FWC Final Report by Avian Research & Conservation Institute

i FINAL REPORT

Project number:

FWC Agreement No .06014

Title:

Enduring effects of hurricanes and additive human threats for five avian Species of Greatest Conservation Need in the Florida Keys

Project director:

Kenneth D. Meyer, Ph.D. Avian Research and Conservation Institute 411 N.E. 7 Street, Gainesville, Florida 32601 Phone/fax: 352-335-4151, e-mail: meyer@arcinst.org

Report authors:

Kenneth D. Meyer 1 Gina M. Zimmerman 1 Peter J. Mahoney 1 Suzanne Beyeler 1 Thomas J. Wilmers 2 1 2

Date submitted:

Avian Research and Conservation Institute United States Fish and Wildlife Service

16 December 2008

ENDURING EFFECT OF HURRICANES AND ADDITIVE HUMAN THREATS FOR FIVE AVIAN SPECIES OF GREATEST CONSERVATION NEED IN THE FLORIDA KEYS KENNETH D. MEYER1, GINA M. ZIMMERMAN1, PETER J. MAHONEY1, SUZANNE BEYELER1, AND THOMAS J. WILMERS2 1 2

Avian Research and Conservation Institute, 411 N.E. 7 Street, Gainesville, Florida 32601 United States Fish and Wildlife Service

Date of final report: 16 December 2008 Agreement number: 06014 ABSTRACT: Hurricanes are the most disruptive natural disturbance encountered by species of the subtropical Florida Keys, which experienced significant damage from 6 hurricanes in 2004 and 2005. We studied the stormsâ&#x20AC;&#x2122; effects in the lower Keys on hardwood and mangrove forests and the abundance of 5 Species of Greatest Conservation Need: reddish egret (Egretta rufescens); Cuban yellow warbler (Dendroica petechia gundlachi); black-whiskered vireo (Vireo altiloquus); mangrove clapper rail (Rallus longirostris insularum); and white-crowned pigeon (Patagioenas leucocephala). Landsat imagery from 2001 and 2006 was used to quantify mortality of these habitats, which are critical for the focal species. There was little change in total areas of individual land covers, which we attribute to two factors. First, the distributions of these habitats changed over the 5-year period, with roughly comparable areas of scrub mangrove (11%) and hammock (14%) being damaged in some places while regenerating (from previous damage) elsewhere. Second, the 30-meter resolution of the Landsat data probably exaggerated this apparent shift in cover types by limiting our ability to detect finely-scaled changes. Most of the damage was near edges, particularly shorelines and roads, from exposure to severe winds, but considerable mangrove mortality resulted from deposition of sediment over the pneumatophores of black mangrove trees of island interiors. Development increased by 432 hectares in our comparative analysis at the direct expense of hardwood hammock (i.e., 427 hectares, or 98.8%, had been classified as hammock in 2001). Unlike what we observed for scrub mangrove and hammock, patches of development did not shift location over the study period. They were newly-created in the 5-year period and represent a loss of 10.1% of the 4,211 hectares of hardwood hammock measured in the 2001 imagery. This rapid transformation of imperiled hardwood hammock to development in the lower Florida Keys starkly contrasts with the dynamic, productive response of native habitats to what many would view as the far-more alarming effects of hurricanes. This natural, landscape-level resilience, furthermore, prevailed despite the above-average occurrence 6 powerful storms during the study period. During the 2006 and 2007 nesting seasons, we repeated point count surveys conducted from 2001 to 2004 to determine relative abundance of the 5 focal species and performed flightline counts of WCPIs to estimate nesting effort. We also banded nestling WCPIs to assess post-storm nesting effort and to determine locations and hunting impacts on Florida migrants wintering in the Caribbean. All species except BWVIs showed reduced breeding effort following the storms. The differences for CLRAs and CYEWAs were not statistically significant, perhaps due to insufficient data, but the change approached significance for REEGs and was significant for WCPIs. Of the 4 species that declined, all but REEGs rebounded by 2007. Most notable was the significant increase from 2006 to 2007 in breeding WCPIs to pre-storm levels. Like their critical habitats, these mainly-Caribbean species exhibit a high degree of resilience to the effects of severe weather. We also assessed range sizes and habitat use by WCPIs and survival of adults based on a large sample of radio-tagged pigeons (the first study using marked ii

iii individuals). Pineland habitats, which include embedded hardwood hammocks used by foraging pigeons, became more important for feeding after the storms, whereas the lower-elevation hardwood hammocks not associated with pine forest suffered more damage and became less frequently used. The increased poststorm use of these feeding sites highlights their value for landscape-level conservation planning for WCPIs. These results will contribute to future demographic modeling of the WCPI in Florida, attempts to track population trends rangewide, and efforts to assess the effects of harvest on Caribbean populations, which are subjected to substantial hunting pressure. We also found significant effects of trapping location and mass at capture on pigeon survival, supporting our observations that pigeons captured on some Keys were consistently lighter than elsewhere. This is the first evidence that variations in habitat quality affect WCPI condition and longevity, and that not all areas with high pigeon abundance on the mainline Keys necessarily represent optimal foraging habitat. Finally, we reviewed progress toward collaborations and support for Caribbean colleagues eager to apply sound data to managing their WCPI populations, particularly in the Bahamas, Cuba, and Jamaica, which support the largest numbers of breeding and wintering pigeons. Our most recent Working Group meeting, in Puerto Rico in 2007, drew participants from island nations not previously involved in the Groupâ&#x20AC;&#x2122;s activities. This is very encouraging and calls for continued communication, including finding all opportunities to provide our support for these emerging programs on WCPI management and conservation planning.

iii

iv ACKNOWLEDGMENTS We are grateful for funding from the State Wildlife Grants program of Floridaâ&#x20AC;&#x2122;s Wildlife Legacy Initiative, administered by Brian Branciforte and Kate Haley for the Florida Fish and Wildlife Conservation Commission. Thanks to the following for their help with banding nestling white-crowned pigeons on mangrove islands: Heather Mahoney, Tatiana Ramirez, Helmut Hiller, Chifuyu Beckett, Brian Mealy (who also provided his boat), Katie Lions, and Tom Wilmers. We also thank John Arnett and Tom Wilmers for the pre-hurricane point count data collection. The project would have been impossible without the access, logistical assistance, use of boats, and supporting white-crowned pigeon data provided by Tom Wilmers (Wildlife Biologist) and Ann Morekill (Refuge Manager), Florida Keys National Wildlife Refuges, Big Pine Key, Florida. We thank Suzanne Beyeler, GeoPlan, University of Florida, for GIS consulting and analyses; and Evan Adams, University of Florida, for assistance with the white-crowned pigeon survival analysis.

v TABLE OF CONTENTS ABSTRACT.................................................................................................................................... ii ACKNOWLEDGMENTS ............................................................................................................. iv INTRODUCTION .......................................................................................................................... 1 OBJECTIVES ................................................................................................................................. 5 METHODS ..................................................................................................................................... 6 Extent of Hardwood and Mangrove Mortality............................................................................ 6 Nesting Surveys, All Species ...................................................................................................... 8 Point Counts.- ......................................................................................................................... 8 Flight-line Counts.- ............................................................................................................... 10 WCPI Activity Ranges and Habitats......................................................................................... 11 Activity Ranges ..................................................................................................................... 11 Habitat use.- .......................................................................................................................... 12 WCPI Survival .......................................................................................................................... 12 Impacts on Winter Range.......................................................................................................... 14 Banding WCPI Nestlings .......................................................................................................... 14 RESULTS ..................................................................................................................................... 15 Extent of Hardwood and Mangrove Mortality.......................................................................... 15 Nesting Surveys, All Species .................................................................................................... 18 Point Counts.- ....................................................................................................................... 18 Flight-line Counts.- ............................................................................................................... 21 WCPI Activity Ranges and Habitats......................................................................................... 22 Activity Ranges.- .................................................................................................................. 25 Habitat Use.- ......................................................................................................................... 26 WCPI Survival .......................................................................................................................... 27 Impacts on Winter Range.......................................................................................................... 31 Banding WCPI Nestlings .......................................................................................................... 35 DISCUSSION AND MANAGEMENT IMPLICATIONS .......................................................... 38 Extent of Hardwood and Mangrove Mortality.......................................................................... 38 Nesting Surveys, All Species .................................................................................................... 42 Point Counts.- ....................................................................................................................... 42 Flight-line Counts.- ............................................................................................................... 43 WCPI Activity Ranges and Habitats......................................................................................... 44 WCPI Survival .......................................................................................................................... 45 Banding WCPI Nestlings .......................................................................................................... 47 LITERATURE CITED ................................................................................................................. 48

1 INTRODUCTION Hurricanes are the most disruptive natural disturbance encountered by species of the subtropical Florida Keys. Although speciesâ&#x20AC;&#x2122; adaptations can accommodate short-term habitat loss and facilitate recovery, the additive effects of human impacts compound the stress for species at risk. Few studies have investigated the impacts of hurricanes on any 1 species or habitat. Although bird populations may be directly impacted by hurricanes, they are typically more adversely affected by the subsequent habitat change (Wunderle et al. 1992, Wiley and Wunderle 1993). The Florida Keys experienced substantial damage from 2 hurricanes (Charley and Ivan) in 2004 and 4 hurricanes (Dennis, Wilma, Katrina, and Rita) in 2005 (Fig. 1) with Dennis and Wilma having the most significant effects. Hurricane Charley passed over the Dry Tortugas in August of 2004 with maximum winds of 177 kilometers/hour (110 miles/hour) and storm-surge flooding in the western Florida Keys. Hurricane Ivan crossed southern Florida and reentered the Gulf of Mexico in late September, 2004. In July 2005, Dennisâ&#x20AC;&#x2122;s winds did not exceed Tropical Storm-force for most of the time it affected the Keys, but it produced considerable damage nonetheless. Katrina made landfall near Miami in August of 2005, moved southwestward into the Gulf of Mexico, and produced 25 to 36 cm (10-14 inches) of rain in the Florida Keys. Hurricane Rita passed within 100 km of Key West in September 2005, causing sustained winds of 122 kilometers/hours (76 miles/hour) but minimal storm surge conditions. Hurricane Wilma moved across southern Florida in October 2005, bringing hurricane-force winds to the Florida Keys and the highest storm surge observed in the Keys since1965. The eye of category-3 Hurricane Wilma passed just north of the Florida Keys and caused the ocean to rise rapidly, inundating many islands (Kasper 2005). Wind resulted in failed breeding attempts for birds and

2 damaged vegetation; saltwater flooding caused mortality of upland trees and shrubs. A storm surge can kill mangroves hundreds of kilometers away by changing water or soil salinity and increasing sedimentation (Smith et. al 1994, McCoy et al. 1996). The excessive sedimentation that often results from hurricane storm surge is particularly detrimental to black mangrove (Avicennia germinans) when the pneumatophores become buried. Preliminary assessments, particularly after the October passage of Hurricane Wilma, indicated mortality in mangroves and West Indian hardwood hammocks throughout the lower Keys, an area with a particularly high occurrence of Species of Greatest Conservation Need (SGCN, per Floridaâ&#x20AC;&#x2122;s Comprehensive Wildlife Conservation Strategy). This project focused on 5 such species, which represent 10% of the Strategyâ&#x20AC;&#x2122;s avian SGCNs for which status and trend are ranked in the lowest or unknown categories. We determined how the stormsâ&#x20AC;&#x2122; impacts on hardwood and mangrove forests in the lower Florida Keys affected abundance of these species in 2006 and 2007, 1 and 2 years after the 2005 hurricane season. The 5 species were: reddish egret (Egretta rufescens); Cuban yellow warbler (Dendroica petechia gundlachi); black-whiskered vireo (Vireo altiloquus); mangrove clapper rail (Rallus longirostris insularum); and white-crowned pigeon (Patagioenas leucocephala). The project capitalized on nesting data systematically collected from 2001 to 2004 to determine the effects of the 2004 and 2005 storms on breeding abundance. The reddish egret (REEG), the rarest wading bird in the U.S. (Paul 1996), is a highly specialized feeder ecologically restricted to a very narrow coastal habitat that is especially vulnerable to human impacts (Lowther and Paul 2002). The Florida population is estimated to be 350-400 pairs, perhaps two-thirds of which occur in the Keys (Paul 1996). The conservation priority is to monitor abundance and determine locations and sizes of breeding colonies and

3 associated foraging areas (Lowther and Paul 2002). Wilmers and Arnett (2002) surveyed nesting pairs of REEGs in 2002 on 102 islands of the lower Florida Keys, confirming 17 nests on 11 islands. A breeding population of Cuban yellow warblers (CYEWA) occurs northward to extreme southern Florida, mainly in the Keys. All yellow warblers summering in Florida are D. p. gundlachi (Stevenson and Anderson 1994). Their conservation status is unknown (Lowther et al. 1999). In the course of the 2002 REEG surveys, J. Arnett (unpublished data) conducted systematic point counts of other breeding birds on 105 islands (the REEG sample plus 3) and detected 193 pairs of CYEWAs (based on singing males) on 65 islands. Arnett also collected point-count data in the same manner for black-whiskered vireos (60 pairs on 33 islands) and mangrove clapper rails (116 pairs on 34 islands). Black-whiskered vireos (BWVI) depend almost entirely on mangrove habitats, so they share the threats faced by this fragile coastal environment. In addition, BWVIs suffer high rates of parasitism by shiny cowbirds (Molothrus bonariensis), which are steadily increasing in the vireoâ&#x20AC;&#x2122;s range (Chace et al. 2002). Among clapper rails, the mangrove subspecies (MCLRA) is the least studied (Eddleman and Conway 1998), and specific threats, apart from human encroachment, have not been identified. White-crowned pigeons (WCPI) nest colonially on mangrove islands and feed on the seasonal fruits of West Indian hardwoods. WCPIs are heavily hunted for sport over most of their range. Some bag limits are as high as 50-100/day, nestlings are taken for food, shot birds are sold for food, and illegal shooting is common (Meyer and Zimmerman 2008). The small U.S. population (~10,000 pairs) is jeopardized by strong hunting pressure on parts of its winter range and the potential impacts of severe tropical weather. Although Hurricane Dennis in July 2005 reached only tropical-storm force in the Florida Keys, WCPIs lost >80% of their annual nest

4 productivity as a result (T. Wilmers, unpublished data). Compounding these impacts is the ongoing anthropogenic loss of native habitat, threatening the future of mangrove forests and tropical hammock systems (Smith et. al. 1994, Ellison and Farnsworth 1996, McCoy et al. 1996). The most recent review of WCPI status and trends lists declines for the largest subpopulations: Bahamas, Cuba, Haiti, Dominican Republic, Puerto Rico, Anguilla, St. Martin, and the Virgin Islands (Arendt et al. 1979). Populations on some islands have been completely extirpated. Because of these threats and the vital role WCPIs play in propagating the region’s seasonal deciduous forests (Bancroft and Bowman 2001), itself a rare and floristically diverse community, the species is of critical international conservation concern. In addition to its inclusion among the Strategy’s SGCNs, the WCPI is listed as Threatened in Florida. Research on WCPIs in Florida by Avian Research and Conservation Institute (ARCI) began in 2002 and received 2 years of support from the Florida Fish and Wildlife Conservation Commission’s (FWC) Nongame Wildlife Grants Program in 2004. ARCI organized an August 2005 meeting of representatives of Caribbean nations harboring key WCPI populations to form an international working group and plan a conservation strategy for WCPIs. The working group, with participants from 7 countries, identified the highest conservation priorities, suggested island-specific research questions, and outlined a range wide collaborative banding project that will help inform hunting policy and management across national boundaries. Hunting impacts are of particular interest regarding the migration ecology of Florida’s small population, most of which migrates to Caribbean islands for the winter and, consequently, faces substantial hunting pressure. Based on our work to date, we are most concerned about conditions in the Bahamas, where the 6-month season coincides exactly with the wintering season of Florida’s migrants, the number of hunters is not regulated, and regulations are poorly enforced. The hunt is deeply

5 rooted in tradition and provides income for economically-depressed communities, but steady nesting declines in the Bahamas are increasing hunting pressure on wintering birds from other breeding populations, especially juveniles. It is important to know how catastrophic natural events influence populations of at-risk species. When we have the benefit of pre-disturbance baseline data, as in the present case, replicated surveys can provide valuable advantages for conservation. They can: 1) illuminate immediate management needs; 2) help us design long-term strategies to counter the impacts of destructive events, particularly where reducing concomitant human threats can buffer critical populations from natural spikes in mortality and temporary loss of essential habitats; and 3) permit ranking of species by their resilience to catastrophes so that we can better allocate limited management resources when events such as hurricanes challenge small populations already stressed by human impacts.

OBJECTIVES 1. Use remotely sensed imagery to quantify the mortality of hardwood hammocks and mangrove forests in the lower Florida Keys as a result of 6 hurricanes in 2004 and 2005. 2. In 2006 and 2007, replicate surveys of the 5 target species conducted from 2001 to 2004 to estimate abundance of nesting pairs and compare with pre-hurricane levels. 3. Determine size and vegetation composition of activity ranges for adult WCPIs in the summer of 2006 using birds radio tagged during our previous study. 4. Gather survival data on adult WCPIs through February 2007 using birds previously radio tagged.

6 5. Continue to investigate the impacts of hunting in the Bahamas and the decline of the native population on the status of migratory WCPIs from Florida and elsewhere. 6. Band 350-500 nestling WCPIs per summer in 2006 and 2007 and solicit returns from hunters in the Caribbean as part of a range wide collaboration to determine migration pathways, wintering destinations, and potential source-sink dynamics among the regionâ&#x20AC;&#x2122;s subpopulations.

METHODS Extent of Hardwood and Mangrove Mortality Our proposal specified that we would systematically fly over the lower Florida Keys and adjacent islands in Florida Bay in a Cessna 172 and use a digital camera with an electronically stabilized lens to take overlapping photographs of all stands of hardwood forest greater than about 0.5 hectare. The photographs would be used to estimate the percent mortality, in 10% increments, of each discernable block of hardwood hammock. A sub-sample of these results would then be ground-truthed to validate our percentage estimates. Early in the project, we photographed several islands of various sizes and stands of hardwood hammock in areas where we had opportunistically taken pre-hurricane photographs. Although our low-altitude pictures could have been used for a coarse assessment of damage, we realized that available Landsat imagery provided a far more accurate and consistent depiction of the hurricane related habitat changes. We decided to use the professionally-produced imagery for our analysis of hurricane effects on vegetation, and to expand this task to include not just the hardwood foraging habitat but mangrove nesting habitats as well. The study area for this

7 analysis extended from north of Big Pine Key to the Marquesas Keys, approximately 50 km west of Key West (Fig. 2). Landsat satellite images with a resolution of 30 meters were obtained for the Florida Keys for 2001 (before hurricanes) and 2006 (after hurricanes) (USGS 2007). The Landsat images were imported into the ERDAS Imagine software package (ERDAS IMAGINE速 V9.2) with a projection of USA contiguous Albers equal area conic USGS version NAD 83. Landsat bands 3 and 4 were used to create a Normalized Difference Vegetation Index (NDVI) layer. The NDVI layer is particularly useful in discerning vegetation vigor and subtler variations in vegetative land cover (Hill et al. 2000). The NDVI layer was then combined with the individual bands 2, 3, and 5 to yield 1 composite image with 4 layers (235N). An Advanced Identification of Wetlands (ADID) vector-based map of Florida Keys habitats was created in 2004 to document and evaluate federal jurisdictional wetlands and other land covers. Importation of the ADID map along with the Infrared Digital Ortho quarterquadrangles (DOQQs) of the region allowed the Landsat data to be compared with the other 2 imagery sets for a general habitat reference. Next, we identified and delineated training sites to define signatures in the 2001 and 2006 235N images for each land-cover category. Training sites are areas of known land cover, determined from inspection of the ADID and DOQQs. These training sites were merged to ultimately produce 6 classes: Ridge Hammock, Hammock, Undefined (combination of developed land and barren land), Water, Mangrove, and Scrub Mangrove. A supervised classification was conducted that utilized the parallelepiped non-parametric rule and the maximum-likelihood parametric rules in the classification (ERDAS IMAGINE速

8 V9.2). From this process, 6 masks were created, 1 for each category and repeated in both the 2001 and 2006 imageries. The 2001 and 2006 thematic images were imported into ArcGIS Desktop (ESRI 2006), where analysis of changes in land cover over time were evaluated. In order to remove any discrepancy between years due to tidal differences, the 2001 water mask layer was overlaid onto the 2006 layer so areas not classed as water from 1 year to the next could be forced into the water classification. A mask layer was applied to the undefined (urban and barren land) class as well, forcing areas of 2006 to match the 2001 classification. This represents the true change from vegetated to non-vegetated habitat between the two years by removing unrealistic pixel classification (i.e. change from barren land to hammock in only a 5 year period of time). Nesting Surveys, All Species Point Counts.- In the summer of 2006 and 2007, point counts of REEGs, CYEWAs, BWVIs, CLRAs, and WCPIs were conducted by boat around mangrove islands found throughout of the Great White Heron National Wildlife Refuge (GWHNWR) and portions of the Key West National Wildlife Refuge (KWNWR). The mangrove islands surveyed were relatively small in size, allowing for accurate surveys to be completed within 1 or 2 counts by a single surveyor. The 108 islands used in this study were the same as those represented in a 2002 dataset (J. Arnett, unpublished data), with 102 in GWHNWR, 5 in KWNWR, and 1 south of (Atlantic Ocean side) the mainline Keys. The majority of the study islands are surrounded by very shallow flats, often holding less than a 0.3 meter of water during a normal low tide. We used a 4-meter Boston Whaler with a 24 hp Yamaha outboard motor (2006) and a 5-meter McKee Craft skiff with a 60 hp Evinrude outboard motor (2007), allowing access to waters that would otherwise be unnavigable by larger boats. Surveys were carried out at 30 to120 meters from

9 each island. The distance was influenced by the size of the island (i.e., the distance was directly proportional to the size of island) as well as the conditions present at the time of the survey (i.e. tide and wind). The point where a survey was conducted was determined in 1 of 2 ways: by quickly circumnavigating the island at a distance in order to find the areas with the highest likelihood of detections; or, when this was not possible, by choosing a position that afforded the best view within audible range of the island. The surveys were conducted following the exact protocol of Arnett in 2002. Each count lasted 30 minutes. The start time was noted upon approaching and settling into the position where the count would be conducted. This was necessary to include any birds that were flushed due to the noise from the boatâ&#x20AC;&#x2122;s motor. Each island was surveyed visually for REEGs; by ear for vocalizations of CYEWAs, BWVIs, and CLRAs; and visually and audibly for WCPIs. Any other species that could be identified and the number of individuals of each were recorded during 0-5-, 6-10-, and 11-30-minute increments. It was also noted whether or not they were visually and/or audibly detected. No birds were recorded that were seen or heard outside of the 30minute survey period. Upon completing the survey, the boat was either repositioned for a second survey or driven directly away from the island to avoid further disturbance to nesting birds. In order to evaluate the potential effects of the 2005 hurricane season on the abundance of the 5 species, we produced generalized linear models (GLMs) with Poisson distributions and log-links, using year (2002, 2006, and 2007) as a categorical predictor variable and individual speciesâ&#x20AC;&#x2122; abundance as a response variable (Seavy et al. 2005). Overdispersion was assessed by comparing the modelâ&#x20AC;&#x2122;s residual deviance with the degrees of freedom (residual deviance/df). Those models that resulted in dispersion values substantially different from 1 were fitted in GLM with Negative Binomial distributions, instead of Poisson distributions, and log-links in order to

10 parameterize the overdispersion (k) and improve model fit. We initially performed a pair-wise comparison between the years 2002 and 2006 using a subset of 169 point counts from each year in GLM in order to assess possible changes in bird abundance immediately following the 2005 season. We later compared 2006 to 2007 in GLM using a subset of 168 point counts from each year to detect any post-hurricane trends in abundance. All statistical models were conducted with R statistical software using glm and glm.nb (MASS package) procedures. We used the model parameter estimates to calculate the predicted number of birds per point count location (mean), according to the following model: mu=(intercept.estimate + effect.size(x)) The value for x was 0 for the first year (i.e. 2002 in the 2002/2006) and 1 for the second year in the model comparison. The standard deviation was calculated using the raw data for each year. Flight-line Counts.- For WCPIs, we extended for 2 years (2006 and 2007) the nesting surveys conducted by T. Wilmers (U.S. Fish and Wildlife Service), whose counts began in 2001 and were scheduled to end in 2005. Although WCPIs are included in the survey counts for all 5 critical species, described above, Wilmerâ&#x20AC;&#x2122;s flight-line counts provided the best overall estimate of abundance in the study area. The flight-line surveys, which included all the major nesting islands within the Florida Keys National Wildlife Refuges, consisted of surveys for breeding males during morning departures from nesting colonies (Strong et al. 1994). They were conducted from May to August, but most effort was focused during the incubation stage in June and July, before nests had a substantial chance to fail. The results were expressed as total estimated nests in each of 4 zones: Key West National Wildlife Refuge; and the western, central, and eastern portions of the Great White Heron National Wildlife Refuge. The count data were recorded by island to identify particularly important colonies and areas within each zone.

11 WCPI Activity Ranges and Habitats We deployed 138 VHF transmitters on adult white-crowned pigeons during previous studies, from 2002 to 2005. Based on a conservative estimate of expected battery life, at least 49 of these transmitters could have been operating during at least part of the 2006 nesting season, although we assumed not all the birds were still alive (estimated n=30-35). Activity Ranges.- The 2-bearing locations collected in the field were processed with the program Locate III (Nams 2005) to estimate approximate foraging locations. Ideally, 3 bearings should be used to make a triangulation. Limited field staff and time, however, made it impossible to collect 3 bearings quickly enough to ensure that the bird had not moved substantially between the time of the first and last bearings. We corrected for the 2-bearing location error by collecting a third bearing on a subset of 30 locations and calculating the standard deviation of all bearings. We then entered this number into Locate III so that the program could construct error polygons and determine the geometric center of each, thus estimating the most likely actual location for each bird that we tracked. Individuals tagged and tracked in 2005 that were alive and tracked for 10 locations or more in both 2005 and 2006 were included in the analysis of activity-range sizes pre- and posthurricane. Using ArcGIS (ESRI, Inc.) and the Hawths Analysis Tools extension (Beyer 2004), we constructed activity-range boundaries (minimum convex polygon) over a vectorized habitat map. We used the Advanced Delineation and Identification System (ADID) habitat map, created in 1991 by the U.S. Geological Survey, as our habitat layer for the GIS analysis. The ADID is a vector map of polygons of 14 habitat delineations in the Albers Equal Area Conic projection and the Clarke 1866 ellipsoid.

12 Activity-range sizes were tabulated by determining the area within the minimum convex polygon created by each individualâ&#x20AC;&#x2122;s locations. Polygon sizes (kilometer2) was compared using a paired t-test between individuals tracked in 2005 versus 2006. Habitat use.- Foraging locations from all 2005 radio-tracked pigeons were compared to all the locations determined for the pigeons tracked in 2006. Once the pigeon locations were overlayed on the land use/land cover shapefile, the attribute tables were joined, allowing each pigeon location to be assigned to 1 of the 14 land covers. Points located over the ocean (due to triangulation and/or mapping error) were assigned the closest land cover, which was mangrove in all cases. A Fisher exact probability test was used to test for differences in habitat use between years. WCPI Survival We used the status of VHF radio-tagged individuals at yearly intervals after marking to estimate annual survival of adult white-crowned pigeons in the Florida Keys a) as the proportion of each annual cohort that survived 1 year; and b) with known-fate models derived from Program Mark (version 4.3) similar to the approach of Kaplan and Meier (1958), modified to allow for the staggered entry of newly radio-tagged individuals as the study progressed (Pollock et al. 1989). The transmitters used in 2003 (n = 26) had an expected life of a little more than 12 months, so we added these birds to our survival sample. The transmitters deployed in 2004 (n = 50) and 2005 (n = 51) had an expected life of 18-19 months. Thus, both types of our survival estimates were based on 3 annual cohorts totaling 127 marked individuals. For the Mark analysis, the field telemetry data were grouped into 3 scenarios reflecting different degrees of certainty about the fate of the marked individuals. These scenarios were labeled 1, 2, and 3, with 1 being the most conservative (i.e., missing birds, not confirmed alive or

13 dead, were assumed to have experienced radio failure or loss rather than to have died) and 3 being the most liberal (i.e., missing birds were assumed to have died) regarding the fates of birds for which we permanently lost radio contact without determining why the signal was lost (i.e., we did not find a carcass, recover an expired or malfunctioning transmitter, or observe a living bird still carrying an inoperative transmitter). Status classes (a-f) were defined as follows: a - confirmed dead at the end of the yearly interval; b - confirmed alive at the end of the yearly interval; c - signal lost before earliestexpected radio failure and during the breeding season in Florida, but no evidence to confirm mortality; d - alive in Florida at end of the breeding season, radio expected to last through return the next spring, but no signal detected; e - alive in winter in Florida and radio not expected to expire before the next spring, but no signal detected; and f - radio expected to expire within 90 days following last detection, and no subsequent confirmation of status. The birds assigned each respective status class were grouped differently in each of the 3 scenarios in our analysis, as follows: For Scenario 1, all a birds were considered dead; b birds were considered alive; and c, d, e, and f birds were censored. For Scenario 2, a and d were dead; b was alive;, and c, e, and f were censored. In Scenario 3, a, c, d, e were dead; b was alive; and f was censored. Three yearly cohorts (2003, 2004, and 2005) were tracked over a total of 4 years (20032006). We associated 2 groups and 2 covariates with each individual pigeonâ&#x20AC;&#x2122;s detection history. The groups were year banded and banding location (Key). The individual covariates were sex and mass. Finally, we compared the probabilities of surviving the length of the study (4 years) for the 3 scenarios.

14 Impacts on Winter Range We conducted telemetry searches for Florida birds within the Everglades, the Florida Keys, and the Bahamas. The aircraft used in the Bahamas was a Cessna O-2 (the military version of the Cessna 337 Skymaster) with twin, center-aligned, push/pull engines. This airplane was well suited for this mission, which entailed many hours of over-water flight to and within the Bahamas (we flew from Fernandina Beach, Florida). We surveyed at 300-500 meters above ground level at an airspeed of 150-190 kilometers/hour. We helped plan the third meeting of the white-crowned pigeon International Working Group, which was held in conjunction with the biennial meeting of the Society for the Conservation and Study of Caribbean Birds (SCSCB), in San Juan, Puerto Rico, 20 July 2007. Banding WCPI Nestlings We banded nestling WCPIs with U.S. Fish and Wildlife Service aluminum bands on their left leg. We also recorded the nest height, whether the nestling was a singleton or had a nest mate, and estimated the nestlingâ&#x20AC;&#x2122;s age (in days). Nesting colonies are very sensitive to disturbance, so we timed our colony visits to times when as many young as possible were near fledging age. Two to 5 people visited islands with the largest colonies for 1-3 hours per morning and quickly moved through the area banding young. In a previous, smaller-scale project, recovery rates in the Caribbean ranged from 3-6% (Paul 1977).

15 RESULTS Extent of Hardwood and Mangrove Mortality Although hardwood damage was apparent in concentrated areas when viewed from the ground immediately following Hurricane Wilma in late 2005, the Landsat analysis detected minimal impact from the 2004-2005 hurricane season on the 6 image-based habitat classes. Our Landsat imagery analysis of the lower Florida Keys showed that mangrove habitat increased by 1.98%, scrub mangrove habitat decreased by 3.05%, hammock habitat decreased by 2.24%, developed land areas increased by 2.21%, and barren and non-vegetated areas increased by 1.11% from 2001 to 2006 (Table 1). In 2001 mangrove habitat covered 5,151 hectares of the terrestrial/semi-terrestrial habitat in the study area. In 2006, 4,848 hectares of this habitat remained mangrove and 303 hectares were reclassified as scrub mangrove. Of the 4,646 hectares classified as scrub mangrove in 2001, 689 hectares were reclassified as mangrove, 469 hectares were reclassified as barren, and 3,487 hectares remained scrub mangrove in the 2006 classification.. There were 496 hectares of barren land classified in 2001, of which 252 hectares were reclassified as mangrove habitat in 2006, while 243 hectares of barren land remained classified as barren land. The 5,507 ha of developed land classified in 2006 represents an increase of 432 hectares since 2001. Of the 432 hectares of newly-developed land, 427 hectares, or 98.8%, was originally classified as hammock. We hypothesized that the 2004 and 2005 hurricane seasons may have impacted only a portion of the keys within the study area, limiting our ability to identify distinct patterns in land cover change. To address this issue we re-ran the land cover change analysis on a subset of the study areas (Figure 4). The location of the subset area was delimited by isolating the set of keys that were determined to be the most directly impacted by the winds and storm surges of the 2004

16 â&#x20AC;&#x201C; 2005 storms (see Figure 1 and Kasper 2005). Again, little change in land cover was detected between 2001 and 2006. We found that mangrove decreased by 1%, scrub mangrove increased by 2% and hammock, and developed and barren land remained relatively unchanged (Table 2).

Table 1. Region-wide summary of land cover change from 2001 to 2006. Hectares represents the total area of a habitat class in 2001 and 2006. Percent habitat represents the total terrestrial habitat cover per habitat class in 2001 and 2006. Relative increase or decrease in habitat represents the percent change within a habitat class from 2001 to 2006. Area change is the change within a habitat class from 2001 to 2006. Percent change is the portion of the total area of terrestrial habitat accounted for in the area change per habitat class from 2001 to 2006.

Habitat Type Mangrove Scrub Mangrove Hammock Developed Barren

Year 2001 % Hectares Habitat 5150.97 26.31% 4645.53 23.73% 4211.28 21.51% 5075.37 25.92% 495.72 2.53%

Year 2006 % Hectares Habitat 5537.88 28.28% 4047.57 20.67% 3773.70 19.27% 5507.19 28.13% 712.80 3.64%

Relative Increase or Decrease 6.99 -14.77 -11.60 7.84 30.45

Area Change (ha) 386.91 -597.96 -437.58 431.82 217.08

% Change 1.98% -3.05% -2.24% 2.21% 1.11%

In addition to our satellite imagery analysis we also conducted field based inspections of apparent storm damage. This assessment indicated significant local scale changes to both the hardwood and mangrove systems. Our field evaluation took into account the measurable areas of hardwood mortality that had already occurred prior to 2004 as a result of sea-level rise and associated salt intrusion. Most of the storm-related damage was associated with habitat edges, particularly roads, and a likely consequence of exposure to severe winds. We found that pockets of more extreme wind damage were limited in extent, mostly less than 0.10-0.15 hectare, and randomly distributed. These results suggest that, while hurricane related habitat loss did take place, the patterns of storm damage were spatially distributed in such a way that may have limited our ability to detect these changes when utilizing 30-meter resolution Landsat imagery.

17 Storm damage to mangrove forest, the breeding areas for WCPIs and the other focal species, was more extensive. Our direct observations suggested that this damage resulted mainly from wind and secondarily from storm surge. The damage to red mangrove was widely distributed and variable from island to island, and within islands, but the extent and degree of damage generally decreased from west to east through the study area. Most of the damage to mangrove forest consisted of mortality to portions of black mangrove within island interiors. Our banding work and prior observations by others indicated that these stands of black mangrove, relatively dense but smaller in stature than the adjacent, more seaward stands of red mangrove, serve as nesting sites for the greatest number of WCPIs. In most cases, the black mangroves were not structurally damaged, as one might expect from the impacts of wind or storm surge. Instead, mortality apparently resulted from deposition of a thick layer of sediment over the pneumatophores of the black mangrove trees. The effects varied locally and were most intense in the western portion of the study area.

Table 2. Northern Keys subset summary of land cover change from 2001 to 2006. Hectares represents the total amount of hectares of a habitat class in 2001 and 2006. Percent habitat represents the percent of the total terrestrial habitat cover represented by that individual habitat class in 2001 and 2006. Relative increase or decrease in habitat represents the percent change within a habitat class from 2001 to 2006. Area change is the total amount of hectares of change within a habitat class from 2001 to 2006. Percent change represents the percent of the total area of terrestrial habitat accounted for in the area change with an individual habitat class from 2001 to 2006. Year 2001 Subset Habitat Type Mangrove Scrub Mangrove Hammock Developed Barren

Hectares 561 174 37 0 21

% Habitat 70.74% 21.94% 4.67% 0.00% 2.65%

Year 2006 Subset Hectares 552 190 37 0 14

% Habitat 69.61% 23.96% 4.67% 0.00% 1.77%

Relative Increase or Decrease 0.02 -0.09 0 1 0.33

Area Change (ha) -9 16 0 0 -7

% Change -1.13% 2.02% 0.00% 0.00% -0.88%

Fig 1. Paths of major storms passing near the Florida Keys during 2004 and 2005.

Nesting Surveys, All Species Point Counts.- We conducted point counts of REEGs, CYEWAs, BWVIs, CLRAs, and WCPIs by boat around mangrove islands From 15 May through 30 June 2006 and from 17 May through 30 June 2007 during 185 hours on 21 survey days. There were 171 counts each year on 111 islands during which we also recorded environmental conditions. In 2002, before the hurricanes, J. Arnett (unpublished data) performed 175 counts on 104 Keys over 25 days between 13 May and 29 June, with 2 additional days in July. In general WCPIs were most abundant; REEGs were least abundant in 2002 and both years after hurricanes (Table 3). With the exception of BWVIs,

19 all species had higher numbers in 2007. Overall, the results were varied across islands, with some sites showing decreases and others unchanged for various species relative to the prehurricane counts (Table 3).

Table 3. Numbers of each of the 5 focal species detected per year in 2002, 2006, and 2007 in the lower Florida Keys, with numbers of islands and counts on which each species was found.

Species Black-whiskered vireo # of Islands # of Point Counts

2002 60 32 41

Year 2006 109 41 56

Cuban yellow warbler # of Islands # of Point Counts

192 66 94

191 58 87

207 69 96

Mangrove clapper rail # of Islands # of Point Counts

116 35 42

86 27 39

121 35 44

Reddish egret # of Islands # of Point Counts

21 18 18

10 10 10

16 9 10

White-crowned pigeon # of Islands # of Point Counts

712 66 97

336 58 59

712 69 99

2007 86 33 43

The model results for comparisons between 2002 and 2006 and between 2006 and 2007 for each of the 5 species are shown in Tables 3 and 4. There was no evidence of overdispersion for BWVIs and REEGs when modeled for 2002 versus 2006 (Table 4). We were able, therefore, to use the Poisson distribution in GLM to assess the effect of year on abundance for these 2

20 species. The other 3 species, however, demonstrated signs of overdispersion and were instead analyzed using a negative binomial distribution in GLM. When modeling 2006 to 2007, all but REEGs showed signs of overdispersion (Table 5). Again, we used Poisson distributions for REEGs, but negative binomial distributions for all other species.

Table 4. Comparison of generalized linear models (specifying Poisson or negative binomial distributions with log-link) and effect sizes for 5 avian species from 169 point counts (338 total) in the Lower Florida Keys National Wildlife Refuges system in 2002 and 2006. Significant year effects are indicated by bold italics. Poisson SE 0.07 0.10 NA

P-value 0.077 0.959 NA

Negative binomial Estimate SE P-value 0.12 0.10 0.180 -0.005 -0.04 0.969 1.54 0.36 336

Species Cuban yellow warbler

Parameter Intercept Year Dispersion (k) Df

Estimate 0.13 0.005 NA 336

Black-whiskered vireo

Intercept Year Dispersion (k) Df

-1.04 0.60 NA 336

0.13 0.16 NA

<0.001 <0.001 NA

Mangrove clapper rail

Intercept Year Dispersion (k) Df

-0.38 -0.30 NA 336

0.09 0.04 NA

<0.001 0.036 NA

-0.38 -0.30 0.22 336

0.19 0.27 0.04

0.047 0.274

White-crowned pigeon

Intercept Year Dispersion (k) Df

1.44 -0.75 NA 336

0.04 0.07 NA

<0.001 <0.001 NA

1.44 -0.75 0.24 336

0.16 0.23 0.03 0

<0.001 0.001

Reddish egret

Intercept Year Dispersion (k) Df

-2.09 -0.74 NA 336

0.22 0.38 NA

<0.001 0.054 NA

Of the 5 species, BWVIs and WCPIs showed significant effects between 2002 and 2006 (p-value < 0.05), with BWVIs increasing and WCPIs decreasing after the 2005 storms (Table 6).

21 Reddish Egrets approached significance (p-value=0.054) for a decline between 2002 and 2006 (Table 6). Only WCPIs displayed a significant difference (p-value < 0.05) in numbers between 2006 and 2007, increasing to the point of regaining their 2002 level of abundance (Table 6). Flight-line Counts.-. In 2006, we conducted flight-line counts of WCPIs in the lower Keys on 30 days from 8 June through 28 July 2006. Seventeen Keys were monitored - 6 of them twice, 2 on 3 occasions, and 1 Key 4 times (Barracouta, Table 7). For islands surveyed more than once, only the highest count is presented (Fig. 6). The estimated number of pigeon nests based on the flight-line counts was 1,763, most of which were in the Great White Heron National Wildlife Refugeâ&#x20AC;&#x2122;s eastern sector (Table 7). The 2007 flight-line counts occurred on 37 days during the period 7 July through 10 August 2007. A total of 16 Keys were surveyed an average of 2.3 times each (range 1-5). We estimated that there were 3,839 WCPI nests within the surveyed area (Table 7, Fig. 6). The greatest number of nesting pairs, 1,973, were in the Great White Heron National Wildlife Refugeâ&#x20AC;&#x2122;s eastern sector; the central sector had the fewest nests, 928 (Table 7). This pattern was consistent with that observed in pre-storm counts. The highest flight-line estimates of WCPI nesting effort for the period 2002 to 2007 occurred in 2004, the year Tropical Storm Dennis and Hurricane Wilma passed through the lower Keys (Table 7, Fig. 6). The counts revealed a drop of 56% in nesting, from 2005 to 2006, following the storms, resulting in the smallest number of nesting attempts in any year from 2002 to 2007. From 2006 to 2007, however, the number of WCPIs on flight-lines more than doubled. A 1-way ANOVA revealed no statistical differences in flight-line estimates of WCPIs over the 7-

22 year period (F = 0.753, P = 0.587), nor between 2005 and 2006 (F = 0.942, P = 0.341). There was however, a statistically-significant increase from 2006 to 2007 (F = 4.987, P = 0.034).

Fig 2. Area of White-crowned Pigeon breeding habitat included in the analysis of land-cover change.

Fig 3. Land cover classifications for 2001 (before hurricanes) and 2006 (after hurricanes) in the lower Florida Keys.

24 Table 5. Comparison of generalized linear models (specifying Poisson or negative binomial distributions with log-link) and effect sizes for 5 avian species from 169 point counts (338 total) in the Florida Keys National Wildlife Refuges system in 2006 and 2007. Significant year effects are indicated by bold italics. Poisson Estimate SE 0.12 0.07 0.09 0.10 NA NA 334

P-value 0.090 0.357 NA

Negative binomial Estimate SE P-value 0.12 0.10 0.204 0.09 0.14 0.494 1.46 0.32 334

Species Cuban yellow warbler

Parameter Intercept Year Dispersion (k) Df

Black-whiskered vireo

Intercept Year Dispersion (k) Df

-0.44 -0.23 NA 336

0.10 0.14 NA

<0.001 0.119 NA

-0.44 -0.22 0.4 334

0.16 0.23 0.08

0.005 0.318

Mangrove clapper rail

Intercept Year Dispersion (k) Df

-0.68 0.35 NA 336

0.11 0.14 NA

<0.001 0.012 NA

-0.68 0.35 0.22 334

0.20 0.27 0.04

<0.001 0.194

White-crowned pigeon

Intercept Year Dispersion (k) Df

0.69 0.76 NA 336

0.05 0.07 NA

<0.001 <0.001 NA

0.69 0.76 0.26 334

0.16 0.22 0.03

<0.001 <0.001

Reddish egret

Intercept Year Dispersion (k) Df

-2.83 0.48 NA 334

0.32 0.40 NA

<0.001 0.232 NA

WCPI Activity Ranges and Habitats We gathered location data on radio-tagged WCPIs continuously from 16 June to 17 July 2006 and again during 2 more tracking periods, from 4 to 6 August and on 24 August 2006. We scanned for 49 pigeons that were assumed to be alive at the end of the 2005 field season. Twenty-seven of the 49 pigeons were detected at least 1 time from 16 June to 24 August 2006. One of the 27 birds was found dead on Big Pine Key in late June.

25 During the summer of 2006, we took 558 bearings on 24 individuals, 547 of which were entered into program LOCATE III to estimate foraging locations. The program determined that 212 of the 547 locations were sufficiently accurate to be used for range analysis.

Table 6. Predicted mean abundance between years for 3 focal species in the Lower Keys.

Species Black-whiskered vireo Reddish egret White-crowned pigeon

Predicted mean abundance (SD) 2002 2006 0.35 (0.68) 0.64 (1.18) 0.12 (0.41) 0.06 (0.24) 4.22 (19.84) 1.99 (5.06)

White-crowned pigeon

2006 1.99 (5.06)

2007 4.26 (8.21)

% Difference 45% -100% -112%

Dispersion (k) NA NA 0.24

53%

0.26

Table 7. Number of white-crowned pigeons estimated in each of 3 sectors in the Florida Keys by flightline counts from 2002 through 2007.

Year 2002 2003 2004 2005 2006 2007

Great white heron central 1113 788 1116 593 425 908

Sector Great white heron east 1251 1475 1822 1020 734 1978

Key West National Wildlife Refuge 904 1766 1504 1313 478 796

Total 3267 4029 4442 2925 1637 3682

Activity Ranges.- Activity ranges were estimated for 7 WCPIs for which we had 10 or more locations in each year before (2005) and after (2006) Tropical Storm Dennis and Hurricane Wilma. In 2005, Minimum Convex Polygon (MCP) estimates ranged from 0.02 to 9.22 km2 (mean: 2.74 km2 (SD = 4.00, Table 8). In 2006, the resulting MCP ranged from 0.06 to 42.74

26 km2 (mean: 10.17 km2 (SD = 16.34, Table 8). Although the trend suggested an increase in the mean size of activity ranges the year after the storms, the difference was not significant (t = 1.37, P = 0.11). Habitat Use.- The habitat analysis was based on 607 locations from 67 individuals in 2005, averaging 9.18 locations per bird (SD = 7.66, range 1 to 34). In 2006 we used 205 locations from 22 WCPIs, an average of 9.45 locations per bird (SD = 4.66, range 1 to 18). White-crowned pigeons in both years foraged in only 9 of the 14 available habitat types. About 30 percent of their radio locations were in developed pinelands with embedded hardwood hammocks. The second most-frequently used habitat differed between years - hammock in 2005 and pinelands in 2006 (Table 9). Buttonwoods were the third most-frequently used habitat in both years, while mangroves and freshwater marshes were occupied the least (Table 9). After calculating proportional habitat use and comparing with a Fisherâ&#x20AC;&#x2122;s exact probability test, there was an overall significant difference between habitat use before and after the 2005 storms (P < 0.001).

Table 8. Pre- and post-hurricane activity ranges of 7 white-crowned pigeons tracked by VHF telemetry. MCP area (km2) Pigeon ID 165.234 165.008 165.152 165.044 165.207 165.262 165.126 Mean Standard Deviation

2005 0.02 0.08 0.29 0.33 1.40 7.85 9.22 2.74 4.00

2006 0.06 1.58 4.31 0.15 21.93 0.42 42.70 10.17 16.34

We determined that the developed pinelands with embedded hardwood hammocks and pinelands were used in significantly greater proportion than expected based on availability. Mangrove and scrub mangrove habitats were used significantly less than expected. WCPI Survival Just prior to fall migration in 2005, we knew the locations of 44 of the 50 radio-tagged white-crowned pigeons marked that year in the lower Keys. From June through September 2006, we accounted for 31 of these birds, of which 22 survived through the first anniversary of their tagging data, which suggests an annual survival rate of 45%. Although it was beyond the scope of this study (which called for monitoring only through the summer of 2006), we used other funds to continue tracking radio-tagged WCPIs marked in 2005. During the fall of 2006, we made 3 aerial surveys over the Everglades (south of US Highway 41) and Florida Keys (Elliott Key to Key West). The flights were on 29 September, 26 October, and 4 December 2006. We detected only 2 radio-tagged WCPIs, both of which were among the 22 birds found during the previous summer. Despite the additional search effort, we did not increase the number of WCPIs known to be alive 1 year after being marked, thus supporting our conclusion that annual survival for the cohort radio tagged in the summer of 2005 was about 45%. Any error in that estimate most likely stemmed from premature transmitter failure, which we believe to have been relatively uncommon, rather than our inability to detect normal radio signals of living pigeons. Previously (Meyer and Zimmerman 2008), we used the Kaplan-Meier method (program Mark version 4.3) to estimate annual adult survival for 2004-2006, expressing the results under 3 scenarios that reflected decreasing levels of certainty regarding each missing birdâ&#x20AC;&#x2122;s fate. The 3 annual survival estimates ranged from 51 to 73%, with a mean of 59%. The 2 years with the

28 largest samples, 2004 (n = 50) and 2005 (n = 51), had very similar estimates â&#x20AC;&#x201C; 51.2 and 52.6%, respectively for the second (middle) scenario. Thus, based on the estimates of annual survival derived from our tracking data and combining all 3 cohorts, 68 of 127 radio-tagged pigeons, or 54%, survived for 1 year after being marked. Overall, the model-based survival estimates were higher than the simpler estimates based on the assumption that a radio-tagged sample in a given year represented an even-aged cohort marked simultaneously. This was true even with a large disparity for 1 of the 3 scenarios. This model-based analysis also revealed 2 important â&#x20AC;&#x201C; and likely related â&#x20AC;&#x201C; correlates of survival. Based on Scenarios 1 and 2, the best model showed that survival varied widely among locations, and mass at capture had a significant (P<0.05) positive effect on survival. Little Torch and Summerland keys had the lowest survival rates among capture sites. The results for Scenario 3 differed substantially from those for scenarios 1 and 2. In the best model, year, not location, was the most important covariate. Mass also was important. In this case, survival varied substantially with time. Overall, however, this model was not strongly supported by the available data.

Fig 4. Delineations of subsets of the study area in the lower Florida Keys for the analysis of habitat change. Red land area represents the subset of mangrove keys selected for the analysis (Table 2). Grey areas represent land included in regional study that was not included in the subset analysis.

Fig 5. Land cover classifications of the Great White Heron National Wildlife Refuge, lower Florida Keys, in 2001 (before hurricanes) and 2006 (after hurricanes).

5000 Mule Crawfish

4500

Big Mullet

Estimated WCPI Nest Density

Barracouta

4000

W. Bahia Honda Teakettle Sandfly

3500

Horseshoe/Mangrove Hardup

3000

E. Bahia Honda Coconut Sawyer

2500

Riding Crane

2000 1500 1000 500 0 2002

2003

2004

2005

2006

2007

Year

Fig 6. Numbers of white-crowned pigeons counted during annual flight-line counts at 14 Keys in the lower Florida Keys from 2002 and 2007.

Impacts on Winter Range During the fall of 2006, we made 3 aerial surveys over the Everglades (south of US Highway 41) and Florida Keys (Elliott Key to Key West) in search of the 22 birds detected during the breeding season, plus any additional marked pigeons that we may have missed during our summer surveys, to continue documenting the extent to which individuals from Floridaâ&#x20AC;&#x2122;s breeding population overwinter in the state rather than migrating to Caribbean islands. Based on our 3 winter aerial surveys in Florida, at least 2 of the radio-tagged birds, or 9% of 22 assumed to be alive in August, at the end of the breeding season, remained in the state

32 rather than migrating to Caribbean islands. One was wintering in the Everglades, while the other was in the lower Keys. It is likely that, between August and early December, 1 or more of the 22 marked birds died. The percentage wintering in Florida, therefore, probably was greater than 9%. In the 4 previous years, 15 to 22% of our marked sample overwintered in Florida.

Table 9. Comparison of proportional habitat use by white-crowned pigeons in the Florida Keys before and after the 2005 hurricane season.

Habitat type Developed pineland with hammock Hammocks Buttonwoods Freshwater hardwoods Exotics Scrub mangrove Mangrove Pinelands Freshwater marsh

2005

2006

0.30 0.22 0.21 0.11 0.07 0.04 0.02 0.01 0.00

0.29 0.10 0.16 0.06 0.08 0.03 0.03 0.23 0.01

In addition to the Florida aerial surveys, we flew intensive telemetry surveys in the Bahamas on 16 and 17 October 2006 to take advantage of radio transmitters that remained operating during the early portion of the wintering and hunting seasons, when pigeon detectability and hunting impacts were greatest. The search transects covered the islands for which band recoveries, the reports of local reliable observers, and our own previous detections of radio-tagged WCPIs indicated the highest probability of finding wintering pigeons. These islands were Bimini, North and South Andros, New Providence, Eleuthera, Abaco, and Grand Bahama. On our first survey flight in the Bahamas, in February 2005, we detected 3 radiotagged WCPIs. All 3 were found in the Florida Keys the following summer (2005), and 2 of the

33 3 were found again in the Bahamas in October 2005. The transmitters on all 3 birds, however, had expired before our most recent flight, October 2006, and we did not detect any other marked WCPIs from our Florida study population. While in the Bahamas, we met with Eric Carey, the director of the WCPI program within the Bahamas National Trust (BNT), a private, government-funded organization that is responsible for managing critical species, including the WCPI. The Trust has been our main collaborator in the Bahamas and, with Carey’s guidance, has taken the initiative to focus on the status and conservation needs of the country’s population. During this recent meeting, we discussed the need to establish hunting licenses and fees in the Bahamas to regulate the harvest more rigorously and to generate funds for monitoring and periodic status reviews. We also reviewed ways in which ARCI can help bring international guidance and influence to the process in support of BNT’s efforts, including recommendations based on our biological research, encouragement for BNT board members, hunting associations, and appropriate government administrators to recognize the need for WCPI management and to devote the needed resources to the effort. We also have ongoing collaborations with Jamaican and Cuban colleagues with a similar twofold aim: to learn the extent to which these islands are used as winter habitat by Florida’s WCPIs; and to assist our colleagues in designing conservation-oriented studies to manage their own native populations – objectives that are mutually supportive. In Jamaica, Brandon Hay (Game Bird Committee of the National Environmental Protection Agency) has begun his own field study using training we provided for him and his assistants in 2005. We are advising on fund-raising efforts that he is advancing within Jamaica and abroad. Hay monitored for radiotagged WCPIs from our Florida study population. No Florida birds were detected, nor did we

34 find any marked pigeons in late 2005 when we flew systematic surveys in Jamaica scanning for a much larger sample of radio-tagged WCPIs. In Cuba, Arturo Kirkconnell, who works as an ornithologist for the Cuban National Museum of Natural History, is collaborating on WCPI research. Kirkconnell represented Cuba at the 2005 WCPI working group meeting in Florida, at which time we provided training in radio telemetry so that he could eventually monitor for radio-tagged birds from our Florida study population. Because it is so difficult for U.S. scientists to work in Cuba due to U.S. State Department and Treasury restrictions, and because of Cuba’s objections to importing radio receiving equipment (even for use by a Cuban scientist), we had been uncertain about how to proceed with our collaborative studies in Cuba. In October 2006, we were able to provide telemetry receiving equipment for Kirkconnell’s use. We also accepted an invitation from the Cornell Laboratory of Ornithology to lead a 1-week program in Cuba training scientists and students in radio-tagging, tracking, and analysis techniques. While there, we would have been able to conduct WCPI fieldwork and further training. Cornell’s plans have met with obstacles and delays imposed by the U. S. State Department, so for now, the scheduling of the program is uncertain. We will continue to seek opportunities to work directly in Cuba or at least to train Cuban ornithologists while here in the U.S. At the WCPI working group meeting in July 2007, ARCI staff presented an update on the Florida project and continued discussions regarding monitoring of hunting programs in the Caribbean and ways in which we can collaborate on further study (including banding nestlings to connect breeding and wintering destinations for conservation planning, described below).

35 Banding WCPI Nestlings From 26 July to 23 August 2006, we banded 136 nestling WCPIs at 94 nests with standard aluminum U. S. Fish and Wildlife Service leg bands. The young, which were found in nests on 5 islands on the Florida Bay side of the lower Keys, ranged in age from 6 to 19 days. Of the 136 nestlings, 52 were the only nestlings present and 84 were in 2-bird broods. Our goal for the season was to tag at least 350 nestlings, but poor and relatively asynchronous nesting effort greatly reduced the number of available young. In 2007 we were able to band 390 WCPI nestlings on 4 trips to the Lower Keys from 23 July to 1 September. There were a total of 10 days of nest searching and tagging where we tagged 1 to 81 pigeons, averaging 39 per day (Table 10). We visited 5 Keys, but returned to 3 of them (East Bahia Honda, Cocoanut, and Teakettle) repeatedly throughout the season, banding over 100 on each Key (Table 10). The most productive island was Cocoanut Key, where we averaged 49 nestlings tagged per visit. We found only 1 nest with young on Secret Key, not surprising due to that islandâ&#x20AC;&#x2122;s abundance of green iguanas (Iguana iguana), typically thought to be voracious nest predators. No iguanas were seen on any of the other islands surveyed. The abundance of iguanas on Secret Key may be due to its close proximity to the mainline Keys, where density is high and increasing. During each of our banding efforts, we recorded numbers of nests and their status (empty, with eggs, with young too young or too old to band, unreachable, or undetermined) (Table 10). The 390 nestlings banded in 2007 came from 264 nests. A few nests had 3 eggs and 1 contained 3 nestlings, only 2 of which were old enough to band. We counted 1,888 nests, although some on Cocoanut, East Bahia Honda, and Teakettle undoubtedly were recounted on subsequent trips (Table 10). We did note that WCPIs were

36 reusing nests that formerly fledged young. Most of the nests found were empty (n = 622) or had eggs (n = 631), and 332 had nestlings that were unbandable (too young, too old, or impossible to reach) (Table 10). There were 15 nests in which 1 of the 2 nestlings was too young to band and where siblings were too old. Some nestlings of bandable age were found with 1 (n=9) or 2 (n=1) unhatched eggs in the nest. In summary, 524 white-crowned pigeon nestlings were tagged from 359 nests on 6 Keys in 2 years. We located a total of 2,323 nests, some of which were recounted within years or contained second clutches. The 3 most productive Keys, in order of descending numbers of young banded, were East Bahia Honda, Cocoanut and Teakettle. We had 2 band returns from our cohort of nestlings banded in 2007. The first report came from Harbor Island, Eleuthera, Bahamas on 23 October 2007. The reported band was for a pigeon banded at its nest on Cocoanut Key on 15 August 2007. We assume that the bird was shot by a hunter. A second report came from Little Torch Key in the Florida Keys on 8 December 2008. This bird was banded at 7 days of age on East Bahia Honda Key on 1 September 2007. The man who reported the recovery was raking gravel in his driveway when he found the band and the birdâ&#x20AC;&#x2122;s remains, which suggested that it had been dead for quite some time. He speculated that a cat may have killed it or that it had struck an overhead electrical line.

37 Table 10. White-crowned pigeon nestlings banded and nests found in 2006 and 2007 in the lower Florida Keys.

Fledged nests

Nests w/chicks too young

2 12

2 10

08/21/06

Hardup Key Teakettle Key East Bahia Honda Key

112

08/22/06

Teakettle Key

08/22/06 08/23/06

Cocoanut Key Barracouta Key 2006 Total

21 7 134

14 5 95

13 8 90

30 65 196

2 0 8

8 3 33

0 0 0

0 0 13

67 81 435

06/23/07

Secret Key East Bahia Honda Key

148

Cocoanut Key Central GWHNWR

206

08/04/07

Teakettle Key Teakettle/E. Bahia Honda East Bahia Honda Key

128

08/04/07 08/05/07

Hardup Key Cocoanut Key

3 32

3 23

29 83

13 51

0 19

4 8

9 0

0 0

58 184

08/05/07

136

08/14/07

Teakettle Key East Bahia Honda Key

196

08/14/07

Teakettle Key

196

08/15/07

Cocoanut Key East Bahia Honda Key

103

259

167

2007 Total

390

264

631

622

189

1888

2 Year Total

524

359

721

818

197

119

2323

Date 07/26/06 07/26/06 07/26/06 08/04/06 08/04/06 08/05/06

06/24/07 06/25/07 06/26/07 06/27/07 06/28/07

09/01/07

Chicks banded

Chicks banded from # nests

Nests w/eggs

Empty nests

Cocoanut Key East Bahia Honda Key

Teakettle Key East Bahia Honda Key

Island Name

Unbandable nests

Nests with unknown activity

Nest total

35 2 10

38 DISCUSSION AND MANAGEMENT IMPLICATIONS Extent of Hardwood and Mangrove Mortality The results of the satellite imagery analysis were somewhat contradictory to our field assessment of storm damage. Data collected in the field clearly indicated localized losses of both mangrove and hardwood hammock, yet this was not detected in the analysis of land-cover change. In fact, the Landsat analysis made it appear that very little change occurred in the total areas of individual land covers between 2001 and 2006 (Figure 3, Table 1). That is, the total number of hectares of scrub mangrove remained virtually the same in 2006 as it was in 2001. The same was true for hardwood hammock. However, 11% of the pixels classified as scrub mangrove in 2001 were classified as some other cover in 2006. The same was true for 14% of the 2001 hardwood hammock coverage. Almost identical areas of each of these land covers appeared some place else on the study landscape in 2006, thus compensating for the losses and resulting in essentially no change in the total areas of scrub mangrove and hardwood hammock in the study area from 2001 to 2006. In fact, every land-cover class, with the exception of development, showed similar shifts in distribution. In a scenario such as this, local-scale changes in habitat will appear dramatic while regional changes will not be as apparent. In our particular analysis of hurricane effects in the Florida Keys, two factors probably account for most of this effect. First, mangrove and hammock habitats appear to shift in location, shape, and extent in a dynamic, natural response to the continuous influences of wind and water. A short-term, finely-scaled view of these spatial changes can make them appear dramatic. However, considered in a larger time context, the total areas occupied by these communities (in the absence of climate change and other human influences) probably remain relatively constant.

39 We found evidence of this impressive resilience even within the 5-year time frame of our comparative analysis. That is, as patches of mangrove and hammock were being damaged by the storms of 2004 and 2005, other portions of these plant communities were in various stages of regeneration from prior impacts of severe weather. However, we believe that our view of this natural process was exaggerated by the technical reality that storm damage of varying degrees was spatially distributed in ways that limited our ability to detect these changes using Landsat imagery with its relatively coarse 30-meter resolution. It is worth noting that development increased by 432 hectares in the 5-year period of our analysis at the direct expense of hardwood hammock habitat. We based this conclusion on the fact that 427, or 98.8%, of these specific hectares of developed land were classified as hammock in 2001. Unlike what we observed for scrub mangrove and hammock, these particular patches of development did not shift location over the study period (i.e., disappearing in some places while appearing in others), nor were they as likely to be misidentified due to the 30-meter resolution of the GIS analysis. They were newly-created in the 5-year period and represent a loss of 10.1% of the 4,211 hectares of hardwood hammock measured in the 2001 imagery. This is a startling rate of permanent loss for this imperiled plant community. Many of the patches of hammock and mangrove damaged by severe weather during this brief period were being compensated for by the simultaneous regeneration of comparable patches elsewhere that had been damaged in previous storms. However, areas converted from hammock to development obviously do not regenerate to hammock or any other natural community. This rapid, dead-end transformation of hardwood hammock to development in the lower Florida Keys â&#x20AC;&#x201C; at least 10% in the short time span of this study - starkly contrasts with the dynamic, productive response of native habitats to what many would view as the far-more powerful and

40 alarming effects of hurricanes. This natural, landscape-level resilience, furthermore, prevailed despite the above-average occurrence 6 powerful storms during study period. Storm damage to hardwood hammocks was more severe near forest edges. Fragmentation of this threatened community, therefore, will make it more vulnerable to future tropical storms and hurricanes. The larger stands that will remain mostly on public lands will increase in importance as a food base for the WCPI. In the long run, more hammock habitat in the lower Florida Keys will be lost to development and the gradual, cumulative effects of salt intrusion and tidal action resulting from sea-level rise. Most damage occurred in the mangrove-island nesting habitats of the 5 focal species. The larger red mangrove trees, mainly around the periphery of the islands, took the brunt of the wind and storm-surge damage. However, the most visually-apparent damage occurred in the more interior black mangrove stands due to overwash and the resulting deposition of sand and silt. Recovery from such damage will be very slow, thus eliminating a large portion of the formerly-available WCPI nesting habitat in the lower Florida Keys. In 2007, WCPI nesting attempts were much greater than in 2006, on a par with nonhurricane years, even though there had been essentially no regeneration of black mangrove stands on the island interiors. These are the stands most densely occupied by WCPIs. The most dramatic example was Barracouta Key, 16 kilometers west of Key West, which traditionally has been the site of 1 of the largest WCPI nesting colonies in the Keys. Nest abundance in 2006 did not exceed 40% of typical historic levels on that island. Conflicting patterns of change occurring in this study system emphasize the importance of considering more than 1 perspective scale when conducting research and devising conservation plans. Close examination of the change in spatial distribution of habitat from 2001

41 to 2006 revealed distinct zones of change taking place within many of the Keys (see Fig. 4). Based on trajectories of hurricanes from the 2004-05 hurricanes seasons, one would expect that the northwestern portions of the study area would have experienced the greatest storm impacts. Habitat change in this subset of islands in fact does suggest a moderate die back of mangrove. However, in other areas, portions of scrub mangrove have shifted to mangrove, while barren land has shifted to scrub mangrove (Fig. 4). It is likely that some areas classified as scrub mangrove are actually locations where dense patches of mangrove can grow but are currently in a state of regeneration or degeneration. Further explanation for these seemingly conflicting patterns is that 2 major forces are at work at 2 different scales. The 2004-05 hurricane storm surge conditions along the northern and northwestern edge of the study area may have thinned the mangrove habitat along the northfacing edge and interior portions of several Keys north of Big Pine Key (Table 2). It is also likely that the transformation of scrub mangrove to mangrove is an effect of regional drought cycles on the larger mangrove system. Research suggests that periods of drought cause an increase in salinity, which can cause die back or leaf drop in mangrove habitat (Armentano, 1995, Carlson, 1997). In wet years, such as those of high hurricane activity, the mangrove system recovers from the drought conditions. Portions of mangrove habitat certainly declined due to storm surge and siltation (as is seen in Figure 4), but region wide, mangroves actually showed a slight net due to regeneration over the 6-year period between images. The patterns of habitat change reported here suggest that the mangrove system, in particular, is highly dynamic and exists in a constant state of change at both the local and regional scales. Conservation planning for all 5 focal species, therefore, should consider smallscale habitat needs as well as regional effects of hurricanes, droughts, and other disturbances.

42 As a result, predictive demographic modeling for species of conservation need must take into account the stochastic nature of severe tropical weather. This need, furthermore, will be compounded in years to come by the concomitant effects of climate change in sub-tropical Florida. Nesting Surveys, All Species Point Counts.- All species except BWVIs appeared to have experienced a decline in numbers following Tropical Storm Dennis and Hurricane Wilma. The differences for CLRAs and CYEWAs were not statistically significant, perhaps due to insufficient data or a slightly poorer fit than for the other species. However, the change approached significance for REEGs and was significant for WCPIs. Of the 4 species that declined, all but REEGs appeared to have begun to rebound by 2007. Again, although the apparent increases by CYEWAs and CLRAs were not statistically significant, it is likely they would have been given another yearâ&#x20AC;&#x2122;s worth of count data. Most notable was the statistically significant increase in nesting WCPIs from 2006 to 2007 (i.e., from 1 to 2 years post hurricane). This response demonstrates strong resilience by this mainlyCaribbean species even to extreme habitat perturbations caused by severe weather. As noted below, the flight-line counts of nesting WCPIs and the availability of nestlings for banding revealed corresponding declines in 2006, after the 2 storms in 2005, followed by a rebound in numbers in 2007. The lack of any apparent increase in REEGs by 2007 should be viewed with caution given the scarcity of this species in the study area and what we have experienced as a high variance in detectability.

43 We have no conclusive explanation for why BWVIs increased in the year immediately after the storms, then dropped by the second year as the other species exhibited at least apparent increases. It is possible BWVIs typically experience competitive displacement when CYEWAs (and other passerine species) are at average abundance and, thus, responded favorably when these competing species were reduced from 2005 to 2006. Overall, the most notable conclusion from our pre- and post-storm comparison for this sub-sample of critical birds of the lower Florida Keys is that they seem adapted for a high degree of resilience to major, natural, weather-related impacts on habitat. In this study, these species were confronted with direct breeding-season impacts from Tropical Storm Dennis plus lingering damage to essential habitat. Then, 3 months later, the same region received even more severe habitat damage from Hurricane Wilma that was extensive enough to persist for many years. Indeed, the negative effects on apparent numbers and breeding effort were clearly discernable the following breeding season. Nonetheless, by the second breeding season following these storms, most had rebounded substantially, and the WCPI, for which we had the most data on nesting attempts, essentially returned to pre-storm levels despite persistence of the damage to vegetation. Flight-line Counts.- These surveys indicated that the number of nesting pigeons declined sharply in the 2006 breeding season following the preceding year’s storms, but returned nearly to normal levels in 2007, despite insufficient time for recovery of the damaged nesting substrate. This was consistent with the response detected by the point counts. Further corroboration came from the greatly increased nesting effort from 2006 to 2007 that we observed directly during our banding fieldwork. All 3 assessments – the point counts, flight-line counts, and banding – clearly revealed the pigeons’ continued use of the major nesting islands in proportions similar to those observed prior to the 2005 storms despite the loss of 58% of the

44 black mangrove nesting substrate. That is, fidelity to the traditional islands was strong even though it meant that the birds had to nest in much greater density than previously. We do not know if this greater density compromised nesting success or productivity, but it clearly demonstrated substantial breeding-site fidelity and the capacity to accommodate more concentrated nesting conditions in the face of temporary loss of breeding sites. WCPI Activity Ranges and Habitats Pineland habitats, which include embedded hardwood hammocks where foraging pigeons were regularly observed, became more important to radio-tagged birds after the storm damage occurred in 2005, whereas the lower-elevation hardwood hammocks not associated with pine forest (i.e., free standing) became less frequently used. This might be explained by the higher probability that the isolated, exposed lowland hammocks endured storm damage that diminished fruiting. The pineland-embedded hammocks in our study area have been encouraged by fire suppression in pine forest on public lands in the lower Keys, where the many private inholdings make prescribed burning risky and difficult. We have previously noted the importance of these embedded hammocks as WCPI foraging sites (Meyer and Zimmerman 2008). The increased use of these feeding sites following the 2005 storms highlights their potential value for WCPIs in regard to long-term, landscape-level conservation planning. The low-frequency use of salt marsh and even mangrove forest by the radio-tagged pigeons is not surprising given that the telemetry surveys focused on foraging birds that spent relatively little time on or in transit to mangrove nesting sites.

45 WCPI Survival This study produced the first estimates of annual adult survival of WCPIs based on a respectable sample of marked individuals. These results will contribute to future demographic modeling of the WCPI in Florida and attempts to determine and track population trends. They also will make it possible to assess the effects of harvest on Caribbean populations, nearly all of which are subjected to substantial hunting pressure. We are unable to comment on the significance of the values of our survival estimates in and of themselves due to our lack of prior, comparative data and the absence of sufficient information in the literature. The model approach to estimating survival revealed significant effects on survival of both trap site and mass at capture. This supports our observations that pigeons captured on Little Torch Key and Summerland Key were lighter than those trapped elsewhere. The survival analysis determined that these birds also experienced above-average mortality. This is the first evidence to suggest that variations in habitat quality might affect WCPI condition and longevity, and that not all areas with high pigeon abundance on the mainline Keys necessarily represent optimal foraging habitat. There also may have been effects of age and/or sex, which were not tested in our survival model. Impacts on Winter Range The U.S., and especially Florida, has a stake in how WCPIs are managed in the Bahamas because of the impacts of the countryâ&#x20AC;&#x2122;s 6-month, 50-bird per day hunting season that coincides exactly with the overwintering period of Floridaâ&#x20AC;&#x2122;s migratory breeding population. The Bahamas National Trust (BNT) was well represented at the meeting we held in Florida in 2005 to launch an international WCPI working group and examine the best ways to manage the species cooperatively, across national borders -a particular challenge for WCPI

46 conservation given the species’ migratory behavior and allure as a game species in countries that can provide little science-based harvest management and enforcement. The BNT’s biologists see it as particularly important that the country initiate an on-going research and monitoring program with recurrent funding that will produce harvest and management guidelines. They recognize the concerns of U.S. biologists regarding the migratory population, but they also are alarmed at the apparent sharp decline in the breeding productivity of the 2 largest breeding colonies in the Bahamas – both near Andros Island. In effect, the intense hunting impact on migratory WCPIs wintering in the Bahamas is no longer diluted by the availability of local stock (not that the hunt was necessarily sustainable even when production was much higher in the Bahamas). This issue is growing in importance and urgency. The strategy proposed by the Bahamas for instituting management of the WCPI, both breeding and wintering populations, includes research on nesting biology and demography to set science-based harvest regulations, a much-increased law enforcement effort to enforce these regulations, and population monitoring. Recent changes in the BNT and generous increases in government funding are encouraging. There also is a strong interest by all parties in involving the hunting community in this work as well as in contributing to its financial support. All of these developments will benefit Florida’s migratory population of WCPIs. Jamaican management and conservation of WCPIs, and other columbids that are popular game species, has had a relatively high profile in Jamaica for some time relative to the Bahamas. This has come about by a combined commitment of government policy makers and biologists with private hunting groups and conservation organizations. Understandably, funding limitations have not supported all of the research and monitoring needed to inform their well-intentioned efforts, but interest and commitment are strong and there appears to be widespread agreement

47 among stakeholders. As in the Bahamas, we believe we can play a role in helping to gather information, training, endorsement of appropriate recommendations, and assistance with fund raising. Without a doubt, Cuba harbors the largest national population of WCPIs and, along with the Bahamas, probably represents the major wintering destination for the Florida population. It is likely, furthermore, that Cubaâ&#x20AC;&#x2122;s WCPIs serve as a source population for the rest of the Caribbean and, perhaps to some extent, for Florida as well. Unfortunately, government restrictions in both countries have stymied some excellent collaborative opportunities that have already emerged. We have been eager to capitalize on these opportunities and will continue to seek ways to do so. Right now, the most important contribution we can make is to advise Cuban ornithologists regarding population survey and monitoring techniques. As for all the islands that are likely wintering destinations of Floridaâ&#x20AC;&#x2122;s breeding population, we will continue to advertise for cooperation with banding returns and to seek updated information on harvests and other impacts. Our most recent Working Group meeting, in Puerto Rico in 2007, drew eager participants from island nations not previously involved in the Groupâ&#x20AC;&#x2122;s activities. This is very encouraging and calls for continued communication, including finding all opportunities to provide our support for these emerging programs on WCPI management and conservation planning. Banding WCPI Nestlings Nestling banding in 2007 was efficient and productive. Collaborative banding by biologists in Florida and the Caribbean can be an effective tool for expanding our knowledge of migratory movements and the potential impacts of managed and unregulated harvests in the

48 Caribbean during the fall and winter. We will continue to promote this effort and to offer guidance and training where helpful. We also will repeat efforts in coming years to solicit band returns from hunters on Caribbean islands and the general public in Florida. Each of these sources has produced 2 returns in recent years (during and prior to the present study). The substantially improved banding success in 2007 relative to 2006 corroborated our conclusions from the flight-line and point counts regarding poor nesting effort in 2006 followed by a rapid return to normal levels in 2007.

LITERATURE CITED Bancroft, G. T., and R. Bowman. 2001. White-crowned pigeon. In the Birds of North America, No. 596 (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, Pennsylvania. Beyer, H. L. 2004. Hawth's Analysis Tools for ArcGIS. Available at http://www.spatialecology.com/htools Chace, J. F., B. L. Woodworth, and A. Cruz. 2002. Black-whiskered vireo. In The Birds of North America, No. 607 (A. Poole and F. Gill, editors). The Birds of North America, Inc., Philadelphia, Pennsylvania. Eddleman, W. R., and C. J. Conway. 1998. Clapper Rail. In The Birds of North America, No. 607 (A. Poole and F. Gill, editors). The Birds of North America, Inc., Philadelphia, Pennsylvania. ESRI AcrGIS 9.2 2006. Redlands, CA: Environmental Systems Research Institute. Kaplan, E. L. and E. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481.

49 Lowther, P. E., and R. T. Paul. 2002. Reddish egret. In The Birds of North America, No. 607 (A. Poole and F. Gill, editors). The Birds of North America, Inc., Philadelphia, Pennsylvania. Lowther, P. E., C. Celada, N. K. Klein, C. C. Rimmer, and D. A. Spector. 1999. Yellow warbler. In The Birds of North America, No. 607 (A. Poole and F. Gill, editors). The Birds of North America, Inc., Philadelphia, Pennsylvania. Meyer, K. D., and G. M. Zimmerman. 2008. Foraging habitats, winter residency, survival, and philopatry of adult White-crowned Pigeons in the lower Florida Keys. Final report, Florida Fish and Wildlife Conservation Commission, Tallahassee, Florida, USA. Nams, V. O. 2005. Locate III Userâ&#x20AC;&#x2122;s Guide. Pacer Computer Software, Tatamagouche, Nova Scotia, Canada. Paul, R. 1977. Banding studies of white-crowned pigeons. In Proceedings of the international white-crowned pigeon conference. Bahamas National Trust, Nassau, Bahamas. _____ . 1996. reddish egret. Egretta rufescens. Pp. 281-294 in Rare and endangered biota of Florida (J. A. Rodgers, Jr., H. W. Kale II, and H. T. Smith, editors.). Volume V. Birds. University of Florida Press, Gainesville. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management. 53:7-15. Stevenson, H. M., and B. H. Anderson. 1994. The birdlife of Florida. University Press of Florida, Gainesville. Strong, A. M., R. J. Sawicki, and G. T. Bancroft. 1994. Estimating white-crowned pigeon population size from flight-line counts. Journal of Wildlife Management 58: 156-162.

50 Wilmers, T. J., and J. Arnett. 2002. A survey of reddish egrets in the lower Florida Keys. Unpublished final report, U. S. Fish and Wildlife Service, Florida Keys National Wildlife Refuges, Big Pine Key, Florida.