4. Discussion

Stable isotope data demonstrate a systematic difference in food-crust δ13C between coastal and inland sites in the Ertebølle and Funnel-Beaker periods, which is consistent with the δ13C differences between marine and freshwater fish in this region. Analyses of absorbed lipids in pottery and food-crust samples from the coastal sites confirm that marine mammals or fish were regularly cooked in Ertebølle pots (e.g. Craig et al. 2011) and that the practice was maintained in the early Neolithic Funnel-Beaker culture. There are no lipid analyses from inland Ertebølle sites in Schleswig Holstein, and the stratigraphic association between pottery and other evidence of the exploitation of freshwater resources (e.g. fish-bones, fishing fences) at these sites is unsatisfactory. Moreover, we have no human remains for stable isotope analysis. The scale of freshwater reservoir effects in these rivers thus means that it is possible for food-crust dates to be highly misleading.

The isotopic data from food crusts are therefore important evidence of the prevalence of fish consumption at inland sites in the late Ertebølle period. It is only possible to compare Ertebølle food crusts to Funnel-Beaker food crusts, as we have no data for foods eaten raw or cooked without the use of pottery. While some of the Funnel-Beaker food crusts appear to be 100% terrestrial in origin, most or even all of the Ertebølle food crusts have an aquatic signature.

Note 6: In addition to the observation above that most Funnel-Beaker sherds at Bebensee were typologically late EN I, there are several unpublished 14C ages from terrestrial samples c. 4700–4600 BP (i.e. second half of the 4th millennium).

The radiocarbon ages of the food crusts are consistent with this interpretation, when we consider the scale of freshwater reservoir effects in modern samples from the Trave and Alster (Philippsen and Heinemeier 2013). Food crusts on Funnel-Beaker sherds at Bebensee LA 76, which are unlikely to be older than the mid-4th millennium cal BC, gave 14C ages of up to 6100 BP, meaning reservoir effects in some cases of more than 1000 14C years [Note 6]. Applying similar corrections to the 'oldest' 14C ages for Ertebølle food crusts at Schlamersdorf and Kayhude would imply that pottery did not appear in the Trave and Alster valleys until c. 5700 BP (c. 4500 cal BC). At several coastal sites in East Holstein there is Ertebølle pottery that apparently dates to the second half of the 5th millennium cal BC, and not earlier (Hartz 2011); thus the apparently 'older' dates for pottery from inland sites may simply reflect the fact that reservoir effects in freshwater fish are much greater than those in marine fish and mammals.