Introduction

Industrial-scale extraction of natural resources by colonial powers is becoming an important theme in historical archaeology (Baisre 2010; Fitzpatrick et al. 2015; Högselius 2020; Kittinger et al. 2015; Van Buren 2021). Pearl oyster fisheries is a prime example of resource extraction in colonized territories with a reported negative impact on biodiversity in historical sources (Aqil 2018; Delgado 2012; Farn 1986; Fletcher et al. 2006; Strack 2008; Valcárcel-Rojas et al. 2020). Archaeological evidence for this significant economic phenomenon remains poorly investigated. Here, we present the first comprehensive archaeological study of the pearl oyster middens of the Gulf of Mannar (hereafter GoM) in Sri Lanka, remnants of long-term exploitation of oyster habitats for pearl fisheries in the Indian Ocean (Fig. 1a-b). We approach the remains of Pinctada pearl oysters in these middens as vessels of information on the exploitation of natural resources by non-local powers and try to understand the impact of this extraction on coastal ecosystems.

Fig. 1
figure 1

(a) Geographical distribution of Pinctada sp. and fisheries; (b) Distribution of pearl banks in the Gulf of Mannar. Study area marked with a square. (c) Distribution of identified pearl shell midden clusters shown in different colored lines; (1) Keeri Beach, Mannar Island (KB), (2) Arippu (DB), (3) Silavathura, (4) Kondachchi (KD), (5) Kayakkuli-Kal Aru, (6) Marichchukaddi. Pearl banks marked are as referred in the text (pearl banks are not in scale, based on Herdman, 1903, 1905) (by Thilanka M. Siriwardana)

Full size image

Shell middens commonly represent subsistence shellfish gathering (Erlandson and Rick 2008; Fitzpatrick et al. 2015; Rick 2023; Robson et al. 2023; Waselkov 1987). Pearl shell middens can be studied using the same methods used to investigate subsistence shell middens, but they are unique because they result from fisheries specifically carried out to obtain pearls for trading globally in an interconnected market (Ali 2006; Arunachalam 1952; Carter 2005; Ferguson 1891; Sessional Papers of Government of Ceylon 1889; Woolf 1963). European demands transformed pearls from Sri Lanka and other regions into luxury items (Carter 2005; Rosenthal 1952). Sri Lanka facilitated pearl trade in the broader Indian Ocean region as one of its crucial producers (Flores 2007).

Previous in-depth analyses of historical pearl fisheries have been based on Dutch and British accounts (Ali 2006; Arunachalam, 1952; Bastiampillai 1980; Donkin 1998; Fernando 2021; Katupotha 2019; Ostroff 2016, 2020; Siriwardana 2014; Subrahmanyam 1996; Vadlamudi 2016). These focus on the economic aspects of the fisheries, though some record the formation of the shell middens as well (Caldwell 1877; Donnan 1885; Ferguson and Ferguson 1888). In contrast to rich historical records of pearl fishing in the GoM, hardly anything is known as to what extent pre-Dutch extraction impacted the Pinctada populations and their ecosystem. Pearl fishery middens in GoM are part of a unique cultural landscape threatened by rising sea levels, coastal erosion, and development, increasing the urgency of their documentation.

Pinctada pearl oysters are widely distributed in the Indo-Pacific Ocean (see Fig. 1a), and they have been harvested at varying scales since prehistory. The archaeology of pearl harvesting has mostly examined midden and settlement distribution patterns, with work around the Persian Gulf (Aqil 2018; Carter 2005; Farn 1986; Hawker et al. 2005; Hellyer 1998; Hellyer and Hull 2002; Krzemnicki et al. 2017), Spanish-American colonies (Antczak et al. 2019; Orche 2009) and Eastern Asia (Valcárcel-Rojas et al. 2020; Southgate and Lucas 2008). Archaeological work on Pacific islands and Australia examined the settlement patterns and ethnic identities of fishers involved in oyster fishing for mother-of-pearl during the nineteenth century (Dooley et al. 2021; Fletcher et al. 2006; McPhee 2004; Paterson 2017; Paterson and Veth 2020; Sissons 1979; Skippington et al. 2022; Strack 2008). None of these studies examined the main midden contents, the oyster shells, in detail.

The study identified significant shell midden deposits through surveys and using historical sources. We describe the archaeological record from four excavated sites with specific attention to data from molluscs. We use molluscan taxonomic abundance data to reconstruct the ecological communities affected by pearl fisheries and shell size to infer the patterns of oyster exploitation and its impact on oyster populations.

Historical Background

The pearl fishing industry was the reason why the GoM was once known as Salubham: the Sea of Gain (Herdman 1903). The earliest mentions of pearl fishing date to the fifth century BCE (Deraniyagala 1992; Donkin 1998; McCrindle 1877; Shamsastry 1929). On the island, the Anuradhapura Kingdom (377 BCE-1017 CE) provided the Roman Empire with pearls (Lee 1847; Pliny 1855). Continued pearl extraction played a crucial role in Sri Lanka’s connectivity with Roman and Sassanid empires (Al-Tabari 1999; Bustamante-Alvarez et al. 2021; Francis 2002; Imam 2003; Schoff 1912; Stern et al. 2008). Although Indo-Arabian networks dominated the Indian Ocean pearl extraction for a while, by the twelfth century, Sri Lanka had developed more protective policies concerning its island and offshore resources (Geiger 1929, 1930). Since then, the pearl industry has been controled by different political entities in Sri Lanka and South India, including the Pandyans, Cholas, and Muslims, and disputes over the rights to pearl fishing were not uncommon (Barbosa 1866; Geiger 1930; Gunasekara 1900; Masefield 1908).

Later on, the colonial administrations of the Portuguese (1597–1658), Dutch (1658–1796), and British (1796–1948) exploited the industry with maximum efficiency (Hornell 1905; Pieris 1992; Queyroz 1930). The British streamlined extraction methods and merged the GoM and the Persian Gulf fisheries. Rigorous marketing increased the global demand for pearls, making the industry even more profitable (Rosenthal 1952; Sessional Papers of Government of Ceylon 1889; Strack 2008). The invention of pearl cultivation in the early twentieth century marked the beginning of the rapid decline of traditional natural pearl production. (Strack 2008). The last pearl fishery in Sri Lanka stopped operating in 1983, signaling the end of this once-prosperous industry (Pernetta 1993).

Environmental and Geographical Setting

The Gulf of Mannar (GoM) is situated between the coasts of Sri Lanka and India. The bay boasts a variety of marine habitats, including sandy shores, mud flats, coral reefs, and deep-sea habitats. These habitats provide breeding grounds for over 4,200 marine species (IUCN 2011; Swan 1983), including 856 species of molluscs (Balaji and Edward 2012).

The pearl fishery industry benefited from the GoM marine ecosystem. Pearl banks, also known as oyster colonies or paars, were found on exposed sedimentary formations, sandy bottoms, and coral reefs (Alagarswami and Chellam 1977; Mohamed et al. 2006; Pernetta 1993; Vamadevan 1995) (see Fig. 1b, c). The majority of the paars were located in shallow waters between Mannar Island and Negombo to a depth of around 15 m (Herdman 1903; Sessional Papers of Government of Ceylon 1879). William Herdman, an English naturalist, classified the Sri Lankan pearl banks of GoM into two categories: economically beneficial northern banks from Mannar Island to Kudiramalai and economically less-productive southern banks below Kudiramalai (Herdman 1903; Hornell 1905). To mitigate the impact of the global pandemic and the political and economic upheaval in Sri Lanka in recent years, our study focused on the north section. This section includes the Arippu and Mannar pearl banks, Kareitivu Island, and Dutch Bay banks, while the southern section contains the Chilaw banks, Marawila banks, and Negombo banks (Donnan 1900; Sivalingam 1961; Vane 1887) (see Fig. 1). The nutrition outflows of the Kala Oya, Moderagam Aru, Kal Aru, and Aruvi-Aru rivers supported the complex ecological system at the northern banks (Miththapala 2012; Pernetta 1993). P. fucata is the keystone species of pearl banks. They provide the habitat and food for many organisms adapted to the paars ecosystem (Herdman 1903).

Mannar’s coastal plains provide marginal conditions for humans despite the rich marine environment the sea offers. This is due to the region’s semi-arid climate (annual rainfall below 1,000 mm). The area experiences frequent droughts due to dry winds from the southwest monsoons (mid-May to September), while the northeast monsoon (December to February) and two intermonsoons bring little rain. Moreover, the region’s mean annual temperature exceeds 28 °C (Mannar meteorological station data from 1890 to 2022). Population density is high in the north thanks to agricultural production (Vidanapathirana 2012; Wijepala et al. 2023) and thin in the southern areas due to dense vegetation and the lack of potable water (Boake 1888; Denham 1912). It is worth noting that European colonial administrations prohibited the establishment of settlements on the southern shores of Mannar island (see Fig. 1c), while several major centers developed along the northern shores (Boake 1888; Pieris 1920). Livelihoods primarily depend on fishing, sustained by estuaries and creeks that provide safe landing spots for boats and suitable settlement sites. The coastline is characterized by Holocene dunes covered by thorny shrubs, palmyra palms, and perennial plants.

Materials and Methods

This research study utilizes multiple approaches such as surface surveys, excavations, chronological sequencing, taxonomic analysis with species richness and diversity, shell fragmentation and size frequency analyses, and chemical weathering assessment to investigate various aspects related to the archaeology of the pearl oyster middens. Specifically, we seek insights into the exploitation patterns of the pearl oysters and the ecological conditions in which exploitation occurred.

Surface Surveys

A surface survey was conducted during January and July 2021 to understand the distribution of archaeological sites along the ca. 45-km coastline between Mannar Island and Marichchukaddi and choose locations to excavate. The surface survey was planned considering factors such as abundance, clustering, visibility, exposure, and accessibility of sites, following Burke and Smith (2004), Gause and Gerike (1997), and Schiffer et al. (1978). Shovel testing along a 1.5-km line was employed to locate sites on Mannar Island, while exposed sections were examined to identify sites on the mainland. Settlements were identified by ceramic scatters and architectural structures, both with and without shells. In Kondachchi, surveys spotted high concentrations of ceramic fragments in close proximity to an oyster midden, the relationship between the two was later tested by excavation. In addition to archaeological fieldwork, colonial maps, and archival records were examined to understand midden stratigraphy and chronology.

Excavations and Dating

We conducted 1 × 1 m test pit excavations in three selected middens and a settlement based on relative dates inferred from surface ceramics types. Excavations were done at Keeri Beach (KB), Arippu (DB), and Kondachchi (KD 1) middens, as well as a settlement mound at Kondachchi (KD 2). We used the context recording method suggested by Drewett (2001). Dry/wet sieving with a 1-mm mesh sieve was employed to retrieve the remains. In Arippu, the midden deposit was dense and compact, so we used a wet sieving method with a 2-mm mesh to prevent excavation-induced fragmentation. All retrieved materials were transported to the Rajarata University of Sri Lanka for study.

To establish a chronological framework, we used the radiocarbon dating of shells and charcoal and the relative dating of ceramics and porcelains. We are aware that several factors, like modern carbon contamination, dead carbon contamination, enriching shells with phosphatic matter, and limestone dilution, could impact the accuracy of dating results (Douka et al. 2010; Dipayan et al. 2016). However, no vertebrate remains are preserved in the contexts excavated. Though land snails are present, they can lead to time-averaging impacts as they are known to dwell in subsoil. Beta Analytics conducted the Accelerator Mass Spectrometry (AMS) radiocarbon analysis. For the pearl oysters, the mean ΔR -98 ± 24 was used to correct for the marine reservoir effect from the Marine20 calibration dataset (http://Calib.org/marine).

We used anatomical features and form types to classify the types of ceramic found for relative dating. We relied on Schenk’s (2001) documentation of the sequence of form types, A through K, at Tissamaharama up to the ninth century CE. However, ceramic sequences since the tenth century have been studied in less detail to aid in establishing accurate ceramic chronologies. Nevertheless, we refer to several works from Sri Lanka and South India to date the ceramics (Abraham 2002; Johal 2023; Ragupathy 1987; Rajan 1994; Sinopoli and Morrison 2007).

Taxonomic Abundance, Diversity, and Richness

The faunal remains were identified to the lowest possible taxonomic level, using the World Register of Marine Species (WoRMS Editorial Board 2022), M. Fernando’s (2021) guide to the shells of Sri Lankan seashore and for terrestrial species, the work of Raheem and Naggs (2006). The Number of Identified Specimens (NISP) was recorded by counting all specimens on the sieve. To calculate the Minimum Number of Individuals (MNI), we applied a Non-Repetitive Element (NRE) based approach, using hinges (for bivalves) and apices (for gastropods) (Giovas 2009; Mason et al. 1998). Left and right hinges were counted for bivalves, and the higher count was used to represent the MNI (Mason et al. 1998). A severely damaged hinge was classified as non diagnostic and was not included in the MNI calculations.

We used the Shannon Index of Diversity (H’) and the Margalef Richness Index (d1) to estimate the diversity and richness of midden fauna and infer the ecological conditions of pearl banks. The Shannon Index of Diversity quantifies the species abundance and richness within excavation units (Cruz-Uribe 1988; Peres 2010). The index is computed using the formula:

$$H{^\prime}= \Sigma{P}_{i} \text{l}\text{o}\text{g} {P}_{i},$$

where Pi represents the relative abundance of the ith taxon within the sample. Lower H’ values suggest lower species diversity, while higher values indicate more diverse ecological communities represented by excavated units. Species richness was calculated using the Margalef index (d1) with the number of taxa (NTAXA) contributed to the faunal assemblage (Cruz-Uribe 1988) and is given by:

$$d1\hspace{0.17em}=\hspace{0.17em}S-1/\text{l}\text{o}\text{g}N\text{,}$$

where S is NTAXA, and N is MNI.

Higher Margalef index values correspond to higher species richness, while lower values correspond to lower species richness.

Shell Fragmentation and Morphometry

Shell morphometry provides crucial information to understand the impact of exploitation frequency and pressure (Crumley 2021; Erlandson and Rick 2010; Jerardino and Navarro 2008; Thurstan 2022; Urlich and Handley 2020). To analyze exploitation patterns in pearl oyster fisheries, determining the size frequencies of the oysters is necessary. Researchers and traditional divers have agreed on the maturity age for P. fucata fisheries: At around four to five years, thick, robust shells with visible growth lines are considered ideal for pearl harvesting (Herdman 1903; Hornell 1905, 1922). Due to the fragmented nature of the oysters in the middens, this study uses Heel Depth (HD) to reconstruct the size frequencies of the pearl oysters. Ecologists commonly use anteroposterior and dorsoventral measurements, hinge length, and thickness to quantify pearl oyster morphology (Alagarswami and Chellam 1977; Mohamed et al. 2006) (Fig. 2). However, these methods are unsuitable for archaeological materials due to the inherent fragility of pearl oysters.

Fig. 2
figure 2

Shell dimensions of pearl oysters as used in the study. APM- Anteroposterior measurement, DVM-Dorsoventral measurement, HL- Hinge Length, HD- Heel Depth (modified after Gervis and Sims, 1992, fig. 4)

Full size image

Differential preservation across successive deposits might affect shell size frequencies reconstructed from HD. This issue was addressed by recording the frequency of different morphometric attributes preserved in each Non-Repetitive element (NRE). The analysis primarily focused on measurable attributes and their relationship to fragmentation. The measurements of Dorsoventral Length (DVM), Hinge Length (HL), and Heel Depth (HD), also known as hinge groove, were taken following Saucedo and Southgate (2008).

The preservation of mollusc shells composed of calcite (CaCO3) also depends on the localized pH conditions, so it is also crucial to consider the potential taphonomic bias in shell size introduced by chemical weathering. To investigate the impact of chemical weathering, we conducted pH measurements using a 1:2.5 soil-to-water ratio on samples from excavated deposits in the Soil Science Department laboratory of Rajarata University. All the studied pearl shell middens have high pH values, indicating alkaline environments. Mollusc shells release calcium carbonate, which combines with groundwater to preserve organic materials in an alkaline environment (Robson et al. 2023).

Studies on the HD-growth correlation of wild populations of P. fucata are limited, but available data indicate that pearl oysters reach an HD of 5 mm after three years of age and grow by 1.5 mm annually (Gervis and Sims 1992; Tranter 1959). Assuming that oyster growth did not change significantly during the Late Holocene, HD values of 6–8 mm should indicate the size/age of oysters exploited for pearl fisheries. Vernier calipers were used to take the pearl oyster measurements, which were taken to 0.1 mm, following the methods described by Gervis and Sims (1992).

We are checking if there is a significant difference between the hypothesized mean value obtained from the population mean of all excavation units and the mean value of each test pit level. If the oysters were all of similar size and age throughout the excavation, their shell sizes would likely cluster around a mean value. Also, if the size and age of exploited oysters changed across different periods due to less selective fishing practices, we would expect a distinct difference between the average size of shells from one pit level deviating from the overall average size. This helps us understand if fishing practices followed selective harvesting, illustrating such practices where lower values may represent fishing juveniles due to more frequent harvestings. Data was analysed using one-sample Student’s t-test at a 95% confidence interval. The analysis is carried out using JASP version 0.16.2 and Microsoft Excel 2013.

Results

Surface Surveys

Field walking revealed 49 new archaeological sites, including six pearl shell midden clusters, nine shell middens, 29 ceramic scatters, and four wells. Shell midden clusters are groupings of mollusc shells that were transported and separated by dunes, with several hundred-meter gaps between the clusters. When multiple middens are in proximity, they are considered part of the same cluster. The pearl shell midden clusters are separated by distances ranging from 1.7 km (Kal Aru to Kondachchy) to 28.5 km (Arippu to Keeri Beach). These were distributed along Keeri Beach (ca. 1.9 km), Arippu (ca. 2.6 km), Silavatura (ca. 2 km), Kondachchi (ca. 1.1 km in smaller deposits), Kayakkuli-Kal Aru estuary (ca. 6.6 km), and Marichchukaddi (ca. 1.9 km). These clusters are mainly located in areas with access to seasonal streams (Āra) or other freshwater sources. Areas of human habitation were accompanied by edible shell middens and wells near pearl oyster middens. Ceramic scatters were found near or away from the shell middens. These findings provide valuable insights into the region’s past human activities and resource use (additional data are given in Online Resource 1). Site visibility was affected by coastal erosion, aeolian deposition, and vegetation (tropical thorns).

Eroded remains on cliff tops suggest that these middens were originally much more extensive, extending inland, but have recently been eroded by the sea. Pinctada fucata shells were the dominant species on all midden surfaces. The middens were concealed with dunes except Arippu, which was covered with soil. Middens lack any visible permanent structures except the Doric Bungalow, Arippu, the residence of the British governor, built-in 1802 (Fig. 3a). The bungalow was constructed using bricks and lime made of pearl oyster shells. The Arippu middens now expand over an area of about 0.4 km2; they must have been more extensive before the sea eroded them away.

Fig. 3
figure 3

(a) The British Governor’s House at Arippu, now famously known as the Doric Bungalow, as it was seen during the early nineteenth-century CE. (Cordiner 1807); (b) Test pit excavation in the midden deposits at Doric Bungalow, with present remains of the bungalow in the background (photo by Thilanka M. Siriwardana)

Full size image

Settlements are defined with ceramic scatters. Some scatters represent extensive settlements, such as the ceramic deposit found on a dune formation at Kondachchi, which extends for about 1 km along the coast and contains plain redware and local Grooved Rim Ware (GRW). Through surface surveys, 74 rims were collected from the middens, showing various distribution patterns and associations in the case of porcelain, GRW, and plain ware (Figs. 4 and 5). Another example is a mound north of Arippu that contains plain redware, edible shells, and a human burial facing west and heading north. The inland dunes of Keeri Beach have potsherds and edible shell fragments, primarily Donax sp. (Additional data of surface surveys given in Online Resource 1).

Fig. 4
figure 4

Selected ceramic form types from the KD2 settlement excavation (a-L4, f-L3P4, g-L2, h-L3, i-L4), KB (b, d) and, DB (c, e). GRW types- b, c, d, e, g, (Drawing by Nadeera Dissanayake). Type forms after (Schenk 2001)

Full size image
Fig. 5
figure 5

A British Period porcelain plate from Kayakkuli/Kal-Aru estuary right bank shell midden

Full size image

Excavations: Stratigraphy and Chronology

Keeri Beach of Mannar Island (KB)

The pearl oyster middens at Keeri Beach are covered by dune sands that contain artifacts such as ceramics and porcelain. The site was selected for excavation because it is threatened by development. A 150-cm deep test pit was dug 181 cm above mean sea level (AMSL), revealing a thick oyster shell deposit at 138–105 cm AMSL in layer C4 (Fig. 6). The local GRW and other plainware ceramics were found in C4, along with five porcelain sherds. The porcelain is European and dates C4 to the post-seventeenth century. Porcelain was found only in Keeri Beach excavations.

Fig. 6
figure 6

The Keeri Beach (KB) shell midden. (a) site distribution and the location of the test pit; (b) test pit profile; (c) elevation profile (through the blue line in the map) (Drawing by Nadeera Dissanayake)

Full size image

Arippu Middens at Doric Bungalow (DB)

A test pit was excavated in Arippu near the Doric Bungalow, which is the early nineteenth century seat of the British governor in the region. Excavations reached a depth of 37 cm (Fig. 7). Middens deposited on the cliff formation of 430 cm AMSL, covered in densely compacted dark brownish clayey sand. The average pH value was moderately alkaline at 8.4. Highly compacted pearl oyster midden deposited at layer C2, 416 –402 cm AMSL. AMS radiocarbon dating of a P. fucata shell from the midden dates to 860 +/- 30 B.P. calibrated to 1424–1673 cal CE at a 95.4% probability range (Beta-611,655) (Fig. 9). The sample for dating was retrieved from 411 cm AMSL. Excavations revealed plainware that looked like GRW, but no rims were found to diagnose or date the ware more accurately. The absence of porcelain and the radiocarbon date may indicate a pre- or early-colonial (Portuguese period) date for this midden, although this dating needs to be handled with caution.

Fig. 7
figure 7

The Arippu (DB) shell midden. (a) site distribution and the location of the test pit; (b) test pit profile; (c) profiles of shell midden from exposed cliff; (d) elevation profile (through the green line in the map) (Drawing by Nadeera Dissanayake)

Full size image

Kondachchi Middens (KD1)

The test pit conducted at Kondachchi shell middens, located within eroded coastal dunes, extends to 85 cm below the surface and reveals five distinct levels based on sand coloration. The average pH value is 8.5. The third level (L3 = 238 –222 cm AMSL), in pale brown angular fine sand, showed a thick pearl oyster shell deposit (Fig. 8a, c). AMS radiocarbon dating of a P. fucata shell from L3 provided a date of 850 +/- 30 BP calibrated between 1430 and 1680 cal CE at a 95.4% probability range (Beta-611,656) (Fig. 9). The sample for dating was retrieved from 229 cm above mean sea level. L3 provided ceramics of plainware as well, with rims. A 17th century date can be cautiously considered plausible for the formation of this level.

Fig. 8
figure 8

The Kondachchi shell midden (KD1) and settlement (KD2) (a) site distribution and the location of the test pits; (b) settlement test pit profile; (c) shell midden test pit profile; (d) elevation profile (through the green line in the map) (Drawing by Nadeera Dissanayake)

Full size image
Fig. 9
figure 9

Calibrated probability distribution of radiocarbon dates from Doric Bungalow shell midden (DB), Kondachchi shell midden (KD1) and Kondachchi settlement (KD 2), dates obtained by AMS (by Thilanka M. Siriwardana)

Full size image

Kondachchi Settlement (KD2)

To study the link between pearl shell middens and settlements in Kondachchi, a test pit was dug just 330 m north of KD1 in a site dubbed KD2 (see Fig. 5). The pit dug into 105 cm revealed five levels from the preserved surface (265 cm + msl) (Fig. 8a, b). The pH values of levels 1 to 5 range from 6.4 to 6.8 (slightly acidic non-saline). Excavations uncovered 5,101 ceramic fragments, including rims, four beads, corroded metal fragments, highly weathered shells (NISP = 10), and coral fragments (NISP = 298). The shell assemblage consisted of Turbinella pyrum (columella only), Chicoreus virgineus, Volegalea cochlidium, and cowrie fragments. Although charcoal indicated the presence of hearths, the absence of architecture shows that any structures would have been like the present-day Cadjan huts made of perishable coconut leaf mats.

A charcoal fragment from Level 3 (L3) at 2.25 m AMSL provided a date 160 +/- 30 BP (95.4% probability), calibrated between 1719 and 1786 cal CE, but at 31.6% probability (Beta-611,657). Another charcoal fragment, this time from Level 4 (L4) at 1.6 cm AMSL provided an AMS date of 1130 +/- 30 BP (at 95.4% probability), calibrated to 875–994 cal CE (with 89.3% probability) Beta-611,658) (Fig. 9).

The excavation of the KD2 settlement has produced 79 rims and 5,022 shard fragments (> 2 mm) weighing 11.5 kg, in great contrast to midden excavations, which have yielded a much smaller assemblage of 120 sherds weighing only 0.3 kg.

It is important to note that various plain-ware vessels have been reported from KD2 settlement, the middens KB, DB, KD1, and Kal-Aru, including Type A cooking pots (see Fig. 3a) and Type F storage vessels. At the middens and settlements, the GRW of these types (see Fig. 3b–e, g) is found, except for the lower levels of KD 2. Instead, at lower levels of KD 2, type H coarse redware shallow conical dishes dominate (see Fig. 3h, i). These dishes have round and straight rims and are multipurpose vessels. During the excavation, 22 rims of this vessel were found. Two have horizontal grooved lines, a wavy or a lattice pattern inside the vessel, and a connection with the rice husking /destoning purposes. The modern local vernacular names of these types are nambiliya and Koraha. Additionally, only middens contained wide orifice cooking vessels (Type B) (see Fig. 3d, e) and thick-rim storage vessels (Type C) (see Fig. 3f). Type B is found in plain ware and GRW forms in such contexts. Small water jars or vessels (Type D) were mostly found in the middens, with only two sherds of this type being found at level 2 of KD 2. On the other hand, we found no large storage vessels with thick walls (Type E), begging bowls (Type G), and lids (Type K) at any of the study sites (type forms after Schenk 2001; GRW form after Ragupathy 1987).

From the middens located on Keeri Beach in Mannar Island, the southern shores of Arippu, Silavatura, Kayakkuli-Kal Aru, and Marichchukaddi were found to contain GRW and colonial-era porcelains more prominently. Among these, a British porcelain plate depicting two steamships sailing in the open sea with a seal reading Colombo Crystal Palace is found at Kal-Aru right bank shell middens (see Fig. 5).

Table 1 Summary of faunal remains found from the three midden test pits (KB – Keeri Beach, DB – Doric Bungalow Arippu, KD-Kondachchi)
Full size table

Taxonomic Abundance, Diversity, and Richness

Excavations produced 12,110 faunal specimens, with molluscs making up 98% of the total (Table 1). There were 81 species of molluscs from 51 families, with P. fucata being the most frequent at 81%. The remaining 2% of the fauna comprised corals, nullipore balls, and polychaetes such as keelworms. The corals found were Turbinaria, Madrepore, Porites, and Pavona; some were attached to oyster shells. Other than six terrestrial gastropods, 75 aquatic molluscs were identified, including 34 bivalves, 39 gastropods, and two scaphopods (For complete data see Online Resource 4: MNIs, habitat ranges, and trophic guilds).

Table 2 displays Shannon’s Diversity Index (H’) and Margalef Richness Index (d1) of malacofauna, along with NTAXA and MNI values at each excavation unit. It is important to note that the MNI of P. fucata directly affects the results, with H’ being less than 2.2 in all levels. The average index values of the three excavations suggest low diversity (< 1), except for moderate H’ values (> 1.5) at levels C2 and C5 of site KB.

Table 2 Species diversity and richness in excavation units
Full size table

At the lowest level (C5) of the KB middens, diversity is moderate (H’=2.24), and species richness is high (d1 = 4.33). The assemblage is characterized by species inhabiting sandy beaches at low tide, including Umbonium vestiarium, Donax incarnates, and shells of the Cerithiidae and Potamididae families. The large amount of pearl oysters at C 4 decreases diversity, but species richness is the highest observed among the nine assemblages (d1 = 5.21).

The malacofaunal assemblage from the DB is less diverse but has a high richness (d1 = 4.09). The KD midden contexts show significantly lower levels of diversity and richness.

Variation in species diversity between the test pits is notable. Figure 10 shows the richness (d1) distribution for different families across the three test pits. Donacidae, Veneridae, Neritidae, Potamididae, and Cerithiidae are restricted to KB, leading to overall higher species richness (d1 = > 2.3) than other midden contexts. Species such as Bistolida hirundo, Crenocrassatella rostrate, Anadara spp., Spondylus spinosus, Corbicula solida, Babylonia zeylanica, and Vexillum amandam are only reported from the KD midden. Magallana bilineata, Pyrene flava, and Nassarius celebensis are known only from DB. However, the NISP of these species from KD and DB are very low (n = < 2).

Fig. 10
figure 10

Mollusc families with NISP > 5 in each excavated midden excluding P. fucata (KB – Keeri Beach, Mannar; DB – Doric Bungalow, Arippu; KD – Kondachchi) (drawing by Thilanka M. Siriwardana)

Full size image

The recovered malacofauna findings indicate that they consist of three types of feeding behaviors: suspension feeders, detritivores/grazers, and predators, with a ratio of NTAXA 25:19:18. Most of the species are suspension feeders and predators, making up 70% of the total. Also, 47% of the aquatic gastropod species are predators.

The oyster shells retrieved show damages by boring sponge species such as Pione margaritiferae as many as 18% (n = 422) of KB, 15% (n = 72) of DB, and 13% (n = 93) of KD. Fouling in the shells suggests the presence of polychaetes.

In addition to the aquatic species, six terrestrial species were found in the midden assemblages. All six were reported from DB, while Keeri Beach dune contexts contained Ariophanta exilis, Pseudotrachia vittata, and Rachis punctata in minimal amounts (NISP = < 3). Kondachchi contained only A. exilis.

Table 3 The morphometric data in non-repetitive elements (NRE) of P. fucata in each excavation unit of case study sites
Full size table

Shell Fragmentation and Morphometry

Although preservation conditions are good due to high pH values, we can see increased fragmentation of the ventral aspect of the shells due to weathering and root action, reducing the availability of complete shells for size measurements.

Our study revealed a significant difference between the abundance of unique elements (NRE) and the percentage of NISP within each excavation layer in many units, except KB/C4-5. The NRE was lower than the NISP percentage, indicating a limited occurrence of distinct elements in the collection and a high level of shell fragmentation. Table 3 summarizes the variation of morphometric attributes of shells, such as preserved hinge lengths (HLs) and heel depths (HDs), within each excavation unit. This variability has implications for studying the range of shell sizes, as it can significantly impact the assessment of shell size distributions. In the samples, the number of NREs of P. fucata was around one-third (n = 3,518) of the NISP (n = 9,838). Due to fragmentation, the number of hinges that aid in diagnosing the left and right valves varied from the reported NRE. Notably, the upper levels of the excavation sites displayed an increased fragmentation of shells.

We found enough shells in the KB/C4 excavation unit to study the relationship between shell size and preservation level. In this unit, the oyster shells had a range of HL from 9 – 63 mm and HD from 0.7 – 16.6 mm. We identified 2,198 hinges out of the 2,846 oyster shells found (77.3%), and most of these hinges (97.5%) were identified as NRE. This allowed us to assess the grouped multivariate frequency of HD, HL, and DVM percentages of measurable shells. Figure 11 shows the frequencies of these shell measurements in KB/C4 grouped into 1-mm segments.

Fig. 11
figure 11

Variability in shell morphometrics: grouped multivariate frequency analysis by heel depth (HD), hinge length (HL), and dorsoventral measurement (DVM) in KB/C4, showing potential features for size reconstructions. Classes in the x-axis are grouped using HD value size ranges in 1-mm intervals (Drawing by Thilanka M. Siriwardana)

Full size image

We can see the most preserved dimension was HD. However, measurements of ventral aspects and anterior and posterior auricles were poorly preserved, resulting in fewer DVM and HL measurements. There is a significant decrease in shell quantities for sizes below 4 mm and above 11 mm, deviating from the overall pattern. If only the NRE of smaller shells were lost due to fragmentation, we must expect the distribution to be left-skewed with a higher frequency of larger shells. The data show a normal distribution with a minor standard deviation (SD = 1.6) from the sample mean (x̄=7), suggesting that the observed distribution is unlikely to be solely influenced by shell fragmentation bias. The middens’ moderately alkaline non-saline pH values showed that the densely accumulated pearl oyster shells created an environment conducive to preserving even small shells. The KD settlement deposits exhibited slightly acidic non-saline pH values, correlated with scarce malacofaunal remains and significant shell deterioration, typical of tropical coastal environments. (Online Resources 2).

Size Frequencies of P. fucata

The HD data for P. fucata were analyzed across all test pit levels (Table 4). The data showed a symmetrical bell curve distribution pattern. The frequency of peaks in each layer was centered around the population mean value of 6.7 mm, with slight positive or negative deviations. The analysis revealed that KB/C4-5 displayed positive mean differences, indicating larger shell sizes, while DB and KD/L2-3 showed negative mean differences, suggesting smaller shell sizes. KB/C2-3 and KD/L1, with p-values greater than 0.05, on the other hand, did not show significant differences from the population mean of 6.7. The standard deviations ranged from 1.4 to 1.9, indicating closely clustered data points with relatively low variability. With these data, the pearl oyster shell HD values show a significant application of the morphometric unit to calculate the size distributions of the oysters. The cloud plot in Fig. 12 visually represents the distribution pattern. (Descriptive statistics given in Online Resource 5).

Fig. 12
figure 12

Raincloud plots of HD distribution and frequency in the contexts (c) of each three test pits. Statistical analysis based on Student’s t-test, dotted line α = 6.7 mm, KB- Keeri Beach, Mannar Island, DB- Doric Bungalow, Arippu, KD- Kondachchi (Drawing by Thilanka M. Siriwardana)

Full size image
Table 4 Descriptive statistics and the student’s two-way t-test results of P. fucata HD values by site and excavation unit
Full size table

Discussion

Chronological Patterns

The middens on Keeri Beach can be dated to the Portuguese and early Dutch periods. Historical accounts offer clues about the potential age and exploitation history of the KB middens. Records suggest that by the Dutch period (eighteenth century), the Anaivilundan and Nadukkuda Banks, associated with Keeri Beach fisheries, were abandoned due to sedimentation (Herdman 1903). Therefore, these could be Portuguese or early Dutch middens. After claiming governance of Mannar Island in 1560, the Portuguese initiated extensive pearl fisheries in 1560 CE, targeting the pearl banks from Sri Lankan to South Indian shores (Hornell 1922). This period of intensified exploitation may align with the possible formation timeframe of the KB middens. Also, a letter sent to the Dutch East India Company in 1658 mentions a fishery at Mannar (Pieris 1929). The KB middens may have formed during this new expedited pearl fishery sessions. According to British sources, there are no records of any British fishery activities on the island (Herdman 1904).

Arippu and Kondachchi have been known as the pearl fishing centers of Sri Lanka since at least the fourteenth century CE (Lee 1829; Pieters 1908; Queyroz 1930), so it is likely that the middens associated with the Doric Bungalow (built around 1804) contain layers that date back to the fourteenth century CE or earlier. The radiocarbon date from Arippu-DB provides a date in the fifteenth-seventeenth centuries. While it is unclear when pearl fishing at Kondachchi started, historical accounts describe it as a temporary pearl fishery camping area and an auxiliary station of Arippu (Cordiner 1807). It is possible to place the DB and KD middens between the last phase of Jaffna-controled fisheries since the early sixteenth century and the commencement of Portuguese fishing during the late decades of the sixteenth century.

The AMS dates from the lower levels of the coastal dune settlement KD2, suggesting that the early phase of the settlement dates to the ninth and tenth centuries CE. This dating is consistent with the sequence of ceramic forms, which means that the impact of any time-averaging effect in the dune matrix is limited. From these levels onwards, there is a significant abundance of conical dishes (Type H) with linear patterns inside. Type H is considered one of the key types postdating the seventh and eighth centuries CE in Sri Lanka (Kuna 1987; Schenk 2001). The AMS date from KD2 L3 indicates a post-seventeenth century date for forming the upper level, but it is also uncertain. However, it is plausible that these levels are related to extensive eighth century pearl fishing camping grounds known from Dutch and British historical records. According to the records, pearl extraction at KD2 must have stopped in 1801 (Cordiner 1807; Herdman 1903). Altogether, the ceramic, AMS, and historical evidence identify KD2 as a settlement that emerged during the late Anuradhapura period of Sri Lanka and existed through the European colonial period. However, it cannot be confirmed that it was continuously used.

The grooved rim ware (GRW) ceramic fragments have been discovered in KD2, along with middens. These are similar to those found at various Indian Ocean sites (Abraham 2002; Carswell 1976, fig. 13; Haour and Jaufar 2022, fig. 5.11; McKinnon 2011; Ragupathy 1987; Rajan 1994, fig. 17; Rougeulle et al. 2015, figs. 171, 172, 173, 174, 175, 176; Sinopoli and Morrison 2007, figs. 6.6, 6.10–6.11). These artifacts date back to the tenth century CE to recent times. The similarity in their decorative elements and shapes indicates that there has been increased activity in these Indian Ocean regions since that period. However, their systematic analysis is lacking, which hinders further comparisons. The discovery of porcelain artifacts in the midden suggests the significant influence of European colonialism and provides a time marker to identify the Dutch-British colonial middens.

Exploitation Patterns

The more extensive middens may indicate the primary processing areas or the lands frequently used to pile the shells. In contrast, thin deposits further away from water sources might suggest small-scale extraction. With the assistance of the pearl fishery records, we can identify these as the remnants of Kottus, the wooden-fenced enclosures along the shore, specifically designed for the storage and processing of pearl oyster harvests in large-scale or smaller enclosures similar to those above, known as toddis/ tottis, where oysters were processed in small scale (Dixon 1939; Heydt 1744; Kunz and Stevenson 1908).

Reliable morphometric features such as Heel Depth demonstrate that small and large pearl oysters rarely occur in the midden deposits. Fragmentation analysis and alkaline pH levels outrule the possibility that size distributions have been impacted by taphonomic processes. Normal HD distributions with minor standard deviations reflect selective harvesting. Unlike shellfish harvesting for consumption, where shellfish are usually foraged optimally (Waselkov 1987), pearl oyster harvesting selects sizes to ensure the gain from pearls. This is further supported by the data points from all excavations that were closely clustered, supporting a consistent size–age relationship in resource use over time. Thus, the recovered shells from the excavations potentially fall within the documented scientific age range of 3 ½ to 4 ½ for harvesting pearl oysters of GoM (Ferguson and Ferguson 1888; Hornell 1905; Sivalingam 1961).

The samples from Keeri Beach contain higher numbers of oysters larger than the population mean, with a mean HD of 7 mm. Shells in KB are significantly larger than those recovered from the other sites. The abundance of large oysters suggests that the industry was prolific from a pearl fisheries perspective. The potential of the banks exploited had not been exhausted yet. Either the harvesting mode was at a scale that allowed sufficient time for the oysters to keep growing, or harvesting was intensive but targeted a relatively pristine bank that had been spared human exploitation for an extended period.

The high species richness and abundance of corals within KB middens support the idea of a long-term undisturbed environment. Within the KB sample, unique juveniles at KB/C4 exhibit a distinct distribution pattern, indicating the challenge of avoiding their collection when nested among mature individuals, as acknowledged by Herdman (1903). The KB site, possibly representing an untouched pearl bank, is a tangible example within this early colonial historical context, showcasing the delicate balance required for sustainable pearl fisheries.

In contrast, the shells from the temporally overlapping Arippu and Kondachchi sites are smaller than average. Young oysters likely ended up in these middens because fisheries were conducted without pre-inspection for maturity, indicating that fisheries did not target large pearls only. Extraction strategies also included young oysters with seed oysters.

As stated in a 1697 Dutch governor’s account, the pearl fishery of Arippu and Kondachchi harvested the oysters, “upon which become in turn ‘fit to be taken up’ (Lee 1847) may refer to such younger oysters. These were fished twice yearly in February and October (Barbosa 1866; De Silva 2009). Therefore, middens can serve as a tangible manifestation of the adverse effects caused by a rise in the intensity and frequency of recurring disruptions.

To contextualize our findings within the fifteenth-seventeenth centuries CE, we need to explore the global pearl market of that time. Historical literature from the third century BCE highlights the significance of pearls from the Gulf of Mannar, particularly larger ones (Geiger 1929; McCrindle 1877; Pliny 1855). Pearls were sorted into classes, with larger ones highly appreciated (Hultzsch 1892; Kunz and Stevenson 1908; S. Lee 1829). Sri Lanka, meanwhile, specializes in finely round seed pearls, less than 2 mm in size (Herdman 1906). Beyond their market value, seed pearls symbolized plutocratic excess, used by kings and elites for offerings and personal adornment (Bertolacci 1983; Hultzsch 1892; Kunz and Stevenson 1908; Vamadevan 1995). Understanding their symbolic and economic importance is crucial for comprehending their multifaceted historical use in Asia (Arunachalam 1952; Krzemnicki et al. 2017; Kunz and Stevenson 1908).

Dynamics of the pearl trade underwent a transformative shift when, in 1498, Columbus’s discovery of abundant rough pearls (aljofar) on Cubagua marked the beginning of the second “pearl age” (Kunz and Stevenson 1908; Strack 2008; Warsh 2018). This discovery unleashed a massive flow of pearls to Europe, fundamentally altering the symbol of luxury in Renaissance art. The focus shifted from larger pearls to the lavish use of seed pearls in garments (Dirlam et al. 1985; Donkin 1998; Isman 2020; Kunz and Stevenson 1908). Seed pearls, now accessible to every social class, disrupted traditional hierarchies (Warsh 2018). Seeking these pearls, Portugal named Sri Lanka in the 1502 chart “Canteno Planisphere,” as a source of pearls, seed-pearls (aljofar), spices, and cinnamon, highlighting its significance in the global pearl trade (Pires 1944 [1515]). Sri Lankan kings were actively involved in producing seed pearls in Chilaw and procured more from South Indian kings who harvested the northern pearl banks of Sri Lanka. Portuguese ships scoured every Indian Ocean port towards China for these coveted seed pearls (De Silva 2009; Pires 1944 [1515]).

Our review connects historical scenarios with archaeological findings, revealing intriguing patterns in the pearl trade’s impact on fisheries exploitation. The shift in preference from larger pearls to seed pearls in Renaissance Europe parallels our observations with younger oysters in the pearl middens of the fifteenth-seventeenth centuries CE. Historical instances, such as the depletion of Cubaguan oysters under Spanish exploitation, underscore the importance of understanding the consequences of intense exploitation (Kunz and Stevenson 1908; Romero et al. 1999; Warsh 2018). Similarly, evidence points to long-term abandonment in Sri Lanka since the 1640s due to Portuguese exploitation (Baldaeus 1672; G. Lee 1847; Valentyn 1726). There are also notable lapses in fisheries, such as the significant depletion of oysters in Arippu and Kondachchi during a pearl fishery event in 1732 (Lee 1847).

Extracting young oysters and not allowing for sufficient recovery time can lead to overexploitation and depletion of resources. Overfishing can harm oyster populations by inhibiting their natural growth and reproduction. Overexploitation contributes to the decline of the oyster population (Herdman 1903), with other disturbances explored below. The resilience demonstrated by Gulf of Mannar (GoM) pearl oysters under pressure suggests a complex interplay of factors. Different studies on dietary consumption of molluscs such as Conomurex fasciatus and Saxidomus gigantea have shown that different frequencies of stress can lead to changes in the size of mollusc shells, while also displaying resilience in their populations (Hausmann et al. 2021; Milner 2013; Toniello et al. 2019). Unexploited colonies of pearl oysters may have played a critical role in sustaining populations over time. They acted as a buffer, ensuring the long-term survival of pearl oysters and providing a sustainable resource for the industry (Herdman 1903). Regarding Indigenous knowledge, selectivity often correlates with consumption rate and population size. In such cases, minimal pressure is exerted on source shellfish populations (e.g., This can be seen in the historical behavior of pearl divers who were mindful of not overharvesting pearl oysters. They understood the need to avoid putting too much pressure on the resource (Pieters 1908). Connecting the dots between the potential resilience of Gulf of Mannar (GoM) pearl oysters and the specific case of Keeri Beach (KB) presents another compelling aspect.

While our study does not cover middens from the last fisheries, we can see a trend set in the fishery industry to be more sustainable (Hornell 1922). The policy of limiting the harvesting frequency was first proposed by the Dutch Governor Jan Schreuder (1757–62) (Lee 1847), yet implemented by the British only after the 1830s (Bertolacci 1817; Herdman 1904; Hornell 1924). The establishment of the Pearl Fishery Superintendent resulted in descriptive records on each fishery. This intricate web of historical and archaeological evidence offers a rich tapestry for understanding the dynamics of pearl fisheries and their long-term ecological impact.

Another way to understand the exploitation patterns of shell middens is by examining the accumulation rates. This measures the size or weight of shell deposits at specific times or between two points in time (Robson et al. 2023). Regarding edible shell middens, accumulation rates are influenced by group size, length of occupation, and frequency of visits (Jerardino 2013; Rick 2023; Robson et al. 2023). These notions are also relevant to pearl shell middens. Specifically, each fishing season involves a varying number of pearl divers and a varying number of days, depending on the number of oysters and banks available for harvesting each year (Donkin 1998; Herdman 1906; Hornell 1922). As we have observed, the fragmentation of shells and the displacement of piles indicate that a comprehensive examination of the cultural, natural, and taphonomic mechanisms associated with pearl oyster middens is necessary. These developments can aid in answering more precise queries.

The findings in DB are significant as they indicate the deliberate movement of shells hundreds of meters from the shore. Considering the historical shoreline of Arippu, it should have been further seaward than it is today (Herdman 1903, fig. 20). The inland boundary of the midden distribution, about 400 m, indicates that the shells or post-processing debris have piled inland. Determining the accumulation rates within and between different sites can be challenging due to various factors, such as enclosed spaces near the coast and the piling up of shells inland. Post-depositional disturbances, such as coastal erosion, dune formation, and shell fragmentation, can further complicate differentiation. The minimum depositional gap is six months, and it can vary for centuries depending on the frequency of camping in the same place for fisheries. However, none of the studied sites have shown varying accumulation levels separated by macro-level sand/soil deposition. Although not considered in the present study, studying these rates can provide valuable information and lead to new questions and interpretations, including insights into how the oysters were exploited. Research on prehistoric shell middens has shown a connection between accumulation rates and the spatial distribution of the middens (Hausmann et al. 2019; Zangrando et al. 2021).

Ignoring these oysters’ nutritional value would negate the impact of the exploitation pattern. Pearl oysters are edible (Zhang et al. 2022) but are not locally consumed during fisheries due to the inability to extract seed pearls until the oyster has decayed. Freshly opened oysters are sometimes sun-dried and eaten by the poorer classes (Cordiner 1807), but this is rarely reported. There is no evidence of consuming pearl oysters or other bycatch species in these middens. Shells in middens are sometimes used for lime manufacturing in regional constructions, which can be considered a form of exploitation. The extraction process leaves visible remnants in major constructions such as the Doric Bungalow, Periyacarisal Dutch Church, and Mannar Fort.

Ecology and Diversity

The pearl banks’ historical ecology revolves around pearl oysters (Bertolacci 1983; Gollenesse 1973), with other organisms such as coral, sponges, other molluscs, foraminifera, worms, starfishes, and holothurians also present (Hornell 1922; Spittel 1969). Though these organisms surfaced with pearl oysters created a curiosity as a representation of the ecosystem of each bank (e.g., Bennett 1843), only the studies since the mid-nineteenth century have shown that the involvement of minor species in pearl production correlates with an increased occurrence of pearls (Fernando 2021; Herdman 1903; Hornell 1922; Kelaart 1857). This highlights the importance of understanding the diverse ecological components that contribute to successful pearl formation. The species diversity and richness revealed in the present study predates these known early modern records.

The molluscan taxonomic diversity is moderate to low in all excavated middens. Species richness varies among middens; Keeri Beach has the highest, and others are low. Studies show that the diversity (H’) and richness (d1) of reefs are low near the surface and increase with depth (Huston 1985). The diversity pattern globally shows an increase from the surface to a depth of 8–12 m and then remains relatively constant until 30 m, after which it decreases again (Hutson, 1985). The diversity and richness of the midden record may be correlated to the depths from which divers extracted the oysters. Keeri Beach fisheries are associated with the Anaivilundan and Nadukkuda banks at 7–11 m, while the Arippu-Kondachhi banks at 7–9 m (Herdman 1903). Since species richness and diversity should be fairly high at these depths under “normal” natural conditions, the low levels represented in the middens require further consideration.

The low species diversity observed in the pearl oyster middens could be partially explained by considering the influence of natural disturbances on these marine ecosystems. The intermediate disturbance hypothesis proposes that species diversity is highest at intermediate disturbance intensity and frequency levels. Severe disturbances prevent the biome from recovering, which leads to lower diversity (Lenz et al. 2004; Marull et al. 2015; Moi et al. 2020). The organisms displaying resilience over thresholds in multiple stable states may persist (Petraitis and Hoffman 2010). Pearl banks facing frequent intense abiotic or biotic disturbances contributed to the formation of middens, such as light penetration due to turbidity, water depth, excessive sedimentation, temperature, wave energy, food availability, and predation control species diversity (Herdman 1903; Huston 1985). These factors likely played a role in shaping the ecological makeup of the pearl banks that contributed to the middens we are studying.

In the reported colonial past, severe storms and wave actions damaged pearl oyster colonies in the pearl banks near the surface (Herdman 1903). Sedimentation affects reefs and can act as a disturbance or regulator to the development of fauna. The pearl banks mentioned above that contribute to midden formation have been destroyed by the early nineteenth century CE due to sedimentation, preventing oysters from growing into large colonies ((Bertolacci 1983; Herdman 1903; Hornell 1922). Sedimentation and decrease of keystone species, like pearl oysters, negatively impact the biotic system. With these, we can infer a generalized interaction between disturbance frequency/magnitude and rate of dominance achievement with species diversity patterns. Biotic disturbances as predatory-competing organisms had been noted to negatively impact oyster colonies’ growth (Herdman 1903). However, such biotic disturbances are challenging to detect naturally (Huston 1985), yet not negligible. Based on available information, the impact of biotic or abiotic disturbance cannot be singularized, yet it represents an impact upon overall low species diversity in the midden.

It is possible that the low diversity of mollusc species found in the middens could be due to cultural impacts as well, such as deliberate removal of by-catch from the collected oysters or industrial exploitation repetitively bringing intense and frequent disturbance to the banks. Examining the early twentieth-century list of mollusc species from pearl banks, which is incomplete as it only includes handpicked larger species, we can see the British-administered banks showed higher species diversity (Herdman 1906; Hornel 1905). There is a significant difference in the dominating taxa between Herdman’s list and the middens examined. Certain species are missing, which is a matter of concern. Divers’ oyster baskets are usually filled with unwanted debris, too (Cordiner 1807), and the species diversity within these accumulations must be broad and should not be missing from the archaeological record. The large species might have been thrown back to the sea during transportation to save space and energy. Secondary processing sorted and removed lesser species, but the debris may have piled separately and should not be ignored.

With these two possibilities, the by-catch in middens may be limited to smaller species encrusted on oyster shells, a notable attribute of the minor species in the middens. Another possibility is the dietary-economic use of by-catch such as Turbinella pyrum (Siriwardana 2012) during the fishery. However, extracting edible portions of such large species on board was unlikely, despite the possibility of gathering their shells to be sold. We can infer such practice did not occur during the oyster fishery process, which was conducted under high surveillance to avoid stealing, so no extractive measures were allowed in between (Ferguson and Ferguson 1888; Sessional Papers of Government of Ceylon 1879; Vane 1887).

The impact of the culturally induced removal of species during the post-processing phase can be mitigated by examining the tropic guild of the remaining species. For example, a relatively higher percentage of suspension feeders and predatory gastropods can be seen from the KB. Similar trophic guilds have been observed in live tropical habitats (Steger et al. 2017), suggesting that the ecological conditions represented by the KB middens may reflect the nature of the source pearl oyster bank. Therefore, the species diversity-richness of the middens can still provide us an insight into the paleoecology of the pearl banks.

We should not forget the published ecological diversity of the pearl banks and the coexisting malacofauna found in the archaeological middens, representing time with a gap of at least three centuries. We can see predatory gastropods of the Fasciolariidae family may have generated more pressure on oysters in these archaic banks, where we do not have traces of small carnivorous gastropods like Purpura, Nassa, and Sistrum reported much later (e.g., Hornell 1922). However, the midden’s missing apex predators and other pests or competitors provide only a partial picture of the ecosystem. If we can examine these species’ coexistence in microhabitat conditions through midden, we can unravel the ecological intricacies that shaped the marine environment in the past. Understanding the interplay between pearl oysters and malacofauna contributes to a nuanced portrayal of the ancient ecosystem and the ecological roles played by each component.

It is worth noting that the small molluscs, including predatory modulus, brood oysters, and gastropods, are not found in large numbers in the middens assemblage. This could be due to the reduced intensity of other fisheries on the pearl banks. To prevent illegal fishing, fishing activities have been prohibited in the Pearl Banks region, which may have contributed to an increase in the population of other fish such as sponge-eating fish (Holacanthus spp), Gymbodonts, Vellamin (Lethrinus spp.), and triggerfish. These fish species play an essential role in controlling the competing organisms in the ecosystem. (Hornell 1922). We must find the place of such patterns as they result from deliberate avoidance of intervention.

While definitively quantifying human impact on past pearl bank ecosystems remains elusive based on these by-catch fauna, the study of archaeological middens provides a valuable window into the ecological dynamics of these historically significant marine environments. This knowledge enhances our understanding of the past and informs future conservation and management strategies for pearl oyster ecosystems.

Conclusions

This study explored the archaeology of oyster middens to investigate the pearl fishery industry in the Gulf of Mannar. The middens primarily contain waste generated by exclusively focusing on a single species of P. fucata. P. fucata and associated taxa recovered from the middens provide information on the status of the marine habitats exploited by the pearl fisheries before and during intensive colonial pearl extraction. Although further research is required to distinguish the impact of colonial activities on the taxonomic richness and diversity of the oyster banks from the effects of natural disturbances such as storms, patterns in oyster size distributions show how continuous intensive extraction disturbed oyster populations.

The present study provides a significant starting point for a historical, archaeological research agenda using pearl oyster shell middens as a tangible source to investigate long-term trends in coastal resource extraction by colonial powers in Sri Lanka. Future research will correlate the stratigraphy of the middens more firmly to the historical sequence of extraction and investigate the temporal and spatial relations between the settlements of the pearl divers and industrial middens more thoroughly. Combining archaeological and ethnoarchaeological studies can further explore the nature of historical and earlier pearl fisheries and socialecological resilience. The highly endangered natural-cultural landscape of pearl oyster middens around the GoM contains solid evidence that can shed light on larger networked socio-ecological systems in the Indian Ocean and beyond.