Distribution and Abundance of Breeding Seabirds Along the Coast of Florida,

Transcription

1 Distribution and Abundance of Breeding Seabirds Along the Coast of Florida, Jeffery A. Gore Julie A. Hovis Gary L. Sprandel Nancy J. Douglass Final Performance Report July 2007 Florida Fish and Wildlife Conservation Commission Tallahassee, Florida

2 Suggested citation: Gore, J. A., J. A. Hovis, G. L. Sprandel, and N. J. Douglass Distribution and abundance of breeding seabirds along the coast of Florida, Final Performance Report, Florida Fish and Wildlife Conservation Commission, Tallahassee.

3 TABLE OF CONTENTS ABSTRACT INTRODUCTION STUDY AREA METHODS Ground Colonies Roof Colonies Data Analysis RESULTS Abundance Distribution Breeding Site Fidelity Important Sites DISCUSSION Abundance and Distribution Southeastern Coast Species Peripheral Species of the Florida Keys and Dry Tortugas Rare or Incidental Species Important Breeding Sites Other Breeding Sites MANAGEMENT IMPLICATIONS ACKNOWLEDGMENTS LITERATURE CITED TABLES FIGURES APPENDICES

4 Distribution and Abundance of Breeding Seabirds Along the Coast of Florida, Jeffery A. Gore 1 Julie A. Hovis 2,3 Gary L. Sprandel 4,5 Nancy J. Douglass 6 1 Florida Fish and Wildlife Conservation Commission, 3911 Highway 2321, Panama City, FL Florida Fish and Wildlife Conservation Commission, SW 10 th Street, Ocala, FL Current address: U.S. Air Force, 345 Cullen Street, Building 218, Shaw AFB, SC Florida Fish and Wildlife Conservation Commission, 620 S. Meridian St., Tallahassee, FL Current address: Kentucky Department of Fish and Wildlife Resources, #1 Sportsman's Lane, Frankfort, KY USA 6 Florida Fish and Wildlife Conservation Commission, 3900 Drane Field Road, Lakeland, FL Abstract: More than a dozen species of seabirds forage in the waters along Florida s coast and nest along the shoreline. The status of breeding populations of these species is of concern because seabird nesting colonies are subject to natural and human-related disturbance along Florida s coastline. In order to gather baseline information on the location and relative importance of seabird breeding colonies in the state, we surveyed all breeding sites of 13 species of seabirds in coastal Florida from and once each year estimated the number of nests (breeding pairs) at each site. We found 132 breeding colonies at ground sites and 288 colonies on gravel-covered roofs of buildings. The number of occupied ground sites ranged annually from over the 3-year period and roof sites ranged from Sooty terns (Onychoprion fuscatus) and laughing gulls (Larus atricilla) were the most abundant species, with each averaging >20,000 breeding pairs annually. Royal terns (Thalasseu maxima), least terns (Sterna antillarum), brown noddies (Anous stolidus), and black skimmers (Rynchops niger) each averaged >1,500 pairs. The 7 other species were much less common. Most roof sites supported only least terns, but black skimmers, roseate terns, and gull-billed terns also nested on roofs. Breeding sites were not distributed evenly among 6 coastal regions and the northeast,

5 2 northwest, and southwest coasts accounted for 77-84% of occupied ground sites annually and 73-75% of roof sites. Least terns, black skimmers, and laughing gulls occupied 399, 65, and 19 different breeding sites over the 3-yr survey. The remaining 10 species each bred at 9 sites and most bred at 3 sites. Most (84%) of the ground sites were on islands, and sites not on islands supported only least terns or black skimmers. Of the 132 ground colonies, 70 (53%) were at locations reported as breeding sites prior to our survey. Most (95%) new sites were used only by breeding least terns or black skimmers. Sites with large colonies (>50 nests) were occupied 2 more often for multiple years than were sites that supported small colonies ( χ = 44.54, 2 df, P < 0.001). We found no difference among habitat types ( χ 2 = 8.73, 4 df, P = 0.68), between 2 isolated and accessible sites ( χ = 4.34, 2 df, P = 0.114), or between ground and roof sites 2 ( χ =0.86, 2 df, P = 0.652) in the number of years a breeding site was occupied. We ranked the 132 ground-nesting sites by relative importance based on number of breeding species and pairs present. The 20 most important sites comprised all ground-breeding colonies and pairs of 9 of the 13 species surveyed and the majority of colonies and pairs for 2 other species. Least terns and black skimmers bred at many sites and they were the only species with most breeding colonies outside the top 20 sites. Least terns were the only species for which most breeding pairs nested at sites outside the top 20. Managers can use the survey results to support continued and enhanced conservation measures for the most important seabird sites. INTRODUCTION More than a dozen species of colonial seabirds regularly breed in Florida (Table 1) and most nesting pairs occupy a small number of sites unevenly distributed along Florida s coast (Stevenson and Anderson 1994). Because these species nest primarily in beach or dune habitats

6 3 that are also prized by humans, their nesting colonies are often disturbed. Human disturbance includes intense activities, such as land development, that severely impact seabird nesting habitat and prevent nesting, but it also includes more subtle activities, such as beach recreation, that can greatly affect nesting and fledging success (Boersma et al. 2002). Because of the popularity of Florida s beaches, disturbance by recreational beachgoers is potentially a serious threat to breeding seabird populations in the state. Common but significant recreational disturbances include people walking or driving vehicles along beaches, people using motorboats and personal watercraft near shore, and people allowing pets to run free along the beach (Burger 1981, Rodgers and Smith 1995, Rodgers and Schwikert 2002). In addition to habitat loss and recreational disturbance, Florida s seabird populations are potentially impacted by natural and human-influenced changes in populations of their fish and invertebrate prey (see Montevecchi 2002). Chemical pollutants and floatable plastics may also impact seabirds, but their impact to populations is usually difficult to ascertain (Burger and Gochfeld 2002). As Florida s human population increases, all these impacts will likely become more common and conserving breeding seabird populations in the state will become more difficult. To assess the need for seabird conservation in Florida and to gauge the effectiveness of conservation actions, managers and biologists need some knowledge of the distribution and abundance of seabird populations. Howell (1932) provided an early account of Florida s seabirds and other authors have provided more recent detailed accounts (e.g., Robertson and Woolfenden 1992, Stevenson and Anderson 1994). Comparing these reports provides an historical perspective of change in the diversity of seabird breeding populations in Florida. For example, at least 4 of the seabird species that currently breed in Florida were not known to be breeding in the state in Howell s time (Stevenson and Anderson 1994). Despite this historical

7 4 value, the various past species accounts were not intended as baseline population assessments and, because they are primarily compilations of records from many sources and years, they do not provide a comprehensive view of the size or distribution of seabird colonies in the state at any one time. Some surveys have provided quantitative estimates of Florida s breeding seabird populations. Spendelow and Patton (1988) compiled estimates of the number of colonies in the contiguous United States from and they relied primarily on data from Clapp et al. (1983) and Portnoy et al. (1981) for data from Florida. Both of those surveys were large-scale multi-state surveys geared toward finding large, well known, or conspicuous colonies and could not provide detailed coverage of Florida. Spendelow and Patton (1988) expressed this concern when they repeatedly noted the lack of comprehensive surveys from Florida, particularly from the Gulf coast, and cautioned that their estimates did not include all colonies. Gore (1991) and Hovis and Gore (2000) provided more comprehensive surveys for single years, but they covered only northern portions of Florida. None of these past surveys attempted to identify and assess all seabird colonies present along Florida s entire coast in a single year or to compare variation among years. Past surveys also typically have not accounted for seabird colonies that use gravel-covered roofs. The use of gravel roofs as nesting habitat has been a relatively recent behavior (Fisk 1978), but colonies on roofs now account for a significant portion of the nesting population of some seabird species, notably the least tern (Gore 1991, Hovis and Gore 2000). Spendelow and Patton (1988:218) acknowledged Fisk s (1978) estimate that 21% of least terns nested on roofs, but they did not include roof colonies in their statewide population assessments and neither did Clapp et al. (1983) nor Portnoy et al. (1981). Gore (1991), Hovis and Gore (2000) surveyed roof

8 5 colonies, but only in north Florida. Therefore, no existing statewide surveys identify the location of roof colonies throughout Florida nor do they estimate the importance of roofs as nesting habitat relative to ground sites. We recently assessed the breeding populations of 13 species of seabirds in Florida (Table 1) by recording for 3 consecutive years the location and estimated size of breeding colonies along the entire coast. Specifically, our objectives were to (1) map all breeding sites used by seabirds in coastal Florida each year, (2) estimate relative abundance of breeding pairs (nests) during 1 visit at each ground site each year, and (3) estimate the relative importance of each ground site based upon the number of species and pairs present and the number of years used. Managers of coastal habitats in Florida can use this information to learn of breeding colonies on property they manage, to evaluate the importance of those colonies relative to others in Florida, and to predict or assess impacts of management decisions on the colonies. On a larger scale, this data represents a baseline measure of seabird nesting populations in Florida that will be valuable in future comparisons and as a foundation of data for developing a comprehensive management plan for seabird colonies in the state. STUDY AREA Florida s coastline extends for >1,900 km (Fernald and Purdum 1992). If estuary borders and other tidal shorelines are included, the total coastal boundary exceeds 3,600 km in length. Not included in these linear estimates are shorelines of the state s >4,500 coastal islands. In addition to having an extensive coastline, Florida covers a broad geographic area that extends >725 km east to west ( degrees longitude) and >680 km south to north ( degrees latitude). As expected for an area of these dimensions, Florida s coast is not ecologically

9 6 homogenous (Johnson and Barbour 1990) and breeding seabird colonies are not evenly distributed along it (see Hovis and Gore 2000). For our survey, we divided Florida s coast into 6 regions based upon ecological conditions that likely influence the distribution or abundance of seabird colonies (Fig. 1). These coastal regions follow closely, but not exactly, those of Johnson and Barbour (1990) and Sprandel et al. (2000), and they reflect broad regional variation in tide height, wave energy, shoreline substrate and vegetation, presence of barrier islands and protected waters, and development of coastal habitat. Johnson and Barbour (1990) and Sprandel et al. (1997) describe differences among these regions. Our survey covered the entire coastline of Florida, including all islands, and extended inland 5 km from the mainland coast. We defined the coast as the boundary between land and salt water, including all bays, sounds, and inlets. Where estuaries or the intracoastal waterway were present, our study area extended 5 km inland from the mouth of the estuary or from the waterway. The coastal boundary was less distinct where urban areas sprawled homogeneously from the coast and manmade canals and waterways were abundant. In these areas, a few colonies may be slightly more than 5 km from saltwater. METHODS We surveyed the coast of Florida primarily for breeding colonies of seabirds in the Family Laridae, i.e., terns, skimmers, and gulls (Table 1). We also collected data on breeding colonies of magnificent frigatebirds and masked boobies, which breed in the United States only in the Dry Tortugas, a small archipelago about 110 km west of Key West. Brown pelicans (Pelecanus occidentalis) often nest independently of other colonial seabirds in habitats such as mangrove islands, and their numbers in Florida are routinely monitored by other surveys (Wood et al.

10 7 1995). Therefore, although we recorded numbers of brown pelicans when nested near other seabird colonies, our survey of pelican colonies was not thorough and results are not presented here. We also did not consider double-crested cormorants (Phalacrocorax auritus) in this coastal survey because they occur in both marine and freshwater habitats and often nest inland. Our intent was to record in each of 3 breeding seasons, , the number and location of all ground-nesting colonies of seabirds in Florida and to estimate the number of breeding pairs of each species in each year. We recorded the number of nests observed rather than pairs and assumed that each nest represented 1 breeding pair. Least terns and a few other seabird species nest not only on the ground but also on gravel-covered roofs of buildings (Fisk 1978). However, roof sites present unique survey challenges, including obtaining access to the roof (Hovis and Gore 2000). Therefore, we limited our survey objective to recording the location of roof colonies and did not attempt to count the number of nesting pairs on roofs, except for species other than least terns. Ground Colonies Surveys of ground colonies proceeded in 3 steps: (1) compiling existing data on known colony locations, (2) conducting aerial surveys for new colonies, and (3) visiting known and potential breeding sites on the ground. We compiled information on ground-nesting colonies of seabirds present in Florida in from the published literature as well as from several sources of unpublished data, including the Florida Breeding Bird Atlas, the Florida Natural Areas Inventory, the Wildlife Occurrence (Observation) System (WOS), a Coastal Wildlife Questionnaire, the Colonial Bird Register, and conversations with knowledgeable biologists and local birders (Sprandel 1999). We considered sites identified in this pre-survey

11 8 search as being historical sites, i.e., sites known to have supported breeding colonies prior to To standardize data collection among observers and regions, we prepared protocols for conducting aerial and ground surveys based on procedures established during a 1993 pilot survey of nesting colonies in north Florida (see Hovis and Gore 2000; Appendices A-B). During April and May 1998, biologists from the Florida Fish and Wildlife Conservation Commission (FWC) flew along the Florida coastline to search for previously undocumented nesting colonies. Aerial surveys were conducted from helicopters or, for parts of Florida Bay, from fixed-wing aircraft, and flights were timed to coincide with initial colony formation so that birds would likely be present. In addition to recording new colonies, aerial observers also noted vacant sites that seemed highly suitable for nesting (e.g., isolated or undisturbed sandy beaches) and worthy of a ground visit. Observers also noted historical nesting sites that were obviously no longer suitable for nesting due to encroaching vegetation, beach erosion, or coastal development. No aerial surveys were conducted in 1999 or To minimize bias among observers, all biologists responsible for surveying a coastal region met in March 1998 to discuss survey and data management protocols and to review nesting seabird identification and biology. We visited all suitable historical nesting sites and all new sites each breeding season in to estimate the status and size of the nesting colony. To match surveys with peak egg-laying periods of most species and to minimize sampling errors due to renesting birds, we conducted surveys between 15 May and 30 June. Due to logistical constraints, we visited a few sites outside those dates. We considered sites to be occupied by a breeding colony if nesting activity was confirmed, i.e., if we observed nest-distraction behavior, adults on a nest, a nest with eggs or egg shells, a nest with downy young, or fledged (out-of-nest)

12 9 young. When distinctly separate groups of birds nested near each other, we typically distinguished groups >300 m apart as separate colonies and considered closer groups to be single colonies. For each confirmed colony, we recorded detailed information on location, colony size, habitat, disturbance, and management (see Appendix C). At each colony site we recorded the number of nests or breeding pairs by species. When possible, counts were made from the periphery of the colony so that nesting birds were not disturbed. This was particularly true for colonies that had mobile young that would be scattered by our activities. At some sites, we walked through the colony and counted active nests. Intrusions into colonies typically were limited to 30 minutes in order to minimize exposure of eggs or young to heat and predators. We worked within very large colonies for >30 minutes if we did not keep birds in any part of the colony off their nests for >30 minutes. We also restricted visits to early morning or evening (before 0900 or after 1600 hrs), although some exceptions were made on overcast days or for small colonies that could be surveyed in a few minutes. In some instances, exact counts of nests or pairs were not practical, and we extrapolated from counts of part of the colony or estimated the size of the entire colony. At most colonies, we used the number of nests as the best estimate of the number of breeding pairs. With few exceptions, we conservatively did not consider chicks out of the nest when estimating number of nests because we were not certain if the fledglings were part of a clutch that was already recorded at a nest. Exceptions were made at 1 site in 1999 and 3 sites in 2000 where observers recorded only fledgling least terns and no nests. In these cases, we estimated the number of least tern nests based upon a mean clutch size of 1.7 (Gore and Kinnison 1991) and an assumption of no mortality.

13 10 In addition to nesting status, we recorded latitude and longitude with a GPS receiver and noted ownership, disturbance (people, pets, vehicles, flooding, predators), presence of solitary nesting shorebirds, and habitat type. We grouped habitat as beach/dune, dredged material (including causeways), or other habitat, a miscellaneous category that included a) densely vegetated, non-beach areas; b) unvegetated, ephemeral sandbars; and c) artificial habitats such as mainland construction sites. Dredged material included causeways built to carry roads across water as well as spoil islands created by the deposition and storage of material dredged from the bay bottom. We also noted whether sites were at mainland or island locations, including the large barrier islands of the Atlantic coast. We sorted island sites into those that were accessible by vehicle or foot from the mainland and those that were isolated and accessible only by boat. Roof Colonies We used existing sources of information on seabird colonies (see Sprandel 1999) to compile a list of buildings that supported a nesting colony in past years. FWC biologists then solicited volunteer observers from a pool of past observers as well as from local nature and birding organizations. Volunteers received a detailed list and map of sites to check as well as data sheets and instructions for submitting records (Appendix D). Volunteers visited all known roof-nesting sites in a defined area to determine whether a seabird colony currently used the roof; they also recorded observations for any new, i.e., unlisted, colonies they encountered. Because of the difficulties of obtaining access to roofs and the potential dangers of walking on roofs, we asked volunteers to view the roof only from the ground. At each site, volunteers looked and listened for birds going to and from the roof. They also looked for fecal whitewash on the building and ground and for dropped fish. If no evidence of nesting seabirds was observed in 15 min, volunteers visited the site again 1-2 weeks later. Sites with no evidence of

14 11 birds on 2 visits were considered unoccupied. Volunteers did not estimate the number of birds but did record the species. When birds other than least terns were observed, an FWC biologist returned to the roof to obtain a count of nests of the uncommon species. Data Analysis Observers entered records into a central electronic database (see Appendix E). Separate databases were maintained for roof data, ground site data, and ground colony count data, and observations were organized by an identification number assigned to each colony site. We estimated change in occupancy of nesting sites between years as colony turnover (Erwin et al. 1981). Turnover between sites (T) was calculated as: T = (S 1 /N 1 + S 2 /N 2 )/2 where S 1 = number of sites occupied only in year 1, N1 = total number of sites occupied in year 1, S 2 = number of sites occupied only in year 2, and N 2 = total number of sites occupied in year 2. Results are presented as a percentage (T x 100). Because we did not count least tern nests on roofs, we estimated size of least tern roof colonies using data from 88 roofs in north Florida where least tern nests were counted in 1993 (Hovis and Gore 2000). We obtained an initial estimate by multiplying the number of observed roof colonies by 58 nests, which was the mean number of nests per roof that Hovis and Gore (2000) found in To better account for skewness in the 1993 data, we fit the nest count data to a generalized linear model with zero-truncated negative binomial distribution. To account for uncertainty in the estimate of mean number of nests per colony, random samples from a multivariate normal distribution with the same mean parameters as the generalized linear model fit were generated and used to produce 1000 sets of ( = 555) simulated statewide nest counts with the assumed distribution. The 1000 sets of 3-years simulated

15 12 statewide counts were each fit to a generalized linear model with the assumed distribution to obtain 1000 estimates of the 3-yr average number of nests per colony, the standard error of the estimate, and the 95% confidence interval for this average. These 1000 sets of values were averaged and backtransformed to the original count scale. These averages were then multiplied by the average number of statewide colonies (185) to obtain a total statewide estimate of the number of nests, standard error, and confidence limits. A statewide estimate of the 3-yr average number of black skimmer nests on roofs was similarly produced, but unlike for least terns, some counts of skimmer nests were made in Nesting skimmers were counted during 24 of 42 visits to the 27 roofs where skimmers nested at least once during , and these data were fit to a generalized linear model with a right censored negative binomial distribution (counts of 0 were included, and some counts were known only to be >0, i.e., nesting birds present). As with the least tern data, the statewide 3-yr average number of nests, standard error, and confidence limits were estimated for roof colonies with black skimmers. To estimate the relative importance of the ground sites, we first ranked each site by the number of nesting pairs of each species. Thus, the site with the most nesting pairs among all sites visited received a rank of 1 for that year and species, while the site with the next most pairs received a rank of 2, and so on. If a species did not nest at a site in any year, the site received no rank score for that species in that year. We calculated a relative importance score for each site by summing the inverse of a site s ranks for each species and year. The sum of the inverted ranks produced the final importance score or overall rank for each ground site visited during the

16 13 survey, as calculated by the following formula: site importance = 13 i= site rank 1999 site rank of species i of species i site rank of species i where i = 1 of the 13 seabird species. Thus, a higher importance score represents a site s greater relative importance to the conservation of breeding seabird populations in Florida. RESULTS Abundance We found breeding colonies of 13 species of seabirds at 132 different ground sites (Appendix F) and 288 different roof sites (Appendix G) along Florida s coast during Not all sites were occupied each year (Appendix H), and across years the number of occupied ground sites ranged from while the number of occupied roofs ranged from (Table 2). A mean of 71,230 pairs of seabirds (range 69,168 74,214) nested annually at ground sites during the 3-year survey, but the number of breeding pairs varied widely among the 13 species observed (Table 2). Sooty terns and laughing gulls were by far the most abundant species, and sooty terns had twice as many pairs as laughing gulls. Royal terns, least terns, brown noddies, and black skimmers were less common, but each averaged >1,500 breeding pairs annually. The remaining 7 species each averaged <500 pairs and annual means ranged from 3.3 pairs for bridled terns to pairs for Sandwich terns. All 13 species nested on at least 1 ground site, but we observed only least terns, black skimmers, roseate terns, and gull-billed terns nesting on buildings with gravel-covered roofs. Of the 288 roofs sites, 260 (90%) supported only least terns. Roofs supported an estimated 10,652 pairs of breeding seabirds each year, and 98% were least terns (Table 2). We initially estimated

17 14 10,711 (95% CI = 9,211-12,210) pairs of least terns nested annually on roofs, based upon the number of roofs occupied each year multiplied by the mean of 58 nests per roof reported by Hovis and Gore (2000). However, we subsequently estimated, based upon simulated counts fit to a generalized linear model, the mean of occupied roof sites supported about 10,491 (95% CI = 9,581 11,487) pairs each year or about 84% of all nesting pairs of least terns (Table 2). Least terns were the only species with more colonies on roofs than on the ground, and roof colonies of least terns outnumbered ground colonies by a ratio of >2.8:1 in each year of the survey (Table 2). In comparison, black skimmers nested on a much smaller proportion of roofs relative to ground sites (0.6:1). We counted black skimmer nests on 24 of 42 occasions when skimmers were confirmed breeding on roofs, and counts at individual sites ranged from 1-75 nests ( x = 10.0, SE ±3.09). Black skimmers nested on an average of 14 roofs annually, and using the mean of 10 black skimmer nests per roof, we initially estimated 140 pairs of skimmers nested annually on roofs. Subsequent estimates based upon count data fit to a generalized linear model produced an estimated annual mean of 5.5 (SE ±1.57) nests/roof and an annual mean of (SE ±42.35) pairs of skimmers nesting on roofs (Table 2). Roof sites supported 8.7% of breeding pairs of black skimmers (Table 2). Roseate terns and gull-billed terns nested at few ground sites and each nested on only 1 roof (Table 2). Gull-billed terns nested on a roof in Bay County and in 1999 that roof supported 4 (17%) of the 24 roof and ground nests recorded. Most roseate terns nested at a single ground site, but 35 pairs (11% of all pairs) nested on a roof in Monroe County in Distribution All 6 coastal regions contained seabird breeding sites (Figs. 2, 3), but the number of sites varied significantly among regions (ground sites: χ 2 = 73.73, P < ; roof sites: χ 2 = 79.71,

18 15 P < ; Table 3). The Northeast, Northwest, and Southwest regions together accounted for 77-84% of the occupied ground sites each year and 73-75% of the roof sites (Table 3). The relative proportion of ground and roof sites also varied among regions (χ 2 = 32.26, P < , expected 1:1 ratio). The Big Bend region had a higher proportion of ground sites than roof sites, while all other regions had a higher proportion of roof sites than ground sites (Table 3). The number of breeding pairs at ground sites also varied widely among coastal regions, ranging from a 3-year mean of 742 pairs in the Big Bend region to >42,000 pairs in the South region (Table 3, Fig. 4). The number of breeding pairs varied widely among coastal regions both for uncommon species that nested at few sites, such as the Caspian tern, and for common and widespread species. For example, least terns ranged from a mean of 22 ground-nesting pairs per colony in the South region to a mean of 740 pairs per colony in the Northwest region (Table 3). The number of least tern colonies on roofs also varied widely among regions from an annual mean of 4 colonies in the Big Bend region to 49.0 colonies in the Northwest region (Table 3). Breeding sites were not distributed evenly within each coastal region (Appendices I-N), and each region supported a different combination of breeding species (Fig. 4) that varied among years (Table 3). The South region supported as many as 8 species, 6 of which nested only in the lower Keys and Dry Tortugas. The Northwest and Southwest regions each supported 7 species, and the less diverse Big Bend and Southeast regions each contained 3 species. Each region had 1 or more species that nested at a single site. Least terns were the only species found breeding in all 6 regions, and they occurred at more sites by far than any other species (Table 2, Fig. 5). We found least terns breeding at 112 ground sites and 287 roof sites (399 total sites) during , but in any single year least terns occupied only ground sites and roof sites (Table 2). In 1999, the year with the

19 16 most colonies, we found a total of 258 breeding colonies of least terns at ground and roof sites in coastal Florida. In contrast, black skimmers and laughing gulls, the species with the next widest distributions, occurred at only 65 and 19 different sites, respectively, during the 3-year survey (Table 2). The single-year maximum was 42 colonies for black skimmers and 16 colonies for laughing gulls. Black skimmers bred in 4 regions (Fig. 6) and laughing gulls bred in 5 regions (Fig. 7). The other 10 species each bred at 9 different sites during our survey and 7 sites in any single year; most bred at 3 sites (Table 2). The few colonies of royal terns, Sandwich terns, Caspian terns, and gull-billed terns were distributed widely across multiple regions (Figs. 8, 9). Colonies of roseate terns, bridled terns, sooty terns, brown noddies, masked boobies, and magnificent frigatebirds occurred only within the South region, in the Keys or Dry Tortugas (Figs. 7 and 10), and at only 1 or 2 locations (Table 2). Of the 132 different ground sites used by breeding seabirds in , 111 (84%) were on islands (Table 4), including causeways and islands made of dredged material. The remaining 21 (16%) breeding sites were on mainland locations and these sites supported only least terns (N = 20) or least terns and black skimmers (N = 1). All mainland sites were readily accessible to people in vehicles or on foot, as were 55 (50%) of the 111 island sites. Most breeding sites occurred in beach and dune habitat on either road-accessible (32%) or isolated (19%) natural islands, or on isolated islands of dredged material (17%) (Table 4). We found no significant difference between mainland and island locations in the proportion of breeding sites present in 2 beach, dredged material, and other habitat types ( χ = 3.18, 2 df, P = 0.204). Islands not only supported more seabird breeding colonies than mainland sites, they also supported most nesting pairs (Table 4). For example, islands accounted for >95% of all least

20 17 tern nests each year and >99% of black skimmer nests. The only mainland site that supported black skimmers, Holiday Island in Okaloosa County, was a narrow peninsula that sometimes separated from the mainland. Laughing gulls always nested on islands (N = 19) with dense low vegetation. Caspian terns, royal terns, and Sandwich terns nested only on small islands (N = 9) within large protected bays, and most of these nesting sites were not accessible to vehicles. Gull-billed terns also nested primarily on small islands (N = 4), but they also nested on 1 mainland roof. Masked boobies, magnificent frigatebirds, roseate terns, bridled terns, sooty terns, and brown noddies nested on small isolated islands (N = 4) in open sea, except for 1 small roseate tern colony on a roof on Vaca Key, an island in the middle Florida Keys. Breeding Site Fidelity Historical and New Sites. We identified 275 ground sites where seabirds had been recorded breeding prior to Of these historical sites, 70 (25%) were occupied by a breeding seabird colony at least once during and 205 (75%) were vacant all 3 years (Table 5). Including the Big Bend region, which had only 4 historical sites, the proportion of previously occupied sites that were vacant during all years of our survey ranged from % among regions. Of the 205 historical sites that were vacant (Appendix O), 53 (26%) were considered to be unsuitable sites for breeding seabirds in because of excessive vegetative cover, erosion or development of the site, or intensive disturbance by people. Observers recorded the remaining 152 vacant sites as suitable for nesting, i.e., they saw no obvious reason why birds would not use the site. In addition to the 70 historical ground sites occupied in , we also found seabirds breeding at 62 new ground sites. The ratio of new to historical ground sites varied markedly among regions and only the Big Bend and Northwest regions had more new sites than occupied

21 18 historical sites (Table 5). Least terns were present at 50 (81%) of the new sites and black skimmers occupied 20 (32%) of the sites. Combined, least terns and black skimmers nested at 59 (95%) of the new sites. Laughing gulls (N = 7 sites), royal terns (N = 2), and gull-billed terns (N = 1) were the only other species that nested at new sites and only laughing gulls occupied a new site without least terns or black skimmers also being present. Temporal Variation. All 6 seabird species that nested solely in the Keys or Dry Tortugas occupied their nesting sites each year of the survey, i.e., there was no colony turnover in Four of the species breeding solely in the Keys or Dry Tortugas occupied only 1 breeding site and the other 2 species, roseate tern and brown noddy, each occupied 2 sites. Roseate terns nested at 1 ground site and 1 roof site each year. Brown noddies nested on Bush Key and adjacent Long Key in In 1998 and 1999 we recorded brown noddies only on Bush Key, but surveys on Long Key were limited and some noddies may have nested there without being detected. Unlike the species that nested only in the Keys or Tortugas, most of the 7 other seabird species in Florida nested at >2 sites and occupied a different number of sites each year (Table 2). Caspian terns nested at the same combination of sites throughout the 3 year survey, but other species varied in colony turnover rates between consecutive years (Table 6). Colony turnover between years was higher for ground-nesting least terns (mean of 2-yr pairs = 42.3%) than roofnesting least terns ( x = 35.2%, Table 6). Black skimmers showed an opposite trend with ground-nesting colonies having lower turnover between years ( x = 42.6%) than roof-nesting colonies ( x = 50.2%, Table 6). Colony turnover for ground-nesting least terns was similar at sites accessible to vehicle and pedestrian traffic ( x = 42.1%) and at isolated islands ( x = 42.4%; Table 7), but the turnover rate for colonies of black skimmers was lower ( x = 37.1%) at

22 19 isolated sites than at accessible sites ( x = 50.4%). Turnover was not calculated for the remaining habitat types or species because of small sample sizes. Species varied in the number of years that they occupied a breeding site during (Table 8). As noted above, all 6 species that nested solely in the Keys and Dry Tortugas nested at 2 sites and they occupied those sites each year of the survey. Among the 7 species that nested at >2 sites, laughing gulls, royal terns, Caspian terns, and Sandwich terns had more colonies that persisted for 2 or 3 years than colonies that persisted for only 1 year (Table 8). In contrast, more colonies of least terns and black skimmers lasted a single year rather than 2 or 3 years. Only laughing gulls most often skipped a year before returning to a site, i.e., they occupied most sites in 1998 and 2000, but not in We compared the number of years a site was occupied by breeding seabirds of any species with habitat type, accessibility to humans, and colony size (number of nests) to determine if 2 these variables influenced temporal nesting patterns. We found no significant difference ( χ = 5.22, 4 df, P = ) among the 3 most common habitat types, i.e., beach/dune, dredged material, and other, in the number of years a site was occupied. Roof and ground sites also did 2 not differ in the number of years occupied by a breeding colony ( χ = 0.86, 2 df, P = 0.652). Isolated and accessible sites did not vary significantly ( χ = 4.34, 2 df, P = 0.114), although isolated sites were occupied most often (43%, N = 56) for all 3 years and accessible sites were occupied most often (49%, N = 76) for just 1 year. Sites that supported large (>50 nests of any species) colonies of seabirds were occupied for more years and small-colony sites for fewer years ( χ 2 = 44.54, 2 df, P < 0.001) than expected by chance. Most (73%, N = 41) large colonies occupied a site all 3 years of the survey, but most (57%, N = 91) smaller colonies were present at a site for only 1 year. 2

23 20 Only least terns and black skimmers nested at enough sites to compare how site and colony characteristics affected the number of years a species occupied a site. Large colonies persisted for more years than small colonies for both least terns ( χ skimmers ( χ 2 = 26.58, 2 df, P < 0.001) and black = 12.80, 2 df, P = 0.002) (Table 9). We found no difference between isolated sites 2 and accessible ground sites in the number years a site was occupied by either least terns ( χ = , 2 df, P = 0.750) or black skimmers ( χ = 1.04, 2 df, P = 0.594). We also found no 2 difference between roof and ground sites in number of years occupied by least terns ( χ = 5.38, 2 2 df, P = 0.068) or black skimmers ( χ = 1.09, 2 df, P = 0.560). Ground-nesting seabirds abandoned 8 of 11 colonies (not including colonies new in 2000) at mainland sites during (Table 10), but the proportion of breeding sites that were abandoned did not differ significantly among habitat types ( χ 2 = 5.35, 3 df, P = 0.148). The proportion of abandoned sites also did not vary significantly between roof and ground sites for 2 2 least terns ( χ = 1.15, 1 df, P = 0.28) or black skimmers ( χ = 0.01, 1 df, P = 0.913). Thirtythree percent (N = 132) of all ground-nesting colonies persisted for 3 consecutive years. Among habitat types, other habitats had the highest proportion sites occupied all years (39%; N = 18) and dredged material sites had the lowest (26%; N = 34). Thirty-three percent (N = 288) of all roof-nesting colonies persisted for 3 consecutive years. Newly established colonies had fewer nests on average than colonies at sites that had been occupied in a previous year (Table 11). Black skimmers, for example, averaged 30.5 nests per colony in the first year they nested at a ground site and nests per colony when they had nested at the site in the previous year (Wilcoxon Rank Sum Test, P = 0.006). 2

24 21 The number of breeding pairs at individual sites fluctuated greatly among years (Appendix H), even at sites with large colonies or colonies that were present in all years. When pairs from all sites were combined, the statewide number of breeding pairs for individual species also varied annually although not in the same manner for all species (Table 2). Bridled terns, frigatebirds, and roseate terns showed the greatest relative change in number of breeding pairs among years, but each of those species nested at only 1 ground site. Species nesting at multiple sites fluctuated less annually than those at single sites, and black skimmers varied the least. To minimize disturbance, we only coarsely estimated pairs in the large colonies of sooty terns and brown noddies; therefore, those estimates did not vary annually (Table 2), even though the actual number of breeding pairs undoubtedly did. Important Sites We ranked the 132 ground-nesting sites by relative importance based upon the number of species and breeding pairs, and Table 12 lists the 20 most important sites. All but 2 of these 20 important sites were occupied each year and all but 4 supported >1 nesting species. For coastal Florida, these 20 sites comprised all colonies and pairs of breeding Caspian terns, Sandwich terns, bridled terns, sooty terns, brown noddies, masked boobies and magnificent frigatebirds, and they also included all ground-nesting colonies of gull-billed terns and roseate terns. The top 20 sites also supported 8 of 9 royal tern colonies (99.9% of breeding pairs), 10 of 19 laughing gull colonies (96.3% of pairs), and 13 of 38 black skimmer ground colonies (90.8% of pairs). The widespread least tern had only11 of 112 ground colonies among the top 20 breeding sites, but these supported 40.5% of all pairs of least terns. An even smaller group of 11 sites within the top 20 contained a large portion of the breeding seabirds in Florida in For example, the 11 most important sites contained all the

25 22 breeding colonies of brown noddies, bridled terns, sooty terns, masked boobies, magnificent frigatebirds, Caspian terns, and Sandwich terns. These 11 sites included the only roseate tern ground-nesting site (93.3% of all pairs), 6 of 9 royal tern breeding colonies (97.4% of pairs), 3 of 4 gull-billed tern ground colonies (96.5% of pairs), and 6 of 19 laughing gull colonies (87.5% of pairs). These sites also included 5 of 38 black skimmer ground colonies (72.1% of pairs), and 5 of 112 least tern ground colonies (18.3% of pairs). The number of species at a site was a component of the importance function, and as expected, sites with low importance scores had low species diversity. When ranked by importance score, the lower 76.5% of sites (N = 132) supported only least terns, black skimmers, and laughing gulls. Of those 101 sites, 79 supported only least terns, 14 had least terns and black skimmers, 3 had only black skimmers, 2 had only laughing gulls, and 3 had gulls along with either least terns or black skimmers. More of the 66 sites with importance scores above the median occurred on islands (96%) than mainland locations compared to the 66 sites with importance scores below the median (71% islands; χ = 13.96, 1 df, P < 0.001). Higher ranked sites also occurred more frequently on isolated islands (50%) than on road-accessible islands compared to lower ranked sites (32% isolated islands; χ 2 2 = 4.512, 1 df, P = 0.034). Sites with scores above and below the median did 2 not differ significantly among beach/dune, dredged, and other habitat types ( χ = 4.09, 2 df, P = 0.130). Of the top 20 sites, all were on islands and 15 were on isolated islands. DISCUSSION Abundance and Distribution

26 23 Because the number of seabirds nesting at a site can vary greatly within a season and between years (Johnson and Krone 2001), identifying changes or trends in local or regional breeding populations is difficult. This is especially true when sampling methods differ among surveys, as is the case for Florida where past surveys of breeding seabirds did not cover the entire coast as thoroughly as our recent survey (Spendelow and Patton 1988). Older surveys concentrated on areas known or likely to support seabird colonies. Statewide population estimates made from these surveys are biased because not all sections of the coast were covered in the same years and some areas were surveyed less intensively than others. Because of these limitations, comparisons among Florida surveys are not useful for closely monitoring population trends statewide or locally. However, comparisons are useful in identifying breeding sites with a long history of use, sites that have consistently supported large colonies, and changes in the statewide distribution of species and breeding sites. Also, although precise estimates of breeding pairs are not always available, changes in the relative abundance of species statewide can highlight species whose populations merit closer monitoring. With these limitations and uses in mind, we compared the survey data with data from the survey (Table 13), and for some species, with earlier accounts. For this comparison, we separated the breeding seabird species in Florida into 2 primary groups. The first group includes those species that occur widely along the coast of the southeastern United States and whose breeding range includes all or most of coastal Florida. These species each breed at more than one site in the state, they each nest along both coasts, and their Florida nesting sites are not on the periphery of the main breeding range of the species. Because these species are not restricted to particular nesting locations or regions, their Florida populations likely can be enhanced by creating or conserving additional suitable habitat at new

27 24 sites. Therefore, conservation plans should consider potential nesting habitat throughout the state, not just at current breeding sites. The second group includes species that breed in Florida only at 1 or 2 sites and only in the Keys or the Dry Tortugas. These species include the most abundant (sooty tern) and rarest (bridled tern) breeding seabirds in Florida, but they have in common that their breeding colonies in Florida are at the northern periphery of the species breeding range. Because they are already at the edge of their ranges, protecting suitable habitat elsewhere in Florida will not likely attract new nesting colonies of these species or expand their breeding range in the state. Therefore, conservation of this second group should focus on suitable sites in the Keys and Dry Tortugas. We also briefly note 3 additional species that rarely or incidentally nest in Florida. Southeastern Coast Species Laughing Gull. Belant and Dolbeer (1993) estimated about 250,000 pairs of laughing gulls bred in the eastern United States in In Florida, laughing gulls occupied 32 colonies in , and the breeding population comprised more than twice as many nesting pairs as we found in (Table 13, Fig. 7). Belant and Dolbeer (1993) concluded that the breeding population in Florida declined an average of % annually from , but the decline was likely not steady and in some years was probably steep. Nevertheless, only 3 other states exhibited any decline in breeding pairs of laughing gulls during the same period. Frohring and Kushlan (1986) reviewed available data on laughing gulls in extreme south Florida and found little evidence of breeding prior to the 1970s. However, when they surveyed Florida Bay and the Florida Keys from , they recorded laughing gulls present at as many as 19 sites, confirmed nesting at 6-14 sites, and counted nests during 3 survey years (Frohring and Kushlan 1986). In contrast, we found only 1 breeding colony in the area, on

28 25 North Nest Key in the far upper end of Florida Bay (Appendices F, L-map 3), and estimated it supported 100 nests. Even though Frohring and Kushlan (1986) were uncertain how well their nest count represented the total nesting population, the breeding population in south Florida was clearly larger in than during our survey. The breeding population of laughing gulls around Tampa Bay has also fluctuated in the past few decades. Patton and Hanners (1984) compiled historical accounts of laughing gull breeding populations in Tampa Bay and showed a rapid increase in the number of breeding pairs during the 1970s, from a few thousand pairs to tens of thousands of pairs. They estimated 48,700 and 50,300 pairs of laughing gulls nested in Tampa Bay in 1981 and 1982, respectively, with the largest colony supporting an estimated 31,000 pairs in In contrast, we estimated between 12,200 and 17,050 pairs nested in the area from (Appendix H), with the largest colony containing 8,200 pairs. The breeding population of laughing gulls in north Florida has never been large relative to south Florida and no sites contained tens of thousands of pairs (Spendelow and Patton 1988). However, the St. George Causeway in Franklin County and Huguenot Memorial Park in Duval County now support 2 of the largest colonies of laughing gulls in Florida (Table 12). Although laughing gulls are often considered a common species and even nuisance (Belant and Dolbeer 1993), our data suggest they are now less numerous in Florida and breed at fewer locations than about 20 years ago (Table 13). The lower numbers of gulls we observed supports Belant and Dolbeer s (1993) conclusion that the number of nesting gulls is continuing to decline. Laughing gull populations in south Florida and near Tampa Bay may have declined with changes in availability of food, particularly agricultural crops and garbage from landfills (Patton and Hanners 1984, Frohring and Kushlan 1986). Conversion from landfills to trash incineration