Pelagic Fishing at 42,000 Years Before the Present and the Maritime Skills of Modern Humans

See allHide authors and affiliations

Science  25 Nov 2011:
Vol. 334, Issue 6059, pp. 1117-1121
DOI: 10.1126/science.1207703


By 50,000 years ago, it is clear that modern humans were capable of long-distance sea travel as they colonized Australia. However, evidence for advanced maritime skills, and for fishing in particular, is rare before the terminal Pleistocene/early Holocene. Here we report remains of a variety of pelagic and other fish species dating to 42,000 years before the present from Jerimalai shelter in East Timor, as well as the earliest definite evidence for fishhook manufacture in the world. Capturing pelagic fish such as tuna requires high levels of planning and complex maritime technology. The evidence implies that the inhabitants were fishing in the deep sea.

Although humans were able to travel hundreds of kilometers over the ocean by 50,000 years ago (ka), as required for the colonization of Australia, global evidence for fishing is rare before about 12 ka (1, 2). Middle Stone Age sites in southern Africa, such as Klaises River Mouth Main Cave, Pinnacle Point, and Ysterfontein 1, contain evidence of shellfish predation and the remains of marine mammals such as seals, but evidence of fishing before the Holocene is absent or exceptionally rare (1, 3, 4). Whether this reflects real behavioral changes or the loss of coastal archaeological sites as sea level rose is unknown. A record of early marine fishing is found at Blombos Cave dating between ~140 and 50 ka, but the fish represented are shallow-water species and would not have required boats or complex technology for their capture (5).

At Jerimalai shelter in East Timor, evidence exists for systematic pelagic fishing from 42 ka, showing the high level of maritime capacity possessed by early colonists moving into, and through, the islands of Wallacea (Fig. 1 and Table 1).

Fig. 1

Map showing Sunda, Wallacea, Sahul, and the location of sites.

Table 1

Accelerator mass spectrometry radiocarbon dates from Jerimalai shelter, East Timor. Dates were calibrated to 2σ with the program OxCal 4.1, using the Marine09 calibration curve, ΔR = 0. cal yr B.P., calendar years before the present.

View this table:

During Pleistocene low sea stands, the southernmost extension of the Eurasian continent comprised Sumatra, Java, Bali, Borneo, and the now-drowned shelf in between (Sunda). New Guinea, Australia, and the Aru islands were joined as a separate continent, Sahul. Between these two landmasses stretches the 1500-km-wide Wallacean Archipelago (Fig. 1), with more than 17,000 islands. It includes the Lesser Sunda islands as well as Sulawesi, Maluku (excluding the Aru Islands), and the Philippines east of Palawan. During the human time span, Sunda, Sahul, and the islands of Wallacea were not connected by a land bridge. The human colonization of Wallacea and Sahul must have been undertaken by boat, but there is little in the way of archaeological evidence to attest to the marine skills of the early colonists.

Along the northern coasts of some Wallacean islands and of mainland and island Papua New Guinea (Fig. 1), cave sites with evidence of Pleistocene coastal exploitation proximal to current shorelines are locally preserved. However, even in these regions, evidence for Pleistocene fishing is absent or sparse. The caves preserve a variety of marine shell species (6, 7), and a few have low quantities of fish and shark remains dating back to 40 ka (8), but the remains are not suggestive of developed maritime skills (9, 10). A more convincing case for systematic fishing comes from Kilu Cave in the northern Solomon Islands, dated to 32 ka. Kilu contains a larger quantity of fish bone, including some pelagic species, but overall fishing at Kilu seems to have been inshore (11).

At the eastern end of East Timor, Pleistocene raised coralline terraces run parallel to the present coastline and contain many caves and shelters (1215). At one such cave, Jerimalai (15) (Fig. 1 and figs. S1 and S2), excavation of two 1 × 1 m test pits (squares A and B) in 2005 produced a rich assemblage of cultural material, including well-preserved faunal remains, a large number of stone artefacts (n = 9752), bone points, fishhooks, and shell beads dating to the terminal Pleistocene (figs. S3 and S4) (15). Much like other assemblages associated with early modern humans in South and Southeast Asia, the lithic assemblage is characterized by faceted radial and rotated core technology and an informal scraper assemblage (figs. S6 to S8 and table S1) (1618). The assemblage is made from high-quality chert and is one of the only large and well-dated palaeolithic assemblages in Southeast Asia dating to the period of modern human colonization. Among the faunal remains, the number and weight of fish bones far exceeds those of other fauna, averaging 56% (range, 16.7 to 80.0%) by weight of the total vertebrate remains between spit 3 and spit 55. Bones from pelagic fish, such as tuna, compose almost 50% of the total fish assemblage in the earliest occupation levels (Fig. 2).

Fig. 2

Graph showing changing ratios of pelagic and inshore fish species through time (in percent based on MNI). cal BP, calendar years before the present.

Where fish are less dominant, the remains of marine turtles predominate. Unlike the marine fauna, the terrestrial faunal remains indicate opportunistic exploitation of the limited range of vertebrates found on Timor at the time. They comprise small quantities of murid rodents, bats, birds, and various terrestrial reptiles (mainly pythons, with small quantities of other snakes and monitors and smaller lizards) (15, 19).

We recovered a total of 38,687 fish bones from square B. 19,252 of these were identified at least to anatomical element. (Table 2). The total minimum number of individuals (MNI) of fish remains identified to family was 796, and the number of individual specimens (NISP) was 2822. Twenty-three taxa were identified, including 22 families and 1 species (Monotaxis granoculis). The MNI of Scombridae (tunas) is greatest in most of the spits, almost to the base of the site, accounting for approximately 16% of total MNI. Scaridae (parrotfish), Carangidae (trevallies), Balistidae (triggerfish), and Serranidae (groupers) follow in MNI and NISP. These species make up about 12 to 15% of the total MNI. Other major fish families identified are Lethrinidae (emperors), Lutjanidae (snappers), Acanthuridae (unicornfish), Labridae (wrasses), and Tetradontidae (puffers). For Elasmobranchi (rays and sharks), both are recognized, and some shark vertebra were identified as Carcharhinidae.

Table 2

Fish bone assemblage from square B Jerimalai shelter by anatomical element.

View this table:

At least 15 fish taxa were exploited during the earliest period of occupation, from 42 to 38 ka (phase I), although the quantity of fish bones from this phase is limited (Table 3). Although this may in part be a consequence of less intensive occupation of the site at this time, it is more likely due to the cemented nature of the deposit, which made recovering and extracting small bones problematic (see methods in the supporting online material). The prominent feature during this earliest occupation phase is the intensive focus on pelagic fish species, particularly Scombrids. The ratio of pelagic species (49%) and inshore species (51%) based on MNI is almost equal (Fig. 2). Sharks and rays were also exploited (Table 3). At present the site appears to have little evidence of occupation between ~38 and 24 ka. This may be due to sea level retreat rendering the shelter farther from the coast during this time; however, research at other nearby coastal sites such as Lene Hara Cave suggests that it more likely reflects limited sampling due to the small size of the excavation (20).

Table 3

MNI and NISP of identified fish taxa from the 1 × 1 m excavation, square B, Jerimalai shelter, East Timor.

View this table:

Between 17 and 9 ka (phase II), sea level rose, and by 9 ka Jerimalai would have almost been as close to the coast as it is today. In terms of offshore fish exploitation, in phase II the ratio of pelagic species in the total MNI remained high (46%), but the number of Carangids (trevallies) dramatically increased, whereas the exploitation of Scombrids (tunas) remained similar. There is also a marked change in the evidence of inshore fishing. Parrotfish were still intensively exploited, but the number of groupers and triggerfish increased.

The mid-Holocene assemblage (phases III and IV) is different. The most prominent change is the dramatic increase (66 to 76%) in the total MNI and lower ratio of pelagic fish species (34 to 24%) (Fig. 2). Pelagic species were still targeted, but inshore fish dominate the assemblage. The increase of inshore species may reflect the stabilization of sea level, warmer temperatures, and the establishment of reef habitats.

Because no direct evidence of fishing technology has been found in the earliest occupation period, it is uncertain how and by which method the occupants at Jerimalai captured pelagic and other species. Tuna can be caught in purse seines or leader nets (21). Simple fish-aggregating devices such as tethered log floats can be used to attract tuna (21, 22). The small size of the pelagic bones (individuals of ~50 to 60 cm in length) indicates that mostly immature individuals were taken, which may favor an interpretation of netting over line fishing as the capture method.

Zooarchaeological and ethnoarchaeological studies in Southeast Asia and Oceania suggest that the parrotfish and unicornfish would likely have been caught by netting or spearing, whereas trevallies, triggerfish, snappers, emperors, and groupers are most commonly captured by angling using a baited hook, but netting and trapping can be successful depending on species and body size (2328). A broken hook in square A, spit 29, is dated between ~23 and 16 ka (specimen lab codes Wk-17830 and Wk-19227) (Fig. 3.). This is the earliest definite evidence for fishhook manufacture in the world. Another hook fragment in square A, spit 26, is dated ~11 ka (Wk-18156) (fig. S5 and Table 1), as is one from nearby Lene Hara Cave (13). The dramatic increase of trevallies and groupers in phase II is consistent with the appearance of angling technology.

Fig. 3

A broken shell fishhook dated between ~23 and 16 ka from square A, spit 29, Jerimalai shelter, East Timor (scale is in millimeters).

All are Trochus shell single-piece baited hooks and do not seem suitable for pelagic fishing; however, it is possible that other types of hooks were developed at the same time. Bone points made on the spines of large fish first appear at Jerimalai in phase II. Their function is uncertain, but they clearly represent a component of a composite tool, such as fine barbs for fish spears, or part of complex hooks for trolling. The number of bone points increases dramatically in the mid- to late-Holocene levels (phases III and IV). No artefacts related to netting have yet been recovered, but the manufacture of strong fiber line is implied by the hooks, so knowledge of netting seems probable.

The finds from East Timor demonstrate the high level of maritime skills and technology possessed by the modern humans who colonized Wallacea. These skills would have made possible the occupation of the faunally depauperate islands of Wallacea and facilitated the early maritime colonization of Australia and Near Oceania.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S8

Table S1

References (2931)

References and Notes

  1. Acknowledgments: Australian Research Council Discovery grant DP0556210 funded this project. Australian Research Council Discovery grants DP0556210 and DPl10102864 to C.C. funded the lithics research.
View Abstract

Stay Connected to Science

Navigate This Article