Survey designs and statistical methods for the estimation of avian population trends

              Biological Report 90(1)
January 1990
Survey Designs and Statistical Methods
for the Estimation of Avian Population
Trends
Fish and Wildlife Service
u.S. Department of the Interior
Biological Report
This publication series of the Fish and Wildlife Service comprises reports on the results of research,
developments in technology, and ecological surveys and inventories of effects of land-use changes on fishery and
wildlife resources. They' may include proceedings of workshops, technical conferences, or symposia; and
interpretive bibliographies.
Copies of this publication may be obtained from the Publications Unit, U.S. Fish and
Wildlife Service, 1849 C Street, N.W., Mail Stop 130 - ARLSQ, Washington, DC 20240, or
may be purchased from the National Technical Information Service (NTIS), 5285 Port
Royal Road, Springfield, VA 22161.
ISSN 0895·1926
This publication may be cited as follows:
Sauer, J. R., and S. Droege, editors. 1990. Survey designs and statistical methods for the estimation of avian
population trends. U.S. Fish Wildl. Serv., Biol Rep. 90(1). 166 pp.
Biological Report 90(1)
January 1990
Survey Designs and Statistical Methods for
the Estimation of Avian Population Trends
edited by
John R. Sauer and Sam Droege
u. S. Fish and TiVildlife Service
Patuxent Wildlife Research Center and
Office ofMigratory Bird Management
Laurel, Maryland 20708
u.S. Department of the Interior
Fish and Wildlife Service
Washington, DC 20240
Preface
On 12-13 April 1988 a workshop on the analysis of avian population trends was held at the Patuxent Wildlife
Research Center in Laurel, Maryland. The workshop was cosponsored by the Branch of Migratory Bird Research
of the Patuxent Wildlife Research Center and the Office of Migratory Bird Management. During this workshop,
we hoped to bring together some of the biologists and statisticians that coordinate and analyze data from major
bird surveys to discuss recent advances in analytical methods of estimating population trends. The workshop had
three sessions: one to describe some of the major surveys used to estimate population trends, one to discuss
analytical methods, and one to consider population trends of a selected species: scissor-tailed flycatchers (Tyrallllus
forficatus), for which a data set from the North American Breeding Bird Surveyhad been distributed to participants
before the meeting.
These proceedings present the results of the workshop. The papers are organized into three parts, following
the design of the workshop. Part I is composed of papers that describe the design of major avian surveys, along
with reviews of the constraints that the designs place on the analysis of population trends. Part II presents some of
the major analytical methods that are used to estimate population trends. There is a good deal of diversity among
the papers in this part, with some papers discussing overall approaches to surveys and their analysis, others
attempting to analyze the relations among the methods, and some presenting only a specific method of analysis.
Several papers broach general questions of sample size allocation for roadside surveys and associated technical
questions. Part III contains three analyses of the scissor-tailed flycatcher data set: two variants of the
route-regression method and a nonparametric analysis.
11
Acknowledgm~nts
We thank R. 1. Smith of the Office of Migratory Bird Management, and R. J. Hall and R. L. Jachowski of the
Branch of Migratory Bird Research, Patuxent Wildlife Research Center, for providing us with financial support
for both the workshop and the preparation of the proceedings. The Patuxent Wildlife Research Center provided
logistical support for the workshop. N. Bushby and N. Hestbeck of the Branch of Technical Services, Patuxent
Wildlife Research Center, coordinated the peer review of manuscripts. M. A. McKeogh assisted with many aspects
of the workshop and the preparation of the proceedings. We also thank D. S. Chu, D. K. Dawson, R. M. Erwin,
P. H. Geissler, and J. D. Nichols for assistance with the workshop.
Most of the participants in the workshop were called on to review at least one manuscript. We thank C. Bunck,
G. S. Butcher, D. Bystrak, J. T. Cary, B. T. Collins, D. K. Dawson, R. T. Engstrom, R. M. Erwin, M. R. Fuller,
P. H. Geissler, J. M. Hagan, G. A. Hall, J. Hatfield, M. A. Howe, F. James, D. W. Johnston, M. K. Klimkiewicz,
W. Link, C. E. McCulloch, J. D. Nichols, B. R. Noon, R. J. O'Connor, G. Pendleton, C. S. Robbins, C. R. Smith,
J. R. Spendelow, S. R. Taub, K. Titus, and J. Verner for their efforts in manuscript review. L. Hungerbuhler and
M. A. McKeogh typed several of the manuscripts.
1lI
Contents
Part I: Surveys Used to Estimate Avian Trends Page
The North American Breeding Bird Survey. Sam Droege 1
Audubon Christmas Bird Counts. Gregory S. Butcher. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Description of the Wisconsin Checklist Project. Stanley A. Temple and John R. Cary. . . . . . . . . . . . . . . 14
Use of breeding bird atlases to monitor population change. Chandler S. Robbins . . . . . . . . . . . . . . . . . . 18
Methodology of the International Shorebird Survey and constraints on trend analysis.
Marshall A. Howe 23
Evaluation of the Colonial Bird Register. R. Todd Engstrom 26
Descriptions of surveys: breeding bird censuses. David w: Johnston 33
Migration banding data: a source of information on bird population trends?
Deanna K. Dawson 37
Sources of migrant hawk counts for monitoring raptor populations.
Mark R. Fuller and Kimberly Titus 41
The Common Birds Census in the United Kingdom. Raymond J. O'Connor
Part II: Methods of Trend Analysis
47
Topics in route-regression analysis. Paul H. Geissler and John R. Sauer . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Estimation of annual indices from roadside surveys. John R. Sauer and Paul H. Geissler. . . . . . . . . . . . 58
Using rerandomizing tests in route-regression analysis of avian population trends.
Brian T. Collins 63
Estimating (relative) species abundance from route counts ofthe Breeding Bird Survey.
Lincoln E. Moses and Daniel Rabinowitz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Smoothed scatterplot analysis of long-term Breeding Bird Census data. Stephan R. Taub . . . . . . . . . . . 80
Methodological issues in the estimation of trends in bird populations with an example:
the pine warbler. Frances C. James, Charles E. McCulloch, and Loretta E. Wolfe. . . . . . . . . . . . . . . . 84
Using checklist records to reveal trends in bird populations.
Stanley A. Temple and John R. Cary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Detecting trends in hawk migration count data. Kimberly Titus, Mark R. Fuller,
and Dan Jacobs 105
Evaluation of the sensitivity of breeding bird surveys using a stochastic simulation model.
Janles Cox 114
Influence of observer effort on the number of individual birds recorded on Christmas Bird Counts.
Gregory S. Butcher, and Charles E. McCulloch 120
Population trends in the least tern (Sterna antillarnm) from Maine to Virginia: 1975-1986.
R. Todd Engstrom, Gregory S. Butcher, and James D. Lowe 130
Trend analyses for raptor nesting productivity: an example with peregrine falcon data.
Paul H. Geissler, Mark R. Fuller, and Lynne S. McAllister. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 139
Current thinking on United Kingdom bird monitoring. Raymond J. O'Connor 144
IV
Part ill: Scissor-tailed Flycatcher Analysis
Analysis of scissor-tailed flycatcher population changes. Brian T. Col/ins. . . . . . . . . . . . . . . . . . . . . . . .. 157
Route-regression analysis of scissor-tailed flycatcher population trends. John R. Sauer 160
Trends in counts of scissor-tailed flycatchers based on a nonparametric rank-trend analysis.
Kinlberly Titus 164
v
Part I: Surveys Used to EstiInate
Avian Trends
For most of the surveys discussed in the workshop, trend analyses have traditionally involved (1) identifying a
dependent variable that indexes population size at one of a group of survey sites (or routes) and (2) estimating
population changes (trends) over time in some way that takes into account the intrinsic features of each site. In
this part'of the workshop, we discussed some of the sources of information used to estimate bird population trends,
and we attempted to identify some of the constraints that the underlying sources of information placed on the
selection of a relevant dependent variable and the analytical method of estimating trends.
In some surveys, the dependent variables used to index population size at a site in a year are "natural," such as
the total number of individuals heard or seen over the 50 stops of a Breeding Bird Survey route, or the population
estimates from a Common Bird Census, Breeding Bird Census, or a colonial bird count. In other surveys, count
data must be adjusted to control for variation in observer effort, such as occurs in the Christmas Bird Count,
migration banding, hawk counts, or the International Shorebird Survey. Finally, in certain surveys, presence or
absence data must be converted to estimates of proportion-of-area-occupied data for analysis, as in breeding bird
atlases or checklist projects. All surveys, however, share a common problem: the yearly index values are not
statistically based estimates of population size, but instead are assumed to be related to population size by a
proportionality constant (Bart and Schoultz 1984). Standardization of survey procedures and a posteriori
adjustments of indices by effort, discussed in this section, represent attempts to minimize variation in this
proportionality constant. We recommend that readers see Lancia et al. (in press) for a detailed discussion of the
consequences of assuming a constant relation between population indices and population parameters.
Literature Cited
Bart, J., and J. D. Schoultz. 1984. Reliability of singing bird surveys: changes in observer efficiency with avian density. Auk
101:307-318.
Lancia, R. A., J. D. Nichols, and K. H. Pollock. Estimating the number of animals in wildlife populations. In T. H. Bookhout,
ed. Wildlife management techniques, 5th ed. The Wildlife Society, Bethesda, Md. (In press)
AVIAJ'I POPULATION TRENDS 1 - ,
The North American Breeding Bird Survey
by
Sam Droege
u.s. Fish and Wildlife Service
Office ofMigratory Bird Management
Laurel, Maryland 20708
Origin and Extent of Survey
The North American Breeding Bird Survey (BBS) is
coordinated and maintained by the U.S. Fish and
Wildlife Service and the Canadian Wildlife Service. Its
purpose is to provide long-term information about the
abundance and distribution of breeding birds in the
United States and Canada. Data exist for more than 500
species. More than 400 of these species have been
recorded on more than 50 survey routes (Erskine 1978;
Robbins et al. 1986; Droege and Sauer 1989).
The BBS was designed and initiated by C. S. Robbins.
In 1966 routes were first run in the eastern States and
Provinces. In 1%7 the survey was expanded to include
the central States and Provinces, and by 1968 the entire
continent was included.
Design of the BBS
Individual BBS routes are 24.5 miles long and
comprise 50 point counts spaced one-half mile apart.
Each point count, or stop, extends for 3 min. During this
period the observer counts all birds heard singing or
calling and any bird sighted within a quarter-mile radius.
The density of BBS routes in a State or Province varies
with the available number of observers. Density is
highest in the East and lowest in the Intermountain West
and northern boreal zones (Figure). Once an
appropriate density of routes has been estimated for a
State or Province, routes are stratified by degree blocks
of latitude and longitude (e.g., 1 per degree block in
Nevada and 16 per degree block in Maryland). Within
a degree block, starting points are chosen at random, as
are directions of travel. The actual roads the route
follows are constrained by their availability and traffic.
With a few exceptions, routes do not cross degree blocks
or physiographic strata boundaries.
Each route is run once a year by a skilled volunteer.
Most routes are run during June. However, routes in
portions of Arizona, California, Florida, Nevada, New
Mexico, and Texas are often run in May, and some
northern routes are run during the first week of July.
Data are submitted by the observer to the Office of
Migratory Bird Management, U.S. Fish and Wildlife
Service, Laurel, Maryland, both on original field sheets
and on summary sheets. These data are checked for
transcriptional, mathematical, and biological errors on
arrival. Each summary sheet is cataloged, entered into
a data base, and then verified. Data are sent through
several computer routines that check for clerical and
mathematical errors. Every route is checked against
regional lists of expected species. Unusual species are
flagged on printouts of a route's data and sent to the
observer who double checks the work and, if necessary,
sends further documentation. Data collected under
poor weather conditions or by observers with
insufficient ability to accurately census birds are coded
but not used in subsequent analyses.
Constraints on the Analysis of the BBS
Differences in regional density of routes and
consistency in coverage (Figure) affect our ability to
track regional populations. Most BBS routes have had
gaps in coverage during the past 22 years. Gaps occur
when observers are unavailable or if a completed survey
is deemed unsuitable. Statistical analysis of BBS data
requires special care because of the relatively large
amount of missing data. Simple yearly means can
produce deceptive results, especially if data are missing
from routes containing very high or low densities of the
species being analyzed (Geissler and Sauer 1990). Every
region has, to some extent, missing data problems.
Additionally, almost every region has had new routes
added and old ones discontinued, further adding to the
problems of unbalanced data. The number of BBS
routes run also varies with year. During the first years of
the program, coverage was of relatively low intensity.
Small sample sizes reduced our ability to accurately
track population changes.
2 BIOLOGICAL REPORT 90(1)
BREEDING BIRO SURVEY ROUTES
3181 TOTAL OBSERVATIONS PLOTT
Figure. Distribution of Breeding Bird Survey routes.
Most of the United States and Canada have an
extensive network of secondary roads. The random
placement of BBS routes in these regions is relatively
straightforward. However, in some portions of the
mountainous west (e.g., the Great Basin) and the far
north (e.g., Northwest Territories), roads are fewer and
tend to occur in valleys and near watercourses. In
regions of low relief, riparian zones are likely to be
undersampled because most roads travel across rather
than within floodplains. The location and design of
regional road networks will influence calculations of
avian relative abundance. Unfortunately, we will not
know how well BBS routes represent regional habitats
until LANDSAT data and associated technologies
become more available.
Roadsides can be considered unique habitats in some
regions. Most roads act as boundaries, creating edge
habitats populated by speeding metallic predators. They
often parallel superb song and hunting perches, such as
telephone lines and fences. How dissimilar roadside
stops are from random geographic points and how
dissimilar rates of roadside bird population change are
from the entire landscape's rate of change are currently
unknown. However, at least in regions oflow elevational
relief, all common diurnal species appear to be well
represented on BBS routes, although perhaps not
relative to their true abundance.
Observers vary in their ability to hear, identify, and
estimate the abundance of birds (Faanes and Bystrak
1984). Even among experienced observers, large
differences exist in how numbers are estimated (Bart
and Schoultz 1984). Wherever possible, observer
covariables should be included in statistical analyses of
BBS data (Geissler and Sauer 1990).
Flocking and colonial birds present special problems
because of their greater variation among yearly counts
(Sauer and Droege 1989). Variation often results from
localized concentrations of feeding birds (e.g., gulls,
herons, vultures) or stops near colonies (e.g., swallows).
This variation can be reduced if the number of stops at
which the birds were seen are used to calculate trends
rather than total number of individuals (Bart and
Klosiewski 1989). However, the number of stops at which
a species is present is a measure of the area the species
occupies rather than a direct measure of its abundance.
The number of individuals recorded on a BBS route
is both a function of a species' detectability and its true
abundance. Detectability varies among species,
observers, and habitats, and with time of year, weather,
density, and the male's breeding status (Emlen 1971;
Berthold 1976; Marten and Marler 1977; Ralph and
Scott 1981). Due to these differences in detectability,
relative abundance is not directly comparable between
species. Detectability can also change with local
population density. Herice, the relation between BBS
AVIAN POPULATION TRENDS 3
counts and true density could be curvilinear or even a
step-function. Because the life histories of the surveyed
populations are almost always poorly known it is
difficult to make generalizations about the BBS's
sensitivity to true population change. Statistical analyses
ofthese data and their subsequent interpretation should
dwell on the patterns of population change rather than
on the magnitudes of calculated trends and variances.
Variations associated with weather conditions and
time-of-year are minimized by only accepting data from
routes run during acceptable conditions.
Several recent studies have sought to corroborate
population trends on BBS routes with data from
independent surveys. Carolina wren (77l!yothonls
iudovicianus) numbers from Indiana May Counts
closely correspond to BBS annual indices calculated
for the same region (E. Hopkins, personal
communication). Raptor counts from migration
stations and Christmas Bird Counts (CBC) show good
corroboration for those species commonly detected on
BBS routes (Schueck et al. 1989; Titus et al. 1989).
State long-term trends for mourning doves (Zenaida
macroura) from the Mourning Dove Survey are
correlated with BBS trends calculated from the same
States (J. R. Sauer, personal communication).
Preliminary work with several species recorded
commonly on both BBS routes and CBC's also yield
similar population trends (Butcher et al. 1990).
Population trends derived from the Quebec checklist
program are in almost complete concordance with the
sign of trends from BBS routes in Quebec (A. Cyr,
personal communication).
We will continue to investigate and catalog the
relative strengths and weaknesses of the BBS data set
and encourage others to do likewise.
Literature Cited
Bart, J., and S. P. K1osiewski. 1989. Use of presence-absencc
to measure changes in avian density. J. Wildl. Manage.
53:847-852.
Bart, J., and J. D. Schoultz. 1984. Rcliability of singing bird
surveys: changes in observer efficiency with avian density.
Auk 101:307-318.
Berthold, P. 1976. Methoden der bestandserfasung in der
ornithologie: ubersicht und kritische betrachtung. J.
Ornithol. 117:1--69.
Butcher, G. S., M. R. Fuller, L. S. McAllister, and P. H.
Giessler. 1990. An evaluation of the Christmas Bird Counts
for monitoring population trends of selected species. Wildl.
Soc. Bull. (In press)
Droege, S., and J. R. Sauer. 1989. North American Breeding
Bird Survey Annual Summary 1988. U.S. Fish Wildl. Serv.,
BioI. Rep. 89(13). 16 pp.
Emlen, J. T. 1971. Population densities of birds derived from
transect counts. Auk 88:323-342.
Erskine, A. J. 1978. The first ten years of the Co-operative
Breeding Bird Survey in Canada. can. Wildl. Serv., Rep.
Ser. 42. 61 pp.
4 BIOLOGICAL REPORT 90(1)
Faanes, C. A, and D. Bystrak. 1981. The role of observer bias
in the North American Breeding Bird Survey. Pages
353-359 in C. J. Ralph and J. M. Scott, eds. Estimating
numbers of terrestrial birds. Stud. Avian BioI. 6.
Geissler, P. H., and J. R. Sauer. 1990. Topics in
route-regression analysis. Pages 54-57 in Survey designs
and statistical methods for the estimation of avian
population trends. U.S. Fish Wildl. Serv., BioI. Rep 90(1).
166 pp.
Marten, K., and P. Marler. 1977. Sound transmission and its
significance for animal vocalization. 1. Temperate habitats.
Behav. Ecol. Sociobiol. 2:271-290.
Ralph, C. J., and J. M. Scott, editors. 1981. Estimating
numbers of terrestrial birds. Stud. Avian BioI. 6. 630 pp.
Robbins, C. S., D. Bystrak, and P. H. Geissler. 1986. The
Breeding Bird Survey: its first fifteen years, 1%5-1979.
U.S. Fish WildI. Serv., Resour. Publ. 157. 1% pp.
Sauer, J. R., and S. Droeg~. 1989. Wood duck population
trends from the North American Breeding Bird Survey.
Pages 159-165 in L. H. Fredrickson et aI., eds. Proceedings
of the 1988 North American Wood Duck Symposium,
SI. Louis, Mo.
Schueck, L. S., M. R. Fuller, and W. S. Seegar. 1989. Falcons.
Northeast Raptor Management Symposium and
Workshop. NationaL Wildlife Federation. Sci. Tech. Ser.
13:71-80.
Titus, K., M. R. Fuller, D. F. Stauffer, and J. R. Sauer. 1989.
Buteos. Northeast Raptor Management Symposium and
Workshop. National Wildlife Federation. Sci. Tech. Ser.
13:53--64.
AVIAN POPULATIO~ TRENDS 5
Audubon Christmas Bird Counts
by
Gregory S. Butcher
Cornell Laboratory of Ornithology
159 Sapsucker Woods Road
Ithaca, New York 14850
Origin and Current Design of the
Christmas Bird Count
The Christmas Bird Count (CBe) is the oldest and
largest wildlife survey in the world. It is sponsored by
the National Audubon Society, and the results are
published inAl7lericall Birds. It began in 1900 when 26
individuals responded to an editorial in Bird-Lore
magazine (Chapman 1900) by spending an hour or two
counting birds in their neighborhood on Christmas
afternoon. Since then, the increase in both the number
of counts (Fig. 1) and the number of participants
(Fig. 2) has been dramatic. In 1986-87, 41,249
individuals participated at 1,544 locations, including
1,508 locations in the United States (excluding
Hawaii) and Canada (Table 1).
All CBC's occur within a 15-mile-diameter circle.
Each local CBC coordinator chooses a single calendar
day within 2 weeks of Christmas Day for each year's
count. Each local CBC is conducted by between 1 and
1800
1600
1400
~ 1200
C
::J
o
U 1000
Q) 800
-0
E
.£ 600
230 field observers, who spend between 1 and 650
party-hours and 0 and 4,100 party-miles looking for
birds (Butcher and McCulloch 1990). Each local CBC
also includes between 0 and 343 individuals recording
birds at home. Most party-miles are covered by car;
party-hours can include hours on foot, bicycle, skis,
snowshoes, snowmobile, car, boat, and airplane. Many
counts include nocturnal party-hours and party-miles.
Every CBC has one or more count coordinators, who
are responsible for recruiting participants, assigning
participants to areas, providing overall guidance to
participants, compiling the results, securing written or
photographic descriptions of rare species, and sending
the results to the National Audubon Society. Most
coordinators divide the count circle into areas and
assign one group to each area. Participants are
instructed not to count birds out of their assigned area.
Rarities may be staked out ahead of time, and known
birding "hot-spots" may be revisited several times
during the day. When areas are revisited (and when
counting at feeders), participants record only the
45000
40000
35000
~
co
30000
CL
~~
""§ 25000
CL
o 20000
~
<1>
-0
E 15000
::J
:z:
400
Year
10000
5000
1900 10 20 30 40
Year
50 60 70 80 85
Fig. 1. The number of Christmas Bird Counts done each year
has grown dramatically from 1900 to 1985.
Fig. 2. The number of participants in Christmas Bird Counts has
grown dramatically from 1900 to 1985.
6 BIOLOGICAL REPORT 90(1)
Table 1. Physiographic regions ofthe United States and Canada, excluding Hawaii.
1 Subtropical
2 Floridian
3 Coastal Flatwoods
4 Upper Coastal Plain
5 Mississippi Alluvial Plain
6 Coastal Prairies
7 South Texas Brushlands
8 East Texas Prairies
9 Glaciated Coastal Plain
10 Northern Piedmont
11 Southern Piedmont
12 Southern New England
13 Ridge and Valley
14 Highland Rim
15 Lexington Plain
16 Great Lakes Plain
17 Driftless Area
18 SI. Lawrence River Plain
19 Ozark-Ouachita Plateau
20 Great Lakes Transition
21 Cumberland Plateau
22 Ohio Hills
23 Blue Ridge Mountains
24 Allegheny Plateau
25 Open Boreal Forest
26 Adirondack Mountains
27 Northern New England
28 Northern Spruce-Hardwoods
29 Closed Boreal Forest
30 Aspen Parklands
31 Till Plains
32 Dissected Till Plains
33 Osage Plain-Cross Timbers
34 High Plains Border
35 Rolling Red Prairies
36 High Plains
37 Drift Prairie
38 Glaciated Missouri Plateau
39 Great Plains Roughlands
40 Black Prairie
53 Edwards Plateau
54 Rolling Red Plains
55 Staked Plains
56 Chihuahuan Desert
61 Black Hills
62 Southern Rocky Mountains
63 Fraser Plateau
64 Central Rocky Mountains
65 Dissected Rocky Mountains
66 Sierra Nevada
67 Cascade Mountains
68 Northern Rocky Mountains
80 Great Basin Deserts
81 Mexican Highlands
82 Sonoran Desert
83 Mojave Desert
84 Pinyon-Juniper Woodlands
85 Pitt-Klamath Plateau
86 Wyoming Basin
87 Intermountain Grasslands
88 Basin and Range
89 Columbia Plateau
90 Southern California Grasslands
91 Central Valley
92 California Foothills
93 Southern Pacific Rainforests
94 Northern Pacific Rainforests
95 Los Angeles Ranges
96 Southern Alaska Coast
98 Willamette Lowlands
99 Tundra
maximum number of individuals of a particular species
seen at one time. Most count coordinators try to cover
as much of the count circle as the number of participants
and accessibility will allow. Most count coordinators
serve for many years, and many participants cover the
same areas year after year.
Evaluation of the CBC Design
There are a number of standards that should be met,
to some degree, if a surveyor census technique is to be
used for studies of changes in relative abundance
through time or space. The CBC does have standardized
techniques, some of which were presented earlier.
However, certain important factors, such as the route
covered and the amount of effort, are not standardized.
In some cases, count locations don't remain constant. In
this section, I list the ideal for a standardized count, then
compare the CBC design with the ideal.
Year-fa-year Changes in Coverage
The most basic point is that a survey should be done
the same way in different areas and in different years.
However, on the CBC, the amount of effort spent
looking for birds (as measured by observers, hours, or
miles) varies dramatically between count locations
within years and within count locations between years
(Butcher and McCulloch 1990). Also, individuals cover
their area of the CBC circle in a variety of ways. For
example, some observers may use boats in some years
and not in other years. This methodological difference
can have a dramatic effect on the number of waterfowl
counted. Similarly, some participants stay near their
cars in cold or wet years, but walk longer distances in
warm, dry years. This difference can dramatically affect
the number oflandbirds counted. Wilds (1980) pointed
out an interesting bias through time: birders avoid urban
areas; thus, when a circle becomes increasingly urban,
birders move their efforts to the less-developed parts of
the circle. Despite all these variations that can occur, in
fact most observers cover the same areas year after year
in essentially the same way. In summary, coverage varies
from year to year, but remains very similar.
Count circles may change location yet keep the same
name. The change may be as little as 1% or as much as
100%. Count circles may stay in the same place, but
report different latitude-longitude coordinates or a
different name. Deciding whether a similar location in
different years is the same location requires detective
work and individual judgment.
Expertise and Diligence ofCBC Participants
In a standardized survey, reported numbers should
be representative of what is in the study area on the
day of the survey. For this to be true for CBC's, birders
should know how to identify birds and should be
competent and diligent at recording numbers of
individuals. In a standardized survey, a coordinator
might screen observers for ability and train them until
they meet minimal criteria. In the CBC, beginners
participate often without training. However, beginners
are almost always assigned to parties with an
experienced birder. Expert birders are patchily
distributed throughout the country; some counts will
have many expert birders, others will have few or none.
Groups that lack expert birders will make errors in
identification; however, gross errors are weeded out
by the extensive editing system supervised by the
National Audubon Society. Editors in this system
include count coordinators, State editors, and national
editors. An interesting bias relating to identification
errors exists through time: birders are getting better at
identifying birds because field guides are getting
better, optical equipment is getting better, and new
identification tips are proliferating. Thus, there maybe
a built-in bias towards increase in numbers for species
that are (or were) even moderately difficult to identify.
We know that participants are more interested in
finding rare species than counting common species;
however, participants do count common species, even
if it is not always their primary interest.
Representativeness ofCBC Locations
If a goal of a survey is to extrapolate results to a larger
region than the area being surveyed, then the areas
included in the survey should be representative of the
AVIAN POPULATION TRENDS 7
areas not included. For this to be true, surveys should be
widely distributed in the area of interest and should
include habitats that are representative of nearby areas
that are not surveyed. Ideally, census-survey locations
should be assigned on a stratified basis, perhaps by
degree block of latitude-longitude, by physiographic
region, or by vegetation type; within the stratified units,
survey locations are often randomly chosen.
CBC locations are chosen by local count compilers. As
Fig. 3 shows, CBC's are not randomly distributed. They
are concentrated near cities; however, they avoid inner
cities. Thus, suburban areas are over-represented and
remote areas under-represented, although some
high-quality areas far from cities are sought out for CBC's.
Bystrak (1981) divided North America into a number
of· physiographic regions. Plants and birds should be
more similar to each other within those regions than
between them. I used an updated version of Bystrak's
regions to determine the percentage of coverage of each
region by CBC's, assuming that the entire CBC circle
was covered (Table 2). In the winters of 1982-83, only
five northern regions were covered less than 1% by
CBC's (Fig. 4).
Timing ofCounts
Birds are most sedentary during the breeding season
and least sedentary during spring and fall migration.
Thus, winter is the second best ~eason for consistent bird
counts. CBC's can occur during early winter from
15 December to 5 January. Bad weather within this
.. ~ ..... . .\
.. ..:...: . .s.. ..,... ~
-' .
......
.... :'
Fig. 3. A map of the locations of all Christmas Bird Counts in the 48 contiguous United States and southern C',anada submitted to
the National Audubon Society for the winter of 1982-83.
8 BIOLOGICAL REPORT 90(1)
period may cause movements ofbirds during the survey;
year-to-year differences in the severity of early-winter
weather may cause dramatic fluctuations in reported
numbers for some species. Bad weather on a local count
day may depress both observer effort and bird activity
(see references in Ferner 1984); however, counts are
rarely rescheduled due to bad weather. Luckily, many
CBC's occur at the same time each year.
Recommendationsfor Using CBC Data
Researchers who use CBC data should:
• be aware of potential identification problems for
their species of interest (e.g., accipiters [Daniels 1975],
Thayer's gull, Lams thayeri [Mark 1981]);
• be aware of potential counting biases (Arbib 1981;
Bock and Root 1981);
• be aware of biases in habitat coverage (Wilds 1980;
Bock and Root 1981);
• be aware of the possible effects of cold or wet
weather either before or during a CBC (see references
in Ferner 1984);
• decide whether to split or lump count locations that
have moved slightly between years;
• determine how effort affects the counts of their
species of interest (Butcher and McCulloch 1990);
• consider the use of reference species, especially
when individuals of a species might be encountered by
more than one type of effort (Raynor 1975; Bock and
Root 1981; Haney 1983); and
• use a number of CBC locations and years for any
study (Bock and Root 1981).
Why Use eRe Data?
The CBC is a birdwatching event, not a scientific event,
and always has been. Why then should scientists pay any
attention to the information collected during CBC's? The
major reason is that the CBC is one of only two surveys in
the world that collects information on relative abundance
over a large proportion of an entire continent for a large
proportion of that continent's species. The other survey
is, of course, the Breeding Bird Survey (BBS) of the U.S.
Fish and Wildlife Service and the Canadian Wildlife
Service. CBC data are useful because they complement
the BBS, which is conducted in early summer.
A number of species are encountered frequently on
CBC's that are rarely encountered on the BBS. These
include raptors (Table 3) and coastal and wetlands
species (Table 4). In 1977, 322 species were seen on 20
or more BBS routes (Robbins et al. 1986); in 1982-83,
362 species were seen at 20 or more CBC locations
(Table 5). A total of 451 species appear on one or both
lists. Based on an arbitrary standard, BBS data are
better for 241 species, and CBC data are better for
210 species (Table 5).
Ninety-nine species were seen on 100 or more BBS
routes and at 100 or more CBC locations; these species
can be used to compare population dynamics for the same
species from the two different surveys to see if similar
trends are produced from both data bases. Butcher and
Fuller (1986) studied population trends of seven species
using the two data bases. Six of the seven species showed
qualitatively similar trends. Only the eastern bluebird
(Sialia sialis) showed different trends on the two data
bases. The eastern bluebird is a weather-sensitive species
with widely varying population levels from year to year;
thus, it is inappropriate to try to summarize its population
dynamics with a single trend line.
Other sources of bird population information also
correlate well with CBC data. Butcher (1986) found that
the American black duck (Anas mbripes) population
trend derived from CBC data was similar to the trend
derived from the Mid-Winter Waterfowl Inventory of the
U.S. Fish and Wildlife Service. Robbins (1970, 1978) and
Robbins and Bystrak (1974) experimented with a
carefully controlled winter survey for 5years in Maryland.
They found that both population trends and yearly
population indices were similar, using data from their
experimental survey and nearby CBC's. Dunn (1986)
Table 2. Proportion ofphysiographic regions covered by CBC's. a
Coverage (%)
>10
1-10
0-1
1963-64
10
41
20
1982-83
a Locations of physiographic regions are shown in Fig. 4; the names of all physiographic regions are listed in Table 1.
The following are the six regions that are less than 1% covered by CBC's:
25 - Open Borcal Forest
29 - Closed Boreal Forest
63 - Fraser Plateau
67 - Cascade Mountains
68-Canadian Rock)' Mountains
99-Tundra.
AVIA"i POPULATIO~TRENDS 9
Fig. 4. This map shows the physiographic regions as determined for the Breeding Bird SUlvey of the U.S. Fish and Wildlife Service.
These regions were assigned by D. Bystrak, based primarily on Aldrich (1%3) and modified by Bureau of Agricultural
Economics (1933), Fenneman (1931, 1938), Kuchler (1%4), Erskine (1978), and other sources. The names ofthe physiographic
regions are listed in Table 1. The areas included within the broken hemylines are covered 1% or more by CBC circles (Table 2).
10 BIOLOGICAL REPORT 90(1)
Table 3. FrequellCY ofellcoullter ofraptors 011 the Christmas Bird COUllt alld the Breedillg Bird Survey.
Christmas Bird Breeding Bird
Count (1982--83) Survey (1977)
Total Total Total Total
~cies circles birds routes birds
Black vulture 229 11,330 102 551
Turkey vulture 384 38,189 490 3,271
Osprey 109 1,479 61 104
American swallow-tailed kite 0 0 4 11
Black-shouldered kite 92 1,818 25 60
Snail kite 7 4 1 7
Mississippi kite 0 0 38 152
Bald eagle
adult 428 4,308 12 20
immature 289 4,965
age unknown 54 768
Northern harrier 806 11,124 169 320
Sharp-shinned hawk 790 2,875 29 31
Cooper's hawk 663 1,908 53 62
Northern goshawk 292 445 5 5
Harris' hawk 28 269 12 35
Red-shouldered hawk 440 3,530 163 309
Broad-winged hawk 25 83 142 181
Swainson's hawk 8 15 124 321
White-tailed hawk 13 43 2 2
Zone-tailed hawk 1 1 2 2
Red-tailed hawk 1,206 33,786 588 1,287
Ferruginous hawk 126 427 21 37
Rough-legged hawk 610 3,491 1 1
Golden eagle
adult 186 516 41 54
immature 109 259
age unknown 48 108
Crested caracara 20 202 8 24
American kestrel 1,109 29,633 665 1,622
Merlin 255 475 10 11
Peregrine falcon 104 166 3 3
Gyrfalcon 21 29 0 0
Prairie falcon 211 516 19 19
compared year-to-year population changes of Ontario
feeder counts and CBC's and found comparable results
for 12 of 25 species. Species that attended feeders
infrequently and species that had relatively small
population changes from year to year showed no
comparable results.
Many other studies of population dynamics that
used CBC data verified long-term trends or dramatic
population eruptions that were known from other
sources. A number of species, especially insectivorous
passerines in the southeastern United States, were
shown using CBC data to be sensitive to severe winter
weather (James 1962, 1963); the weather sensitivity of
these species was confirmed by BBS data
(Robbins et a1. 1986).
In summary, for all its difficulties, the CBC has an
impressive record for producing useful analyses of the
population dynamics of North American birds. Taken
as a group, these studies prove that there is a mother
lode of useful information in the CBC data base.
Creation of a Computerized CBC
Data Base
The Cornell Laboratory of Ornithology completed
computerization of a 3D-year data base of CBC
information in 1990, which will greatly aid the analysis
of population dynamics of species that are frequently
encountered on CBC's (the 362 species of Table 5).
Trends from the BBS data base are currently being
AVIAN POPULATION TRENDS 11
Table 4. Frequency of encounters ofcoastal and wetlands species on Christmas Bird Count and the Breeding Bird
Survey.
Christmas Bird Breeding Bird
Count (1982--83) Survey (1977)
Total Total Total Total
Species circles birds routes birds
Red-throated loon 161 4,656 0 0
Common loon 384 9,771 114 321
Pied-billed grebe 621 13,803 76 157
Horned grebe 320 10,557 11 29
Red-necked grebe 112 2,708 13 47
Eared grebe 185 52,813 22 443
Western grebea 137 70,078 8 449
Double-crested cormorant 375 139,924 45 379
American bittern 123 411 108 316
Great blue heron 894 26,525 481 1,305
Great egret 242 19,565 83 561
Snowy egret 160 13,974 37 263
Little blue heron 102 5,295 97 645
Tricolored heron 97 4,751 18 109
Cattle egret 138 35,426 149 5,582
Green-backed heron 182 1,650 493 1,158
Black-crowned night-heron 226 8,530 51 202
Tundra swan 171 55,018 0 0
Mute swan 122 4,486 3 15
Snow goose (white) 207 973,524 0 0
Snow goose (blue) 118 245,815 0 0
Canada goose 825 1,315,990 92 1,621
Wood duck 410 7,860 160 475
Green-winged teal 484 170,066 41 135
American black duck 567 151,686 54 234
Mallard 1,156 1,577,194 438 5,470
Northern pintail 469 743,705 62 636
Blue-winged teal 168 9,292 150 1,313
Northern shoveler 349 154,089 53 320
Gadwall 530 75,585 60 642
American wigeon 526 381,199 52 282
Canvasback 381 120,135 24 122
Redhead 313 40,026 35 234
Ring-necked duck 523 45,729 9 25
Greater scaup 253 156,303 1 10
Lesser scaup 541 115,772 41 830
Oldsquaw 203 176,165 1 31
Black scoter 150 14,528 0 0
Surf scoter 157 62,552 1 20
White-winged scoter 192 42,401 2 80
Common gOldeneye 752 92,937 23 84
Barrow's goldeneye 106 6,688 6 23
Bufflehcad 628 77,522 13 70
Hooded merganser 508 10,649 8 31
Common merganser 623 119,841 36 97
Red-breasted merganser 408 67,616 6 12
Ruddy duck 413 154,634 3 549
Osprey 109 1,479 61 104
Bald eagle
adult 428 4,308 12 20
immature 289 4,965
age unknown 54 768
12 BIOLOGICAL REPORT 90(1)
Table 4. Continued.
Christmas Bird Breeding Bird
Count (1982-83) Survey (1977)
Total Total Total Total
Species circles birds routes birds
Virginia rail 192 1,063 17 24
Sora 138 956 75 285
Common moorhen 157 10,337 21 131
American coot 657 451,716 112 1,747
Black-bellied plover 173 26,095 0 0
Greater yellowlegs 220 4,446 6 24
Lesser yellowlegs 131 2,492 21 165
Willet 103 34,611 69 573
Spotted sandpiper 196 1,660 203 441
Ruddy turnstone 118 7,538 0 0
Sanderling 167 38,156 0 0
Western sandpiper 143 103,135 0 0
Least sandpiper 207 47,906 2 2
Dunlin 211 192,570 1 1
Long-billed dowitcher 105 20,139 0 0
Common snipe 559 9,004 249 1,394
American woodcock 140 481 36 65
Laughing gull 105 148,781 33 831
Bonaparte's gull 282 139,919 4 31
Ring-billed gull 758 829,705 110 2,929
California gull 123 88,128 32 1,156
Herring gull 725 893,756 129 3,529
Iceland gull 91 6,600 0 0
Kumlien's gull 17 85 0 0
Glaucous-winged gull 101 138,918 11 672
Glaucous gull 143 1,232 1 160
Great black-backed gull 289 104,018 35 296
Forster's tern 156 16,882 24 115
Belted kingfisher 1,027 11,478 462 718
Marsh wren 341 5,680 85 355
Swamp sparrow 617 24,661 278 811
" Includes Clark's grebe.
Table 5. Species freque/lt~y encountered on the Breeding Bird Survey and the Christmas Bird COUllt.
Breeding Bird Survey (data from 1977)
Encounters
20-49
50-99
100+
20+
Number of species
84
62
176
322
Number of species BBS > 1/2 x CBCa
48
45
148
241
Christmas Bird Count (data from winter 1982-83)
Encounters Number of species Numbcr of species CBC > 2 x BBSa
20-39
40-99
100-199
200+
20+
57
100
73
132
362
33
58
41
78
210
a This column is meant to indicate which species are better covered by the BBS and which by the Cnc. Because the nns is better standardized
than the cnc, a species is considered better covered by the cnc only when it is encountered twice as frequently on the cnc as on the BBS.
recalculated each year; thus, when the CBC analyses can
be done, we will have an unprecedented amount of
quantitative information on thc population status and
trends of North American birds. This information will
be of tremendous value for establishing priorities for
North American bird conservation~In addition, the data
will prove useful in improving our understanding of the
population biology of North American birds.
Acknowledgments
The Christmas Bird Count Population Project at
Cornel1 has been generously supported by J. KixMiI1er,
G. W. Knight, N. B. Delavan Foundation, E. I. du Pont
de Nemours and Company, Exxon Corporation, FMC
Corporation, IBM Corporation, Johnson Foundation
(Trust), MARS Inc., Merck Family Fund, Williams
Companies Foundation Inc., and many others. J. Lowe
helped with data improvements and summaries; he
prepared the figures. T. Engstrom, S. Droege, J. Trapp,
and an anonymous reviewer helped to improve the
manuscript. In addition, I would like to thank the many
CBCvolunteers who colIected, recorded, and edited the
data; the National Audubon Society and the staff of
American Birds who coordinate the CBC and publish
the results; and C. Bock and J. Shipman, who provided
computerized CBC data.
Literature Cited
Aldrich, J. W. 1%3. Life areas of North America. J. Wildl.
Manage. 27:530-531.
Arbib, R. S. 1981. The Christmas Bird Count: constructing an
"ideal model." Pages 30-33 in C. J. Ralph and J. M. Scott,
eds. Estimating numbers of terrestrial birds. Stud. Avian
BioI. 6.
Bock, C. E., and T. L. Root. 1981. The Christmas Bird Count
and avian ecology. Pages 17-23 in C. J. Ralph and J. M.
Scott, eds. Estimating numbers of terrestrial birds. Stud.
Avian BioI. 6.
Bureau of Agricultural Economics. 1933. Natural land use
areas of the United States. U.S. Dep. Agric., Bur. Agric.
Econ.
Butcher, G. S. 1986. Populations of black ducks and mallards
in winter, 1950-1983. Report to U.S. Fish and Wildlife
Service, WaShington, D.C.
Butcher, G. S., and M. R. Fuller. 1986. Bird populations in
winter and summer: an evaluation of the Christmas Bird
Count and a comparison with the Breeding Bird Survey.
Report to U.S. Fish and Wildlife Service, Washington, D.C.
AVlAo,-" POPULATIOS TREl'iDS 13
Butcher, G. S., and C. E. McCulloch. 1990. Intluence of
observer effort on the number of individual birds recorded
on Christmas Bird Counts. Pages 5-13 in J. R. Sauer and
S. Droege, eds. Survey designs and statistical methods for
the estimation of avian population trends. U.S. Fish WildI.
Serv., BioI. Rep. 90( 1). 166 pp.
Bystrak, D. 1981. The North American Breeding Bird Survey.
Pages 34-41 in C. J. Ralph and J. M. Scott, eds. Estimating
numbers of terrestrial birds. Stud. Avian BioI. 6.
Chapman, F. M. 1900. A Christmas bird-census. Bird-Lore
2:192.
Daniels, G. G. 1975. An inquiry into Christmas Bird Count
Accipiter reports. Am. Birds 29:634-637.
Dunn, E. H. 1986. Feeder counts and winter bird population
trends. Am. Birds 40:61--66.
Erskine, A. J. 1978. The first ten years of the cooperative
Breeding Bird Survey in Canada. can. Wildl. Servo Rep.
Ser. 42. 61 pp.
Fenneman, N. M. 1931. Physiography of Western United
States. McGraw-Hill, New York.
Fenneman, N. M. 1938. Physiography of Eastern United
States. McGraw-Hill, New York.
Ferner, J. W. 1984. The Audubon Christmas Count as a
method of monitoring nongame bird populations. Pages
332-341 in W. C. McComb, ed. Proceedings of the
Workshop on Management of Nongame Species and
Ecological Communities, University of Kentucky,
Lexington.
Haney, J. C. 1983. The usc of reference species as a technique
in evaluating Christmas Bird Count data. Am. Birds
37:363-364.
James, D. 1962. The changing seasons. Audubon Field Notes
16:306-311.
James, D. 1963. The changing seasons. Audubon Field Notes
17:300-304.
Kuchler, A. W. 1964. Potential natural vegetation of the
conterminous United States: manual to accompany the
map. American Geographical Society, New York.
Mark, D. M. 1981. Thayer's gulls from western Christmas Bird
Counts: a cautionary note. Am. Birds 35:898-900.
Raynor, G. S. 1975. Techniques for evaluating and analyzing
Christmas Bird Count data. Am. Birds 29:626-633.
Robbins, C. S. 1970. Winter bird survey technique tested in
Maryland. Md. Birdlife 26: 11-20.
Robbins, C. S. 1978. Census techniques for forest birds. Pages
142-163 in Proceedings of the Management of Southern
Forests for Nongame Birds Workshop, Atlanta, Ga.
U.S. For. Serv., Gen. Tech. Rep. SE-14.
Robbins, C. S., and D. Bystrak. 1974. The winter bird survey
of central Maryland, USA. Acta Ornithol. 14:254-271.
Robbins, C. S., D. Bystrak, and P. H. Geis.~ler. 1986. The
Breeding Bird Survey: its first fifteen years, 1965-1979.
U.S. Fish Wild!. Serv., Resour. Pub!. 157. 196 pp.
Wilds, C. 1980. The Washington, D.C., Christmas Bird Count
as an indicator of environmental change. At I. Nat.
33: 10-11.
14 BIOLOGICAL REPORT 90(1)
Description of the Wisconsin Checklist Project
by
Stanley A. Temple and John R. Cary
Department ofWildlife Ecology
University ofWisconsin
Madison, Wisconsin 53706
Introduction
A birding checklist is a simple record of the species
of birds that a field observer found at a particular place
and time. This form of record-keeping has long been a
standard practice among American bird watchers
(Hickey 1943). Checklists have traditionally provided
information on geographic ranges and seasonal
occurrences of migrants. Only recently have checklists
been used to monitor temporal and spatial variation in
the abundance of birds (Temple and Temple 1976). In
theory, when manyobservers in an area keep concurrent
checklist records, their checklists will differ because of
differences in the abundance of species and in observer
effort. Some species will be reported on every observer's
checklist, whereas other species will occur on those of
only a few observers. The reporting frequency for each
species in an area can be calculated as the percentage
of checklists on which the species has been reported.
Reporting frequencies are assumed to be primarily a
function of each species' relative abundance. A high
reporting frequency indicates that a species is relatively
abundant, whereas a low reporting frequency indicates
that a species is relatively uncommon.
To test whether or not an analysis of systematically
maintained checklist records could actually be used to
monitor variations in the abundance and distribution of
bird species, the Wisconsin Society for Ornithology
(WSO), under the guidance of S. A. Temple, began the
Wisconsin Checklist Project in 1982 (Temple 1982).
This paper described the details of how the Wisconsin
Checklist Project operated and summarized briefly
some of the project's major accomplishments.
Data Collection
The first and most important requirement for using
checklist records to monitor bird populations is a
group of reliable and competent field observers who
are willing to maintain systematic records. In
Wisconsin, many such skilled individuals are members
of the WSO, an organization with a long tradition of
volunteer participation in ornithological research
programs. In response to the call for participants
(Temple 1982), 431 of the approximately 1,200 WSO
members have volunteered to take part in the
Wisconsin Checklist Project. Of these volunteers, 257
have been regular contributors to the project; some
have submitted records for every reporting period
since the project began.
We asked participants to keep careful records of the
bird species that they detected during each week of the
year and in which county they had observed them. We
provided them with special forms on which to record
information for each week (Fig. 1). These forms are
designed so that information recorded on them can be
read directly by an optical scanning device. This
innovation made it possible for us to transfer large
amounts of information rapidly and easily from the
forms to magnetic tapes and disks that could then be
used by computers.
On the weekly reporting forms, participants
identified themselves and provided information on the
county in which they birded during the recording week,
the date of the Sunday that began the recording week,
and the level of intensity with which they searched for
birds. They then indicated which of 265 bird species they
had detected during the week by filling in the "bubble"
adjacent to the bird's name. They were not asked to
provide any assessment, either objective or subjecti~e,
of the abundance of the birds they detected. Completed
forms were submitted for annual analysis. Since the
project began in 1982 we have received more than 30,000
weekly checklist forms - about 6,000 forms each year.
Analysis ofRecords
After the forms were scanned and the information
converted to a computer-accessible format, we
checked records for accuracy. We looked for
AVIAN POPULATION TRENDS 15
•• Illes Trans-Opl1C MP 18·17665·54 •••
~"'Jl- ,
-J>---Z-LARK
SWIFT
KINGFISHER
FLYCATCHERS
NIGHTHAWKS
WOODPECKERS
o HOrllf'r!
OBdlf~d
oRcd-hciICj\:cI oReel-heilled o Yel·11 S,lPSlKkel o DOV'I;'11','
QHillrY
ON Flicker o Plle~1\cd
oOlive-sided oE V'vnod-Ppwep
O':'dl(l\;\-lJpIIiC(1 o ACilcll:111 o Alder
OWlllm'
QL','cJSl oE,lstc>rn Phoetw oGre;l! C'('slee!
OE:: KlnqhlHI
OCUT1l1lK1I' o Whip-poor-vl,llI
HUMMINGBIRD oRlilly·tl1lu;ltecl
GULLS
TERNS
DOVES
CUCKOOS
ORo"k
OMmjfl)'11~)
OC<tSPI<Hl
OC0l11111011
OFurslCr'~
OSlclt'\..
o BI;lCk-billpd o Yr:llov. -hilled
OFrdrlklins o BOlldP;U It' c,
ORlng-tJllll>cI
OH\'mI1~1
OGLIlI'.()\I"
COOT
CRANE
PLOVERS
MOORHEN
SANDPIPERS
OClllnl1l0n
o BI,lC~-iJpllll~(j ol_~~<.;<.;,,, l-:;()I(I('I'
OSt'll111)..,lfn;-l\l'l!
OKilld\'"
olil Ydl()wle~J~ o Lf'"o.;(~r Yellowlc(:o.;
OSoj't;lI'i
OVVtllcl
OSOO1tC(l o LJpl<i11(' OWLS oHllCls\JIJ(dI1 0 (om BDrl' o M,IrI)I,~(i Gml' 0 E SCf('i~ch oRu(id.,T lIIllo.;lUr1, 0 Glr:dl Horllf'(l o H"II Kllll' 0 Snuw-,
_,_~_~__.._O S,lIl(It,!1·2 ~0~B~dl-'lr-',,(I_
BG0@@@eG@@@@@@e0GG@@GG@@@00
18 G 0@@@eG@@@@@@e0GG@@0e@@@00 1'0
!8G0@@@GG@@@@@@@0GG@@GG@@@00 .~
i8G0@@@GG@@@@@@@0GG@@GG@@@00 I::!
18G0@@@GG@@@@@@e0GG@@GG@@@00r--­,8@
0@@@GG@@@@@@e0GG@@GG@@@0Oj I
1
8G0@@@GG@@@@@@e0GG@@GG@@@00l
8G0@@@eG@@@@@@e0GG@@00@@@00~-i,
18G0@@@GG@@@@@@e0GG@@GG@@@00L
GG0@@@eG@@@@@@e0GG@@0G@@@00
1
8G0@@@GG@@(3)@@@e0GG@@GG@@@001 •
GG0@@@eG@@@@@@e08G@@0G@@@00 I
HAWKS
PHEASANT
PARTRIDGE
OM!'llll\ o Pt'I~'~jlllli
OO:-,prt"v
OS,ild E'-lui,
ONllIUH'ln Hilrrl("
OSlldrp-SiHl1Iwr:
OCI\l11)"I';-,
0,
OR('d-c,h(HII{I!'If"l1
OSI(,.lO-,·II\(I\'(1
OR,'d-ldl led
OHullql !','I!()I>I: o
o
zco
I
I
I
III
I
I
I
IIIIII
II
I
I
I
I -' LOONS DUCKS (con'tl GROUSE SANDPIPERS (con'tl OWLS Icon'tl I f--'O""R-"-rl.=tI="c="'.:,,e='-1---l-"""---"==:..c:='"---+--OA"'R-,-'fl-e::::'I===----L--;O~s::::e=n;c.:'p'-,'='n'::""-'C"'d=.'.'--.l-?O"'G"-re:':,,'::, =G''"'':':y="'--~
I 0 Common 0 Gr Pr;llflP (luck I'" 0 Led'i! 0 Llll1(1-e,lred
I - 0 Rlll(j·neckec! 0 S!1;II!l·tilde(1 0 Whllt~'r\!llll)(~(1 0 Shorl-I'iln;(1
I _I-i------,G::R::E==B=E=S---' oGn'dlE>' S'~<llln OSillrd,<; Or'\,] S:l\'\':.l1et
I -: 0 Plcd-billed 0 Le$sP[ 51"dUIl TURKEY 0 Per:lOr,\1
t -II 0 Horned OOklsqlliJ\,. 0 V\illd 0 Dunilll
I _I 0 Red-necked 0 BiclCk Scnter 0 Sidt
I 0 Sur! SCUll'! BOBWHITE 0 S -1)111 D(lv\'lt(~hu I-\-\---,""C"'O::.R::.:M::.:O::.:R.::.A::::N::::T:.:S'---_-' OWhllt'-wlllqecl SC01('1 oNnitilHTl OL-lJill DO\/\"11Cht'l
1--- oDoubl,,-crested OCI1111 Gtlld~;n('\'" Oe()ll1 SnIlJ(;
I -, 0 Bldtll'll(~d(i RAILS 0 Am WOOl!cn,:":
I _\-!---=-BI=T=T=E=-RN=S----, 0 H(lIl{J{'11 M!~rqilnsl'r 0 VlrqHlid 0 Wilson's PllilliJrCi!Jt
I -i 0 ArnenCiln 0 ((1111111(111 1\·1erq'Il)S!'1 0 Surd 0 Rccl-Il Phdl:JrOpL
I 0 Leosl 0 Rc(H,r Ml'f(l;mScr
i oRuddy
I -I HERONS
I-\-'-'O""G-,e-,;l-'-B-',,,-,,----+----:-:V"'U"'LT::U"'R::E=----,
I -' oGrC<ll Eqrel o Turkl"'v
I 0 Cillllr: Eqrel
1 0 Green-bJcked
I OSlcl(:k-n Nlqht-
I 0 Yellm'_-(I Nlqht
II
-1-1-,~-=S::::W:.:.A.::.N.:.:S,-__
I - o Tundr"
I OMuu·
II -_r-!---'CGccE-ES.,-E----
lOs,,,,,..,
I OCH1drl;,
II -_r-!---=D"'UC:C"'K:::S---
I OWCH)d
I - 0 Gre~n-\'\'Inq!'d T! ill
lOAm SliWk
I oMdllard
I O'\lnnlwfrl Plnl,nl
I OSIL!(:-v','II"1()<'{11(.,,1
I Oi'\!lHthprl'Sh()Vt,it'l
I -' oGdllv,;lIl o All1Pf!ldl:' 1IV11If'(JI'
Fig. 1. A sample of the optically-scannable checklist form on which volunteers recorded their observations.
16 BIOLOGICAL REPORT 90(1)
misspellings of an observer's name, erroneous dates
for the start of the recording week, nonexistent county
codes, and highly improbable reports of a bird species
outside o! its normal range or season of occurrence in
the State. Errors of the last type were of most concern
to us, and when they were found, we deleted them from
subsequent analysis. Over the first 5 years of the
project we caught 78 such suspicious records. Most of
those seemed to have been simply the result of a
participant accidentally filling in the wrong bubble on
the form. For example, we looked on with suspicion
and deleted a July record of a gray jay (Perisoreus
canadensis) from a southern Wisconsin county (far
beyond the jay's normal range), especially because
during that week the observer had failed to report a
blue jay (Cyanocitta cristata; the region's common jay).
The two jay species appear next to one another on the
checklist form, so we suspected a simple slip of the
pen. To double-check the accuracy of the data, we also
prepared yearly summaries of each participant's
reports and returned these summaries to the
participant for review. This review process allowed the
original observer to help us further in detecting and
correcting erroneous records.
Our analyses of the edited data followed several
paths. In order to detect seasonal variation in
abundance of birds in the State, we calculated weekly
reporting frequencies for each species. Results of this
type of analysis have been presented by Temple and
Temple (1984) and Temple and Cary (1987a). An
example is shown in Fig. 2.
In order to detect geographical variation in the
abundance of birds, we divided the State into 43
Northern Wisconsin
100%
50%
o F M A M 0 0 A SON 0
Southern Wisconsin
100%
50%
o F M A M 0 0 A SON 0
Fig. 2. An example of the seasonal variation in abundance of the
ruby-crowned kinglet (Regulus satrapa) keep in Wisconsin,
as revealed by an analysis of checklist records.
regions, each of which included an adequate number
of observers and checklists. We calculated each
species' reporting frequency in each region over a
specified period, such as the breeding season or the
entire year. Temple and Temple (1987) and Temple
and Cary (1987b) presented some preliminary results
of these types of analyses. An expanded analysis of
trends in relative abundance revealed by the checklist
records is included elsewhere in this volume (Temple
and Cary 1990). An example is shown in Fig. 3.
Important Assumptions
The Wisconsin Checklist Project is based on a
number of assumptions that underlie the use of the data
to detect variations in abundance.
• Volunteer observers are accurate in their reporting
habits, following instructions rigidly and doing their best
to optimize their efforts to detect birds.
• Reporting frequencies are a function of the relative
abundance of birds. Temple and Cary (1990) address
this assumption later in this volume.
• Reporting frequencies are more affected by
changes in abundance than they are by other factors
(e.g., the behaviors of birds or observers) because of the
dynamic mixing of these other effects.
• At least over the short term, the efforts of the
observers remain relatively similar between years (i.e.,
their birding habits and skills change slowly).
Dabsent
~ < 2.7 %
~2.7-57 %
~> 57 %
Fig. 3. An example of the geographic pattern of relative
abundance of the common raven (Corvus corm:) keep, as
revealed by an analysis of checklist records; shading
categories portray reporting frequencies.
Conclusions
Our experiences with the Wisconsin Checklist
Project convince us that the approach has considerable
merit. The technique is readily accepted by volunteer
cooperators, and it produces interpretable data
reflecting seasonal, geographic, and year-to-year
variations in bird populations. Temple and Cary (1987a)
produced a book that presented many of the results of
the project in a format that has proved useful and
appealing to bird watchers, thus demonstrating to
volunteers that their efforts have been worthwhile.
Acknowledgments
We are grateful to the 431 members of the Wisconsin
Society for Ornithology who have submitted checklist
records during the first 5 years of the Wisconsin
Checklist Project. A. J. Temple kept track of these
volunteers and the nearly 30,000 checklists they
submitted for analysis. The Wisconsin Checklist Project
received financial support from the A. W. Schorger
Fund of the Department of Wildlife Ecology, University
of Wisconsin - Madison.
AVIAN POPULATION TRENDS 17
Literature Cited
Hickey, J. J. 1943. Aguide to bird watching. Oxford Univcrsity
Press, London, U.K. 262 pp.
Temple, S. A 1982.-A Wisconsin bird survey based on field
checklist information: a WSO research project. Passenger
Pigeon 44:56-60.
Temple, S. A, and J. R. Cary. 1987a. Wisconsin birds: a
seasonal and geographical guide. University of Wisconsin
Press, Madison. 364 pp.
Temple, S. A, and J. R. Cary. 1987b. Climatic effects on
year-to-year variations in migration phenology: a WSO
research project. Passenger Pigeon 49:70-75.
Temple, S. A, and J. R. Cary. 1990. Using checklist records to
reveal trends in bird populations. Pages 98-104 in J. R.
Sauer and S. Droege, eds. Survey designs and statistical
methods for the evaluation ofavian population trends. U.S.
Fish Wild\. Serv., Bioi Rep. 90(1).166 pp.
Temple, S. A, and A J. Temple. 1984. Results of using
checklist records to monitor Wisconsin birds: a WSO
research project. Passenger Pigeon 46:61-70.
Temple, S. A., and A. J. Temple. 1986a. Geographic
distributions and patterns of relative abundance of
Wisconsin birds: a WSO research project. Passenger
Pigeon 48:58-68.
Temple, S. A, and A J. Temple. 1986b. Year-to-year changes
in the abundance of Wisconsin birds: results of the WSO
checklist project. Passenger Pigeon 48: 158-162.
Temple, S. A, and B. L. Temple. 1976. Avian population
trends in central New York State, 1935-72. Bird Banding
47:238-257.
18 BIOLOGICAI_ REPORT 90(1)
Use of Breeding Bird Atlases to Monitor Population Change
by
Chandler S. Robbins
U.S. Fish and Wildlife Service
Patuxent Wildlife Research Center
Laurel, Maryland 20708
Introduction
Breeding Bird Atlases are cooperative projects
designed to map the breeding distributions of all species
of birds within a prescribed area, such as a county, a
Nation, or even a continent, within a defined period of
years. Mapping is done using a grid of usually 5 or 10
km, and either every block in the grid or a random or
systematic sample of blocks is visited, with the intent of
recording presence or absence of each species during
the nesting season. Primary emphasis is on determining
if a species is present and assigning the highest category
of nesting evidence: Possible, Probable, or Confirmed.
The categories of evidence under each of these three
terms were defined in the first major grid-based avian
atlas (Sharrock 1976), endorsed with minor alterations
by the European Ornithological Atlas Committee in
1974, and accepted, again with very minor alterations,
by the Northeastern Breeding Bird Atlas Conference in
Woodstock, Vermont, in 1981, and by the North
American Ornithological Atlas Committee in 1987.
The original purposes of Breeding Bird Atlases were
to map breeding distribution on a fine scale in such a
way that future expansions and contractions of range
could be documented. Additional benefits identified
with the early atlases were: discovery of ecologically rich
or unique habitats that deserve preservation, location of
new sites for rare and endangered species,
accumulation of a data base for further research, and
provision of a useful and exciting project that would
involve amateur birders during a season when they often
are not heavily involved with other birding activities.
History
The concept of grid-based natural history atlases
originated in the British Isles with the publication in
1962 of Perring and Walters' atlas of the British flora,
which used the national lO-km grid. The Biological
Records Centre, Abbots Ripton, England, subsequently
promoted the systematic mapping of flora and fauna
throughout the British Isles and assisted many other
European countries in initiating atlas projects.
After a trial avian atlas project in three British
counties (Lord and Munns 1970), the British Trust for
Ornithology undertook a 5-year atlas ofbreeding birds
in Britain and Ireland (Sharrock 1976). This was a
tremendous effort that included visits to all 3,862
10-km blocks in the British Isles and involved more
than 10,000 participants. Not only did this atlas
provide direct comparisons with the botanical atlas,
but also the bird atlas came with 12 habitat and climatic
overlays that greatly assisted interpretation of the bird
distribution maps.
The European Ornithological Atlas Committee was
formed in 1971 to standardize methodology and provide
assistance to other Nations. This committee adopted
standardized codes in 1972. During the next decade
national atlases were published for France, Denmark,
West Germany, Netherlands, and Switzerland, as well
as regional atlases for parts of England, Germany, and
Spain. At least two African atlases have been published
(Cyrus and Robson 1980; Nikolaus 1987) as well as a
national atlas for New Zealand (Bull et a1. 1985), a
national atlas for Australia (Blakers et a1. 1984), and
some more-detailed regional atlases for parts of
Australia (Thomas 1979; Emerson et a1. 1987).
Atlasing in the United States began with a
preliminary mapping of bird distribution in Montana by
its 47 one-degree blocks oflatitude and longitude (Skaar
1975). In the same year R. Stewart (1975) published a
much more detailed book on North Dakota breeding
birds based on townships (6 X 6 mi = about 9.7-km
grid); although he did not use atlas categories and
procedures, he personally visited nearly all townships in
the State and mapped every breeding species by
township. The first published North American atlas to
use the standard atlas codes and procedures was of two
adjacent counties in Maryland (Klimkiewicz and Solem
1978). Currently, State and Provincial atlases have been
published for Maine (Adamus 1988), New York
(Andrle and Carroll 1988), Ontario (Cadman et al.
1988), and Vermont (Laughlin and Kibbe 1985). Atlas
projects are in progress or completed for several
Western States, all Canadian Provinces, and for all
States east of the Mississippi River except Mississippi,
Alabama, South Carolina, and Wisconsin. For more
details on the history of atlasing see Robbins (1982).
Procedures
Organization and Recording Forms
Most atlas projects are sponsored by a State or
provincial ornithological or Audubon society in
collaboration with the conservation department. In
most States there is a full-time salaried State
coordinator for the duration of the fieldwork and
manuscript preparation. A network of volunteer county
coordinators solicits, trains, and encourages
cooperators as work progresses.
Participants are supplied with a map of their block
and a handbook. The handbook contains instructions,
explains codes, and often includes a table of "safe dates"
during which one can assume birds of that species are
on nesting territory. Definitions for all the codes are
given in Sharrock (1976), Laughlin (1982), and the North
American Atlas Handbook (Smith, in preparation).
Newsletters are mailed to participants once or twice
each year to maintain their interest by reporting
progress, providing helpful suggestions, and cautioning
against common errors.
Forms vary from project to project because each
State or Province has its own list of breeding species.
The field cards have columns for the Possible, Probable,
and Confirmed codes. Generally there is an annual
summary sheet that is appropriately coded for data
entry. Some States and Provinces have additional
columns for numerical estimates, dates of observations,
or quarter-block designation.
Grid Size
The original lO-km grid of the British Isles was
adopted in much of western Europe except in the small
countries, which used a 5-km grid, and in countries such
as France, Spain, and Portugal, which had national
topographic maps to their own unique" scales. Canada
adopted a lO-km grid, but most of the eastern United
States has used a close approximation to a 5-km grid. In
many northeastern States (Connecticut, Delaware,
Maryland, Massachusetts, New York, Pennsylvania, and
Rhode Island) every 5-km block is visited. In others
(Illinois, New Hampshire, Vermont, Virginia, West
Virginia, and many others), a random or systematic
AVIAN POPULATION TRENDS 19
sample of blocks is visited. In addition, some other
blocks are targeted for special biological or politicaI
reasons, but this special set is not included in the
statistical summary of results. A few States (Maine and
Texas) are using entire 7.5-min topographic maps as
their grid; but using such a coarse scale will make it
difficult to detect future changes. Even if 90% of the
habitat in a block were to disappear, the remaining 10%
probably would retain at least one pair of most or all of
the original species. For this reason, the North American
Breeding BirdA tlas Handbook recommends a 5-km grid
for the United States, even if this means covering only a
sample of atlas blocks instead of all of them. A random
or systematic sample of 2.5-km quarter-blocks is
recommended to enhance opportunities of detecting
future changes or to correlate bird distribution with
other environmental variables. Quarter-blocks are
discussed further in connection with applications.
Field Techniques
Participants are urged to study their topographic
map and to visit all habitats that occur in their block.
They are urged to concentrate their fieldwork in the
early morning when most species reach their peak of
activity. Participants also are encouraged to make some
nocturnal trips for species that are most active at night.
Initial efforts are to record as many species as possible.
On subsequent trips an attempt is made to upgrade each
species from Possible to Probable status or from
Probable to Confirmed. Most observers continue to visit
their block until additional trips provide few new species
and few upgrades of status.
Observers are encouraged to keep working within
a block until they have found at least 75% of the
expected species. Coordinators in many States try to
estimate the number of species that occur in each
block, or at least an average number for the State or
Province. Some States have a goal of getting
confirmation for 50% of the species recorded in each
block; other States have put more emphasis on
Probable records and try to get 75% of the species into
Probable or Confirmed status. For much of the East
there has been a goal of 70 or 75 spec'ies, with no more
than a quarter of these in Possible status.
Some atlas projects have set a minimum goal of 16 h
of coverage. Because of differences in observer
expertise, either a species goal or reaching the limit of
one's competence (failing to add other species or to
achieve additional upgrades) seems more realistic.
Numerical E5timates
There is no compulsory procedure for estimating the
number of birds of each species within atlas blocks. The
20 BIOLOGICAL REPORT 90(1)
North American Breeding Bird Atlas Handbook (Smith, in
preparation) lists various methods that have been used.
For the United States and southern Canada, Breeding
Bird Survey(BBS) data for the more common species give
index values by physiographic regions within each State
and Province. These indices should provide better
comparisons with future years than estimates of actual
numbers made by unskilled amateurs.
Miniroutes (Bystrak 1980) were developed
specifically for adding a quantitative dimension to atlas
studies. These are abbreviated BBS routes of 15 to 25
standardized 3-min counts at half-mile intervals along
secondary roads. The counts must be made by
experienced observers who are familiar with songs and
calls of all species likely to be encountered. When
miniroutes are run in all blocks or in a specified sample
of atlas blocks, they can provide an index of either
abundance or distribution (depending on whether birds
are counted or just checked for presence and absence).
For a discussion of application of miniroutes see
Robbins and Dowell (1986).
The atlases of France (Yeatman 1976) and Ontario
(Cadman et al. 1988) have used observer estimates of
number of pairs of each species in each block. These
estimates are made in powers of 10: 1, 2-10, 11-100,
101-1,000, and so forth. The estimates required
considerable interaction between coordinators and
observers because of a strong tendency to
underestimate the number of birds in an atlas block.
The British, in their winter atlas (Lack 1986),
successfully used actual counts obtained per 6 h of
fieldwork in winter, and these counts were adjusted by
use of a conspicuousness factor that was generated for
each species. So far, no similar procedure has been tried
in North America.
Editing and Computerization
Records are submitted annually at the close of the
breeding season, through the county coordinators to the
atlas coordinator, for critical review before data entry.
After computer editing, listings and maps are prepared
and the data reviewed again at various levels. After
fieldwork is completed, species accounts to accompany
the maps are written by selected authors.
Standardization
General procedures, including codes and
terminology, have been fairly well standardized from the
beginning. With a few exceptions the grid used has been
close to 10 km or 5 km, depending largely on the size of
the country, State, or Province. However, it has not been
feasible to adopt the Universal Transverse Mercator
(UTM) grid worldwide, as some had hoped. The
number of years of fieldwork has usually been 5 or 6.
So far, numerical treatment has not been
standardized, In some countries extensive bird
population information from other sources, such as
counts of colonial waterbirds or breeding densities by
habitats, has been incorporated into the atlas book. At
the other extreme, some atlases have made no attempt
to present numerical data beyond the percentage of
blocks with each species.
The text accompanying the atlas maps differs widely
in scope and content, ranging from no species text at all
in the Maine atlas to major State bird books that include
all species (not just breeding birds) in Delaware and
Massachusetts (in various stages of compilation).
Applications
Mapping Range Expansions and Contractions
This continues to be one of the major objectives of
atlas programs. Atlas work already has demonstrated
the potential for showing changes in breeding
distribution at the county as well as the State level. In
Maryland, for example, the 198J-87 atlas showed major
changes in many species from the range maps published
by Stewart and Robbins (1958), These changes included
more than doubling of the Maryland breeding range of
the tree swallow (Tachycineta bic%r), fish crow (COl'\lUS
ossifragus), golden-crowned kinglet (Regu/us satrapa),
veery (Catharus fuscescens), Nashville warbler
(Vennivora mjicapilla), blue-winged warbler (V. pinus),
and swamp sparrow (Me/ospiza georgiana); substantial
increases in breeding range of the mallard (Anas
p/atyrhynchos), black vulture (Coragyps atratus), cliff
swallow (Himndo pyrrhonota), and blue grosbeak
(Guiraca caem/ea); and extirpation of the yellow-bellied
sapsucker (Sphyrapicus varius) , Bewick's wren
(Thryomanes bewickii), Bachman's sparrow (Aimophi/a
aestiva/is), and the native peregrine falcon (Fa/co
peregrinus). In addition, the Canada goose (Brama
canadensis) and the house finch (Carpodacus
mexicanus) , unknown as Maryland breeders in the
1950's, now nest in almost every county.
Detecting and Moniton'ng Population Changes
The BBS does a great job of monitoring the common
roadside species by major regions of the continent, but
it cannot be depended on at the county level, where
sample sizes are too small,
The two atlases of Howard County, Maryland, have
demonstrated that even over a period as short as 10
years (1973-75 to 1983-87) many changes in bird
distribution and abundance occur, and these can easily
be detected with a quarter-block (2.5-km) grid, even
without the use of numerical counts. It is hard to obtain
a meaningful estimate of the number of hours spent on
atlas fieldwork, especially when atlasers reside in the
block to which they are assigned, or make casual
observations on the way to work each day. Thus, total
hours reported spent on atlasing is not as good a basis
for comparison over a period of years as the sum of the
bird lists for all atlas blocks in the county. In the Howard
County example, the total number of records (sum of
the species lists) for the 136 quarter-blocks in 1973-75
was 8,297, compared to 9,683 for 1983-87, a 16.7%
increase. This increase in efficiency of coverage (1.167)
was multiplied by the number of blocks in which a
species was recorded during the first atlas period to
obtain an estimate of the expected number of blocks for
the second period, assuming no change in population.
A major departure from this estimate indicates an
increase or decrease in the local population of the
species in question. Thirty-four Howard County species
showed significant changes by chi-square tests
(P < 0.05); a few of these are shown in the Table. The
entire list appears in Acta Zoologica Fe/llzica (Robbins
et al. 1989).
By breaking a county into regions, or by combining
data from several nearby counties, one can apply
nonparametric statistics and assign confidence limits to
the probability of change. Use of miniroutes, which is
recommended in the N011h Americall Breedillg Bird
Atlas Halldbook (Smith, in preparation), will further
AVIA.1IJ POPULATION TRENDS 21
increase the potential for detecting changes within a
State, Province, or county.
OtherApplications
There are many other applications of atlas data that
were not conceived when atlas programs began. These
include:
• documenting the effects of habitat fragmentation,
• defining boundaries of ecological regions (bird
districts) on the basis of their bird populations,
• land use planning so as to preserve areas of special
conservation value,
• detailed correlation of bird distribution with forest
cover types and other forms of land use,
• compilation of automated data banks that can be
used for a multitude of research and conservation
applications,
• providing an impartial means of defining rarity, and
• combining data from many adjacent States and
Provinces to obtain a larger perspective on current
population status and on changes in distribution and
abundance.
The Future
Atlasing has proven to be extremely popular among
tens of thousands of amateurs, who give freely of their
time and expertise. Not only are many of the most
sophisticated amateurs in North America contributing
their talents, but also they are training countless
Table. Challges ill lIumber of quarter-blocks ill Howard COUllty, Marylalld, where selected species were detected ill
1973-75 alld 1983-87. Expected values for 1983-87 are 1.167 times the 1973-75 values, represellting the increase
ill totallllll1zber of records received.
1973-75 1983-87 1983-87
Species Observed Expected Observed
Increases
Canada goose (Branta canadensis) 8 9 46
Mallard (Anas platyrhynchos) 49 57 86
Black vulture (Coragyps atratllS) 18 21 73
Pileated woodpecker (DryOCOPllS pileatllS) 29 34 86
Cliff swallow (Hinmdo pyrrhonota) 6 7 20
Hooded warbler (Wilsonia citrina) 43 50 74
House finch (CarpodaClls mexicanlls) 0 0 129
Decreases
American black duck (Anas rnbripes) 10 12 3
Whip-poor-will (Caprimulgus vocifems) 30 35 21
Horned lark (Eremophila alpestris) 47 55 18
Bank swallow (Riparia riparia) 9 11 2
Vesper sparrow (Pooecetes graminells) 51 60 22
Grasshopper sparrow (Ammodramus savannarnm) 103 121 78
Eastern meadowlark (StIlmella magna) 126 147 103
Change
(%)
+411%
+ 51%
+248%
+ 153%
+186%
+ 48%
-75%
-40%
-(,7%
-82%
-(,3%
-36%
-30%
22 BIOLOGICAL REPORT 90(1)
thousands in the techniques ofatlasing and are diverting
the attention of many birders into serious ornithological
data gathering. The rather crude atlasing of today will
probably be followed by more sophisticated atlas
endeavors in the future, with a more quantitative
approach. I expect atlases of the next century to be as
much advanced over today's atlas endeavors as today's
Christmas Bird Counts are over the initial published
Christmas Bird Censuses of the early 1900's.
Acknowledgments
I thank the members of the Howard County Chapter
of the Maryland Ornithological Society, especially D. C.
Dupree, J. Farrell, M. K. Klimkiewicz, and J. K. Solem,
for the intensive atlas coverage of their county during
1973--75 and 1983-87. I also thank S. Droege, J. R.
Sauer, and C. R. Smith for their constructive comments
on the manuscript.
Literature Cited
Adamus, P. R. 1988. Atlas of the breeding birds in Maine,
1978-1983. Maine Department of Inland Fisheries and
Wildlife, Augusta. 366 pp.
Andrle, R. F., and J. R. Carrol!. 1988. The atlas of breeding
birds in New York State. Cornell Laboratory of
Ornithology, Ithaca, N.Y. 551 pp.
Blakers, M., S. J. J. F. Davies, and P. N. Reilly. 1984. The atlas
of Australian birds. Royal Australasian Ornithologists
Union. 738 pp.
Bull,P. c., P. D. Gaze, and c.J. R. Robertson. 1985. The atlas
of bird distribution in New Zealand. Ornithological Society
of New Zealand, Wellington North. 2% pp.
Bystrak, D. 1980. Application ofminiroutes to bird population
studies. Md. Birdlife 36:131-138.
Cadman, M. D., P. F. J. Eagles, and F. M. Helleiner. 1987.
Atlas of the breeding birds of Ontario. University of
Waterloo Press, Waterloo, Ont. 617 pp.
Cyrus, D., and N. Robson. 1980. Bird atlas of Nata!. University
of Natal Press, Pietermaritzburg. 320 pp.
Emerson, W. B., C. M. Beardsell, R. I. Normand, and R. H.
Lyon. 1987. Atlas of Victoria birds. Melbourne, Australia.
271 pp.
K1imkiewicz, M. K., and J. K. Solem. 1978. The breeding bird
atlas of Montgomery and Howard Counties, Maryland.
Md. Birdlife 34:3-39.
Lack, P. 1986. The atlas of wintering birds in Britain and
Ireland. British Trust for Ornithology, Tring, U.K.,447 pp.
Laughlin, S. B., editor. 1982. Proceedings of the northeastern
breeding bird atlas conference. Vermont Institute of
Natural Science, Woodstock. 122 pp.
Laughlin, S. B., and D. P. Kibbe. 1985. The atlas of breeding
birds of Vermont. Vermont Institute of Natural Science,
Woodstock. 456 pp.
Lord, J., and D. J. Munns. 1970. Atlas of breeding birds of the
west Midlands. Collins, London, U.K. 276 pp.
Nikolaus, G. 1987. Distributional atlas of Sudan's birds with
notes on habitat and status. Bonner Zoo!. Monographien
25, Bonn, West Germany. 322 pp.
Perring, F. H., and S. M. Walters. 1%2. Atlas of the British
flora. Botanical Society of the British Isles. T. Nelson,
London, U.K. 432 pp.
Robbins, C. S. 1982. Overview of international at lasing. Pages
3-10 in S. B. Laughlin, ed. Proceedings of the Northeastern
Breeding Bird Atlas Conference, Vermont Institute of
Natural Science, Woodstock.
Robbins, C. S., and B. A. Dowell. 1986. Use of miniroutes and
Breeding Bird Survey data to estimate abundance. Pages
28-40 in S. M. Sutcliffe, R. E. Bonney, Jr., and J. D. Lowe,
eds. Proceedings of the Second Northeastern Breeding
Bird Atlas Conference, Laboratory of Ornithology, Cornell
University, Ithaca, N.Y.
Robbins, C. S., S. Droege, and J. R. Sauer. 1989. Monitoring
bird populations with Breeding Bird Survey and atlas data.
Ann. Zoo!. Fennici. 26:297-304.
Sharrock, J. T. R. 1976. The atlas of breeding birds in Britain
and Ireland. British Trust for Ornithology, Tring, UK.
477 pp.
Skaar, P. D. 1975. Montana bird distribution. P. D. Skaar
(privately printed), Bozeman, Mont.
Stewart, R. E. 1975. Breeding birds of North Dakota.
Tri-college Center for Environmental Studies, Fargo,
N. Dak. 295 pp.
Stewart, R. E., and C. S. Robbins. 1958. Birds ofMaryland and
the District of Columbia. U.S. Fish Wildl. Serv., N. Am.
Fauna 62. 401 pp.
Thomas, D. 1979. Tasmanian bird atlas. Fauna of Tasmania
Handbook 2, University of Tasmania, Australia. 171 pp.
Yeatman, 1... 1976. Atlas des oiseaux nicheur~ de France. Soc.
Ornithologique de France, Paris. 282 pp.
AVIAN POPULATION TRENDS 23
Methodology of the International Shorebird Survey and
Constraints on Trend Analysis
by
Marshall A. Howe
u.s. Fish and Wildlife Service
Patuxent Wildlife Research Center
Laurel, Maryland 20708
Origin and Extent of the Survey
The International Shorebird Survey (ISS), managed
by Manomet Bird Observatory (Massachusetts) since
1972, enlists the services of volunteer observers to
conduct surveys of migrating shorebirds in the Western
Hemisphere (Manomet Bird Observatory 1980). In
1974, the definition ofISS was broadened to include the
Maritimes Shorebird Survey (MSS; Morrison 1976)
conducted during the 1970's by the Canadian Wildlife
Service. However, because of difficulties in obtaining
the MSS data, the ISS includes only the Manomet survey
for the purposes of the present discussion. The purpose
of the ISS has always been to identify and document
areas of major importance to shorebirds during fall
migration (in recent years some spring surveys have
been conducted). The ISS has focused on the eastern
United States, especially between Maine and North
Carolina, but includes sites as distant as Kansas, the
Caribbean, and Latin America.
Design of the Survey
Participants in the ISS are recruited on the basis of
recommendations from persons known to be experienced
at shorebird identification. In this way, a reasonable level
of quality control is assured. Unlike the Breeding Bird
Survey and some other surveys, sampling areas in the ISS
are not preselected randomly. In part, this is because the
survey was not designed to obtain estimates of population
change. Furthermore, the difficulty of access to many
wetland habitats used by shorebirds makes random
selection impractical. Observers are encouraged to select
sites on the basis of traditional shorebird use and
convenience of access. A special effort is made to ensure
coverage of sites known to attract large numbers of
shorebirds every year. Observers are asked to derme the
limits of the survey site and visit it at lca~t three times per
month between 1 July and 31 October, the period that
encompasses the majority of the migration period for
most species. Few observers achieve such intensive
coverage, however, and many sites are surveyed only a few
times annually or missed entirely in some years. The total
number of each species is determined by either direct
count or subjective estimation and coded appropriately.
In tidal areas there is an effort to conduct the survey
during the same segment ofthe tide cycle to minimize the
influence of local movements in response to tide
fluctuations. Each survey is summarized on forms
provided by Manomet Bird Observatory. The forms are
then checked and edited by Manomet staff, and the data
are computerized.
To date, more than 600 volunteers have collected
data for the ISS at more than 500 sites in the United
States. Annually, an average of 1,500 surveys at 150 sites
has been conducted. Because turnover rates of
migrating birds at different sites are poorly known, it is
difficult to estimate the total number of shorebirds
censused annually. In a recent analysis of only 64
Atlantic coastal sites, the minimum annual estimate
(assuming no turnover) for the commonest species, the
semipalmated sandpiper (Calidris pusilla), was 34,500
(Howe et al. 1989). The actual number observed at this
subset of sites was probably 3-4 times this number.
International Shorebird Survey sampling design and
certain properties of the data present difficulties for
population trend estimation. I summarize those
difficulties and evaluate the degree to which they can be
overcome. Some of these issues are addressed in more
detail elsewhere (Howe et al. 1989).
Problems Associated with Sampling
Design
To be suitable for trend analysis, the data must meet
the assumption that the birds sampled are
representative of the total population or some definable
24 BIOLOGICAL REPORT 90(1)
subset. Although the nonrandom distribution of ISS
sites threatens this assumption, there are potentially
more serious problems. For example, many species
prefer inland, freshwater sites, opportunistically
selecting wetlands that have suitable water conditions at
any given time. Also, the number of ISS sites is limited,
and the influence of major stopovers not included in the
network is unknown.
Finally, during migration, shorebirds move through
stopover sites at varying rates determined by
physiological and climatological factors. Major flights
could be overlooked if the interval between samples is
too long. All of these factors potentially influence the
accuracy and precision of ISS counts, introducing
variation that is independent oftrue population change.
By accepting certain assumptions and using the ISS
data least sensitive to the sources of variation listed
previously, it seems that these concerns can be
minimized. In a preliminary study (Howe et al. 1989) we
confined trend analyses to a subset of 12 species we .
believe to be obligate users of coastal habitats
(eliminating opportunistic species) and examined only
ISS sites on the Atlantic coast. We felt that this approach
would reduce the risk of overlooking major segments of
the population. Concern about the nonrandom
distribution of sites is allayed somewhat by our present
knowledge of the ecology of migrating shorebirds.
Several studies (e.g., Smith and Houghton 1984; B. A.
Harrington, unpublished data) showed that at least
some individuals of some species showed between-year
fidelity to migration stopovers. Also, many stopover
sites attract large numbers of shorebirds annually. If
traditional use of these sites is the rule for many species
of shorebirds, our confidence in the sample counts as
indicators oftrue population trends is greatly increased.
With regard to the temporal aspect of migration,
analysis ofISS data shows that most species exhibit fairly
sharp population peaks. If none of the samples occurs
during the peak period, the populations for that site will
be seriously underestimated. Several studies of
semipalmated sandpipers (Page and Middleton 1972;
Lank 1983; Dunn et al. 1988) suggested that, early in
southbound migration, lengths of stay of individual birds
at stopovers averaged 2 to 3 weeks. If we assume this is
typical of migrating shorebirds, a regime of sampling at
10-day intervals should on average yield one to three
counts during the peak period.
Problems Associated with Data
Structure
Conducting route-regression trend analysis (Geissler
and Noon 1981) requires a single count representing the
population at a given site in a given year. Some ISS sites
may have several dozen censuses conducted during a
year, covering a period in which populations may
fluctuate dramatically. These counts must be converted
to one value representing an annual site index. We
examined four approaches to calculating site indices
and determined that the best measure was the log of the
average of all counts within a 21-day period centered at
the regional migration peak (Howe et al. 1989). Sites
with less than two counts during that period were
eliminated from analysis. This approach yielded the
smallest variances for population trend estimates. All
indices that incorporated data from both peak and
nonpeak periods yielded extraordinarily high variances.
Even the best population index yielded within-site
variances much higher than those typical of surveys like
the BBS, which samples populations that are more or
less uniformly distributed in space and time. This is a
potential problem for trend analysis because of the
importance of the variance in the back-transformation
from the log scale. However, analyses of simulated data
sets with comparably high variances yielded trend
estimates not significantly different from the trends
assumed for the simulation (Howe et al. 1989). This is
an encouraging sign, suggesting that the variance
problem may not be critical.
A final concern about the ISS data set analyzed for
1972-83 was the large number of site-year combinations
for which data were unavailable. This situation could
result in certain years contributing disproportionately to
the trend estimate. We analyzed the entire data set for
the sanderling (c. alba), which was recorded at an
average of 52 sites each year, and found that all possible
pairs of years were uniformly represented despite the
spotty coverage. Inconsistent coverage in the future,
however, could occasionally lead to biases toward
certain years and produce erroneous trend estimates.
Conclusions
The International Shorebird Survey, though not
rigorously designed to monitor shorebird populations,
produces data that can presently be used to derive
imprecise population trend estimates for obligate
coastal species. It is unlikely that species that often use
transitory wetlands during migration can be monitored
effectively. For the present, it is assumed that population
changes at coastal ISS sites are representative of
population changes throughout the coastal system (this
assumption is presently being field-tested). Observer
variation in estimating the size of large flocks is also a
potential problem, the magnitude ofwhich is not known.
More precise estimates of population trends could be
achieved by increasing the frequency of surveys at the
most important sites, particularly during peak periods,
and by assuring continuity of coverage across years.
Formal incorporation of coastal Canadian sites into the
ISS would help improve population trend estimates for
several species.
Literature Cited
Dunn, P.O., T. A. May, M. A. McCollough, and M. A. Howe.
1988. Length of stay and fat content of migrant
semipalmated sandpipers in eastern Maine. Condor
90:824-835.
Geissler, P. H., and B. R. Noon. 1981. Estimates of avian
population trends from the North American Breeding Bird
Survey. Pages 42-51 in C. J. Ralph and J. M. Scott, eds.
Estimating the numbers of terrestrial birds. Stud. Avian
BioI. 6.
Howe, M. A., P. H. Geissler, and B. A. Harrington. 1989.
Population trends of North American shorebirds based on
AVIAN POPULATION TRENDS 25
the International Shorebird Survey. BioI. Conserv.
49:185-199.
Lank, D. B. 1983. Migratory behavior of semipalmated
sandpipers at inland and coastal staging areas. Ph.D.
dissertation, Cornell University, Ithaca, N.Y.
Manomet Bird Observatory. 1980. Study ofautumn shorebird
migration in the Atlantic corridor 1974-1978-a study of
red knot status and migration routes in the United States
east of the Rocky Mountains. Final report to U.S. Fish and
Wildlife Service, contract 14-16-0009-79-075.
Morrison, R. 1. G. 1976. Maritimes shorebird survey 1975,
preliminary report. Canadian Wildlife Service, Ottawa.
Page, G., and A. L. A. Middleton. 1972. Fat deposition during
autumn migration in the semipalmated sandpiper.
Bird-Banding 43:85-96.
Smith, P. W., and N. T. Houghton. 1984. Fidelity of
semipalmated plovers to a migration stopover. J. Field
Ornithol. 55:247-49.
26 BIOLOGICAL REPORT 90(1)
Evaluation of the Colonial Bird Register
by
R. Todd Engstrom
Comell Laboratory (Jf Omithology
159 Sapsucker Woods Road
Ithaca, New York 14850
History of the Colonial Bird Register
In the early 1970's E. J. Fisk and R. Downing started
counting least terns (Sterna antillamm) along the
eastern seaboard because they were concerned that
the species was declining (Downing 1973; Fisk 1975).
They organized a loose network of volunteers to count
terns, although they did not have a standardized
system of data collection.
At about the same time, the prospect of offshore oil
drilling and increasing human use of coastal habitats in
the eastern United States posed threats to wildlife,
especially colonial birds. In response to these threats,
the U.S. Fish and Wildlife Service initiated surveys of
colonial waterbirds on the Great Lakes (Scharf 1979)
and along all coastal areas of the United States (Erwin
1979; Erwin and Korschgen 1979; Korschgen 1979;
Sowls et al. 1980; Portnoy et al. 1981; Keller et al. 1984;
Spendelow and Patton 1988).
A central data repository was an attractive idea when
interest in waterbird colonies was high and financial
support was available. The surveys of Fisk and Downing
provided a model for an organization of volunteers to
collect data. The Colonial Bird Register (CBR) was
established on 1 August 1975 by joint agreement
between the National Audubon Society and the Cornell
Laboratory of Ornithology. The CBR relied on a
network of volunteer organizations and individuals to
contribute data.
From 1975 through 1984, the National Audubon
Society paid most of the expenses of the CBR, including
the salary of the director. The Cornell Laboratory of
Ornithology covered administrative costs and provided
office space. Between 1979 and 1986, the U.S. Fish and
Wildlife Service also supported the CBR through a
series of cooperative agreements. User fees for data
retrieval were expected to defray operational costs of
the CBR; however, the fees, averaging a few hundred
dollars per year, never covered the costs.
CBR Design and Content
One of the first priorities of the CBR was to devise a
form to standardize data reporting. This form had to be
detailed enough to provide data for a variety of analyses
and easy to use. Data categories on the form have been
modified twice over the years (McCrimmon 1978; Erwin
et al. 1984; Appendix).
The latest CBR form (Appendix) has space for
information on the date of colony visit, colony location,
habitat, colony disturbances, colony-site ownership and
management, survey technique, and nesting stage. The
vehicle from which birds are counted (e.g., airplane,
helicopter, boat) is also recorded.
The CBR contains data on locations and estimates of
abundance for gull, tern, heron, and egret colonies,
some of which are composed of several species
(McCrimmon 1978; Erwin et al. 1984). A single record
is made for each visit to a colony, although a colony may
be visited more than once in a year. An estimate of
abundance for each colony can be a count ofactive nests,
adults, or the number of breeding pairs. Records for 48
colonial seabird and wading bird species are included in
the CBR (Table). The CBR also contains records for
some swallows and more than 30 noncolonial species
counted incidentally among truly colonial species.
Data are edited and entered into the computerized
data base, which is in the ISIS data base management
system on the Cornell University mainframe computer.
Data can be retrieved by various geographical divisions
(e.g., colony, county, State) or by species.
CBR data come primarily from State wildlife
agencies, the U.S. Fish and Wildlife Service, and private
organizations such as The Nature Conservancy and
State Audubon societies. The survey of colonial
waterbird colonies from Maine to Florida in 1976 and
1977 forms the foundation of the computerized portion
of the CBR (Erwin and Korschgen 1979; Portnoy et al.
1981) - data from Texas from 1973 to 1980 comprise
another large portion (Texas Colonial Waterbird
AVIAN POPULATION TRENDS 27
Table. Names ofspecies included in the Colonial Bird Register (CBR).
Common
Leach's storm-petrel
American white pelican
Brown pelican
Great cormorant
Double-crested cormorant
Olivaceous cormorant
Magnificent frigatebird
Least bittern
Great blue heron
"Great white" heron
Great egret
Snowy egret
Little blue heron
Tricolored heron
Reddish egret
cattle egret
Green-backed heron
Black-crowned night-heron
Yellow-crowned night-heron
White ibis
Glossy ibis
White-faced ibis
Roseate spoonbill
Wood stork
Common eider
Black-necked stilt
American avocet
Laughing gull
Franklin's gull
Ring-billed gull
california gull
Herring gull
Glaucous-winged gull
Great black-backed gull
Gull-billed tern
caspian tern
Royal tcrn
Sandwich tern
Roseate tern
Common tern
Arctic tern
Forstcr's tern
Least tern
Sooty tern
Black tern
Black skimmer
Razorbill
I31ack guillemot
Atlantic puffin
Scientific
Oceanodroma leucorhoa
Pelecanus erythrorhynchos
Pelecanus occidentalis
Phalaerocorax carbo
Phalaerocorax auritus
Phalacrocorax olivacellS
Fregata magnijicens
lxobrychus exilis
Ardea herodias
Ardea herodias occidentalis
Casmerodius albus
Egretta thula
Egretta caemlea
Egretta tricolor
Egretta mfescens
Bubulcus ibis
Butorides striatus
Nycticorax nycticorax
Nycticorax violaceus
Eudocimus albus
Plegadis falcinellus
Plegadis chihi
Ajaia ajaja
Mycteria americana
Somateria mollissima
Himantopus mexicamlS
Recurvirostra americana
Lams atricilla
Lams pipixcan
Lams delawarensis
Lams californiws
Lams argentatus
Lams glaucescens
Lams marinus
Sterna nilotica
Sterna caspia
Sterna maxima
Sterna sandvicensis
Sterna dougallii
Sterna hinmdo
Sterna paradisaea
Sterna forsteri
Sterna antillanun
Sterna fuscata
Ch/idonias niger
Rynchops niger
Alca torda
Cepplllls wylie
Fmtercula arctica
CBR records'
33
133
163
43
862
106
8
31
4,514
40
2,382
1,791
1,273
1,322
459
1,610
313
1,708
281
520
395
413
333
152
483
27
21
1,296
109
307
181
1,800
9
980
794
361
365
217
216
2,623
116
824
4,097
34
136
1,659
7
237
11
, The number of records over all years. This is not the same as the number of colonies; some colonies were visited more than once in a season.
Society 1982). Most recently, organized counts such as
the annual survey of Long Island, New York (MacLean
et a1. 1988), have generated the most data for the CBR,
followed by reports from many national wildlife refuges.
The number of records in the CBR ranges from 0 in
West Virginia to nearly 2,200 in New York (Fig. 1).
28 BIOLOGICAL REPORT 90(1)
Number of records
Do to 9
EJ 10 to 99
~ 100 to 499
m500 to 999
.>1,000
Fig. 1. The number of computerized CBR records of visits to colonies by State. This does not include 263 records from Canada.
Some colonies are visited several times in a season.
Coverage among States varies from complete annual
surveys of all colonial birds to spotty reporting of only a
few colonies. The greatest number of CBR records
came from 1976 (3,100) and 1977 (2,600). Data
contributions dropped in 1978, and since then average
1,000 to 1,500 records per year (Fig. 2).
Evaluation of the CBR
The CBR was designed to consolidate the results of
local and regional colonial waterbird surveys.
Accessibility of the data for researchers was a high
priority. As data accumulated, examination of
population trends became possible (Erwin et al. 1984).
3,500
Quantitative data must be collected in a consistent way
over time for population trend analysis. Usefulness of
the CBR depends on the quality and consistency of the
monitoring effort and data base accessibility.
Monitoring Effort
The CBR includes data for a heterogeneous group
of species from a large geographical area for an
indefinite period. The diffuse data aren't extensive for
any single purpose, although some species in some
regions have enough data for preliminary analysis
(Engstrom et al. 1990). Most contributors to the CBR
were concerned primarily with bird populations for a
3,000
2,500
en
-0.....
au
.a..>.. 2POO
..... a
.....
a> 1,500
..Cl
E
::>
z:
1,000
500
a
Fig. 2. The number of computerized CBR
records by year. Note: only records of least
tern colonies are represented in 1986.
197374 75 76 77 78 79 80 81 82 83 84 85 1986
Year
single State and used custom-made sampling schemes.
The assortment of goals and methods at the State level
made the centralized CBR data base a patchwork. The
ultimate uses of the data need to be articulated clearly
from the origin of the data base (Moroney 1984), then
monitoring must adhere to standard practices of data
collection and recording.
The technical problems associated with the
monitoring effort that supplies the CBR (Erwin et al.
1984) include: (1) observers vary in quality; (2) counting
techniques vary among colonies, States, and years; (3)
colony names are inconsistent among years and
observers; (4) inactive colonies are not recorded as
"zeros," making trend analysis problematic; and (5)
geographical coverage of colonies varies among States.
The Data Ba5e
Centralization and computerization have made CBR
data easily accessible to conservation organizations,
academic researchers, and State wildlife management
agencies. The data base greatly speeds organization of a
wide array of counts and independent studies into
regional summaries. Standardized reporting aids
comparison of data from different sources, if the forms
were filled out completely. Data that were collected by
different methods can be compared by using correction
factors, although these factors should be used with
caution. For example, empirically derived correction
factors have been used to compare aerial counts with
ground counts (Erwin 1979; Thompson 1982; Engstrom
et al. 1990).
Contributors want to see the results of contributing
their data to a centralized data base. Products of the
CBR, such as an analysis of trends in the population of
a species or maps of colony locations, are expensive.
Data contributors may be reluctant to go through the
burden of filling in standardized forms and then be
asked to pay for an analysis of their own data.
Additionally, some contributors, especially university
biologists, want to protect their data from use by other
researchers because they intend to publish their results.
In the extreme case, the data base is left in the untenable
position of gathering data that cannot be used. This
problem can be circumvented by making an agreement
between the contributor and the data base to protect the
data for a specified time.
Recommendations
1. The goals, methods of analysis, and census or survey
guidelines that relate to those goals should be clearly
defined at the outset of any colonial waterbird
monitoring program. Technical instructions for
counting seabirds have been established in Great
AVI~'Ii POPULATlo;-,; TRENDS 29
Britain and Canada (Anonymous 1969; Nettleship
1976). A summary of colonial waterbird counting
techniques similar to that for terrestrial birds could be
written by an individual or organization or could be
produced in a symposium (Ralph and Scott 1981;
Verner 1986).
2. State programs to monitor colonial waterbirds
should be coordinated. As nongame wildlife
departments have developed, coastal surveys have
become important State programs. The goals, field
techniques, and common types of analysis ofthe various
State surveys should be identified and standards
adopted. The precision of counts should be improved to
permit rigorous analysis. Any common goals and
cooperative projects that emerge would help to define
the structure of a useful database that is compatible
among States.
3. Given a focus on which species to monitor and the
purpose for monitoring, the operational requirements
of a colonial bird monitoring program should include
(a) conducting field tests of census methodology, (b)
coordination of survey efforts within a region, (c)
training participants, (d) promoting the program, and
(e) periodically publishing results.
4. Operational requirements of a colonial waterbird
data base should include (a) data entry and editing
(especially checking the data for duplicate colony names
and errors), (b) maintaining the data in an accessible
data-storage system, (c) communicating with
cooperators, (d) publishing results, and (e) performing
retrievals for data users.
5. Ifa centralized data base is desirable, the products of
the data base and sources of funding for those products
should be clearly identified from the beginning.
Conclusions
The Cornell Laboratory of Ornithology discontinued
the CBR in 1988 because of insufficient funds to manage
the data base or to provide the guidance necessary to
ensure high quality data. The CBR was the cornerstone
of an ambitious landmark effort to survey and monitor
colonial birds. Although new data are not being added,
the extant CBR data base is available for research
purposes. It is important to learn the lessons provided
by the CBR beeause a high quality monitoring program
for colonial birds is needed now as much as it was when
the CBR was initiated (Parnell et al. 1988).
Acknowledgments
I thank G. S. Butcher, R. M. Erwin, and J. D. Lowe
for helpful comments on the manuscript.
30 BIOLOGICAL REPORT 90(1)
Literature Cited
Anonymous. 1969. Operation seafarer, a guide for observers.
The Seabird Group. 14 pp.
Downing, R. L. 1973. A preliminary nesting survey of least
terns and black skimmers in the east. Am. Birds
27:946-949.
Engstrom, R. T., G. S. Butcher, and J. D. Lowe. 1990.
Population trends in the least tern (Sterna antillamm) from
Maine to Virginia: 1975-1986. Pages 130--138 in J. R.
Sauer and S. Droege, eds. Survey designs and statistical
methods for the estimation ofavian population trends. U.S.
Fish Wildl. Serv., BioI. Rep. 90(1).166 pp.
Erwin, R. M. 1979. Coastal waterbird colonies: Cape
Elizabeth, Maine to Virginia. U.S. Fish Wildt. Serv.,
FWS/OBS-79110. 212 pp.
Erwin, R. M., P. H. Geissler, M. L. Shaffer, and D. A.
McCrimmon. 1984. Colonial bird monitoring: a strategy for
regional and national evaluation. Pages 342-357 in W. C.
McComb, ed. Proceedings to the Workshop on
Management of Nongame Species and Ecological
Communities, Lexington, Ky.
Erwin, R. M., and C. Korschgen. 1979. Coastal waterbird
colonies: Maine to Virginia, 1977-an atlas showing colony
locations and species composition. U.S. Fish Wildl. Serv.,
FWS/OBS-79/08. 647 pp.
Fisk, E. J. 1975. Least tern: beleaguered, opportunistic and
roof-nesting. Am. Birds 29: 15-16.
Keller, C. E., J. A. Spendelow, and R. D. Greer. 1984. Atlas of
wading bird and seabird nesting colonies in coastal
Louisiana, Mississippi, and Alabama: 1983. U.S. Fish Wildl.
Serv., FWS/OBS-84/13. 127 pp.
Korschgen, C. 1979. Coastal waterbird colonies: Maine. U.S.
Fish Wildl. Serv., FWS/OBS-79109. 83 pp.
MacLean, D. c., T. S. Litwin, A. Ducey-Ortiz, and R. A. Lent.
1988. 1987 Long Island colonial waterbird and piping
plover survey. Seatuck Research Program, Cornell
Laboratory of Ornithology, Islip, N.Y. 174 pp.
McCrimmon, D. 1978. The collection, management, and
exchange of information on colonially nesting birds. Pages
187-196 in A. Sprunt IV, J. Ogden, and S. Winckler, eds.
Wading birds. Natl. Audubon Soc. Res. Rep. 7. 381 pp.
Moroney, M. J. 1984. Principles of wildlife management.
James A. Bailey, ed. John Wiley & Sons, N.Y.
Nettleship, D. N. 1976. census techniques for seabirds of
arctic and eastern Omada. Can. Wildl. Servo Occas. Pap.
25.
Parnell, J. F., D. G. Ainley, H. Blokpoel, B. Cain, T. W. Custer,
J. L. Dusi, S. Kress, J. A. Kushlan, W. E. Southern, L. E.
Stenzel, and B. C. Thompson. 1988. Colonial waterbird
management in North America. Colon. Waterbirds
11:129-169.
Portnoy, J. W., R. M. Erwin, and T. W. Custer. 1981. Atlas of
gull and tern colonies: North C.arolina to Key West, florida
(including pelicans, cormorants and skImmers). U.S. Fish
Wildt. Serv., FWS/OBS-80/05. 121 pp.
Ralph, C. J., and J. M. Scott, editors. 1981. Estimating
numbers of terrestrial birds. Stud. Avian BioI. 6.
Scharf, W. C. 1979. Nesting and migrating areas of birds of the
United States Great Lakes 30 April-25 August 1976. U.S.
Fish Wildl. Serv., FWS/OBS-77/2. 113 pp.
Sowls, A. L., A. R. DeGange, J. W. Nelson, and G. S. Lester.
1980. Catalog of California seabird colonies. U.S. Fish
Wildl. Serv., FWS/OBS-37/80. 371 pp.
Spendelow, J. A., and S. R. Patton. 1988. National atlas of
coastal waterbird colonies in the contiguous United States:
1976-82. U.S. Fish Wildl. Serv., BioI. Rep. 88(5). 326 pp.
Texas Colonial Waterbird Society. 1982. An atlas and census
of Texas waterbird colonies 1973-1980. Caesar Kleberg
Wildlife Research Institute, Kingsville, Tex.
Thompson, B. C. 1982. Distribution, colony characteristics,
and population status of least terns on the Texas coast.
Ph.D. dissertation, Texas A&M University, College
Station.
Verner, J. 1985. A'>SCssment of counting techniques. Pages
247-302 in R. F. Johnston, ed. Current ornithology, Vol. 2.
Plenum Press, New York.
AVIAN POPULATION TRENDS 31
Appendix. Colonial Bird Register form.
COLONIAL BIRD REGISTER
at the Laboratory of Ornithology • Cornell University
159 Sapsucker Woods Road • Ithaca NY 14850·1999 • (607) 255·4999
GENERAL INSTRUCTIONS: Please fill out form In pencil, so that we can easily make necessary changes during our editing, Shaded
areas are for editor's use only. We appreciate any information you can provide us concerning nesting bird colonies. Please send a report
even If you cannot provide all the requested mformation. If you know the colony has been previously censused and you know the official
colony name or 10 number, then some of the requested information can be omitted. If you visit a colony more than once during a year,
you may staple the forms together and omit duplicated mfQrmatlon on subsequent forms,
VISIT AND LOCATION INSTRUCTIONS
NAME If you don't know an official name, use a nearby landmark (town, lake, river, etc.). Please provide a colony map/sketch that
mcludes the landmark used for the name. COLONY/SUBCOLONY When in doubt about whether two groups are separate colonies, we
suggest that you lump them together for reporting purposes and perhaps identify them as subcolonles, Please show subcolony locations
on a colony map/sketch. TIME: Use military time, e.g. 1:00 PM =1300, 9:00 AM = 0900,
COLONY VISIT INFORMATION
Date (mm/dd/yy) _
Time started (military) __-'--__
COLONY LOCATION INFORMATION
Colony name
Length of visit _"_ hours
Recorder's name
Latitude
__ O __ 'N
Longitude
__ O __ 'W Subcolony #
Recorder's address
Other observers _
Colony visit # (thiS year)
Nearest community _
County ~ _
State/Province/Territory Country _
Colony 10 numbers: Federal State Other _
COLONY SITE INFORMATION
General habitat: (CIRCLE ONE) penmsula(O) barrier Island(1) saltwater non-barrier island(2) freshwater island(3)
shoreline /lake,pond,lmpoundment(4) shorel ine / ocean,estuary(5) riparian / rlver.stream(6) other(7)
SpeCifiC habitat' (CIRCLE ONE) salt marsh(2) fresh marsh(4) mangrove(34) shrub swamp(36) wooded swamp(11) bog(10)
spoil/fill area(6) sand bar(20) salt lIat(21) sandy beach(37) gravel beach(38) dune(39) rocks(40)
deciduous forest(41) evergreen forest(42) mixed lorest(43) rooftop(27) parkmg lot(44)
pier / Jetty / dock/breakwater(45) other(16) _
Nest substrate (CIRCLE ONE) evergreen trees(1) deciduous trees(2) mixed trees(3) dead trees(37) shrub(4)
grass/sedge/rush/herb(38) dead herbaceous/wrack(39) salt Ilat(25) soil(34) sand(12) gravel(32)
cobble(14) rocks/crevlces(15) rocks/cllff(16) burrow(1?) man-made structure (explain 19)
other(20) _
% Vegetation cover within colony %
COLONY DISTURBANCE
For each possible factor indicate whether disturbance is absent. current, or potential.
"Current" means that a disturbance probably influenced your survey and population
estimate and IS probably affecting this year's productivity. "Potential" means that although
there was no eVidence of current disturbance, It IS still possible that the disturbance could
affect this year's productiVity For each factor present prOVide more details in the
REMARKS section
c E c c
'" 0 .~ "' 2 0 c <ii co OJ rn ~ 2:- '6 "0 " OJ <;;
0 "0 "' "' c E U ii => g aOJ. "iii '" "0 r: OJ ~ '" c 0 0- > > r:
none C C C 0 0 0 0
current [j C C 0 0 0 0 0
potential C [J C 0 C 0 0 0
O,d poor weather Influence your population estimate this visit?
Yes C No C If yes, give details in REMARKS section.
COLONY MANAGEMENT
Owner
Owner's address
Colony Management (CIRCLE YES or NO
or UNKNOWN)
posted Y N U
fenced Y N U
patrolled Y N U
predator-control Y N U
fencing (CIRCLE)
none string electric snow
other _
32 BIOLOGICAL REPORT 90(1)
INSTRUCTIONS FOR REPRODUCTIVE INFORMATION: We are asking for two types of survey information: (1) what did you actually see
(adult count, active nest count) and (2) what is your best estimate of the number of breeding pairs using this site this year. Please give
details of how this estimate was obtained in the REMARKS section. Please list each species on a separate line (do not write in shaded
areas). Choose codes for "counted from" and "survey technique" and write them in the boxes. Adult count (individuals) and active nest
count refer to actual numbers seen on the day of the visit. Estimated # of breeding pairs refers to the total size of the colony (this mayor
may not be the same as active nests) e.g. a tern colony of 1apairs is washed out shortly before a visit, during this visit 15 adults are seen
but only 2 pairs have started to renest. Adult count (individuals) = 15, active nest count = 2, estimated # of breeding pairs = 1a. For nesting
stage check the appropriate box or boxes, if unknown please write "unknown" across the boxes. If you band birds, please specify adult,
young, both, or none in the last column. Extra information should be put in the REMARKS section. Thanks' COUNTED FROM:
3-airplane, 4-helicopter, 5-boat, 6-motorized land vehicle, 7-foot (periphery of colony), 8-foot (within colony), 9-other (describe in
REMARKS section) SURVEY TECHNIQUE: 1-total adult count, 2-total nest count, 3-photography, 4-partial adult count, 5-partial nest
count, 6-quadrat sample, 7-line strip sample, 8-visual estimate, 9-best guess, a-other (describe in REMARKS section)
nesting stage (check appropriate
boxes & circle most frequent stage)
OJ c:
0 ,.,c:
1 =...
,."
OJ I al 0_
.0 u ,., > c..c: ~ <Il_
.!!l 0> '" c: Icn~
<Il c: ~ 0 '" OJ 0> " "0 ." u~O>
- <Il ..: c: 0 0> "
,
0> DC:
0.= 0> .$ 0> c: "
,., c: u c: 0>"
""~ c: c: 0 0 u ';;' 0> c: OJ 0> £g, ,., OJ -= 0 c: .~
<ii Uo>
~ 0> ';;' :;; c: ,., a; 0> ." OJ 0> 0 OJ -g£f OJ c: .!!' .0 i' c: <ii -" c: u c: 2c: c: ." " '" ~
.c: c: OJ c: <ii "'- OJ- "'.- OJ ';;; 0> Eil 0> " 1ii " c: <ii '" .0-6 '" > c: C. 0 0 .0 0 C. OJ -" .c: u 2 ,., ~ 5! '" :;:;:l ".;:; Q) c. :-::m
"0 <Il~
"''' OJ.o 10 20 22
u
OJ
E
,.,
.c OJ ,g >
C<Il " :s species u 2 <Il 5§
2 ,., ci. "u
c: OJ 0 ~s
" ~ c. ".- 0 " J' -g~ " <Il
Colony name _
Date ~
Visit # __ of __ for this year
Sheet # __.of __ for this visit
REPRODUCTIVE INFORMATION
I
I
i I I
I, i
ii
REMARKS: attach additional sheets if necessary COLONY MAP/SKETCH: (should show colony in relation
to landmarks or Lat. & Long.) attach additional sheets if
necessary
AVIA."I,' POPULATIO:"l TRE:"IDS 33
Descriptions of Surveys: Breeding Bird Censuses
by
David W. Johnston
5219 Concordia Street
Fai1ax, Virginia 22032
Introduction
In recent years many community ecologists have
strongly emphasized the need for ecological
investigations that extend over long periods, an
approach also embodied in the Long-term Ecological
Research Program (LTER) of the National Science
Foundation (Callahan 1984). Such long-term records
are perceived as providing valuable insights into a
predictive capability for population and community
changes over time. For example, studies of birds,
particularly those conducted during the breeding
season that span several consecutive years or even
decades, should provide good indicators of long-term
population changes and dynamics. Changes in some
North American bird populations were suggested as
early as 1969 by Aldrich and Robbins (1970). More
recently, based on a small number of long-term census
data sets, Briggs and Criswell (1979), Hall
(1984 a,1984 b), and Johnston and Winings (1987)
reported declines in breeding bird populations at some
eastern forest sites. Nonetheless, Bohlen (1984:27)
correctly cautioned that "not enough long-term
censuses have been conducted to provide conclusive
evidence that...species ...are seriously dwindling
overalL"
By searching the published literature, I found 15
long-term data sets of breeding bird censuses for
deciduous forest sites, mostly in the eastern United
States. The sets range from 10 to 50 years, one extending
back to 1927 and about half including the 1980's. The
present report identifies these sites and their
characteristics. Breeding bird data from the sites are
discussed as being appropriate for use in monitoring
avian population changes over long periods.
Sources of the Data Sets and Site
Characteristics
Censuses of breeding landbirds from the United
States date back to the 1914-20 surveys of the Bureau of
Biological Survey (Cooke 1923), and since 1937, the
National Audubon Society has sponsored annual
breeding bird censuses at scattered sites and in different
habitats in North America. In this paper, only data from
deciduous forests in the eastern and midwestern United
States have been analyzed, chiefly because of the
availability of their long-term sets (at least 10 years;
Table 1). I have not found long-term quantitative sets
for other major ecological community-types in North
America except for the reports of Hall (1984 a,1984 b)
from eastern spruce forests.
Most of the data sets have been published in
Audubon Field Notes, American Birds, and The Atlantic
Naturalist (see Table 2 for additional published
sources). The 15 sites were geographically widespread,
ranging from south Georgia to New Hampshire and
from Maryland to Illinois. Some censuses date back
50 years (Table 1; Figure).
These census sites were in ecologically similar
habitats. Only mature forests (from the authors' or
compilers' original descriptions) were used, and none
were used in which forest succession was evident.
Some sites contained a few scattered pines (and thus
might not have been a true climax), but all sites were
reportedly dominated by mature hardwoods of various
species (e.g., oak, hickory, beech, maple) in various
combinations. The estimated ages of sites and
dominant tree species differed somewhat among the
sites (Table 2).
The environments of some sites reportedly remained
unchanged throughout the duration of the censuses, both
inside and outside the site (Table 2). Other sites in Table 2
either incurred some nonsuccessional habitat changes,
the magnitude of which might be questionable, or lacked
convincing documentation for changes over the duration
of the censuses to determine actual or possible
perturbations. I contacted as many of the compilers or
authors as possible, thus obtaining first-hand, crucial
information on the continuity ofthe site within contiguous
forests. For the most part, the same site and area were
censused in subsequent years, and boundaries remained
34 BIOLOGICAL REPORT 90(1)
Table 1. Deciduous forest sites for long-temz breeding bird censuses andyears ofcensuses.
First Last Numbcrof Span
Sites year year censuses (years)
Unchanged-parts of large forest tracts,
remaining unisolated during census periods
Wilkesboro, North carolina 1954 1973 20 20
Hubbard Brook, New Hampshire 1%9 1984 16 16
Dranesville, Virginia 1973 1986 14 14
Wormsloe Plantation, Georgia 1%3 1973 11 11
calhoun County, Michigan 1938 1947 10 10
Clayton, Georgia 1%9 1978 10 10
Changes known or undetermined
Trelease Woods, Illinois 1927 1976 42 50
cabin John, Maryland 1947 1986 34 40
Rock Creek Park, District of Columbia 1948 1986 29 39
calvert County, Maryland 1960 1986 26 27
Glover-Archbold Park, District of Columbia 1959 1986 25 28
Cleveland, Ohio 1932 1947 15 16
Greenbrook Sanctuary, New Jersey 1949 1983 11 37
Connecticut Arboretum, Connecticut 1953 1985 11 33
Licking County, Ohio 1937 1947 10 11
unchanged over the census years. At only one site was the
census area changed: Rock Creek Park, where the
original census area of 32.4 ha established in 1948 was
reduced to 26.3 ha in 1%3.
Census Methods
Although some censustakers and annual compilers
might have changed at a given site over the years,
methods for censusing and tabulating the breeding
bird populations did not; census methods reportedly
followed the widely used spot-mapping-singing
male-territory mapping technique. This technique, as
described by Williams (1936) and Stewart and Aldrich
(1949), is based on the spacing of (most) forest
breeding birds into territories identified chiefly by the
locations of males, their conspicuous vocalizations,
and behavior patterns. By mapping locations of all
(singing) males (and pairs) in a census area on
different days during the breeding season, one can
identify territories reasonably accurately. By assuming
that each territory contains or will contain a pair of the
species, the number of occupied territories can be
translated into the number of pairs of each species
breeding on a given site.
The Data Sets
For the purpose of a comparative analysis, the
number of breeding pairs at each site as reported by
the compiler or author has been standardized to the
number of pairs per 100 acres. For each year the total
breeding pairs at a site can be subdivided into two
components: neotropical migrant species and
short-distance migrant-resident species. Thus, for
each census-year at a given site, three numbers are
available (all in breeding pairs per 100 acres): total
birds, neotropical migrants, and short-distance
migrants or residents.
Some of the data sets are clearly periodic, that is,
censuses are equally spaced over the years (Table 1).
For the sets in Table 1 and the Figure, periodic ones
are those in which censuses were reported for every
year over the time span; other sets had gaps in the years
censused (i.e., aperiodic).
Potential Limitations of the Data Sets
Despite the perceived value of these long-term data
sets, the following potential limitations for estimating
population trends are recognized (see also discussions
by various authors in Ralph and Scott 1981):
• the aperiodicity of some of the data sets;
• variation in census effort (man-hours expended);
• inadequate vegetation-site description at the
inception of the first census; thus, the true ecological age
of the site might be unknown;
• biased estimates, such as overestimati