Bridging Archaeology and Marine Ecology: Coral Archives of Hellenistic Coastal Change

Abstract

Stony corals are long-lived, calcifying cnidarians that can be preserved within archaeological strata, offering insights into past seawater conditions, anthropogenic influences, and harbor dynamics. This study analyzes sub-fossil Cladocora sp. colonies from ancient Akko, Israel, dated to the Hellenistic period (~335–94 BCE), alongside modern Cladocora caespitosa from Haifa Bay, Israel. We employed micromorphology, stable isotope analysis, and DNA sequencing to assess species identity, colony growth form, and environmental conditions experienced by the corals. Comparisons suggest that Hellenistic Akko corals grew in high-light, cooler-water, high-energy environments, potentially with exposure to terrestrial waste. The exceptional preservation of these colonies indicates rapid burial, possibly linked to ancient harbor activities or extreme sedimentation. Our results demonstrate the utility of scleractinian corals as valuable paleoenvironmental archives, capable of integrating both biological and geochemical proxies to reconstruct past marine conditions. By linking archaeological and ecological records, this multidisciplinary approach provides a comprehensive understanding of historical coastal dynamics, including ancient harbor use, climate variability, and anthropogenic impacts.

1. Introduction

Stony corals are calcifying cnidarians that are responsible for the formation of large, shallow-water, coastal reef ecosystems [1] that house ~25% of described marine fishes and invertebrates [2,3]. They have adapted to thrive across relatively large gradients of environmental/abiotic factors including temperature, turbidity, flow, light and nutrient availability [4]. At the species level, stony corals exhibit abundant examples of phenotypic plasticity, or visible responses to environmental factors despite shared genotypes. Plasticity of coral growth location, morphology, physiology, and gene expression has been documented in response to light [5], water flow [6,7], nutrient availability [8], and seawater chemistry [9], and aims to maintain energy acquisition across environmental gradients. As such, stony corals have persisted, as a taxonomic order (Scleractinia), for more than 240 million years, enduring multiple catastrophes in Earth’s history [10], and leaving them as records of not only fairly static environmental growth conditions over the long term (e.g., presence/absence, macromorphology, and bulk skeleton analysis) but also of decadal, seasonal, and impending terminal ecosystem changes. For instance, it has been well-documented that the trace element and stable isotope composition of scleractinian coral skeletons records environmental signals from the surrounding seawater in species-specific ways (review by [11]), allowing for paleoceanographic reconstructions of seawater temperature, cation concentrations, and pH [12,13]. Recently Gothmann et al. [14] extended the use of fossil corals as geochemical archives into the Mesozoic. In addition, the symbiotic relationship between scleractinian corals and photosynthetic dinoflagellates can be detected from the isotopic composition of organic matter embedded within the coral skeleton (SOM), which reflects photosymbiotic activity (giving some clues towards light conditions) [15,16,17]. These isotopic signatures also provide insights into the relative contributions of auto- versus heterotrophy to both fixed carbon and nitrogen nutritional sources [15,16,18]. Multiple lines of evidence suggest that the nitrogen isotopic composition (δ¹⁵N) of SOM embedded in skeletons of corals and other calcifying organisms can remain stable over geological timescales [15,19,20,21,22], providing a valuable archive of ecological information without a priori knowledge of surrounding environmental conditions [23,24]. As such, coral-bound δ¹⁵N serves as a robust proxy for assessing photosymbiotic status in fossil corals [15,25], with well-preserved aragonitic skeletons retaining these isotopic signatures for over 200 million years following the death of the holobiont (the coral host and its associated symbiotic community, e.g., [25]).

Cladocora caespitosa (Linnaeus, 1767; family Cladocoridae) is a colonial endosymbiotic stony coral associated with dinoflagellates of the family Symbiodiniaceae, endemic to the Mediterranean Sea [26]. It is considered a remnant reef-building species and a vital evolutionary link between tropical coral reefs and the ancient reef ecosystems that existed in the Mediterranean prior to the end of the Messinian period [27]. Although large reef structures formed by this coral are frequently observed in the fossil record [27], they have become exceptionally rare in the modern Mediterranean [28,29]. Due to its longevity and distinctive calcification patterns, C. caespitosa is increasingly studied as a natural archive for reconstructing past environmental and climatic conditions in the Mediterranean region [30,31].

The Mediterranean of the Hellenistic period (232–63 BCE) was heavily populated along the coasts [32], with a sea level and sea surface temperatures (SSTs) probably lower and cooler than today, respectively [33]. Recent studies have revealed significant changes in SST in the eastern Mediterranean over the past millennia, recording a warming trend in the region, with the 20th century being the warmest period in the last millennium [33,34]. The Hellenistic period was politically unstable in the human history of the Syria-Palestine region, with the “Syrian wars”, the longest sequence of wars in the ancient world [35], spanning from 301 BCE or earlier to 103–101 BCE. Modern Akko (also called Acre and known at the time as Ptolemais-Ake; Figure 1) became a Greek polis at the end of the fourth century BCE, and was an important Egyptian military port and city during the Syrian Wars [36]. The city became active in regional and maritime trade, and it fairly quickly became one of the most important and prosperous cities in the region. During the Roman period, the harbor was the most important gateway to Europe and Rome and was favored by the Roman troops as the regular landing place for over two centuries [37]. Surveys and reconstructions by the British Mandate Department of Antiquities (1920s–1940s) revealed and then confirmed the importance of greater Akko to past geopolitical conflicts in the region [38]. Later archaeological surveys and excavations of Akko then began in the mid-1960s [39], revealing a variety of underwater breakwaters, foundations, and shipwrecks. Underwater excavations of coastal Akko in the 2010s under the auspices of the Israeli Antiquities Authority established six archaeological strata from the Hellenistic period (i.e., sub-fossil) to the British Mandate [37,40]. During these underwater excavations, eight square (2 × 2 m) pits were dug; among them only three were dug up to the bedrock, in which several well-preserved corals of genus Cladocora were identified as colonizers of the ancient bedrock [37,40]. The colonies were radiocarbon dated to 335–94 BCE [41]. Based on their size, corallite length and known modern growth rates of Cladocora [42,43], they were likely decades old at the time of burial. The relatively large colonies (~10–15 cm in diameter) and attachment to the bedrock at a depths that have remained shallow (~10 m) over the past two millennia led the authors to suggest that the site was minimally turbid—allowing sufficient sunlight for coral growth—prior to 94 BCE, with subsequent burial following an acute sedimentation event leading to excellent macro- and microscopic skeletal feature preservation [37,40]. These Cladocora corals thus have the potential to reveal the underwater environment and human uses of the Akko harbor from ~2300 years ago.

In this study, we examined Cladocora sp. corals, separated by approximately 2000 years for preservation of comparable ecological and environmental signatures that can be used to reconstruct past marine conditions. To evaluate this, we conducted multiple analyses of several shallow sub-fossil Cladocora sp. colonies excavated from portions of the Akko port dated to the Hellenistic period as well as to live Cladocora caespitosa from 10 and 30 m depth in Haifa Bay (~10 km north and south, respectively, of Akko Bay; Figure 1). Both micromorphology and DNA sequencing (modern and sub-fossil) support taxonomic identification. Colony growth form (macromorphology) and SOM amino acid stable isotope analysis from both modern and sub-fossil specimens suggests that they experienced roughly similar water energy environments and trophic position, although the sub-fossil Cladocora appears to have experienced higher light levels, at shallow depths, and exposure to anthropogenic terrestrial waste (i.e., sewage). Further, the exceptional preservation of the sub-fossil colonies suggests a rapid burial event. By integrating biological, geochemical, and archaeological data, this study offers a more integrated and comprehensive reconstruction of the eastern Mediterranean’s historical physical and social environments than analyses conducted in isolation.

2. Materials and Methods

2.1. Collection

Modern Specimens: Small fragments (approximately 2 cm each) from five live colonies identified as Cladocora caespitosa were collected via SCUBA from the Israeli Mediterranean Sea in January 2021 at 30 m depth in Haifa Bay, Israel, from the Leonid ship-wreck (32.86975N 34.95405E) (Figure 1). The corals were found on the top part of the bow area of the wreck, facing the light. The colony sizes ranged between 10 and 20 cm in diameter. One fragment was immediately frozen in a liquid nitrogen dry shipper for genetic analysis, and the rest were transported on ice and stored in −20 °C until fragmentation for ex situ analyses. Additionally, in September 2023, four colonies were found in Naharia, attached to the limestone substrate at a depth of 10 m and oriented toward the light. The corals ranged in size from 3 to 10 cm in diameter, and these colonies were examined in situ for photophysiology as described below.

Sub-Fossil Specimens: Five Sub-fossil Cladocora sp. colonies were collected via SCUBA diving during the excavation of the Hellenistic-early Roman Harbour of Akko in May-June 2012 and 2014 at 2.5 m depth. Excavation pits 1, 2, and 7 are located at (32.92094′ N, 35.07138′ E) [37,40]. All excavations were carried out under the auspices of the Israeli Antiquities Authority under permits A6557, A6829, and A7120. Colonies were documented in situ, detached carefully from the bedrock and put together with seawater in a closed plastic box. Boxes were tagged with the following data: permit no., dive and date, pit no. depth; samples of sediments from the surrounding area were taken separately. The coral boxes were transferred to the lab and stored at 4 °C until further analysis [40].

2.2. Photosynthetic Parameters Assessment

In situ physiological assays were conducted 20 September 2023 at 10 and 30 m, targeting live Cladocora colonies at each site: Naharia, Israel (10 m) and Leonid shipwreck, Haifa Bay, Israel (30 m). A Diving-Fluorescence Induction and Relaxation (FIRe) fluorometer [44] was employed, following methods previously described [45]. To capture colony-wide photophysiological responses and account for variability in light exposure within each colony, five readings were taken per colony and averaged, with data further analyzed as averaged readings per-colony. The kinetic nature of the Diving-FIRe technique enables real-time assessment of ETRmax without requiring dark adaptation, allowing in situ measurements during the daytime. Photophysiology data were determined to display normal distribution and homogenous variance between depths so that parametric statistical tests (Welch’s two sample t-tests) were conducted in R 4.0.3, with significance determined from p < 0.05. In addition, a Conductivity-Temperature-Depth (CTD) cast was taken on 5 September 2023 off the coast of Haifa Bay at a 100 m station. Photosynthetically Active Radiation (PAR) measurements from this cast revealed approximately 30% depletion in light intensity between 10 and 30 m depths, decreasing from ~200 to 137 PAR (µmol m⁻² s⁻¹) (Figure S1).

2.3. Skeletal Morphology

2.3.1. Scanning Electron Microscopy

Modern coral fragments were airbrushed and then submerged in 3% sodium hypochlorite for one hour to remove any remaining tissue, rinsed in freshwater, and dried at 60 °C. Sub-fossil coral fragments were sonicated in deionized water three times for five minutes each to remove debris and loosely cemented sand and then dried at 60 °C. Skeletal micro-morphology was visualised using a Thermo Fisher (Philips) XL20 scanning electron microscope (SEM, Waltham, MA, USA) at the Institute of Paleobiology, Polish Academy of Sciences, Warsaw, Poland. All samples were sputter coated with platinum and photographed. SEM was also used to visualize the micro-structural features of polished sections that were lightly etched in Mutvei’s solution following Schöne et al. [46].

2.3.2. Computed Tomography Scanning

3D visualization of the internal structure of the coralla was made with a Zeiss XRadia MicroXCT-200 system (Jena, Germany) at the Institute of Paleobiology, Polish Academy of Sciences, Warsaw, Poland. Scans were performed using the following parameters: voltage: 60 kV, power: 10 W, exposure time: 6 s, pixel size: 19.47 μm, 1201 projections. Three-dimensional images were obtained by processing with the AVIZO 7.1 Fire Edition software.

2.4. DNA Extraction

2.4.1. Modern Corals

Genomic DNA was extracted from each of the five samples of modern C. caespitosa using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA). PCR amplification was carried out with Kodaq 2X PCR MasterMix (ABM, Richmond, BC, Canada). To confirm the taxonomy of both the coral host and its photosymbionts, we targeted the highly conserved cytochrome oxidase subunit 1 (COI) gene using FOL-LDEG (forward) and FOL-HDEG (reverse) primers (modified from [47]) (Table S1). Additionally, the internal transcribed spacer (ITS2) region of Symbiodiniaceae rDNA was amplified using CS1F (forward) and CS2R (reverse) primers taken from [48], respectively (Table S1). The host COI region was sequenced via Sanger sequencing on ABI 3730xl DNA Analyser (Thermo Fisher, Waltham, MA, USA), while Symbiodiniaceae ITS2 libraries were sequenced on the Illumina Miseq platform (San Diego, CA, USA) using a v2-500 cycle kit to generate 250 × 2, paired-end reads. COI sequences from the five samples were aligned using ClustalW to create a consensus sequence, which was then BLASTed against NCBI’s GeneBank database for species identification. Symbiodiniaceae ITS2 data were demultiplexed by the Illumina software v4.0, and the paired forward and reverse fastq.gz files were submitted to SymPortal [49] to assess the diversity and relative abundance of endosymbiont species within each sample.

2.4.2. Sub-Fossil Corals

Specimens were first examined by light microscopy; no large exogenous biological material was observed while clays and some recrystallization was apparent as dense, shiny regions similar to secondary calcite (Figure 2). Specimens were sonicated for 2.5 h in deionized water with frequent water changes (every 10–15 min) to remove debris and loosely cemented sand. Cleaned fragments were dried at 40 °C. After cleaning, sub-fossil coral fragments were only handled in UCLA’s dedicated ancient DNA (aDNA) lab under established ‘clean’ conditions [50,51]. Standard contamination prevention protocols employed included: UV irradiation of the entire lab space before and after each use; only unopened consumables packages were brought into the aDNA lab and boxes were wiped with bleach and UV irradiated before opening and use; PPE always included disposable smocks, hairnets, foot covers, surgical masks, and double-gloving with nitrile gloves; and no modern specimens are ever allowed into the aDNA lab.

aDNA was extracted from cleaned Cladocora sp. using a Qiagen PowerSoil kit (Germantown, MD, USA), per manufacturer’s instructions, after sub-samples of four individual polyps were manually broken into ~1 cm² fragments to fit into the extraction tubes or ground without solvent into <1 mm² fragments in a mortar and pestle. Briefly, broken or ground sub-fossil Cladocora was vortexted in Qiagen Solution CD1 (Germantown, MD, USA) at maximum speed for 10 min. After centrifuging to pellet PowerBeads and coral pieces, supernatant was transferred to clean microcentrifuge tubes. Subsequent steps to centrifuge tube contents and transfer supernatant avoided bead and sediment pellets. Each polyp’s fragmentation method was extracted independently and DNA was eluted in Qiagen solution C6 or sterile water (four sub-samples of each) into clean microcentrifuge tubes. Extracted DNA was stored at −80 °C.

Prior to sample shipment for library preparation and sequencing, aliquots of all extracted DNA plus blanks were analyzed by Qubit (Thermo Fisher, Waltham, MA, USA) and TapeStation (Agilent, Waldbronn, Germany) to assess DNA quantity and quality (Supplementary Materials Table S4). No differences were observed in Qubit or TapeStation data; additionally, these pre-screening methods do not distinguish target versus contaminant DNA. Hence, Cladocora sp. aDNA was pooled according to fragmentation type (manual breakage versus grinding) and elution buffer (Solution C6 versus sterile water), yielding four fossil coral DNA samples in total which were shipped on ice (Supplementary Materials Table S4).

Pooled aDNA was prepared at the UC Santa Cruz Paleogenomics Laboratory into libraries via the “Spotlight” single stranded library preparation method Santa Cruz Reaction (SCR) optimized specifically for ancient (i.e., degraded) DNA [52], using Illumina P5 and P7 adapters, followed by Qubit and fragment analyzer assessment, and then sequenced on a NextSeq v2.5—150 cycles, mid-output. Forward and reverse paired-end reads were imported together to Geneious Prime software v2024.0 (GraphPad Software, Boston, MA, USA) for assembly and alignment. Paired reads were trimmed of adapters using the BBDuck plugin with a Kmer value of 27 and a quality score of 20 (99% likelihood of correct base calling), yielding 76-nt reads. Paired and merged reads from all four pooled aDNA libraries were mapped to mitochondrial genomes from fellow faviids Cyphastrea, Platygyra, and Dipsastraea downloaded from NCBI. Because of the low return rate of reads mapped to faviid mitogenomes, all merged reads from the most promising pooled library (SCS284) were also extracted for further analysis. Mapped reads and all SCS284 merged reads were then separately BLASTed against the NCBI core nucleotide database and best BLAST hit taxonomy was assigned using the taxonomizr package in R 4.0.3

2.4.3. Species Relationships

Modern C. caespitosa CO1 DNA sequences were BLASTed in NCBI and then aligned with all isolates from the same study as the best BLAST hit [53]. These were then aligned with other scleractinian CO1 sequences from the same portion of the gene, as well as a Nematostella sp. (class Anthozoa, order Actiniaria) CO1 sequence as an outgroup (NCBI Accession No. NC_008164.1, complete mitochondrial genome). An ancient Cladocora CO1 DNA sequence pulled from the full aDNA dataset was aligned with other scleractinian CO1 sequences from the same portion of the gene—further downstream from the location of the modern C. caespitosa sequence, as well as the Nematostella sp. CO1 gene as for the modern alignment. All alignments were performed in T-Coffee [54]. Best model fit was determined using default settings in the ModelFinder tool of IQTree 2.3.0 [55] and then trees were predicted in IQTree using 1000 bootstraps; best models based on BIC scores were HKY +F +G4 and K2P +G4 for modern-only upstream and ancient DNA-included downstream CO1 sequences, respectively.

2.5. Compound Specific Stable Isotope Analysis of Amino Acids (CSIA-AA)

2.5.1. Sample Preparation

Tissue from the live coral nubbins collected from Haifa Bay was removed from the skeleton using an airbrush with ultra-pure water, homogenized with an electric homogenizer, and centrifuged three times at 500× g for 10 min at 4 °C to separate host tissue from endosymbionts. Endosymbiont pellets were further washed with ultra-pure water after each triplicate centrifugation. Both fractions were freeze-dried before subsequent analysis. Sub-fossil and modern (after tissue removal) coral skeletal fragments were incubated in 3% bleach overnight, then washed in ultra-pure water and sonicated to remove all debris before decalcification for analysis of skeletal organic matter. Approximately 1.5 g of cleaned skeleton from each colony was hydrolyzed in 0.9 M HCl at room temperature until complete decalcification, concentrated via 15 mL Amicon centrifugal filtration units (30 kd), and freeze-dried.

2.5.2. CSIA-AA

The nitrogen and carbon isotopic composition of amino acids from tissue fractions (modern corals) and skeletal organic matter was analyzed using gas chromatography/combustion/isotope ratio mass spectrometry (GC/C/IRMS). Acid-hydrolyzed samples from the host, endosymbionts, and skeleton were derivatized using the EZ:faast kit (Phenomenex, Aschaffenburg, Germany) [18], with minor modification; dichloromethane was used in place of reagent 6 as the solvent. Approximately 4 mg of each hydrolyzed sample was processed. Amino acid separation was performed on a Zebron ZB-50 column (Phenomenex, Sutter Creek, CA, USA; 30 m, 0.25 mm, and 0.25 µm) using a Thermo Scientific Trace 1300 Gas Chromatograph (Milan, Italy) with helium as the carrier gas at a constant flow of 1.5 mL/min. For carbon isotope analysis, 1.5 µL of sample was injected in split mode (1:15) at 250 °C, while nitrogen analysis involved injecting 2 µL in splitless mode at the same temperature.

The separated amino acids were directed via a MicroChannel device (Thermo Scientific, Waltham, MA, USA) into two analytical streams: one to a Thermo Scientific ISQ quadrupole (Waltham, MA, USA) for compound identification, and the other to a Thermo Scientific Delta-V Advantage (Bremen, Germany) for isotopic analysis. Combustion of amino acids occurred in a GC Isolink II at 1000 °C to generate CO₂ and N₂. Prior to nitrogen isotope measurement, samples passed through a liquid nitrogen cold trap to remove interfering gases.

Each sample was analyzed in duplicate for carbon and triplicate for nitrogen. Isotope ratios were reported in δ notation, relative to Vienna PeeDee Belemnite (VPDB) for carbon and atmospheric N₂ for nitrogen. To correct for carbon introduced during derivatization, adjustments were made following Docherty et al. [56]. Since the Industrial Revolution and associated increase in burning of fossil fuels, the release of CO₂ with isotopically lighter carbon compared to that of the past has altered the overall atmospheric carbon isotope signature, which becomes reflected in lower δ¹³C values of primary producers. Because the samples examined here originate from time periods on either side of the Industrial Revolution, we corrected for these differences using an established Seuss correction factor [57]. Trophic position (TP_Glu-Phe) was calculated using glutamic acid and phenylalanine with the predefined equation of Chikaraishi et al. [58] and constants from Martinez et al. [18].

3. Results

In 2012–2014, during three seasons of underwater excavation at Akko Port, within the larger Haifa Bay area (Figure 1; Supplementary Materials Figure S2) eight metal molds were placed on the harbor sea floor at a distance of 20 to 40 m from the city wall in order to locate the edge of the wharf. During the excavation, stones up to 1.8 m in diameter were moved away to expose the bedrock in three square pits to a depth of 2.2–2.6 m, while only wharf pavement stones were exposed for cleaning in the remaining squares. Coral colonies and fragments were found in situ in the three fully exposed pits. Across pits 1, 2, and 7, a large preserved coral colony 20 × 20 cm remained affixed to the bedrock, one smaller colony remained connected to the bedrock, and some pottery shards covered with preserved corals were found, respectively [40]. ¹⁴C tests of several coral samples dated them to the mid-second century to early 1st century BCE during the Hellenistic—Early Roman Period [40]. The preservation of organic fragments, most likely remnants of the polyp (Figure 2A), along with the aragonitic mineralogy and the preservation of the finest skeletal structural elements, such as the nano-granulate structure in rapid accretion deposits (RADs) (Figure 2B,C), confirms that the corals were rapidly buried alive and tightly sealed by a protective layer of sediment.

3.1. Morphology

The almost identical skeletal structure, from the colony level to the micrometer scale, between the sub-fossil coral and the modern Cladocora caespitosa suggests that the sub-fossil coral represents the same taxon (Figure 2). For both specimens, small clumped colonies are composed of tubular, compact, upward facing corallites, each several mm in diameter, with their own distinct corallite wall (i.e., phaceloid) (Figure 2E,H). Microtomographic reconstructions of modern and sub-fossil corallites, along with their virtual sections (Figure 2F,G,I,J, respectively), reveal identical internal organization in both specimens. Interestingly, some calices of sub-fossil Cladocora show clear constrictions known as rejuvenescence (arrows in Figure 2I,J). SEM micrographs of the distal view of calices show well-developed septal granulations on the septal faces in both Recent (Figure 2K,L) and sub-fossil (Figure 2O,P) specimens. Etched transverse sections of septa and wall show an identical microstructural organization in Recent (Figure 2M,N) and sub-fossil (Figure 2Q,R) skeletons. The mid-septal zone consists of closely spaced rapid accretion deposits (RADs), which are also visible as part of the wall (the so-called trabeculotheca). Regular, fine-scale banding of thickening deposits (Figure 2N,R) is typical of symbiotic corals [25,59]. The sub-fossil Cladocora’s septa hexamerally arranged in approximately four cycles are well-preserved, with excellent preservation of RADs and lower inner edges of S1 septa merging with the columella (Figure 2A,C,D), as is observed for modern C. caespitosa. To support the visual taxonomic identification, we extracted DNA from both the recent intact Cladocora tissue from the Leonid shipwreck and the sub-fossil Akko Cladocora’s preserved skeletons with remaining tissue. In addition, we compared these exceptionally-preserved sub-fossil coral remains to living Cladocora corals immediately north (Naharia; 10 m specimen for in situ FIRe analyses) and south (Leonid shipwreck; 30 m specimen for in situ FIRe, external DNA sequencing and CSIA-AA analyses) to better understand the underwater environment of Akko Port during the Hellenistic-Roman period.

3.2. DNA Results

Analyses of cytochrome c oxidase subunit 1 (CO1; Table S1) gene sequences of the modern coral host (n = 5) collected at Haifa Bay from 30 m depth approximately 12 km south of Akko Bay showed 100% similarity to each other and 100% match to vouchers for eastern Mediterranean Cladocora caespitosa mitochondrial partial COI gene via NCBI nucleotide Blast (accession numbers: MW032500.1-MW032523.1) (Table S2; ref. [53]), confirming the taxonomic assignment based on macro-morphology. The dominant ITS2 (Table S1) endosymbiont type was identified as Breviolum psygmophilum (synonym of Symbiodinuium psygmophiulm, clade B2; Table S3).

Ancient DNA extracted from the best quality pooled sub-fossil Cladocora sp. sample (SCS284; Figure 3A; Supplementary Materials Table S4) was dominated by bacterial sequences, with cnidarians, and more specifically scleractinians, making up 0.3% (non-redundant) and 0.06% (total) BLAST taxonomic assignments of 706,328 merged reads. Of those merged reads, 50–90 mapped to one of the three Faviid mitogenomes used for comparison (Table S4); 10 of these reads that mapped to Cyphastraea had best BLAST hits to anthozoan nicotinamide adenine dinucleotide (NADH) subunits and cytochrome oxidase 1 (CO1) (Figure 3C). One read that mapped to Oculina sp. CO1 (accession no. OQ625380) clusters with the modern Haifa Bay Cladocora caespitosa (Figure 3D).

3.3. Photophysiology

Photophysiological measurements of modern C. caespitosa growing in Haifa Bay at 10 (n = 4 colonies) and 30 m (n = 5 colonies) (Naharia and Leonid shipwreck, respectively; Figure 1, Supplementary Materials Figure S3), and therefore different light intensities, show that the connectivity parameter, p, a measure of photoprotection (energy transference between photosystems II and I) and the functional absorption of light, σ_PSII (a measure of efficiency) did not differ between the two depths (p = 0.5098, 0.3194; Welch Two Sample t-test, respectively; Supplementary Materials Figure S4A,B). However, total photosynthetic efficiency (F_v/F_m) was significantly lower in 10 m corals and maximum photosynthetic yield (P_max) was significantly lower in 30 m corals (p = 0.007876, 0.002829; Welch Two Sample t-test; respectively Figure 4A,B).

3.4. CSIA-AA

Carbon CSIA-AA analysis of four essential AA (valine, leucine, isoleucine, and phenylalanine) indicates that there is no significant difference between the carbon isotopic signature of the host, symbiont, and skeleton of the contemporary Cladocora (PERMANOVA Monte-Carlo p = 0.53, Figure 4G) indicating resource sharing between the coral host and its endosymbionts. The sub-fossil coral skeleton δ¹³C of 3 out of 4 of the essential AA is at least 2‰ higher than all the contemporary compartments (Figure 4C–F).

Similarly to carbon CSIA-AA analysis, the nitrogen CSIA-AA shows no significant difference between the host, symbiont, and skeleton of the contemporary Cladocora (PERMANOVA monte-carlo p = 0.66, Figure 5). However, the sub-fossil coral skeleton δ¹⁵N values of individual amino acids are higher than the contemporary ones by an average of 2–13‰ (Figure 5A–I) except for threonine and isoleucine, which fall within their range. Interestingly, the highest difference (>13‰) is observed in source AA phenylalanine (Figure 5G) which is not affected by trophic discrimination in the food web. There is also no significant difference in the trophic position (TP_Glu-Phe) between the host, endosymbionts, and skeleton of the contemporary coral, and the sub-fossil coral TP_Glu-Phe skeleton is similar to the contemporary one (Figure 5K).

4. Discussion

The Mediterranean Sea is recognized as a long term hotspot of biodiversity, shaped by geological events, climatic fluctuations, and its limited connection to the Atlantic Ocean [60,61]. The region harbors a diverse array of temperate and subtropical species, along with a significant number of endemics [62]. Extensive coral reef systems thrived in the Mediterranean Sea during warmer geological periods, as part of the ancient Tethys more than 200 million years ago [63,64]; however, a major shift in salinity at the end of the Miocene (e.g., six million years ago), termed the Messinian salinity crisis [65] resulted in the near-complete loss of zooxanthellate corals and the disappearance of established shallow-water coral reefs [66]. The sudden opening of Gibraltar filled the Mediterranean again 5.5 million years ago, introducing new scleractinian species to the region [67]. In the present day, shallow rocky seabeds are predominantly covered by frondose algae while long-lived, filter-feeding organisms prevail in light-deprived benthic communities [68]. The scleractinian coral diversity of the Mediterranean today is relatively low, with 37 representatives [69] compared with over 700 species in the Caribbean and Indo-Pacific [4].

Since the Messinian crisis, the colonial coral Cladocora caespitosa (Linnaeus, 1767; family Cladocoridae) is the only remnant of ancient Mediterranean reefs, resembling tropical reef-builders in its ability to form reef-like structures named “reef banks” [28,70]. Ancient dead banks of C. caespitosa (dated from around 2500–3000 yr ago) are known from the coasts of Tunisia and Corsica (central Mediterranean) [71]. While it was abundant during the warmer Pleistocene, still forming reef-like structures [43], modern distribution of this taxon has declined dramatically in the eastern Mediterranean, while retaining a fairly abundant presence in the central-western Mediterranean (mainly found near the Tunisian coast, Aegean Sea, Spain, France and Croatia [4,26,28,72,73,74]), occurring from shallow waters to depths of about 50 m [29,75], on both hard and soft bottoms [76]. In the eastern Mediterranean, surveys conducted since 2014 at eight sites deeper than 25 m along the Israeli coast by the Morris Kahn Marine Research Station Long-Term Ecological Research program (https://med-.haifa.ac.il/data-base/; accessed 1 February 2025; access outside of Israel at https://haifauniversity.shinyapps.io/Marine_Invertebrates/) did not detect any C. caespitosa. However, in autumn 2020, C. caespitosa colonies were recorded for the first time at the Leonid wreck, and three years later additional colonies were observed off the shore of Naharia. C. caespitosa forms small-sized globose to hemispherical colonies ranging from a few hundred polyps to larger banks [72,77], typically of compact arrangements of separated polyps that retain relatively short corallite walls [4], as observed in both the modern and sub-fossil corals examined here (Figure 2). However, C. caespitosa colonies can grow as extended and diffuse corallites under extremely low flow conditions (Ref. [78]; Supplementary Materials Figure S5). The frequency for smaller colonies may result from high mortality rates or fragmentation due to climate changes [79]. It is a slow-growing coral (2.55 ± 0.79 mm year^–1 [29]), and is differentiated from other Cladocora taxa both by its upward-only facing polyps and its limited geographic distribution [4,80]. Hence, macro and micro-morphology, as well as DNA sequencing (Supplementary Materials Table S2) supports identification of the modern Cladocora examined here as C. caespitosa. While modern C. caespitosa can be abundant in the Mediterranean between the surface and 50 m [43,81], the sub-fossil Cladocora likely grew at <5 m depth [40], and shallower depths are typically sites of higher flow and elevated temperature. The compact clustering of the sub-fossil Cladocora colonies’ polyps suggests that they grew in a higher flow environment (Figure 2D,G versus Supplementary Materials Figure S5; ref. [78]). Additionally, the presence of calicular constrictions known as rejuvenescence suggests that this colony survived a past stressor event (Figure 2I,J) [82]. The ancient specimen’s excellent preservation is evidenced in the nano-granular structure of the septa (Figure 2), also observed in rapid accretion deposits of modern coral skeletons collected live. Both macro- and microstructure of the sub-fossil Cladocora adhere to descriptions for modern C. caespitosa. The intact microstructure and appearance of tissue, combined with evidence of rapid encasement in sediment, support its use in molecular analyses.

Ancient DNA (aDNA) sequencing has been successfully used in recent years to assign or confirm species identifications to organisms ranging in age from 100s or 1000s of years old in other invertebrates [83] to >1 million years old in mammoths [84]. Initial aDNA studies in scleractinian corals have relied on bulk skeletal analyses of reef-building species, e.g., Acropora palmata fragments dated 4215 BCE–1099 CE [85]. Our study is the first to identify preserved soft-tissue fragments and subject them to analysis. aDNA sequencing in the present study, along with macro- and micromorphology (Figure 2), supports the sub-fossil Cladocora specimen’s identification and highlights the increasing role that aDNA analysis can play in historical ecological understanding [86], particularly when species preservation is sufficient. For invertebrates, this can be made particularly difficult by the lack of tissue occlusion in the mineral; however, it is not without precedent as aDNA has been sequenced from ornamental mollusk shells and allowed taxonomic identification [87,88]. It is not surprising that the majority of the sequences from even the best quality aDNA extraction in the present study are off-target (Figure 3A,B). In other aDNA studies, as little as <10% of the mapped reads have been aligned to target taxon nuclear and mitochondrial genomes [89,90], including in the original validation of the library preparation method used here [52], with much of the remaining reads mapping to microbial DNA even when steps are taken to remove microbial DNA accepting sacrificial loss of target DNA as well [91]. These off-target sequences are typically due to post-preservation contamination in conjunction with degradation of target sequences over time, processes also observed in ancient protein studies on marine invertebrates [92,93].

Of the two species of Cladocora found in the Mediterranean, C. caespitosa is the only one with documented photosymbiotic ability [76]. Its plasticity in its ability to shift between heterotrophy and autotrophy for fixed carbon acquisition is often related to strong seasonal variability in light availability [94,95], water motion as waves and currents, type of substrata, and seafloor morphology [26,28,29]. Highly variable physiology and morphology, and unresolved genetic relationships of the Favidae family, have left genus Cladocora—and many other taxa in the family—grouped loosely together and considered ‘problematic’ in terms of classification [96]. However, recent study supports a close phylogenetic relationship between Oculina patagonica and Cladocora caespitosa, suggesting their placement within the same family, and potentially even the same genus [97]. This variability and unstable phylogenetic classification, combined with its relative scarcity in the eastern Mediterranean, has precluded its previous utility in reconstructing past shallow water ecology in the Levant. However, the regular, fine-scale banding of thickening deposits, typical of symbiotic corals [59] observed here both in the modern and sub-fossil coral (Figure 2) suggest that the sub-fossil coral is indeed C. caespitosa.

The coral diet appears to have remained remarkably consistent over the past 2000 years, with a high trophic position (TP) indicating a significant reliance on heterotrophic feeding (here ~2.5 compared with TP ~1 for autotrophy [98]. However, the skeletal δ13C values of sub-fossil Cladocora are notably heavier (more positive) than those of their modern counterparts, likely due to differences in light environment [18]. The decoupling between high F_v/F_m and low P_max observed at the deeper site (Figure 4A,B) suggests that while the photosynthetic apparatus remains functionally efficient, light or downstream metabolic limitations may constrain the overall carbon fixation capacity [99,100,101,102,103,104]. Under low-light conditions, isotopic discrimination against ¹³C by Rubisco remains high, favoring the assimilation of ¹²C and resulting in lighter δ¹³C values in the coral skeleton. However, the observation of heavier δ¹³C in the sub-fossil samples (Figure 4C,D) indicates that the corals were not experiencing light limitation at the time of growth. Instead, the enriched ¹³C signature suggests that photosynthetic rates were sufficiently high to reduce isotopic discrimination, consistent with an environment characterized by ample light availability. This interpretation is supported by the paleoenvironmental context, as sub-fossil Cladocora colonies likely grew in shallower and less turbid waters, where higher light availability facilitated faster photosynthesis and heavier δ¹³C signatures [102,103,104]. Furthermore, the stable δ¹³C values we observe between the two time points ~2000 years apart indicate that the balance between autotrophic and heterotrophic nutrition in the sub-fossil Cladocora has remained fairly constant, despite environmental and human-induced changes.

In contrast, significant shifts are evident in nitrogen isotope signals, with nitrogen being heavier in sub-fossil specimens. This difference can be attributed to two interacting factors: (1) historical documents reveal that during the Middle Ages (Crusader period), the city of Akko’s sewage was discharged directly into the harbor, creating a nutrient-rich environment that is typically turbid; in general, Hellenistic and Roman cities had impressive sanitary systems that collected the sewage and discharged it away from population centers [105,106]. The presence of heavier nitrogen isotope in the sub-fossil coral indicates that sewage was discharged into the port of Akko as early as the Hellenistic period, making anthropogenic land-based nitrogen the primary source for corals then, similar to observations near modern cities experiencing recent rapid growth [107]. (2) The widespread adoption of synthetic fertilizers after 1913, produced using the Haber–Bosch process, which converts atmospheric nitrogen into ammonia, introduced nitrogen with isotopic values close to 0‰. This has contributed to the lowering of nitrogen isotope values observed in modern corals [108,109].

The δ¹⁵N distribution in organic matter preserved within coral skeletons has proven to be a valuable tool for reconstructing historical changes in nitrogen sources. For example, examination of skeletal δ¹⁵N in modern Indonesian poritid corals grown between 1970 and 2003 revealed that corals from fertilizer-affected reefs experienced a steady decline in bulk δ¹⁵N values, from +10.7‰ to +3.5‰, reflecting increased use of chemical fertilizers with low δ¹⁵N values and a reduction in organic manure application in upstream agricultural fields [110]. In contrast, corals and algae from sewage-affected reefs maintained consistently high bulk δ¹⁵N values (>8‰), underscoring the isotopic difference between chemical fertilizers and organic nitrogen sources such as sewage [110]. While changes in bulk δ¹⁵N in coral can be attributed to changes in their heterotrophic diet, CSIA-AA demonstrates that the coral maintained its trophic position. Furthermore, using the source amino acid phenylalanine, which remains unchanged between TP, we show that the modern corals exhibit a relatively low δ¹⁵N value averaging 0.3‰, aligning with the isotopic signature of the Haber–Bosch process. Conversely, sub-fossil corals, unaffected by synthetic fertilizers, show much heavier δ¹⁵N-phenylalanine values, approximately 13‰, resembling those found in sewage-affected environments.

The remarkably high degree of preservation of the sub-fossil C. caespitosa—intact microstructures and lack of secondary infillings [111], the fact that the corals were found below the Hellenistic quay, and the presence of organic fragments likely representing polyp remnants suggests a rapid burial process [40]. This burial was possibly triggered by a sudden singular sedimentation event caused by anthropogenic activity, such as the foundation of the harbor, that could cause changes in sediment movement or other changes inside the harbor, as supported by archaeological evidence from the region [40]. Corals under environmental stress retract their polyps into the corallites, and in cases of necrosis, remnants of polyp tissues can settle on internal skeletal structures. In Cladocora, this phenomenon has been documented by Kersting & Linares [82], who described ring-shaped remnants of polyp tissues. The structures identified in our subfossil samples closely resemble these features, supporting the hypothesis that local polyp necrosis and subsequent regeneration occurred during colony growth. This is further corroborated by evidence of rejuvenescence in the colony’s astogeny. Accordingly, these well-preserved fragments were selected for ancient DNA (aDNA) analyses. It should be noted, however, that due to post-mortem bacterial degradation of soft tissues, the recovered sequences are predominantly bacterial in origin.

Additionally, the discovery of multiple Cladocora specimens in all the pits that reached the bedrock during the excavation in Akko suggests a historically larger population. In contrast, no Cladocora colonies were documented by the surveys of the Morris Kahn Marine Research Station Long Term Ecological Research program (MKMRS LTER; established in 2014—https://med-lter.haifa.acil/index.php/en/data-base; accessed on 1 February 2025; access outside of Israel at https://haifauniversity.shinyapps.io/Marine_Invertebrates/)) conducted along the Israeli coasts which indicate a limited distribution with only two known sites harboring a few colonies each.

The discrepancy between historical and current Cladocora population, along with the rapid increase in temperature in the eastern Mediterranean [33] suggests that the population reduction may be related to climate change, pollution, and habitat destruction [112]. Persistence of the species in the eastern Mediterranean today therefore suggests that the population was able to adapt, migrate locally, or repopulate from sites further west, or some combination of these. The species may continue to use these mechanisms today in response to the effects of Anthropocene climate change. Minimal reef accretion observed in the eastern Mediterranean over the past several thousand years precludes coring for a fine-resolution temporal study of ecosystem shifts as performed in tropical locations (e.g., [113]); however, the present study provides important timepoints and geographic locations of comparison in an anthropogenically dynamic environment [114]. Further, the present and similar recent studies comparing between data sets from modern organisms and ancient environments, utilizing newly collected specimens as well as emerging sources in museum-held collections and government- and NGO-directed cultural excavations may also serve as lessons for furthering our understanding of how other coral taxa will meet climate change impacts (e.g., [107,115]).

5. Conclusions

Through the interdisciplinary comparison of a 2000-year-old coral skeleton with modern taxonomic brethren, we propose a framework of the ecological structure of Akko port ~2000 years ago. The present datasets support that at that time, shallow water Cladocora caespitosa, a zooxanthellate stony coral, settled on appropriately hard substrates at a relatively shallow depth with sufficient water flow to promote the formation of compact colonies. Our results suggest that these corals thrived in a high-flow, nutrient-rich, and turbid environment; however, despite the turbidity, light availability was adequate to sustain efficient photosynthesis, likely due to the colony’s position in water shallow enough to allow sufficient light penetration. The colonies possibly survived one or more stressor events. These corals were mixotrophic, like their modern counterparts, and occupied a similar trophic position, with port water clear enough to support photosymbiosis supplemented by phyto- and zooplankton communities, possibly supported by nutrient-rich effluent from the city of Akko, similar to anthropogenic coastal nutrient loading observed today in urbanizing coastal systems [109,116,117]. A sudden catastrophic event entombed the colony studied here, leaving it akin to a hard drive containing much and varied data about the past ecology of the port. Because C. caespitosa colonies are rarely found in the eastern Mediterranean today under similar conditions, the present comparison shows how temperate stony corals, and the structures they provide, respond (adapt, migrate, or die out) to local physical conditions. Similar ecological reconstructions in other locations will also require very well-preserved specimens at both the microscopic (e.g., mineralogical for geochemical analyses) and macroscopic scales, as well as modern local analogs for holistic comparison.