The shells of Trinil

I want to share a paper that might not get a lot of attention but that I think makes an interesting contribution to understanding the ecology of early Homo after its initial dispersal from Africa. The paper is by José Joordens and colleagues, in the early bin at Journal of Human Evolution, titled, “Relevance of aquatic environments for hominins: a case study from Trinil (Java, Indonesia)”.

The authors plowed through the old collections of fossil fauna from Trinil, originally from Dubois’ excavation, looking to characterize the paleoenvironment in terms of hominin habitat preferences. The fauna are Early Pleistocene in age, possibly as old as 1.5 million years ago, but some uncertainty surrounds that date assessment, which is possibly under a million. What they found was some strong hints that the Trinil humans may have been eating shellfish and using other resources from the swampy lowlands in which the site was formed.

If aquatic resources such as molluscs and ?sh were available for hominins on Java, what is the probability that they indeed consumed these resources? In coastal areas, terrestrial predators often consume aquatic foods and have a considerable impact on the local aquatic ecosystem (Polis and Hurd, 1996; Roth, 2003). Systematic, often seasonal, predation by non-human terrestrial mammals on freshwater and marine fauna occurs widely. Carlton and Hodder (2003) reviewed occurrence of terrestrial mammals as predators in marine intertidal communities and documented 121 records of intertidal predation among 38 species of terrestrial mammals. For instance, mice, rats, pigs, chacma baboons, brown bears, black bears, striped and spotted hyenas, coyotes, domestic dogs, grey and red wolves, jackals, and foxes in coastal habitats catch and consume crabs, molluscs, ?sh, and other aquatic fauna (Carlton and Hodder, 2003; Smith and Partridge, 2004).

Terrestrial predators and omnivores eat fish, crabs, crayfish, turtles and shellfish when they get the chance. But the long list here muddies the water, so to speak. We often find hominins in lacustrine and riverine contexts, both because they inhabited those places and because of preservation biases. Those environments often yield evidence of consumption of aquatic organisms, especially shellfish but also fish, crocodiles and aquatic mammals. Somebody ate those animals, and the list above gives a bunch of suspects that aren’t hominins.

So maybe we should redirect our null hypothesis – instead of demanding proof of every instance of Early Pleistocene exploitation of aquatic foods, we should assume they ate the foods available to them. But with 30 or more species noshing on clams, crabs or fish at the shoreline, it’s going to be tough to diagnose cases where humans may have been involved.

Well, anyway, what would this reorientation mean for our understanding of the behavioral breadth of early Homo?

The literature cited above shows that systematic aquatic exploitation (either year-round or seasonal) by terrestrial mammals is normal and predictable mammalian behavior when the mammal is 1) omnivorous, 2) living in a coastal marine or freshwater habitat, 3) where nutritious and catchable aquatic prey is available. Considered in the perspective of aquatic exploitation by terrestrial mammals in coastal habitats, the systematic and seasonal aquatic exploitation by Homo sapiens (Marean et al., 2007) and H. neanderthalensis (Stringer et al., 2008) does not differ from that of other mammals. Also, transport of aquatic prey to a base (such as a cave, in the case of H. sapiens and H. neanderthalensis) is not modern behavior. For example, Navarrete and Castilla (1993) reported that Norway rat burrows in coastal Chile contain remains of w40 intertidal prey species such as limpets, bivalve molluscs, crabs, and ?sh. Erlandson and Moss (2001) provide many more examples of terrestrial omnivorous animals transporting aquatic food (remains) to dens, nests, burrows, and caves on land. A label of modernity, if applicable at all to aquatic exploitation, should perhaps be reserved for aquatic exploitation with evidence of advanced technology such as ?sh hooks and boats. The assumption, that early hominins living in a coastal habitat with catchable nutritious aquatic fauna were restricted to eating terrestrial resources, does not agree with published accounts of common mammalian behavior. Therefore, instead of having to provide evidence of aquatic exploitation before it is considered as a realistic option, we propose that the default assumption in hominin evolutionary research should be that omnivorous hominins who lived in coastal habitats with catchable aquatic fauna could have consumed aquatic resources (Joordens et al. 2009).

I like this point a lot: It is a bad sign when archaeologists use a definition of behavioral modernity includes rats but not Neandertals.

Joordens and colleagues suggest that the Trinil faunal collections may already contain evidence of waste heaps (they say, “midden-like” accumulations) of shells:

[T]he presence of a relatively large number of only adult, large-size Pseudodon shells, excavated from a very limited area (Hauptknochenschicht in Trinil), in both the Dubois and Bandung collections, is a discrepancy in the aquatic assemblage that merits further attention for these shells.

But there aren’t literally heaps of shells in the records; they have the shells and can only infer their original locations within the excavation at a relatively course grain. So the persuasive parts of the assemblage require us to see through the differnet biases that might have affected the collection. After dismissing a number of possible objections, they continue:

The fact that many of the Pseudodon valves are still paired and well-preserved would suggest that the molluscs were not dead and transported by water before fossilization but were buried in live position. However, the complete absence of small, juvenile shells as well as the mixed occurrence of two different (but equally large-sized) shell forms argues against interpretation of burial of a live population (Van Benthem Jutting, 1937). Instead, the discrepancies suggest that the Pseudodon shells could have been brought together, prior to fossilization, by a size-selective collecting agent who may have used them for consumption of molluscan flesh (Joordens et al. 2009: 13).

They found a similar pattern for another species:

The Elongaria assemblage from Trinil, just like Pseudodon, appears to indicate collection by a selective agent for the purpose of mollusc consumption. The Pseudodon and Elongaria assemblages from Trinil have the characteristics of shell middens (e.g., Waselkov, 1987; Rosendahl et al., 2007): large adult shells only, many complete shells, no signs of damage due to water rolling, signs of damage due to being deliberately opened, presence of human (hominin) bones in the same layer. We conclude that they represent a subtle clue of possible aquatic predation by non-hominins or by hominins (ibid.).

This hypothesis may not be testable further, unless signs of deliberate modification are found on one or more shells. Joordens and colleagues write that such a study is “currently underway”. The only other thing to do is apply a higher standard of rigor to possible shellfish features in other Early Pleistocene contexts. Early Pleistocene surface collections dug by vertebrate paleontologists (as opposed to archaeologists) sometimes discard or leave fish bones unidentified, and it is not always clear whether a loose association of shells would be recognized as a possible hominin-accumulated feature.

The paper cites the work of Stewart (1994), who argued for fish consumption at Olduvai Gorge. From that abstract:

Fish remains are associated with many early hominid sites, and five sites at Olduvai Gorge are examined here in detail. The patterns of fish exploitation seen in Late Pleistocene archaeological sites are manifested in three of the Olduvai Gorge sites, making a strong, although not absolute, case for early hominid fish procurement. The implications for early hominid behaviour of fish procurement are several, and include timing of the early hominid seasonal round to exploit spawning or stranded fish, and group size larger than the nuclear family unit, with greater social interaction.

These examples bring to mind the challenge of identifying chimpanzee nutcracking archaeologically. The chimpanzee pattern of behavior is barely systematic enough to pick out from the background. Yet archaeologists have devised some ways to find that slim signal, at least in contexts where they expect chimpanzees to have been active. Late in prehistory, some shell middens were vast and highly-recognizable. Those populations put together the elements of recurrent (or constant) occupation of a site and recurrent transport of shellfish to that site for processing. When we look further into the past, even Neandertals rarely put together those elements in a recognizable way. At the Italian sites where Mary Stiner showed Mousterian shellfish consumption, the presence of shellfish is just at a level where the signal can be picked out due to the lack of other credible transport agents for shellfish remains.

A hitch: Suppose we accept that early humans commonly exploited aquatic resources, even in the absence of specialized archaeological traces of such dietary sources. That does seem to create a problem for the interpretation of stable isotope ratios in fossil humans. I wrote about this issue regarding Neandertal diet (“Neandertals: gone fishin’ or not?”, see also “Shellfish use by Neandertals” and “Neandertal diet was not dolphin-safe”). It doesn’t take much fish to increase the nitrogen-15 composition of bone, making it hard to test hypotheses about the proportion of different terrestrial prey species, and even behavioral interpretations such as weaning age may be thrown off. The problem is too many independent dietary parameters to test with only one estimate. Ignoring the possibility of aquatic resource use helps simplify the interpretation, but that doesn’t necessarily make it better.


Joordens JCA, Wesselingh FP, de Vos J, Vonhof HB, Kroon D. 2009. Relevance of aquatic environments for hominins: a case study from Trinil (Java, Indonesia). J Hum Evol (in press) doi:10.1016/j.jhevol.2009.06.003

Stewart KM. 1994. Early hominid utilisation of ?sh resources and implications for seasonality and behaviour. J Hum Evol 27:229245.