.909

Pennsylvania

New England/Mid-Atlantic Coast

2

2

1.00

0.000

Oregon

Northern Pacific Rainforest

657

851

0.77

2.364

Louisiana

Mississippi Alluvial Valley

315

410

0.77

2.333

Tennessee

Mississippi Alluvial Valley

16

21

0.76

1.513

Michigan

Eastern Tallgrass Prairie

31

46

0.67

2.023

Minnesota

Prairie Hardwood Transition

273

497

0.55

1.216

Oregon

Great Basin

529

1156

0.46

2.818

Wisconsin

Prairie Hardwood Transition

446

998

0.45

1.363

Illinois

Prairie Hardwood Transition

13

32

0.41

1.012

Washington

Northern Rockies

93

242

0.38

0.996

South Carolina

Southeastern Coastal Plain

206

543

0.38

1.040

Montana

Northern Rockies

503

1596

0.32

1.012

Florida

Southeastern Coastal Plain

167

584

0.29

0.794

New Jersey

New England/Mid-Atlantic Coast

42

147

0.29

0.672

South Dakota

Eastern Tallgrass Prairie

9

33

0.27

0.944

Nevada

Sierra Nevada

2

8

0.25

0.463

Illinois

Central Hardwoods

42

183

0.23

0.622

Michigan

Prairie Hardwood Transition

139

611

0.23

1.119

Iowa

Eastern Tallgrass Prairie

245

1082

0.23

0.955

Connecticut

Atlantic Northern Forest

2

9

0.22

0.441

Minnesota

Eastern Tallgrass Prairie

22

111

0.20

0.989

Ohio

Eastern Tallgrass Prairie

94

527

0.18

0.742

Pennsylvania

Lower Great Lakes/ St. Lawrence Plain

15

89

0.17

0.482

California

Great Basin

62

400

0.16

0.471

Virginia

Piedmont

65

421

0.15

0.815

38

SE

Nests

Plots

Density

Nebraska

State

Eastern Tallgrass Prairie

Bird Conservation Region

34

221

0.15

0.480

Missouri

Mississippi Alluvial Valley

15

102

0.15

0.534

Nebraska

Prairie Potholes

23

159

0.14

0.488

New Hampshire

New England/Mid-Atlantic Coast

6

42

0.14

0.417

Ohio

Lower Great Lakes/ St. Lawrence Plain

32

230

0.14

0.416

Indiana

Central Hardwoods

49

355

0.14

0.391

Iowa

Prairie Potholes

42

308

0.14

0.758

Montana

Badlands And Prairies

186

1395

0.13

0.681

Arkansas

Mississippi Alluvial Valley

52

395

0.13

0.573

Kentucky

Southeastern Coastal Plain

5

39

0.13

0.522

Maryland

Piedmont

9

71

0.13

0.412

South Carolina

Piedmont

34

273

0.12

0.492

Minnesota

Prairie Potholes

81

699

0.12

0.472

Idaho

Northern Rockies

122

1059

0.12

0.458

Georgia

Southeastern Coastal Plain

109

949

0.11

0.491

Georgia

Piedmont

48

424

0.11

0.452

Colorado

Northern Rockies

10

92

0.11

0.346

Pennsylvania

Piedmont

13

121

0.11

0.337

Louisiana

Gulf Coastal Prairie

39

367

0.11

0.545

California

Sierra Nevada

55

520

0.11

0.459

Oklahoma

West Gulf Coastal Plain/Ouachitas

29

295

0.10

0.445

Arkansas

West Gulf Coastal Plain/Ouachitas

63

650

0.10

0.435

Texas

West Gulf Coastal Plain/Ouachitas

67

712

0.09

0.676

Missouri

Central Hardwoods

79

871

0.09

0.345

Kentucky

Central Hardwoods

62

699

0.09

0.536

North Carolina

Southeastern Coastal Plain

63

714

0.09

0.308

Alabama

Appalachian Mountains

32

374

0.09

0.363

Washington

Great Basin

86

1006

0.09

0.410

Mississippi

Mississippi Alluvial Valley

17

201

0.08

0.397

Oklahoma

Eastern Tallgrass Prairie

13

154

0.08

0.322

Massachusetts

New England/Mid-Atlantic Coast

16

201

0.08

0.366

New Jersey

Appalachian Mountains

3

38

0.08

0.273

California

Northern Pacific Rainforest

38

485

0.08

0.355

South Dakota

Central Mixed Grass Prairie

1

13

0.08

0.277

Wyoming

Northern Rockies

127

1651

0.08

0.410

New Jersey

Piedmont

3

40

0.08

0.267

Tennessee

Central Hardwoods

30

403

0.07

0.330

Alabama

Piedmont

4

55

0.07

0.262

Utah

Northern Rockies

2

28

0.07

0.378

Oregon

Northern Rockies

38

541

0.07

0.378

Connecticut

New England/Mid-Atlantic Coast

8

118

0.07

0.252

Wisconsin

Eastern Tallgrass Prairie

1

16

0.06

0.250

Nebraska

Central Mixed Grass Prairie

73

1220

0.06

0.247

39

SE

Nests

Plots

Density

New Hampshire

State

Atlantic Northern Forest

Bird Conservation Region

11

200

0.06

0.287

SE

Illinois

Eastern Tallgrass Prairie

66

1238

0.05

0.288

Alabama

Southeastern Coastal Plain

43

845

0.05

0.250

Ohio

Central Hardwoods

1

20

0.05

0.224

Tennessee

Appalachian Mountains

19

412

0.05

0.221

Idaho

Great Basin

50

1100

0.05

0.269

Missouri

Eastern Tallgrass Prairie

35

825

0.04

0.264

Wyoming

Badlands And Prairies

27

646

0.04

0.249

Oklahoma

Central Hardwoods

3

75

0.04

0.197

North Carolina

Piedmont

17

458

0.04

0.231

Alabama

Central Hardwoods

3

82

0.04

0.189

Ohio

Appalachian Mountains

11

310

0.04

0.202

Massachusetts

Atlantic Northern Forest

2

58

0.03

0.184

South Dakota

Prairie Potholes

28

880

0.03

0.205

Louisiana

West Gulf Coastal Plain/Ouachitas

15

474

0.03

0.175

Oklahoma

Oaks And Prairies

13

418

0.03

0.199

North Dakota

Badlands And Prairies

17

549

0.03

0.265

Arizona

Sierra Madre Occidental

30

972

0.03

0.179

Colorado

Southern Rockies/Colorado Plateau

45

1473

0.03

0.184

Kansas

Eastern Tallgrass Prairie

20

662

0.03

0.180

Arkansas

Central Hardwoods

10

332

0.03

0.256

Indiana

Eastern Tallgrass Prairie

13

448

0.03

0.168

Montana

Prairie Potholes

24

869

0.03

0.248

California

Coastal California

46

1773

0.03

0.263

Mississippi

Southeastern Coastal Plain

27

1047

0.03

0.219

Nebraska

Shortgrass Prairie

9

350

0.03

0.159

Louisiana

Southeastern Coastal Plain

2

78

0.03

0.159

Colorado

Shortgrass Prairie

28

1129

0.02

0.167

Rhode Island

New England/Mid-Atlantic Coast

North Dakota

Prairie Potholes

Maryland

1

42

0.02

0.154

31

1316

0.02

0.256

Appalachian Mountains

1

46

0.02

0.147

Pennsylvania

Appalachian Mountains

20

966

0.02

0.142

South Dakota

Badlands And Prairies

21

1037

0.02

0.148

Nebraska

Badlands And Prairies

1

52

0.02

0.139

Georgia

Appalachian Mountains

3

162

0.02

0.135

Delaware

New England/Mid-Atlantic Coast

1

57

0.02

0.132

Kentucky

Appalachian Mountains

5

302

0.02

0.190

Virginia

Appalachian Mountains

6

401

0.01

0.122

Arizona

Sonoran And Mojave Deserts

15

1060

0.01

0.159

Tennessee

Southeastern Coastal Plain

3

251

0.01

0.109

Texas

Gulf Coastal Prairie

5

521

0.01

0.158

Vermont

Atlantic Northern Forest

2

211

0.01

0.097

Wyoming

Shortgrass Prairie

1

126

0.01

0.089

40

Nests

Plots

Density

Indiana

State

Prairie Hardwood Transition

Bird Conservation Region

1

133

0.01

0.087

New Mexico

Sierra Madre Occidental

2

285

0.01

0.084

Utah

Southern Rockies/Colorado Plateau

9

1320

0.01

0.113

Oklahoma

Central Mixed Grass Prairie

4

757

0.01

0.073

Utah

Great Basin

3

851

0.00

0.059

New Mexico

Shortgrass Prairie

2

677

0.00

0.054

New York

Appalachian Mountains

1

379

0.00

0.051

Kansas

Central Mixed Grass Prairie

2

1102

0.00

0.043

Texas

Oaks And Prairies

2

1521

0.00

0.036

New Mexico

Chihuahuan Desert

1

878

0.00

0.034

Nevada

Great Basin

2

2466

0.00

0.028

New Mexico

Southern Rockies/Colorado Plateau

1

1326

0.00

0.027

West Virginia

Appalachian Mountains

0

628

0.00

0.000

Delaware

Piedmont

0

1

0.00

0.000

Arizona

Chihuahuan Desert

0

12

0.00

0.000

Texas

Chihuahuan Desert

0

1044

0.00

0.000

Utah

Sonoran And Mojave Deserts

0

4

0.00

0.000

Connecticut

Appalachian Mountains

0

11

0.00

0.000

Texas

Shortgrass Prairie

0

1047

0.00

0.000

Arizona

Southern Rockies/Colorado Plateau

0

937

0.00

0.000

Vermont

Lower Great Lakes/ St. Lawrence Plain

0

45

0.00

0.000

California

Sonoran And Mojave Deserts

0

1056

0.00

0.000

Texas

Tamaulipan Brushlands

0

719

0.00

0.000

Texas

Edwards Plateau

0

588

0.00

0.000

New York

New England/Mid-Atlantic Coast

0

84

0.00

0.000

Mississippi

Gulf Coastal Prairie

0

3

0.00

0.000

Wyoming

Southern Rockies/Colorado Plateau

0

108

0.00

0.000

Massachusetts

Appalachian Mountains

0

7

0.00

0.000

Nevada

Sonoran And Mojave Deserts

0

389

0.00

0.000

Nevada

Southern Rockies/Colorado Plateau

0

3

0.00

0.000

Maine

New England/Mid-Atlantic Coast

0

30

0.00

0.000

New York

Atlantic Northern Forest

0

294

0.00

0.000

Kansas

Shortgrass Prairie

0

372

0.00

0.000

Kansas

Oaks And Prairies

0

3

0.00

0.000

Kansas

Central Hardwoods

0

1

0.00

0.000

Texas

Central Mixed Grass Prairie

0

888

0.00

0.000

North Carolina

Appalachian Mountains

0

213

0.00

0.000

Ohio

Prairie Hardwood Transition

0

1

0.00

0.000

Oklahoma

Shortgrass Prairie

0

113

0.00

0.000

Idaho

Southern Rockies/Colorado Plateau

0

6

0.00

0.000

South Carolina

Appalachian Mountains

0

19

0.00

0.000

South Dakota

Shortgrass Prairie

0

23

0.00

0.000

New York

Lower Great Lakes/ St. Lawrence Plain

0

581

0.00

0.000

41

SE

Table 3. The range of percent of the total sample allocated to both the list and area frame.

The percentages are those within each frame. The first line indicates the allocation of the
flight miles (cost) to each frame. The percentages are reversed for the area frame to
emphasize that resources are the complements of each other. The variation in list
coverage causes a large variation in allocation to each frame. Improved list coverage
results in increased allocation to the list frame.
Stratum
FL
IA
IL
LA
Low
MD
ME
MI
MN
MN
MT
OR
OR
SC
Trace
VA
VA
WA
WI
WI
.

BCR Name
Peninsular Florida
Prairie Hardwood Transition
Prairie Hardwood Transition
Mississippi Alluvial Valley
New England/Mid-Atlantic Coast
Atlantic Northern Forest
Boreal Hardwood Transition
Boreal Hardwood Transition
Prairie Hardwood Transition
Northern Rockies
Great Basin
Northern Pacific Rainforest
Southeastern Coastal Plain
New England/Mid-Atlantic Coast
Southeastern Coastal Plain
Northern Pacific Rainforest
Boreal Hardwood Transition
Prairie Hardwood Transition
Percent allocation to each frame

42

List
2.5- 8.8
0.2- 0.6
0.0- 0.1
0.5- 1.9
5.2-18.5
0.6- 2.3
1.8- 6.4
2.1- 7.6
1.8- 6.4
0.9- 3.2
0.9- 3.1
0.4- 1.3
0.8- 2.8
0.6- 2.3
0.0- 0.0
0.3- 1.2
0.5- 1.7
1.0- 3.7
1.4- 4.9
1.0- 3.6
23-80

Area
7.8-2.1
0.8-0.2
0.1-0.0
3.0-0.8
16.1-1.8
1.7-0.4
7.5-0.9
7.1-1.8
5.2-1.3
2.1-0.5
3.5-0.9
4.8-1.3
3.5-0.9
2.1-0.6
0.3-0.1
0.8-0.2
2.4-0.6
5.3-1.4
5.6-1.4
4.2-1.1
77-20

Table 4. Flight miles (thousands) needed to obtain a sample with given precision
requirements for an estimate of the total for the four survey options (DF is dual-frame).
The standard errors are those needed to obtain the required effect size, power, and
significance level requirements. The list-only survey is not continued beyond the first
survey requirements because even censusing the list does not reduce the bias due to
incomplete coverage. Thus, a list-only survey cannot meet any of the survey
requirements.
Precision

0.25
0.25
0.25
0.25
0.25
0.25
0.15
0.15
0.15
0.15
0.15
0.15
0.1
0.1
0.1
0.1
0.1
0.1

Power

Sig-Lvl

SE

List

Plot

DF

Comb-DF

0.8
0.8
0.8
0.9
0.9
0.9
0.8
0.8
0.8
0.9
0.9
0.9
0.8
0.8
0.8
0.9
0.9
0.9

0.10
0.05
0.01
0.10
0.05
0.01
0.10
0.05
0.01
0.10
0.05
0.01
0.10
0.05
0.01
0.10
0.05
0.01

1,133
1,006
825
963
869
731
680
604
495
578
522
438
453
402
330
385
348
292

16

186
230
327
249
298
402
453
550
745
589
689
883
843
989
1,252
1,044
1,180
1,419

40
46
59
48
55
70
78
94
129
101
119
158
149
181
251
195
230
305

39
45
58
48
54
69
77
93
127
99
117
155
147
178
247
191
226
300

43

Table 5. The list and area columns show the mean and (range) of the sample sizes for
each stratum for the proposed survey goal. Results are based on simulated list coverages
and percent occupied nests because we did not have information for each stratum. Note
that part of the list sample will be from the area plots that have list nests. List nests in the
area plots become part of the list sample.
Stratum
FL
IA
IL
LA
Low
MD
ME
MI
MN
MN
MT
OR
OR
SC
Trace
VA
VA
WA
WI
WI

BCR Name
Peninsular Florida
Prairie Hardwood Transition
Prairie Hardwood Transition
Mississippi Alluvial Valley
New England/Mid-Atlantic Coast
Atlantic Northern Forest
Boreal Hardwood Transition
Boreal Hardwood Transition
Prairie Hardwood Transition
Northern Rockies
Great Basin
Northern Pacific Rainforest
Southeastern Coastal Plain
New England/Mid-Atlantic Coast
Southeastern Coastal Plain
Northern Pacific Rainforest
Boreal Hardwood Transition
Prairie Hardwood Transition

44

14
1
1
3
28
4
10
12
10
5
5
2
5
4
1
2
3
6
8
6

List
( 8-24)
( 1- 2)
( 1- 1)
( 2- 6)
(16-49)
( 2- 6)
( 6-17)
( 7-20)
( 6-17)
( 3- 9)
( 3- 9)
( 2- 4)
( 3- 8)
( 2- 6)
( 1- 1)
( 1- 4)
( 2- 5)
( 4-10)
( 5-13)
( 4-10)

18
2
1
7
29
4
17
16
12
5
8
11
8
5
1
2
6
12
12
10

Area
(2-61)
(1- 7)
(1- 2)
(1-26)
(2-94)
(1-14)
(2-62)
(2-53)
(2-42)
(1-18)
(1-29)
(2-41)
(1-29)
(1-19)
(1- 3)
(1- 7)
(1-20)
(2-43)
(2-43)
(2-37)

Figures

Figure 1: State boundaries and Bird Conservation Regions (BCRs) in the
contiguousUnited States

45

Figure 2: Strata after aggregating adjacent strata with similar nest densities. The
aggregate strata are: Trace with only 14 nests in the stratum (light green); Low (the result
of aggregating the similar low density strata, yellow); Washington and Oregon Northern
Pacific Rainforest; Washington, Idaho, and Montana Northern Rockies; North and
Central Florida (South Florida is part of the Low stratum even though it is not adjacent to
it); Iowa, Illinois and Wisconsin Prairie Hardwood Transition; Minnesota, Wisconsin,
and Michigan Boreal Hardwood Transition; Mississippi Alluvial Valley in Louisiana;
South Carolina Southeastern Coastal Plain; Delaware; Virginia and Maryland New
England/Mid-Atlantic Coast; Virginia Southeastern Coastal Plain; and two Maine strata.

46

Figure 3: Distribution of bald eagle nests as determined from 2004-2006 State lists.

47

Figure 4: Missouri area frame consisting of 10 km square plots. Shaded squares are the
sample plots selected. Diamonds are the list nests selected.

48

Figure 5: Number of miles needed to fly samples of a given number of list-plot clusters.
Each line represents an aggregate stratum. Low and Trace are aggregate strata with low
densities of bald eagles.

49

Figure 6: Number of miles needed to fly samples of a given number of plots. Each line
represents an aggregate stratum. Low and Trace are aggregate strata with low densities
of bald eagles.

50

Figure 7: The standard error for a given flight miles for each sample design. The listonly design standard error is a mean square error that includes the bias squared. This bias
is the major source of error and cannot be reduced by any amount of sampling, not even a
census of the nest list.

51

Figure 8: An example sample selection with sample sizes for the proposted survey
requirements. The red samples are the list plot clusters, the blue are the area plots, and
the green are the area plots also with the list nests. The sample sizes are not fixed for
each State, only for the 20 strata. More information on the proportion of occupied nests,
list coverage in each stratum and efficient ways to sample the Low and Trace strata will
modify the allocation to better target sampling.

52

Figure 9: Standard errors for a given list coverage for each of the survey design options.
These were calculated given the survey requirements are the stated survey goals. Note
that when the list coverage is 40 percent the dual-frame design is equivalent to the plotonly design. The list-only design converges with the dual-frame as the list coverage
approaches 100 percent. Unfortunately with the list-only design, there is no measure of
actual measure of list coverage.

53

Figure 10: Proportion of the flight miles allocated to the list frame vs. the area frame
given different list coverages. As the cost of the area frame increases and/or the
variability decreases, more is allocated to list-frame as the two quicker rising curves
show.

54

Appendix 2
Bald Eagle Post-delisting Monitoring Standard Operating Procedures
Based on the Maine Bald Eagle Pilot Project
by John R. Sauer and Mark C. Otto
Introduction

There are three potential types of aerial survey plots for the bald eagle post-delisting
monitoring survey: list, area, and combined plots.
List plots will be sampled if the State has not recently updated its nest list. In such cases,
the status of all known nests will need to be observed and recorded for selected list plots.

Potential bald eagle habitat in area plots will be scanned for any nests and on- or off-thenest bald eagles. The dual observer protocol described below is used for area plots to
assess observer detection.
Combined plots are area plots that also contain known list nests. These plots will first be
flown as area plots (i.e. scan habitat for any nests using dual observer protocol), and then
subsequently flown as list plots (i.e. return flight through plot to record status of all
known nests). To avoid the potential for observers to search for known nests during the
area plot portion of the survey, pilots and observers should not know which area plots are
actually combined plots until after the area plot portion of the survey is complete (i.e.
they complete the area plot survey and then turn over the map to reveal known nest
locations).

Area plot sampling requires that observers conduct independent counting in silence.
Communication among observers waits until after an observation has been passed by the
plane to identify what observers actually saw: the bird or nest. For each observation,
information on who saw it and who did not see it should be recorded. If one of the
counters is only participating in some of the data collection, they should be recorded as
“not counting” for observations that occurred when they were not counting.
This Standard Operating Procedure (SOP) is written assuming the rear seat observer is on
the right side of the plane. Sitting behind the pilot is acceptable if needed, but it is
preferable that the rear-seat observer be on the right side of the plane and count in tandem
with the front-seat observer. In either case, the plane should be oriented, so the observers
in tandem (front-back seat) are on the shoreward side or the side with better eagle habitat.
Orienting the plane this way will facilitate counting in critical eagle habitat but should
only be done when it does not require undue maneuvering or a safety hazard. To maintain
independence of counts, the pilot should not orient the plane to make it easier for the rear
observer to see eagles that the pilot has seen. If necessary, a screen should be added to
prevent observers from seeing each other during counting. Also, front seat observers
should not point out sightings as they occur so as to keep the observation independent.
The observers sitting in tandem will implement the double observer counting procedure.

55

Observations will not be noted by either observer until it is clear that the other observer
has “missed” the observation. For the front-seat observer, this position should be when
their view of the observation becomes obscured, which we define as an angle of 135
degrees behind the observer. For the back-seat observer, this position will be after the
observation passes 90 degrees (opposite their position). At this point, the observation
will be described by the observer who made the observation, and it will be noted who
among the crew (front-seat observer, rear-seat observer, pilot) saw the observation (of a
single, pair, nest, juvenile). Note any discrepancies in the identifications of sightings. If
needed, the aircraft can bank to verify the observation and collect additional information.
For accurate GPS observations it is useful to circle before taking locations.
If possible, the pilot will also be included as a third participant in the observations,
although in interior transects the pilot will collect independent data (see below).
Maps of area plots and list plots should be studied before the survey flights to plan
potential flight routes. This will minimize potential confusion and maximize observation
time during the survey.
Information to be Collected

Information to be collected includes: nest, adult or immature eagle observations; which
observers were looking, seeing, or missed the sighting; activity of individual birds
(flying, perching, on nest incubating or not incubating); and nest status, (see sample data
sheet, p. 61). A “capture history” format will be used to record whether each of the three
observers actually saw an observation. The field will have three, single-character values
that reflect the status of the observation for each of the three observers: pilot, front-seat
observer, and rear-seat observer respectively. Status from an observer will be indicated
as a “0” for not seen, a “1” for seen, and an “x” for not observing. For example, the entry
"x01" would be entered when the pilot was not observing, the front observer did not see
the observation, but it was seen by the rear observer. Observations by the pilot that were
recorded on the left side of the aircraft would be coded “1xx.” Observations made by the
dual observers when the pilot was not attempting to observe on the right side of the place
would be either “x10,” “x01,” or “x11,” where x01 represents pilot not observing, not
seen by front observer, but seen by rear-seat observer.

56

Information to be collected has been coded into the data entry program (Table 1). Eagle
ages and nest condition are to be entered as separate species codes: adult, immature, and
unknown for off the nest, and empty, incubating, occupied (but not incubating), eggs, or
eaglets for a nest sighting. Location off the nest can be flying, perching, or on nest, and
the location on the nest identifies the nest tree type to more easily find the nest again.
The locations can be tailored to the survey area. Examples are pine, spruce, conifer,
deciduous, ground, or not applicable (na). In Florida, we added cell phone tower and
stadium lights. Information may be included in the comment field such as approximate
height of nest (5 m increments), nest contents (fresh material, egg shell fragments), or
whether the nest was seen while cruising or during a secondary search. Eagles on nests
should be recorded with the nest record, but eagles observed separately from the nests
should be recorded separately. Any ambiguity in recording eagles at nests should be
discussed in the comments field of the record.
Table 1. Variables and acceptable values for the data entry files.
Year
Month
Day
Pilot
FObserver
RObserver
Plot
CapHist

2004
4-5
1-31
Consistent
initials
Number
001
010
011
01x
0x1
100
101
10x
110
111

Location
(off nest)

fly
perch
na

Location
(on nest)

pine
spruce
conifer
deciduous
ground
na

The Record and Transcribe programs automatically record time and location
(latitude and longitude) from the GPS used. The time and location are recorded both in
the track files describing the flight path and in the observation files described above for
each observation. The flight path information is important in reconciling area
observations with known nests and to determine how effectively the plots were searched.
If another method of recording the transcribing is used, these should also be part of the
data collected. Instructions on using Record and Transcribe are in BPManual.pdf
special instructions for these surveys are included in FileInstructions.txt.

57

Survey Notes

Observers and pilots should make a text file of notes about weather, general observations,
problems in the protocol, and ideas how to improve the survey. Unusual seating
positions (pilot not in the left front seat, etc.) and who is recording the data should also be
noted. Observations for each day can be made in the same file. The file should include
the time zone used to record the times and the datum the GPS is recording the locations.
Recording Procedures

The front-right observer will reconcile and record observations for all three observers
during the flight. Jack Hodges program, or a predetermined alternative, will be used to
record and transcribe the sightings. At the beginning of the flight, give the date and
weather. When entering and leaving each plot, say “beginning” or “ending list or area
plot” and give the plot identification number. For contiguous plots say “leaving [list or
area] [number] and entering [list or area] [number]” every time the plot changes (or in
Combined plots, when the search changes from searching the habitat to checking status of
known nests). The codes recorded will be as follows in Table 2:
Table 2: Codes for Jack Hodges program to record sightings
Code
ADULT
IMMATURE
UNKNOWN
REPAIRED
UNREPAIRED
EMPTY
DESTROYED
NOTFOUND
INCUBATING
OCCUPIED
EGGS
EAGLET
BEGAREA
ENDAREA
BEGLIST
ENDLIST
GPS
COMMENT

The first three codes are for sightings off the nest: ADULT, IMMATURE, UNKNOWN.
The next nine are nest condition codes. REPAIRED is a nest with fresh sticks or boughs
on top (sign of an occupied nest), UNREPAIRED if not. If the nest condition cannot be
determined, use EMPTY. INCUBATING, OCCUPIED, EGGS, EAGLET all are codes
of an occupied nest. Put number of adults, eggs or fledglings, nest contents not described
by the codes, experienced observer knew location of nest, etc. in the comment field.
BEG[LIST or AREA] and END[LIST or AREA] are recorded when entering and
leaving each plot. The number field should have the plot number without leading zeros.
58

That no observations were seen in a plot can be noted in the comments of an END[LIST
or AREA] record. This will confirm that no records were missed. From the plot codes,
ferrying and time within plots can be determined. GPS is used to record the location of a
nest or other geographic feature. CapHist and Location are not used in GPS records. The
comment field is used to record what location is recorded. The COMMENT record is
used to make a note, such as fog over the coastal part of a plot. The location may or may
not be important to the note.
Header variables are Year, Month, Day, Pilot, FObserver, RObserver, and Plot. These
are variables that stay constant for the number of observations and are treated differently
than the Code, CapHist, Location, Comment variables. In the resulting dataset, the fields
should be separated by commas. Any comments should follow after the fields. Commas
should not be used in the comments; use slashes or semicolons instead. Acceptable
values of the header variables are shown in Table 1 in the format used by the
Transcribe program.
The time will be set to the local time zone at the start of the survey, for example Eastern
Daylight Time and noted in the survey notes. GIS locations must be recorded. Locations
of flying eagles will be recorded after the counts have been reconciled understanding that
they will be farther down the track from the sighting. Locations of nests should be taken
over the nest to reconcile sightings with known nest locations. When transcribing,
remember to save the new header variable values every time the header variables
change. The new values will not be recorded unless you do.
A checklist for the observers to use in the plane is included at the end of this appendix. It
is on a separate sheet and may be laminated.
Flight Procedures
Seating Positions: Pilot, front-seat observer, rear-seat observer.
Altitude and speed: Fly at 200-700 feet above ground level at around 100 miles per
hour. Make adjustments to give the best visibility to the tandem observers when possible.
Timing and sequence: Time of day should not matter unless the sun is too low and
affects visibility. Observers should not continue when overly fatigued. Surveys should
be run before the trees leaf out.

59

Defining flight paths within area plots: The surveys should only include the parts of
plots that occur within the defined survey area (i.e. within the contiguous 48 states,
excluding areas within large bodies of water, etc.). Because eagles nest within 1 mile of
water (i.e., coastlines, islands, inland ponds and lakes of >35 acres, and rivers >330 feet
in width or similar definition appropriate for the geographic area), searching should be
constrained to these locations. Flight paths should be defined on maps prior to surveys,
preferably in consultation with an experienced eagle survey biologist. Transects will be
flown along shorelines and interior habitats along shorelines.
Defining flight paths within list plots: Observers should note identifying features and
locations of known nests on maps prior to the flights. This will improve the success of
nest searches. Record nests that are not found and ones that are destroyed.
Defining flight paths for combined plots: Known nests in the area plots will not be
known to the observers prior to the flight. The combined plots will be treated as area
plots during the pre-flight planning. Then, after the plot habitat is flown, the front-seat
observer will turn to the next map. If there are known nests, the map will contain the
locations. The plot will then be treated as a list plot and the status of known nests will be
obtained. If the status of the nests was obtained during the preceding area survey, then
the nest does not need to be checked again. A non-observer will need to insert the known
nest maps after the area plot maps for the combined plots in the order that they will be
flown. It is important to the double observer protocol that the observers scan the habitat
during the area plot part of the survey and not look for particular known nests.
Shoreline transects: Surveys should be conducted first along edges of water bodies in
plots with the right side of the plane (containing front-and and rear-seat observers) on the
landward side of the aircraft. The pilot navigates and can act as a third observer for land
sightings, or can observe additional land areas (e.g. islands) in the area. The plane should
be oriented to facilitate consistent viewing regions for both observers, without tight turns
or banking. Single observations by the pilot can be recorded separately from the tandem
observers and entered as “1xx,” unless the pilot is counting the same field of view as the
other front-seat observer. Areas surveyed on the shoreline pass will be one-quarter mile
inland from edge of water, and the plane should be located close to shore.
Interior transects: If it appears that the shoreline transect did not cover adjacent eagle
habitat, the aircraft should turn and fly a transect with the aircraft approximately one-half
mile inland from the shoreline after the shoreline transect is completed. The front and
rear-seat observer will conduct the double-observer procedure in the one-quarter milewide strip that extends from the edge of the shoreline to the aircraft. The pilot will
survey the area on the shore-ward side of the aircraft, from the midline of the aircraft
(one-half mile from shore) to one mile from the shoreline.

Islands should be surveyed by shoreline transects and if necessary (e.g., if island is >1/2
mile wide) by interior transects.

60

Observations: Known nests may be observed outside the plots and recorded, as may
new nests within list plots. Information on the locations and status of these nests can be
used to update the nest list of the next sample period but will not be used in the statistical
estimation. These should be marked as “out of sample” in the comments. Nest searching
in conjunction with eagle observations can be conducted, but should be limited to a single
additional pass. During these searches, the double observer counting protocol should be
maintained (i.e., independent counting until nest is passed).

61

Eagle and Nest Observation Checklist
1. Begin plot #___
2a. If an eagle off the nest

a. Age: adult, immature, or unknown
b. Capture history: [x, 0, 1] codes for left-front, right-front, then right-rear
c. Distance: estimate distance in meters of a line parallel to the aircraft track
d. Behavior: flying, perched
e. Location: if perched on tree or ground
2b. If a nest

a. Nest condition: empty, incubating, occupied, eggs, eaglets
b. Capture history: [x, 0, 1] codes for left-front, right-front, then right-rear
c. Behavior: on nest or perched
e. Nest tree: pine, spruce, (generic) conifer, or deciduous
3.

Comment on nest contents including number of eagles, uncertainty about
species’ nest, whether observed while cruising or on a secondary nest search,
observer knowing nest location. Indicate observation off the sample plots or
new nests in list plots

4.

End plot #___

62

Appendix 3

Bald Eagle Nestlist Website
by Mark Otto

The Service is developing a website for the States to update their bald eagle nest list data
and download copies of their data and other coverages needed for the bald eagle postdelisting survey. The survey team will use the nest lists to sample from and to determine
the overall survey design. The main purpose of the website is to make it easy for the
States to incorporate their nest list data into a national database, to insure consistency and
integrity of the data collected from many sources, and to allow the States access to their
own data and information they would need for the survey. The website and database
would be the single, secure site for this data. Initially, the site will be hidden and
protected from unauthorized users. A test version of the site was developed in July 2008.
The site will have several secondary uses. Our Service Migratory Birds and Ecological
Services programs may access it in the future to assist evaluations of applications for bald
eagle permits under the Bald and Golden Eagle Protection Act. Knowing the proximity
of bald and golden eagle nests to potential sources of disturbance will facilitate informed
permitting decisions. Permit issuers may also be a source of information on new nests.
Additionally, the Service’s Law Enforcement program will have access to the website in
order to use the database information to trace any possible links to depredation on nests.
State biologists would input data using an input form for single or small numbers of new
nests. They could also edit past data and upload a text file or other database table.
Uploaded data will be scanned for viruses and checked for validity. The user will
immediately be informed of any problems with the import, so that the data can be
corrected and imported again without undue frustration.
The data would keep information on the locations and the status of the nests over time.
There would be two main tables: one to record information on the nest locations, and
another to track the sightings/status of the nest within and among breeding seasons.
We would use the most current information to develop the survey design. Historical
records would be used for research purposes. Since currently abandoned or destroyed
nests would remain in the database, we would need to be diligent about recording nest
status.
The nest list table would include Nest ID, State, State ID, Site (e.g. Georgetown dump),
Position (Latitude, Longitude in decimal degrees), Datum (1927, 1983, …), Status,
Observer, Tree or site description (spruce, cell phone tower, ground nest, …), Position in
structure, Comment. The sighting table would include Nest ID, year and date, status
(adult, destroyed, deteriorated, eaglet, eggs, empty, immature, incubating, not found,
occupied, repaired, unknown, unrepaired), territory ID if that is recorded, how the
observation was reported, comment, and owner of the record.
Future features might include: (1) an interactive map of nests similar to the Alaska
website which allows information on selected nests to be displayed.

63

(2) Displaying and printing maps of selected 10 km2 plots with aerial photos or
topography identifying bald eagle habitat, with or without list nests. (3) A public part of
the website with information showing survey results and a nest map with imprecise or
fuzzed nest locations.
The users of the website would be the administrators, developers, bald eagle management
team, State coordinators, State biologists, Service Migratory Birds, Ecological Services,
and Law Enforcement programs, USGS and possibly outside researchers. All would
access the website and database with a username and password.






Administrators (1 to 5 people): They would be able to add, modify permissions,
and remove users. They could commit data to the database, edit data in the
database, modify some web pages and code, and run audits.
Developers (1 to 5 people): They would create, test, and maintain the website,
database and code. They would have the same permissions as administrators.
(one or two people)
FWS Bald Eagle Management Team (10 to 15 people): They would be able to
add, remove, and track (list) State coordinators and State biologists. We would
need to keep good contact information for the Team and post that on the site for
biologists and State coordinators to get help.
State Coordinators (50 to 100 people): They would be able to add, remove, and
track (list) biologists within their State. We would need to keep good contact
information for the Coordinators and post that on the site for biologists to get
help.

64





State Biologists (50 to 150 people): They would be able to add, edit, and
download information from their State. They will be able to download coverages
and other information about the survey.
FWS Migratory Birds and Ecological Services (about 100 people): They
would be able to define an area and get the nests possibly affected by a proposed
disturbance activity.
FWS Law Enforcement (1 to 10 people): They will need to be able to trace
activity by user, view when nest list records were accessed and by whom.

For each survey period each of the States in the contiguous 48 States will be requested to
update their list of nests to the extent that data is available. This will include adding any
new nests that have been reported and identifying old nests that were destroyed. This
data is already collected by many of the States. If no data is available, no changes to the
nest list need to be made. To receive consistent information from the States, the website
provides two ways to make changes to their nest lists: edit the nest data by nests and
annual observations by each observation or upload the changes in a comma delimited text
file. The biologists for each state will only be able to view and edit their own data.

65

Bald eagle database users will log on with user name and password. The website is
designed to avoid unintentional traffic as much as possible.

66

The nests and annual observation on those nests can also be viewed and edited.
Respondents may also upload or e-mail the nest list changes in a comma delimited file
with the information described below.

67

The database contains two main tables: one for nest location and information and another
table for the annual observations. Some States may make multiple sightings within each
breeding season. Only the summary records for a season will be recorded: one for the
occupancy status, and possibly one for the productivity status if the State collects it. The
productivity status can stand in for the occupancy if there is not an occupancy record for
that year because a productive nest is by definition occupied. The columns for the nest
table are:
Column
NestID

State
StateID
Lng

Lat
LocAcc

Location
Access
Owner
DatEntered
Comment

Description
Unique index among all the States. This we generate to keep track of
nests over all States
Two letter state abbreviation, using the standard FIPS codes
Text or number index that the State uses for their nests. Should be
unique within the State. This allows the State to identify nests as they do
in their own record system.
Longitude in decimal degrees. We need the locations of the nests to
assign nests to plots, our list cluster sampling unit. We also need the
location to reconcile the nest with the area observations. If the do not
have accurate locations the burden is greater on the States to assign the
lists to plots and reconcile the area and list observations. We will display
to five decimal places but it depends on the accuracy of the data received.
The values will be negative to keep the values within the required -180 to
+180 range. We do not use the recommended variable name long
because Microsoft Access confuses this with the data type long.
Latitude in decimal degrees to 5 decimal places.
Location accuracy: choose the number of meters where less than 10
percent of the locations will be off by more than that distance. Usual
GPS accuracy is 200m. Just guess at the distance. Note: National Map
Accuracy Guidelines, rockyweb.cr.usgs.gov/nmpstds/nmas.html, use less
than a 10 percent error or 1/30" in a 1:20,000 map or within 16.93
meters. (Optional but we need to need to assign the nests to the correct
10 km2 plot, otherwise the State will need to assign the nests to plots)
Description of the tree (these are not used for vegetation classification),
site description, anything to identify the site for reconciliation.
(Optional)
Public, Tribal, Federal, State, Local managed, Private, indicate if it is
accessible to the public. A citizen observer could legally observe the nest
for the survey. (Optional)
ID of the person responsible for the data (Login name in the website)
Date the data was entered. (Entered automatically when uploaded.)
Miscellaneous but useful information (Optional). Information about
territories may be added here.

We keep multiple annual observations on the same nest, so we can construct the list
frame using the annual observation information. For States that collect multiple

68

observations in a breeding season, we can estimate the chances of missing an occupied
nest by only visiting the nest one time. The columns for the annual observation table are:
Column
NestID
State
StateID
Status
NEg
Source
Owner
DatEntered
Comment

Description
Unique index among all the States (Assigned by the database)
Two letter state abbreviation, using the FIPS codes
Text or number index that the State uses. Should be unique within the
state
Nest status (see below)
Number of eggs or eaglets, NULL or not used if Status other than egg or
eaglet (Optional)
Public call in, State biologist, Confirmed (Optional)
ID of the person responsible for the data (Login name in the website,
user does not enter)
Date the data was entered. (Entered automatically when uploaded.)
Miscellaneous but useful information (Optional)

Nest status in the Status column of the annual observation table takes on the following
values:
Code
Description
destroyed Nest destroyed or gone
eaglet
1 or more eaglet in nest
egg
1 or more eggs in nest
empty
Empty nest, repair status unknown
immature Immature on nest
incubating Eagle or pair on nest displaying incubating
notfound Not found on previously known location
occupied Occupied but not incubating
repaired Repaired nest
unknown Location known but status unknown
unrepaired Unrepaired nest

The main information we need to know to set up the list frame sampling is what the
status of a current nest is. The important codes are nest destroyed and not found. Nests
with other status codes make up the list frame to sample from. After a nest list is
sampled or censused during one of the five year sampling periods starting in 2009, we
only need to know whether a nest is occupied or not. The incubating, empty (if not able
to ascertain repair status), repaired, and unrepaired, and unknown refine the occupied
determination. Unknown status nests are considered unoccupied to be conservative but
give added information on secondary analyses. If we help measure productivity from a
sample or census of nests, we need the eggs, fledgling, and eaglet categories. We will
use productivity data that that States are interested in collecting. Keeping and analyzing
the productivity data is an opportunity to provide additional information to the States
69

(and to the Service). We put it in our database to make it easy to process surveys for a
number of States.
The State nest lists can be updated by making the changes to the web forms above or
uploading a table of the updates. The nest table is only updated if the data is different
from the current information. The file would have the following information. The State,
Owner, and DatTim (date the information was entered) will be entered automatically,
using your username, the State that username is associated with, and the computer date
and time.
The location accuracy, nest access, source, and nest and annual observation table
comments are optional fields. Almost always there are oddities about some observations
that are not easily recorded on standard forms, but can save hours of detective work if
they are known. It also provides a way to find out what fields should later be added to
the formal input. We were not going to collect territory information because we cannot
assign territories for new nests or use the information in our estimation. We will also add
the State FIPS code, owner ID, and the data and time the owner entered the information.
These will be tracked and added automatically using their logon information and the
current time. We will work with the States to convert their data to a file that they can
upload to our database.

70

Appendix 4
Regional Bald Eagle Monitoring Coordinators
*For current contact information, go to http://www.fws.gov/midwest/Endangered
Fish and Wildlife Service Regions
Region 1: Regional Coordinator – Suzanne Audet
Suzanne Audet
Upper Columbia Fish & Wildlife Office
11103 E. Montgomery Drive
Spokane, Washington 99206
Tel: (509) 893-8002
Fax: (509) 891-6748
Suzanne_Audet@FWS.gov

States - CA, ID, NV, OR, WA, HI, Guam, American Samoa, Commonwealth of the
Northern Marianas
Region 2: Regional Coordinator – Greg Beatty
Greg Beatty
Arizona Ecological Services Field Office
2321 West Royal Palm Road, Suite 103
Phoenix, Arizona 85021
Tel: (602) 242-0210 x 247
Fax: (602) 242-2513
Greg_Beatty@FWS.gov
States - AZ, NM, OK, TX
Region 3: National Coordinator – Jody Millar
Jody Millar
Rock Island Ecological Field Office
1511 47th Ave.
Moline, Illinois 61265
Tel: (309) 757-5800 x 202
Fax: (309) 757-5807
Jody_Millar@FWS.gov
States - L, IN, IA, MI, MN, MO, OH, WI
Region 4: Regional Coordinator – Candace Martino (Jacksonville) and Al
Begazo (South Florida)
Candace Martino
Jacksonville Field Office
7915 Baymeadows Way, Suite 200
Jacksonville, Florida 32256-7517 Tel: 904 731-3142
Fax: 904 731-3045
Candace_Martino@FWS.gov
71

Alfredo Begazo
South Florida Ecological Services Office
1339 20th Street
Vero Beach, Florida 32960
Tel: (772) 562-3909
Fax: (772) 562-4288
Alfredo_Begazo@FWS.gov
States - AL, AR, FL, GA, KY, LA, MS, NC, SC, TN, PR,VI
Region 5: Regional Coordinator – Craig Koppie
Craig Koppie
Chesapeake Bay Field Office177 Admiral Cochrane Drive
Annapolis, Maryland 21401
Tel: (410) 573-4534
Fax: (410) 269-0832
Craig_Koppie@FWS.gov
Steve Mierzykowski, Environmental Contaminants Coordinator, Bald Eagle
Monitoring Literature Database
U.S. Fish & Wildlife Service
1168 Main Street
Old Town, Maine 04468
Tel: (207) 827-5938 ext. 17
Fax: (207) 827-6099
Steve_Mierzykowski@fws.gov
States - CT, DE, ME, MD, MA, NH, NJ, NY, PA, RI, VT, VA, WV, and Washington,
D.C.
Region 6: Regional Coordinator – Dan Mulhern
Dan Mulhern
Kansas Ecological Services Field Office
2609 Anderson Avenue
Manhattan, Kansas 66502
Tel: (785) 539-3474 ext 109
Fax: 785 539-8567
Dan_Mulhern@FWS.gov
States - CO, KS, MT, ND, NE, SD, UT, WY

72

Region 7: Regional Coordinator – Phil Schempf
Phil Schempf
Migratory Bird Management-Raptors
3000 Vintage Blvd., Suite 240
Juneau, Alaska 99801
Tel: (907) 586-7331 or 7243
Fax: (907) 586-7378
Phil_Schempf@FWS.gov

State – AK
Region 8: Regional Coordinator - Elizabeth Willy
Elizabeth Willy
USFWS-Klamath Falls Field Office
1936 California Ave
Klamath Falls, OR 97601
Tel: (541) 885-2525
Fax: (541) 885-7837
Elizabeth_Willy@fws.gov
States - CA, NV
Region 9: Washington, DC Endangered Species
Mary Klee
USFWS, 4401 N. Fairfax Drive
Mail stop 420 ARLSQ
Arlington, Virginia 22203
Tel: (703) 358-2171
Fax: (703) 358-1735
Mary_Klee@FWS.gov

73

73

Appendix 5
Glossary
Word Definition
Allocation How much of the sample is portioned to each stratum.
Bias The difference between the true value and the estimate. An
estimator is biased if as you increase the sample size the estimates
do not approach the true value.
Cluster A convenient division of the sample into primary and secondary units, so the
sample is chosen from the primary units then all or a sample is taken from the
secondary units of the primary units selected. This usually increases the
variance but makes the sample more cost effective. For example, schools are
chosen as clusters because it is difficult to get permission to sample each
school. Then, students only within the selected schools are sampled.
Confidence intervals Interval that we hope the true value of the estimate falls within a given percent
of the time.
Detection probability The probability that an observer or observers will see the object they are
counting.
Estimate The value of a parameter from a statistical model determined from data. The
formula is the estimator.
Frame A list of all possible elements in the population. The list of all known nests or
the list of all plots in the contiguous 48 States. We choose the sample from the
frame.
Independence Where one event does not give any information on the chance of another event
occurring. For example, having more list nests does not indicate that there
will also be more (or less) new nests found.
Lagrange multiplier Method of optimizing a function when there are constraints, e.g., minimizing
the variance given that only so much can be spent sampling, or minimizing the
cost given that we obtain a precise enough estimate.
Mean Characteristic of a population that is the sum over the number of observations;
average.
Normalize Make so that all the values sum to 1. It may also mean to make the values sum
to zero and scale the values so the standard deviation sums to 1.
Optimal The best of all possible choices. Usually choosing the parameter so the value
of the function is maximized or minimized.
Population The aggregate from which the sample is chosen. The target population is what
we want to sample: active bald eagle nests. The sampled population is what
is actually sampled: may be the active nests in our State nest lists.
Power The chance of detecting a significant difference when a significant difference
exists. Increasing the sample size usually increases the power.

Precision How much estimates of the same sample size will vary. Technically it is the
74

Word Definition
1/Variance. As the variance of the estimate decreases, the precision increases.
Proportional Allocation Allocating the sample among strata in proportion to the size the stratum.
Sample The part of the population that is randomly selected and used to represent and
measure the population.
Significance level The percentage you will accept saying there is a true difference when it only
occurred by chance. At a 5 percent significance level, you will get an estimate
from a sample that is significant, not because there is a difference, but just by
chance. The smaller you make the significance level, the greater sample size
you are going to need to detect a significant difference.
Standard error The standard error is the standard deviation of the estimate, not the population
value. As the sample increases the standard error will decrease but the
estimate of the standard error will just fluctuate less but remain around the
same level.
Stratum Partition of the population so that the units within each stratum are as similar
as possible. You may stratify where sub-estimates are required. We divide
the contiguous 48 States into strata where the density of nests differs greatly.
Variance Measure of the scatter of the population around its mean value. The standard
deviation is the square root of the variance. The standard error is the standard
deviation divided by the square root of the sample size.

75