Home : Wildlife & Habitats : Species Profiles : Breeding Bird Atlas : Introduction and Methods

Breeding Bird Atlas

The Atlas Survey:  INTRODUCTION AND METHODS

by Herbert W. Kale II, Bill Pranty, and Bradley M. Stith
1993 (revised 1998)

You also may download this document.
(PDF format, free Adobe Acrobat Reader © required) Free Adobe Acrobat Reader

Contents

Introduction
Sponsors and Funding
Administration
Atlas Coordinators
Format of the Species Accounts

Methods
Grid and Sampling Design
Adequacy of Coverage
Survey Procedures
Breeding Criteria and Codes
Data Collection
- Field Data Card
- Casual Observation Card
- Verification Report Form
Data processing and Analysis
- Computer Database
- Data Input Routine
- Quality Control Printouts
- Nearest Neighbor Analysis
- Data Conversion
Mapping
Final Quality Control
Safe-Dates
Limitations and Biases

A Blockbusting Postscript

In Memoriam


Introduction

The goal of a Breeding Bird Atlas (BBA) project is to map the distribution of each species that breeds (or might breed) in a defined area during a selected period of time.  These data provide a historical baseline that can be compared with subsequent Atlas projects to detect changes in the distribution and abundance of birds.

The first Breeding Bird Atlas project was conducted in Great Britain and Ireland from 1968 to 1971 (Sharrock 1976).  It was inspired by the earlier Atlas of the British Flora (Perring and Walters 1962).  Atlas projects for several European countries soon followed.

In North America, Atlas projects in 36 states and 8 provinces were planned, underway, or completed by 1990 (Smith 1990).

The idea of a Florida Breeding Bird Atlas project arose from a suggestion made by Erma J. Fisk in 1981 to the President of the Florida Audubon Society, Peter Mott, and its Vice President for Ornithology, Dr. Herbert W. Kale II.  In her usual style, Jonnie, as she was known to her friends, offered to supply the seed money to cover planning costs.  Although she no longer resided in Florida, Jonnie maintained an interest in the Florida Atlas and continued to provide financial support each year until her death in January 1990. Thus, it is fitting that this Florida Breeding Bird Atlas be dedicated to the memory of this nurturing and dedicated amateur ornithologist.

During the early 1980s, presentations were made to members of various Audubon Society chapters and the Florida Ornithological Society (FOS) to raise enthusiasm for an Atlas project in Florida.  A search for major funding also was initiated.  In 1983, the Florida Legislature established the Nongame Wildlife Program of the Florida Game and Fresh Water Fish Commission [Editor's note: now the Florida Fish and Wildlife Conservation Commission] (hereafter the Commission).  In 1984, funding for the Nongame Program was provided by the creation of the Florida Nongame Wildlife Trust Fund.  A $4.00 fee was added to the $6.00 first-time registration of motor vehicles brought into the state by new residents.  With a daily immigration of approximately 900 people into the state, an annual income of 1.3 - 2 million dollars was projected.

Beginning in 1985, the Commission established a grants and contracts program to fund non-Commission projects that supported nongame wildlife management and conservation.  A proposal to conduct a Breeding Bird Atlas project was submitted by the Florida Audubon Society (FAS) and approved by the Commission.

On 12-13 March 1985, the Commission and FAS sponsored an Atlas planning conference in Orlando to acquaint Florida birdwatchers with the project and to select volunteers to serve as regional and county coordinators.  Resource leaders in attendance were Chandler S. Robbins of the U.S. Fish and Wildlife Service (USFWS), an authority on bird population surveys and Atlasses; Janet Carroll, state coordinator of the New York BBA, then in its final year of fieldwork; and Reed Noss of the Florida Natural Areas Inventory, who had been involved with the Ohio BBA.  An Atlas Advisory Board was appointed and met in April 1985 to select the project's State Coordinator.

Fieldwork for the Florida Breeding Bird Atlas began 1 January 1986 and ended 31 December 1991.  More than 1,880 participants were involved and more than 136,000 records were compiled.  Approximately 200 dedicated participants helped throughout most of the 6 years of the Atlas survey.

Sponsors and funding

The Florida Breeding Bird Atlas project was sponsored by the FAS and the FOS.  Primary funding came from the Nongame Wildlife Trust Fund, administered by the Commission.  Members of the FOS and FAS chapters provided the majority of regional and county coordinators and fieldworkers for the project. Supplemental funding and services also were provided by the above-mentioned organizations and by private corporations, foundations, garden clubs, and individuals.

In 1986, the FAS sent a fundraising appeal, specifically for the Atlas project, to each of its members. Many of those who responded were contacted for additional support in 1988, 1990, and 1991.  All contributors are listed in the acknowledgments.

Administration

The FAS's Vice President for Ornithology and part-time Project Director, Dr. Herbert W. Kale II, was responsible for the overall direction of the project and supplemental fund-raising.  He served in this capacity until his death in August 1995.  A fulltime State Coordinator, C. Wesley Biggs, was hired to coordinate fieldwork and data collection and review.  He served from late 1985 to May 1991, when he became incapacitated.  Biggs was replaced by Bill Pranty, who served as the State Coordinator until the contract terminated in February 1992.  A part-time secretary completed the paid staff at FAS.

The computer database for the project was subcontracted to Dr. Stephen R. Humphrey of the Florida Museum of Natural History at the University of Florida.  Dr. Humphrey hired Bradley M. Stith, a graduate student and computer programmer to develop programs and procedures for a personal computer database.

Atlas Coordinators

            To coordinate the Atlas fieldwork, data collection, and review, Florida was divided into 9 regions (see Figure 1.1).  Nine regional coordinators were appointed and most counties appointed a county coordinator. Regional coordinators were responsible for maintaining contact with county coordinators and ensuring that fieldwork proceeded.  County coordinators also assisted with recruiting and training cooperators.  For those counties without a county coordinator, the regional coordinator assumed responsibility. We are grateful for the work and efforts of all of our coordinators, who were the heart of the Florida Atlas project.

Figure 1.1
map showing bba regions
Click for larger Image

The experience of other Atlas projects alerted us to the need to hire field personnel to survey Atlas blocks in remote or inaccessible parts of the state where few birdwatchers lived.  Provisions for "blockbusting" teams in years 4 and 5 were included in the original application for funding but were not granted by the Commission.  A grant from the Frank Stanley Beveridge Foundation provided funds to support 4 field-workers for several weeks in the summer of 1988.  Other funds enabled the Atlas project to support 1 person for several weeks in 1989.

In early 1990, faced with the likelihood that the project would not be completed without supplemental atlassing, the Commission agreed to provide funding for a sixth year of fieldwork.   Additional funds provided by the Commission enabled the project to hire 4 fieldworkers in the spring and summer of 1991 and to put the headquarters' staff in the field fulltime during the 1990 and 1991 breeding seasons.

In addition, many of our regional and county coordinators and atlasers, aware of our funding difficulties, put forth extraordinary efforts during the 2 final years of the project to reach as many incompletely surveyed blocks and quadrangles as possible, either alone or with the help of cooperators.

Each year during the Atlas project, the FOS sponsored a blockbusting event during the Memorial Day weekend at a location in north Florida.  Concentrated survey efforts by expert Atlasers in 1 or 2 counties over a 3-day period enabled a considerable amount of work to be accomplished.

Format of the Species Accounts

There is currently a wealth of publications dealing with Florida's avifauna.  In 1992, the FOS published its sixth special publication, Florida Bird Species: An Annotated List, by William B. Robertson, Jr., and Glen E. Woolfenden.  This book summarizes the ornithological record of every species found in the Florida from the state's discovery by Europeans through 1991.

The Birdlife of Florida, by Henry M. Stevenson and Bruce H. Anderson, was published in 1994 by the University Press of Florida. It is the long-awaited replacement for Arthur Howell's 1932 classic, Florida Bird Life, revised by Alexander Sprunt II in 1954.  The Birdlife of Florida is the culmination of more than 20 years of work by the late Henry Stevenson and also was funded by the Commission.  It presents a detailed account of the taxonomy, history, distribution, and biology of every bird species reported in the state through 1993.

The Endangered Biota of Florida series published by the Florida Committee on Rare and Endangered Plants and Animals (FCREPA) was recently revised.  The bird volume, edited by Rodgers et al., a revision of Kale (1978), was published in 1996.

Existence of these published references eliminates the need to be repetitive, hence, the species accounts in this Atlas are intentionally brief.  Each account includes an introduction and paragraphs dealing with breeding occurrence and distribution in reference to the accompanying map, as well as brief discussion of habitat, diet, and nesting biology.

Concurrent with the Atlas project, a systematic statewide inventory of breeding bird colonies was conducted in 1987-1989 by the Nongame Wildlife Section staff (Runde et al. 1991; Runde 1991).  The goal of that project was to revisit colony sites surveyed in 1976-1978 (Nesbitt et al. 1982) and to resurvey the entire state to locate all extant colonies.  Originally, we planned to add those data into the Atlas database.  However, most of the colony data was obtained by aerial surveys, which estimated numbers of adults but did not specify whether the adults were on a nest or attending young, the criteria necessary to confirm breeding in the Florida BBA.  Undoubtedly, in most cases it is valid to presume that a species is breeding in a colony containing numerous individuals of that species, but because we could not verify these criteria for all of the surveyed colonies they were not included in the Atlas database.  It is interesting that the Atlas database includes colonies not recorded by the Commission survey; some of these may represent colonies found outside the Commission's survey period.

Methods

Grid and Sampling Design

            U.S. Geological Survey topographic maps, which are mapped in minutes of latitude and longitude, provided the basic grid for the Florida BBA.  Each quadrangle map is 7.5 min (equal to 1/8 of 1 degree). Florida is covered by approximately 1,037 quadrangles, several of which overlap into Alabama and Georgia and contain so little area in Florida that they were excluded from the project, as were 3 quadrangles in Florida Bay that contain no land forms.  Hence, the actual number of quadrangles surveyed was 1,028.


Each quadrangle was divided into 6 numbered blocks (see Figure 1.2).  Initially, our goal was to survey every block in a quadrangle, and in some parts of the state this was accomplished.  Aware of the enormity of such an endeavor in a large state with remote areas and a limited number of qualified workers, we established a minimum goal of 2 priority blocks per quadrangle.  These blocks represent a combination of stratified-random and nonrandom sampling.  One priority block was the southeastern corner block (Block 6).  In those cases where Block 6 was inaccessible or contained more than 50% open water, the next lower numbered block that qualified became the priority block.  The second priority block in each quadrangle was selected by the cooperator in consultation with the county or regional coordinator. 

Figure 1.2. 
The Florida BBA grid system: the quadrangle and its 6 numbered blocks.

1

4

2

5

3

6

Adequacy of Coverage

Atlas work in other states had shown that 75% or more of the breeding species present in a block could be found in 16 to 20 field hours by checking all the major habitat types present.  Additional time might have raised the percentage, but in cases where many blocks needed to be surveyed and workers were scarce, atlasers moved to another block when they reached or approached 16 to 20 field hours. 

In Florida, with its extended breeding season and few experienced Atlasers, we suggested that a minimum of 25 hours be spent in a block before moving to another block.  This 25-hour rule of thumb is insufficient if the cooperator is inexperienced or if access to the block is restricted.  Conversely, in areas of habitat homogeneity (i.e., the vast marshes of the Everglades), fewer than 25 hours were needed to finish a block.

A second guideline, and probably the one most used in Florida by experienced Atlasers, was to list all the species likely to be nesting in the block and then work the block until 75% or more  of the species were recorded and, secondarily, at least 50% confirmed. This secondary goal was not pursued when blockbusting an area.

Survey Procedures

 Fieldwork was conducted from 1986 through 1991, a scheduled 5-year effort, followed by a 1-year cleanup. During the fall of 1985, training workshops were held in each region to familiarize coordinators and cooperators with materials and techniques.  Twelve issues of the Florida Breeding Bird Atlas Newsletter were sent to nearly 1,500 participants and interested parties.

In addition, contact was maintained with coordinators through letters, phone calls, and personal visits, and in January or February during 1987-1990, 2 coordinator conferences --- 1 in Tallahassee for north Florida and 1 in Lake Placid for south Florida --- were held to discuss progress, procedures, and problems and to exchange ideas and information.

During the first several years, many coordinators conducted workshops with their cooperators, and some of them produced their own area newsletters.

Each cooperator received a packet containing a Handbook for Cooperators, A Guide to Breeding Ranges, Seasons, and Habitats, Field Data Cards (Figure 1.3 and Figure 1.4 ), Casual Observation Cards (Figure 1.5), and Verification Report Forms.

Figure 1.3
Field Data Card Back
Field Data Card - Back><br>
              <span class=Click for larger image
Figure 1.4
Field Data Card Front
Field Data Card Front
Click for larger image

Figure 1.5
Casual Observation Card
Casual Observation Card

Breeding Criteria and Codes

The Florida BBA project borrowed freely from other Atlas projects, and the 20 breeding criteria codes selected represent what we considered to be those most appropriate for use in Florida.  These are listed below.  Except for the multiple male codes (discussed below), our codes and criteria follow the standardized behavior code system recommended for North America (Laughlin et al. 1982a).  Categories are arranged in a hierarchical order from least certain (Observed) to most certain (Confirmed).

Observed
O
Species observed in its breeding season but outside of nesting habitat, and with no evidence of breeding.
Possible Breeding
SHSpecies (male or female) observed in its breeding season in suitable nesting habitat.
SMSinging males present or breeding calls (or drumming by woodpeckers) heard, in suitable nesting habitat in breeding season.
Probable Breeding
SSeven or more singing territorial males observed only once in the breeding season.  (Note: This code was downgraded to the possible category when the Atlas maps were produced.)
PPair (male and female) observed in suitable nesting habitat in breeding season.
TBird (or pair) apparently holding territory (e.g., a singing male present on at least 2 days, a week or more apart, at the same place) or territorial behavior observed (e.g., chasing conspecifics or responding to tape playbacks).
CCourtship behavior or copulation.
Visiting probable nest site by any species, or nest building by woodpeckers and wrens.
AAgitated behavior or anxiety calls from adults, suggesting probable presence of nearby nest or young.
BBrood patch on adult female (or male when applicable) or cloacal protuberance on adult male.
Nest building or excavation of a nest cavity (except by woodpeckers or wrens).  A bird carrying nesting material, excluding far-ranging species.  (Note: This code was upgraded to the confirmed category when the Atlas maps were produced.)
Confirmed Breeding
SESeven or more singing territorial males observed singing on at least 2 days, a week or more apart, in a block. This code is limited to territorial males only and is to be used rarely.  (Note: This code was downgraded to the probable category when the Atlas maps were produced.)
DDDistraction display or injury-feigning, or attacking humans or other predators.
NUUsed nest or egg shells, correctly identified.
FEFemale with egg in oviduct (for banders).
FYRecently fledged young or downy young (i.e., young incapable of sustained flight).
ON Adults entering or leaving a nest site in circumstances indicating an occupied nest.  High nests, nest holes, or nest boxes, the contents of which cannot be seen, or an adult sitting on a nest, contents of which are unknown.
FS Adults carrying fecal sac or food for young.  Could not be used with raptors, gulls, terns, or crows.
NENest with eggs.  If the nest also contains a cowbird egg, record NE for both host and cowbird species.
NYNest with young seen or heard, or identifiable dead nestlings.  If the nest contains a young cowbird, record NY for both host and cowbird species.

Initially, we included a multiple male code (S) for 7 or more singing territorial males present in suitable nesting habitat during the breeding season.  This was the lowest code in the probable category, and, if it was repeated in the same area a week or more later, we allowed this to be upgraded to SE, the lowest code in the confirmed category.  After the first year of fieldwork, the North American Ornithological Atlas Committee voted to not recommend use of these codes (Smith 1990).  Our decision was to continue to accept them but to discourage their use as much as possible.  Upon termination of fieldwork we decided to downgrade these codes, assigning "S" to the highest code in the possible category and "SE" as the highest code in the probable category and restricting its use to only those species with relatively small territories.

Data Collection

Field Data Card.

            Cooperators surveying a block recorded observations on a field data card (Figure 1.3, Figure 1.4) that listed virtually all the species known to breed in Florida.  Appropriate breeding criteria codes, quadrangle name, block number, county, year, recorder's name, address, phone number, and date and hours of each visit to the block were recorded.  Because quadrangle boundaries sometimes overlap county lines, we recorded the name of the county to which the quadrangle was assigned, which was not necessarily the county in which the observer was recording data.  Space was provided on the cover page to record up to 12 visits, showing day, month, hours, number of observers, and number of new species. During the first year many observers did not record their hours.  In some instances it was difficult to calculate hours spent atlassing when one lived in the atlas block and recorded species while traveling between home and work, for example.  In such cases we asked atlasers  to estimate.

The Species Observed column (O) was originally designed for 2 reasons:  1) to make a record of a species seen in a block, even though no habitat or evidence of breeding existed; and 2) to serve as a reminder to the atlaser to search for that species in appropriate habitat somewhere in the block or quadrangle.  Unfortunately, this category created more confusion than clarity, especially on the first edition of the field card where it was listed under the Breeding Evidence heading.  Cooperators often used it as a catchall, even listing common breeding species that should have been placed under the possible category as a species in suitable habitat (SH).  During final data review, many O codes that were misused were upgraded to SH.

Because we realized the value of obtaining quantitative data on bird populations, we included an abundance column (A) and established codes for estimated numbers of pairs in a block.  It quickly became apparent that we were asking too much of our volunteers to survey as many blocks as possible and, at the same time, to count or estimate the total number of individuals in the block.  Few cooperators recorded abundance data; of those who did, some listed actual numbers, some used the codes, and in some cases we could not determine what the numbers represented.  This Atlas omits all references to estimated abundances.

In 1987, we revised the field data card to add, in phylogenetic order, several new species that had appeared the first year.  Subsequently, this caused difficulty in data checking and verification because the order birds were listed differed among cards.

In hindsight, a major shortcoming of the field data card was the lack of a column in which the recorder could list the date of an observation.  This would have saved us considerable time and effort spent in contacting atlasers to verify the presence of a species within its safe-dates, especially 3 or 4 years after the data were collected.

Casual Observation Card.

The idea for the Casual Observation card (Figure 1.5) came to us from the Ontario Atlas project.  It was designed to encourage reporting of breeding evidence for 10 or fewer species in a block that the observer was "casually" visiting or passing through.  Ironically, Ontario curtailed its use for technical reasons, but we found it a convenient and helpful tool in Florida.  An improvement for a future Florida BBA would be to design the identification heading to conform to that on the field data card.

Verification Report Form.

 Cooperators were requested to submit Verification Report Forms (VRF) for any species considered to be of restricted occurrence as breeding birds or otherwise of uncertain or unclear status in Florida.  These species were asterisked on the field data card.  Also, any species found breeding outside of its known breeding range required a VRF.

Data Processing and Analysis

Computer Database.

            Breeding Bird Atlases provide a natural application for a computer database system.  Older atlases were done by hand or on mainframe computers.  At the beginning of the Florida Atlas project, we used an IBM PCAT microcomputer with Dbase III+ software.  This system had to be upgraded as the project progressed, but advances in computer technology were so rapid that near the end of the Atlas project, a relatively inexpensive home personal computer could handle the entire database system.  The final corrections, printouts, and original maps all were made on an Intelbased 486/33C Gateway computer with a Hewlett-Packard LaserJet IIIp printer.

Data Input Routine.

Considerable programming effort was directed at developing the data entry software.  The primary goal was to reduce data entry errors while permitting rapid touch-typing.  Display screens were designed to mimic the field card in appearance to provide intuitive data entry.  Extensive error checking was built into the software.  Spellings of quadrangle names, bird codes, breeding evidence, and years were checked against data dictionaries while the typist was entering the data.  We found it necessary to switch from Dbase to Foxbase software to handle a typing speed of 80 words-per-minute.  Certain data, such as the region and county, did not have to be entered since the data dictionary for the quadrangles contained this information.  Whenever a typing error was detected, the software would produce an audible alarm and an error message, and the user could browse the data dictionaries to find the misspelled entry.

To reduce the size of the final database, only a single observation code was allowed for each bird species in each block. Upgrading breeding evidence for a species in a given block simply required typing in the new breeding code.  We put rules in place to prevent accidentally downgrading the breeding code for a species already in the database.

Quality Control Printouts.

As part of the data checking procedure, a program was written that produced a legal-sized page printout of the data that exactly overlaid the original field cards.  Errors of omission and commission were detected by placing the printout on top of the card on a light table.  Unfortunately, this procedure was only effective if all of the data for each block were on a single field card.  After the third year it became evident that this requirement could not be met because it created a lot of extra work for volunteers and staff, introduced another source of transcription errors, and because data were arriving from other sources, such as casual observation cards.  The overlay printout was abandoned in favor of a more condensed version that fit all of the data for the entire quadrangle on a letter-sized page.  This reduced the number of printouts from several thousand pages to a more manageable 1,028 pages.  These printouts also provided a column showing the highest breeding code for each species for the entire quadrangle and summarized the total number of confirmed, probable, and possible breeders in each analysis.

Nearest Neighbor Analysis.

To facilitate the planning of blockbusting and to help identify species that were likely to breed in each block, the quality control printouts made after the third year included information about the species found in the 4 neighboring quadrangles (some quadrangles had fewer than 4 neighbors).  This summary quickly identified quadrangles with weak coverage compared to their neighbors.  The final printout retained a summary table for the surrounding quadrangles.  Also, a "7th block" was included on the final printout to include species recorded in the quadrangle by AtlaserAtlasers who did not record block numbers.  This situation arose during the last year of the Atlas project, when some Atlasers went out "quadrangle busting" for missing species, and, contrary to instructions, did not record block numbers.

Data Conversion.

Software was written to incorporate data from outside sources into the database.  Mickey Wheeler, the coordinator for Region 9, devised her own database system, and a conversion program was written to handle her data.  Aerial survey data from state and federal agencies for wading birds, Snail Kites, and Bald Eagles had to be converted from latitude and longitude coordinates, obtained from LoranC, to block locations within quadrangles.  Conversion to a block location was fairly simple because the quadrangle dictionary contained the latitude/longitude centroids for each quadrangle.  Software was written to read in an ASCII file or to accept keyboard entry of latitude/longitude coordinates.

Throughout the project, the central office where data were received and corrected and the computer facilities where data were entered were in different locations.  This contractual arrangement sometimes created delays but generally worked satisfactorily. Copies of the database, stored electronically and on paper, have been deposited at the Commission headquarters in Tallahassee, the Florida Museum of Natural History in Gainesville, and the FAS headquarters in Florida.

Mapping

Maps form the heart of the Breeding Bird Atlas, and various mapping options were explored.  An inexpensive mapping program called AtlasGraphics was used initially to produce work maps.  The mainframe version of Sasgraph was evaluated for producing final maps but proved to be too expensive and inflexible.  Toward the end of the project Arc/Info, a well-known geographic information system (GIS), was acquired to produce work and final maps.  Programs were written to generate sets of maps with Arc/Info on a laserjet printer.  This worked satisfactorily, except that several days of continuous processing were required to produce a set of maps for all species.  In the end, a program was written in Dbase to produce the final maps.  This program generates HPGL/2 standard plotter commands, which are available in many plotters and in the Hewlett Packard LaserJet III family of printers.  By working directly with the database and the printer, this program provided maximum control over map corrections and was much faster than the other alternatives that were evaluated.

Although all data were recorded and computerized at the block level, species maps in this Atlas show distribution by quadrangles.

Final Quality Control

A Breeding Bird Atlas project reliant upon hundreds of volunteers, including backyard birdwatchers and beginning birders, must maintain strict (even ruthless) control over the submitted data.  Otherwise, the value of the database will be compromised by inaccurate data. Insomuch as the database is the sole source of new information published in this Atlas, tainted data could destroy the project's basic premise, which is to accurately map the breeding distributions of every species of bird that breeds or might breed within Florida.

A defense against the submission of bad data would be to limit the project only to birders with proven identification skills.  However, this is not feasible in a state as large as Florida, and for a project that relies almost entirely on volunteers. There are simply an insufficient number of experienced birders to thoroughly survey the state in only 5 or 6 field seasons.  Therefore, we accepted all the help that was offered.

The 5 main types of errors committed by Atlas observers were: simple transcription errors resulting from a "slip of the pen," errors in identification, incorrect usage of BBA codes (e.g., nest-building for Common Nighthawk [N] or singing male [SM] for Red-shouldered Hawk), problems associated with the safe-dates policy, and undocumented extralimital records.

Transcription errors were commonly committed by anyone who filled out an Atlas data card.  Misidentification was largely restricted to less-experienced observers.  When BBA codes were incorrectly used, records were changed to the most reasonable code.  Within the safe-dates policy, editing to comply with safe-dates periods for species was unambiguous.  The more challenging task was handling the "safe" records that probably referred to late spring or early fall migrants rather than locally breeding birds.  Finally, dozens of out-of-range breeding reports (sometimes the result of misidentification) were not accompanied by VRFs or other written documentation.

It is difficult to safeguard against transcription errors, and many were committed. Some transcription errors were easy to spot because they showed up on the preliminary maps as species far out of their known breeding range and/or species not in suitable breeding habitat.  For example, in one instance a Mangrove Cuckoo was reported from an inland county.  In another, a Spot-breasted Oriole was recorded in a Panhandle county.  In reality, these were a mistranscribed Yellow-billed Cuckoo and an Orchard Oriole, respectively.  However, not all transcription errors involve extralimital records.  A Red-bellied Woodpecker sighting, for example, might be mistranscribed as a Downy Woodpecker. This type of transcription error would not be noticed because both species share the same distribution in the state.  Undoubtedly, minor transcription errors of common species appear on the maps in this Atlas.

The only transcription errors to be avoided are the ones that would extend a species' breeding range into an area where it does not breed.  We are confident no such transcription errors remain in this Atlas. 

Another type of error committed was misidentification of a bird, sometimes resulting in an extralimital breeding report. For example, a Pine Warbler might be misidentified as a Yellow-throated Vireo.  Because the Yellow-throated Vireo is not proven to breed south of central Florida (although its range is increasing southward) and Pine Warblers breed throughout most of the state, a misidentified Pine Warbler carrying food in south Florida would extend the vireo's "confirmed" breeding range.  This one misidentification would alter the Yellow-throated Vireo's range map.  Because this Atlas will be referenced (and quoted) and because many users of the book may use the maps exclusively rather than in conjunction with the text, such errors could decrease its value.

A multifaceted system was created to protect the Atlas database from erroneous identifications and extralimital records.  The first safeguard was the VRF.  This form was required for a species that was observed out of its known breeding range.  If the observer failed to submit a VRF, the appropriate county and/or regional coordinator was expected to contact the observer, and request that a form be submitted.  If this was not done, the state coordinator needed to contact the observer.

For whatever reason, a VRF was not supplied for many of the sightings that required verification.  Because it was not feasible to ask for details of a sighting that occurred up to 6 years ago, in late 1991, Pranty and Kale reluctantly chose to delete all records reported by observers with unknown field skills.  For rarities reported by knowledgeable observers, we contacted as many of the observers as possible to rule out the possibility of transcription errors.  When an observer could not be contacted, we deleted the report.  This conservative approach, a standard practice (see guidelines by the FOS Records Committee), was truly our only alternative. 

We added some breeding records that were not (but should have been) documented by atlasers; in these cases, the species accounts indicate these records were not verified.  "Confirmed" breeding records were edited the most severely, as only this code implies positive breeding in a quadrangle.  For the most part, "possible" and "probable" codes of species not too far out of their normal breeding range have been retained on the maps.  We deleted a number of other records from the Atlas database because there was insufficient time to verify them.  Some of these may indeed be accurate and documentable by the observer, and should be published.  The Florida Field Naturalist, the official journal of the FOS, is the obvious outlet for publishing these reports.

Even though each year's data were computerized within months of being received, maps for every species were not produced until the end of the sixth field season.  We wholeheartedly recommend that future Atlas projects generate maps for all species yearly in order to facilitate quality control.

In the absence of species maps, printouts of the data were generated by block (1986 and 1987) and by block and quadrangle (1990 and 1991).  Due to problems with the data-entry program and quality-control methods, printouts were not produced for the years 1988 and 1989, except for one county in the state (Pasco), which was the "guinea pig" for the quality-control procedures being enacted.


Although full sets of maps were not printed until August 1991, sample maps were produced in earlier years.  Inspection of these made it obvious that many migrants and winter residents were being listed as potential breeding birds.  Atlasers were starting too early in the spring and perhaps continuing fieldwork into the fall.  This was most evident for species with large numbers of migrants or winter residents but few breeding birds (e.g., Gray Catbird), for species with lengthy migratory periods (e.g., Barn Swallow), and other birds that begin fall migration as early as late June or early July (e.g., Louisiana Waterthrush).

A data Quality Control Committee (QCC) was established to assist in cleaning up the database.  The QCC was staffed by several of the most active coordinators.  Their primary task was to determine ways to purge the database of records that did not refer to breeding birds, and to prevent such records from being submitted by atlasers during the remaining years of the project.  The most significant policy enacted was the creation of "safe-dates" described below.

Safe-dates

In June 1988, the "Guidelines for Safe-dates" was published in the BBA Newsletter Number 6.  These dates were established for most of Florida's breeding species to delineate when lower BBA codes (i.e., those that did not indicate strong evidence of local breeding) could be used safely (i.e., after most spring migrants have departed but before most fall migrants have arrived).  Simply, each species that bred in the state was given a pair of dates that restricted its inclusion in the Atlas database.  If the species was seen outside this period, it was not considered to be a breeding bird unless specific breeding behaviors were observed.  The codes restricted by safe-dates were: SH (suitable  habitat), SM (singing male), S (7 or more singing males), P (pair), T (territorial behavior), and SE (7 or more males singing on at least 2 dates, at least 1 week apart).  The basic concept of safe-dates was straightforward, but implementing it turned out to be a tremendous undertaking.

The following procedure was required to clean up the database for safe-dates.  Each Field Data Card or Casual Observation Card was reviewed to determine if any of the records were outside the species' safe-dates period.  If all the atlassing dates on the card were within a species' safe-dates period, the record was retained.  If all the atlassing dates on the card were outside a species' safe-dates period, the record was deleted.  If the atlassing dates overlapped these periods, the county or regional coordinator was notified and asked to contact the observer.  If the observer knew the record was outside the safe-dates, the record was deleted.  If the observer could remember no details of the sighting, it was our plan to delete it.  However, this proposal caused a major uproar among various coordinators.  They argued, for example, that a Northern Cardinal listed as T should not be deleted, even if some atlassing was conducted in April (outside the bird's safe-dates period of 1 May to 1 August). 


To solve this problem, the QCC allowed atlasers to retain these questionable codes using 1 of 2 methods.  The first method involved the atlaser locating a specific date for an observation from a journal, calendar, or other means.  The second method created a "memory code."  This code was to be used solely for species that the atlaser remembered seeing or hearing repeatedly throughout the breeding season.  Insomuch as this denotes territoriality, the memory code was to be used only for the T or SE codes.  It could not be used for SH, SM, or P codes, because these codes implied a single observation (i.e., if the bird or pair was observed more than once, it would have been upgraded by the atlaser to T or SE).  Unfortunately, the revised guidelines sent to QCC members and county and regional coordinators failed to make this clear.  As a result, the memory code was misused, and hundreds of SH, SM, S, and P records that should have been deleted were instead retained.  This became apparent in August 1991 when a complete set of maps and printouts was produced, and it was obvious that many errors still existed in the database.  The formidable task of reviewing each of the 136,000 codes in the database was undertaken by Pranty and required 300 hours between August 1991 and January 1992.  First, all 15,000+ Atlas data cards were inspected to ensure that all the safe-date sensitive codes contained on the cards were within the species' safe-dates. If they were not, and could not be rescued via the memory code, they were deleted.  More than 8,000 changes were made, representing almost 6% of the codes in the database.

When new printouts were generated in late January 1992, the data from 27 (of a total of 1028) randomly selected quadrangles (3 from each region) were reviewed again, comparing the data present on the cards with the revised database in the computer.  The remaining error rate was found to be 1.7%.  This error rate, representing about 2,000 potential errors out of a database of 136,000 records, was considered unacceptable.  In April 1994, Pranty was contracted by the Commission to conduct a second review of the BBA database and, assisted by Stith, to produce new, color quadrangle maps and new, black-and-white block maps.  This second review, completed in October 1994, required 4,000 additional corrections to the database. Of the 10,344 quadrangle-based codes (in a sample of 162 quadrangles) rechecked, 9 errors remained.  If the 162 quadrangles are representative of the entire database, the database should be about 99.91% error-free.

Limitations and Biases

Several factors influence the results of any BBA project, including varying degrees of ability and experience and the distribution of volunteer participants.  Problems with access to property and habitats played a major role in achieving complete coverage in the Florida BBA.  Sometimes owners or managers of land denied access to atlasers, fearing litigation in the event of accident or injury.  One owner of several large ranches in central Florida, who frequently welcomes Audubon groups to visit and birdwatch, denied access because the project was associated with the Commission.  Two of the largest private landowners in the state, Lykes Brothers, with huge holdings in Glades County, and the Mormon Church, owner of Deseret Farms in Orange and Osceola County, would not allow access because they feared the project might find endangered or threatened species.          

Often, we were able to obtain some breeding information by interviewing managers or field-workers knowledgeable about the local avifauna, by driving public roads bordering the lands, or by surveying waterways penetrating the lands.  Some areas that were legally accessible required the use of 4-wheel drive vehicles, which were not available to most atlasers.

Florida has an abundance of professional biologists and ornithologists, but it was greatly disappointing that so few of them were active participants in the Atlas project.  Many state and federal biologists freely shared information when interviewed but rarely volunteered it.  Surprisingly, a number of Florida's "high-intensity" birders showed a lack of interest in the Atlas.  These resources could have made a major contribution to the Atlas.  We urged county coordinators to visit parks and fire towers and other similar facilities to personally contact people who spend considerable time in the outdoors.

Most Atlasers conducted their work during the peak of breeding, April-June, and during daylight hours, hence early breeders and nocturnal species were probably missed in some blocks.  Unless coordinators made special efforts to survey these species, blank spots on data cards and maps could mean lack of coverage, species absence, or both.

One limitation of the Florida BBA needs to be mentioned here.  Because the database contains only the highest code per species per block, it is not possible to search the database in order to retrieve every record.  To do this one must search all the original data cards.

A Blockbusting Postscript

Every Atlas project has its share of anecdotes --- humorous, dramatic, even hair-raising --- long to be remembered and recounted at gatherings of Atlasers.  We hope that none is as frightening or tragic as the experience of Florida's State Coordinator, Wes Biggs, as he was blockbusting for nocturnal species on the night of 30-31 May 1991.  At about 12:30 A.M., Wes pulled over to the side of U.S. Highway 90, about 5 miles east of Lake City in Columbia County.  He parked in the well-lighted parking lot of a tavern to record an Eastern Screech-Owl and Chuck-will's-widow he had just found in that quadrangle.

A car pulled up alongside and Wes looked up as an AK-47 assault rifle was fired through the window.  The bullet tore into Wes' left forearm, which had been resting on the steering wheel, destroying muscle, nerves, and arteries. His assailants (someone in the tavern looked out and saw 2 men) sped off to the east.  Realizing that he was bleeding to death, Wes, a former Marine Corps medic, stopped to remove his belt and use it as a tourniquet.  He then drove on to Lake City, where he was flown to Jacksonville for emergency surgery. He has recovered partial use of his left arm, but will bear the handicap for the rest of his life.

Police speculate that his assailants were the same 2 men who had shot up a house in Lake City with an AK-47 earlier that evening and then fled town.  Upon seeing Wes with his inside light on and the car engine running, they may have thought he was a police officer waiting for them.

In Memoriam

During the course of a multiyear project involving more than 1,700 individuals, it was perhaps inevitable that we should lose some of our active participants through death.  We remember them here to commemorate their contributions to the Atlas project and to ornithology in Florida.                           

Lawrence Alexander, Franklin County Cooperator
Ray Gilbert, Hillsborough County Coordinator    
Erma J. Fisk, Benefactor
Herbert W. Kale II, Project Director   
Margaret Lynch, Citrus County Coordinator          
Virge Markgraf, Duval County Altasser      
Perry W. Oldenburg, Region 3 Atlaser 
John Sauer, Putnam County Atlaser      
Henry M. Stevenson, Atlas Advisory Board Member