Faire gras à Molène: dairy products and ruminant fats detected by lipid and isotopic analysis of pottery dating to the Final Neolithic-Early Bronze Age from the island site of Beg ar Loued (Molène, western Brittany, France)

.

Faire gras à Molène: dairy products and ruminant fats detected by lipid and isotopic analysis of pottery dating to the Final Neolithic-Early Bronze Age from the island site of Beg ar Loued (Molène, western Brittany, France)

INTRODUCTION
The subsistence strategies of early farming communities and consumption of dairy products in Europe have been highlighted since the beginning of the Neolithic period, thanks to numerous studies on lipid residues conducted in Southern, Central and Northern Europe (Craig et al. 2005a;Evershed et al. 2008;Cramp et al. 2014a, b;Roffet-Salque et al. 2015;Debono Spiteri et al. 2016;Cubas et al. 2020).Along the Atlantic coast, the importance of terrestrial animal products, such as ruminant meat and milk in vessels, have been reported, with an increase in dairy products apparent along a northerly latitudinal gradient (Cubas et al. 2020).Similar results are reported for the Bronze Age from sites across the Channel in England (Copley et al. 2003;Šoberl 2011).Despite a short distance to the sea, a general absence of evidence of aquatic products in vessels has been noted (Schulting 2005;Cubas et al. 2020;Evershed et al. 2022).
From the Final Neolithic and into the Bronze Age, evidence of subsistence strategies is still lacking.In Brittany, northwestern France, this period is marked by climatic fluctuations and landscape transformations (Stéphan et al. 2019b), changes in building methods and settlement patterns (reduction in the size of the settlement/population, Blanchet et al. 2018), the introduction of metal production, and the spread of new technical and decorative styles of ceramics (Salanova et al. 2011;Ripoche 2017Ripoche , 2022;;Blanchet et al. 2019;Nicolas et al. 2019;Favrel 2020).Due to the acidic pH of the region's sediments, faunal remains rarely survive, and as a result, subsistence strategies during this period are not well documented.Slaughter profiles at sites where faunal remains have been found show an increase in the breeding of ovicaprines and pigs alongside cattle, underlining the importance of dairying from the Final Neolithic to the Bronze Age in food production (Braguier 2000;Hanot & Tresset 2019).
The study of lipid residues in pottery has the potential to add deeper insights into past subsistence and culinary practices, as demonstrated by previous studies (Craig et al. 2005a, b;Evershed 2008;Cramp et al. 2014aCramp et al. , b, 2019;;Debono Spiteri et al. 2016;Roffet-Salque et al. 2016;Evershed et al. 2022).Most of the data on lipid residues in ceramic vessels in the region is available for inland sites (Cramp et al. 2014a;Cubas et al. 2020), while island contexts, which are unique due to their proximity to the sea, have received little attention.
The site of Beg ar Loued is one of the few EBA settlements on the coast of Brittany which shows exceptional preservation of faunal material (Fig. 1; Pailler et al. 2011;Pailler & Nicolas 2019a, b).It was excavated between 2008 and 2011 by some of us (YP and CN).The first occupation of the site corresponds to a shell deposit dating back to the Final Neolithic (FN: c. 3300-2800 cal.BCE).The most significant remains on the site date to the Early Bronze Age (EBA1: c. 2200-1950 cal.BCE; EBA2: c. 2000-1800 cal.BCE) and consist of two drystone domestic buildings, interpreted as houses.There is also a phase of site desertion (EBA3: c. 1665-1302 cal.BCE).In some areas of the site, this phase is difficult to separate with a layer of older colluvium from an upper area (FN: c. 3000 BCE).The chronology is based on the dating of 39 remains of bones, charcoal and charred residues on ceramic vessels (Pailler & Nicolas 2019a).The faunal remains and location of the site provide clear evidence that the relationship with the ocean was important for populations at Beg ar Loued (Pailler & Nicolas 2019a).Nevertheless, the ceramics, lithics and faunal remains indicate a strong link with the mainland.
This paper explores the cultural choices of the population for the exploitation of coastal/inland resources in pottery vessels, culinary practices, and surface treatment through lipid and isotopic analyses carried out on 129 potsherds.We focus our research on animal products used in ceramic vessels, specifically dairy products, and also investigate whether aquatic resources were processed in ceramics.In addition, we open up new questions about the use of organic products in surfaces/ coating layers for the finishing treatment of specific potsherds.
Beg ar Loued and its Landscape in the finaL neoLithic-earLy Bronze age Environmental studies of the coastline of Brittany have shown that the end of the Neolithic was concomitant with rapid environmental transformation, including major marine transgressions (Bond et al. 1999;Magny 2004; Joly & Visset 2005; David 2014), climate change (e.g.Bond Event 3; Bond et al. 1999) and cooling weather (Martín-Chivelet et al. 2011).Changes in land use and occupation are detected with Neolithic oak stands (Corylus L., Quercus L., Tilia L.,

MOTS CLÉS
Lipides, poterie, Néolithique final, âge du Bronze ancien, France, côte atlantique.particularly migratory species such as gilthead bream, sea bass, garfish, eel and conger eel, demonstrates that fishing was carried out throughout the year, perhaps more intensively during the summer, since the majority of the remains identified are of gilthead bream (Dréano 2019).Based on a study of the burns and traces on the fishbones, as well as their spatial distribution and deposition across the site, it has been suggested that a system of procurement and processing (preparation, drying, smoking and/or salting) of fish was carried out on site, perhaps to transport or trade the fish (Dréano 2019).
Additionally, no less than 21 species of shellfish have been found, including limpets, mussels, thick top shell, and in lower proportions carpet clams, abalone, and scallops (Dupont 2019; Mougne 2019).Around 1% of the limpets show charring marks.A variety of bird remains, both freshwater, shorebirds and marine, are also attested at the site (Hanot & Tresset 2019), suggesting that they could have contributed to the diet of the island's inhabitants.
Terrestrial mammals, such as cattle, pigs, sheep, and perhaps goats, were also found at the site (Pailler et al. 2014;Hanot & Tresset 2019).The presence of all age categories and a homogeneous anatomical distribution of the faunal remains suggests that at least ovicaprines were raised at the site for meat production, and also for secondary product use such as dairying (Hanot & Tresset 2019).The faunal remains Ulmus L.) replaced by various riparian taxa such as ash, elder, hazel, hawthorn and beech (Gaudin 2004;Martín-Chivelet et al. 2011;David 2014).The causes of these changes are still debated (Stéphan et al. 2019a, b).
Beg ar Loued is located on the island of Molène (Finistère) which belongs to the Molène archipelago that was attached to the mainland during the Last Glacial Maximum (Billard et al. 2020).With the gradual rise of sea level, the archipelago was transformed into a variety of smaller islands and islets by the early Neolithic; 4800 cal.BCE (for a better overview, see Dréano et al. 2013;Gandois et al. 2013;Stéphan et al. 2013;Pailler et al. 2014;Bernard et al. 2020;Billard et al. 2020).During the occupation of the two dry-stone buildings of Beg ar Loued, depending on the tide, the sea was between 0.1 and 2 km from the site (Pailler et al. 2014;Stéphan et al. 2019a).
Marine and terrestriaL food strategies at Beg ar Loued An abundance of faunal remains, both marine and terrestrial, have been discovered at Beg ar Loued (Pailler et al. 2004;Hanot & Tresset 2019;Pailler & Nicolas 2019a).Nearly 40 species of marine fish, particularly sea bass and gilthead bream, have been identified, as well as cuttlefish, crabs, and sea urchins (Pailler & Nicolas 2019a).The range of species, have been found on fine vessels, which is unique to the Beg ar Loued corpus, as the use of coating techniques is relatively unknown in the northern French Atlantic zone (Hamon 2008;Salanova et al. 2011).The red colour, reminiscent of the red-brick Campaniforme/Bell Beaker of Europe (Lemercier 2002;Salanova 2004;Favrel 2020), has a brilliant lustre and very fine porosity contrasting with the brown-black ceramics and coatings observed for Middle and Final Neolithic period ceramics (Fig. 2).The site also provides a good example of the dynamic between ceramics and lithics from the mainland versus the islands.While the lithics and pottery were produced locally, ceramic styles follow the same trends observed on the mainland.from Beg ar Loued provide some of the first evidence of animal husbandry in Atlantic island contexts during the Final Neolithic-Early Bronze Age (Hanot & Tresset 2019).Archaeobotanical remains of 11 species of plant taxa were also recovered from the site, which included hulled and naked barley, emmer, and wheat, as well as fava bean and pea (Pailler et al. 2014;Pailler & Nicolas 2019a).A crop weed, wild radish, is also present on the site, as well as three wild fruit trees (hazel, common hawthorn, and common dogwood), possibly linked to a gathering activity (Pailler et al. 2014).These remains suggest that coastal fishing was associated with the use of crops and livestock, thus linking prehistoric human groups to the mainland (Pailler et al. 2014).

ceraMic and Lithic reMains
Many lithic tools and ceramics have been discovered at Beg ar Loued (Pailler & Nicolas 2019a), although the ceramic assemblage is highly fragmentary, and many sherds do not have a precise chronological or typological classification.Despite its limitations, the ceramic corpus is important as it provides one of the few examples of ceramics from a settlement dating through the 3rd and 2nd millennium BCE in the region (Salanova 2019).Final Neolithic Conguel style, Bell-Beaker pottery (Salanova 2000) as well as cordoned urns and handled jugs, which are typical of the Early Bronze Age, have been identified in the assemblage.All the usual forms are present.However, glossy red coating surfaces (engobe)  We also sampled topsoil as a control for lipid residue analysis from a beach scarp north of the site in 2022, several years after excavations and the initial sampling of vessel fragments.
Combining the samples from the Regert & Mazuy 2019 study, the sample-set for lipid residue analysis includes 129 sherds and 30 visible carbonised residues adhering to the external ('RE'; n = 10/30) or the internal surfaces ('RI'; n = 20/30), which are most often found on body fragments from pots with medium thickness walls.Among the 129 sherds, some had abraded surface layers and discolorations, with a few examples of lustrous red coating on the exterior and white coating on the interior part of some sherds (e.g.MR6516, Fig. 2).Where feasible, the surface/coating layers were sampled separately for lipid analysis.Table 1 provides details of the sherds ('T'), visible charred residues ('RI'/'RE') and surface/ coating layers ('SI'/'SE') samples analysed in the study grouped by chronological period.
Despite the fragmentary nature of the ceramic assemblage of Beg ar Loued, several unique elements of the vessel form (rims, bases, decoration) allow a typological attribution and, more rarely, reconstructions of vessel profiles.Most of the pottery is attributed stylistically, as they have been rarely found in well-dated layers (Pailler & Nicolas 2019b).A large number of the vessel fragments (n = 37/129; 29%) came from a sandy colluvium unit, which includes Early Bronze Age objects in situ, and rolled sherds from a scree of an area uphill of the EBA houses and dated to the Final Neolithic, in secondary deposition.When the sherds from this layer are not typical of any cultural group, they are classified as 'Unspecified (FN/ EBA3)' as a precautionary measure.
Prior to lipid extraction, the external layer of each potsherd was removed, and when present, surface/coating layers were gently cleaned with a modelling drill in order to reduce the potential of contamination from the burial environment or during post-excavation (Stern et al. 2000).In total, 174 solvent  extractions were carried out on the 129 sherds collected (sherd, visible charred residues and surface/coating layers).
A selection of 24 solvent-extracted samples were prepared for compound-specific isotopic analysis (CSIA) of palmitic and stearic acid in the extracts to refine the characterisation of ruminant and dairy fats, and poorly preserved molecular assemblages (Dudd & Evershed 1998;Copley et al. 2003).
As the question about the use of aquatic resources at the coastal site of Beg ar Loued was important, we looked for specific evidence of these substances in the solvent-extracted compounds.But due to the challenges for detecting aquatic markers through solvent extraction alone (Garnier et al. 2018), we further conducted acidified methanol extractions on 31 samples (including 3 external surface/coating layers).These samples were selected on the basis of high lipid concentrations and the presence of long-chain fatty acids, phytanic acid or ketones.In summary, we prepared samples (DCM/MeOH; n = 174) for chromatographic (GC-FID, GC-MS) and Compound-Specific Isotopic Analysis (CSIA; n = 24).We also conducted acidified methanol extractions on selected samples (MeOH/ H 2 SO 4 ; n = 31) and analysed them via GC-MS in SIM mode.The methods for lipid extractions (Evershed et al. 1990;Correa-Ascencio & Evershed 2014), GC-MS analysis (Cramp & Evershed 2014;Garnier et al. 2018) and CSIA (Dudd & Evershed 1998;Copley et al. 2003) are based on previously published protocols.See Appendices 1-4 for full details of our analytical strategy, methods, control analysis and data.
In Table 2, we provide the details of lipid yields for each sample category: 'Individuals vessels'; 'T'; 'RI'; 'RE'; 'SI'; 'SE'.Comparisons of the lipid yields from the absorbed sherd residues and visible charred residues revealed that they were broadly similar (Wilcoxon Signed-Rank Test, Z = 2252.5,p = 0.16; Table 2; see also Appendix 5), and external charred residues had higher lipid yields than internal charred residues (Wilcoxon Signed-Rank Test, Z = 118.5,p = 0.4; Table 2; see also Appendix 6).This pattern is contrary to other studies, were lipid yields from visible charred surface residues are significantly higher than those from absorbed residues (Mukherjee et al. 2008;Courel et al. 2020a, b).Visible residues may have been exposed to various diagenesis processes during burial, while absorbed residues are protected from microbial attack by the inorganic ceramic matrix.This could explain the poor conservation of visible residues in this context.We also noticed that external visible residues and external coating layers have higher amounts of lipids than the internal visible residues and internal surfaces/coating layers (see Appendix 6).This could be related to the use of pottery or specific coating treatment.The comparison of the TLE and the percentage of TAGs preserved against the year the potsherd was excavated, or the chronological period of the pottery did not reveal any patterns (see Appendices 7-9).

MoLecuLar characterisation of soLvent-extracted saMpLes
The obtained lipid profiles of all the solvent-extracted samples in this study are dominated by even-chain free fatty acids (C 10:0 -C 30:0 ), half of them including long carboxyl chains (≥C 20:0 ).Other fatty acids are present in low amounts, such Prévost C. et al. as: 1) odd-chain fatty acids (C 11:0 -C 25:0 ) including samples with long carboxyl chains (≥C 17:0 ); and 2) even (C 14Br -C 20Br ) and odd (C 15Br -C 21Br ) branched-chain fatty acids with C 15Br and C 17Br being the most frequent.A third of the samples contain unsaturated fatty acids (C 18:1 , C 16:1 , and some with C 18:2 , C 20:1 , or C 23:1 ).Full details are provided in Appendix 4.
Cholesterol derivatives constitute the major part of the preserved sterols in comparison to plant sterols.Many sherds 'T' contained monoacylglycerols (MAGs), diacylglycerols (DAGs), and triacylglycerols (TAGs), with a third of them containing well-preserved TAGs.
Odd, asymmetrical long-chain ketones (K 29 -K 35 ), formed by condensation of fatty acids (C 16:0 , C 18: 0 and C 20:0 ), and indicating heating of the containers and their contents to high temperatures or for an extended period (Evershed et al. 1995;Raven et al. 1997), were observed in a total of 8 vessels (e.g.MR6553 in Fig. 3C; Appendices 5-7).
The 30 visible charred residues ('RI'/'RE') did not have high lipid survival, and most had lipid profiles consisting of free fatty acids (C 16:0 -C 18:0 ).A single 'RE' demonstrated the preservation of MAGs, DAGs and TAGs (MR6521RE), while another (MR6506RE) had odd-chain n-alkanes, n-alcohols and wax esters ranging between C 42 -C 50 with a higher proportion of palmitic over stearic, indicating a possible mixture of different waxes (Fig. 3D, E).
Extracts of surface coatings/layers ('SE'/'SI') were dominated by even-chain free-fatty acids (C 14:0 -C 26:0 ), but some samples also contained odd-chain fatty acids (C 13:0 -C 23:0 ), branched-chain fatty acids (C 15Br -C 21Br ), and unsaturated fatty acids (C 18:1 or C 20:1 and C 23:1 ).Phytanic acid, MAGs, DAGs and TAGs, as well as n-alkanes, alcohols and wax esters were detected in a smaller subset of samples.TAGs are present in both SIs and SEs and represent between 1.0-26.4% of the TLE.Esters were detected in 3 samples, representing between 3.1-8.6% of the TLE.At least five SEs and some SIs show saturated TAG profiles which are typical of the presence of ruminant dairy fats (T 42 -T 54 , dominated by T 50 (Dudd & Evershed 1998;Dudd et al. 1999), as well as n-alkanes, n-alcohols and wax esters (E 40 -E 48 , majority composed of C 16:0 ), suggesting that the organic component of the layer was a mixture of animal fats (possibly dairy products) and waxes (see Appendices 4; 10-12).For three vessels (MR6516T/SE/ SI, MR6517T/SE/SI and MR6520T/SE/SI), the external and internal coating layers were more lipid enriched and better preserved than the central part of the sherd (see Vessel use and technology and Fig. 6).
In parallel, plant sterols (β-sitosterol and undetermined phytosterol; four vessels), different kinds of betulin derivatives, lupeol derivatives (three samples; two vessels) and terpenoids were also occasionally identified.The presence of unusual compounds in 20% of the sherds and some amides made it necessary to carry out a control analysis on sediments well after the sherds had been excavated and analysed (see Appendices 1-4; 12).The results obtained show that these unusual compounds are indeed present in the sediment (see Appendix 12).Both compounds are not commonly found in archaeological contexts and were randomly detected in the assemblage, sometimes in an unexplained high quantity.Consequently, they have been excluded from our interpretations and considered as contaminants.In contrast, wax esters, n-alkanes and n-alcohols, were mainly found within the absorbed lipids in the clay matrix and rarely in surface/ coating layers or visible charred residues.Thus, they are not considered as contaminants.It's assumed that if they came from the surrounding sediment they would have been detected in the visible charred residues and in the other surface layers analysed, in higher concentrations and systematically associated with other soil contaminants.Furthermore, the ester distribution in the archaeological samples was more often restricted than the one in the sediment (Appendix 12).
Measurements of pH, electrical conductivity (CE), redox potential (Eh) and total dissolved solids (TDS) were performed on the same sediment.These measurements reveal that the sedimentary context is weakly acidic (pH = 6.8) with a tendency for water saturation or soil compaction (Eh = -46 mV).The conductivity is too low to effectively support vegetation growth (CE = 192 µS).The soil is low in salts and dissolved solids (TDS = 88.9 ppm) (see Appendix 2 for details on the method).These values are typical of a leached soil, a condition that could explain the infiltration of certain molecules into the matrix of the ceramics.detection of aquatic products Some solvent-extracted samples had traces of phytanic acid, found in aquatic products and ruminant fats, as well as long chains of saturated and unsaturated fatty acids, found in aquatic products and plant material (Cramp & Evershed 2014;Bondetti et al. 2021).Thus, none of the samples provided clear aquatic markers.
Five samples (MR6560TH, MR6573TH, MR6583, MR6593TH, MR6599TH) had enough concentrations of (APAAs C18 ), to calculate a part of the distribution of the profile, out of which only one (MR6560TH) presented a complete distribution (Fig. 4D).
One sample (MR6534TH) contained the most of the criteria mentioned (PRI, PHY (SRR%: 75%), TMTD and traces of APAA C18 ), without complete evidence for the heating markers APAA C18-C20 , and three other samples had SRR ratio >70% also indicative of the putative presence of aquatic products.

identification of dairy products and ruMinant Meat
Several solvent-extracted samples suggested the presence of degraded ruminant fats, and particularly dairy products, as evidenced by saturated fatty acids, cholesterol derivatives, MAGs, DAGs and TAG profiles (T 40/42 -T 54 ; Regert 2011).
To refine the characterisation of ruminant products and expand the information available for less well-preserved samples, we measured the stable carbon isotope (δ 13 C) values of two main saturated fatty acids (C 16:0 and C 18:0 ) in 24 samples (Craig et al. 2012).The differential routing of dietary carbon and fatty acids during the synthesis of adipose and dairy fats in ruminant animals enables ruminant milk fats to be distinguished from carcass fats by calculating Δ 13 C values (δ 13 C 18:0δ 13 C 16:0 ) and plotting them against δ 13 C 16:0 (Dudd & Evershed 1998;Copley et al. 2003).
Of the 24 solvent-extracted samples prepared for CSIA, eight samples had lower lipid concentrations with only fatty acids, while sixteen had between 1.1 -14.7 TAG% of the TLE, with profiles indicative of ruminant carcass/ dairy products.Of all samples analysed, 17 had Δ 13 C values ranging from -6.4‰ to -3.1‰, indicative of dairy products (e.g.Dudd & Evershed 1998;Copley et al. 2003), and two (MR6605T and MR6618T) had Δ 13 C values ranging from -1.8‰ to -2‰, suggestive of ruminant carcass fats (Fig. 5; see also Appendix 4).Four samples have values that do not fall within established references and may be the product of mixtures (MR6507T, MR6516SI, MR6527T, and MR6575T).
A single sample had enriched 13 C values for C 18:0 , which may indicate marine input, but had a Δ 13 C value of -2.9‰, which falls within the range of ruminant fats (MR6545T).This sample has the possible presence of TMTD, and traces of PRI, PHY (SRR%: 61%) and APAAs C18 , which may explain the stearic acid enrichment of the 13 C.As the sample only has traces of aquatic markers and SRR% indicates ruminant products, it might be a mixture of mainly ruminant meat and marine products.

DISCUSSION
The preservation of lipids in potsherds from the Beg ar Loued, particularly TAGs, is exceptional, although lipid yields are low and lipids from external residues appear to have been leached.The content of ceramic vessels from Beg ar Loued is often the result of mixtures of several products but dominated by terrestrial animal fats (Table 3).The combination of molecular and isotopic data clearly indicates that most of these fats come from ruminant carcass, and/or dairy products.Despite the island context of the site, the incorporation of aquatic products in pottery is debatable.Plant input, possibly mixed with beeswax in a few samples, is also suspected, usually associated with animal fats (details in Appendix 4).
Combined with zooarchaeological information available at the site, we now discuss the food systems and cooking practices at Beg ar Loued.The use of pottery and the role of organic products for coating layers in the manufacture of pottery is also hypothesised.The content of ceramic vessels on the site of Beg ar Loued was dominated by the processing or use of animal fats (n = 79/174; 46%; which comes to 70/129 vessels, 54%).The combination of molecular and isotopic data clearly indicates that most of these fats came from ruminants (n = 28/174; 16%; 28/129 vessels; 21%) and that dairy (n = 34/174; 19%; 30/129 vessels; 23%) made up a significant proportion of these animal products.Mid-chain ketones (K 31 to K 35 ; Raven et al. 1997) are occasionally found in vessels with ruminant carcass and dairy products, suggesting the continuous heating of these products in vessels (Evershed et al. 1995).
Despite the site's close relationship with the sea, this data confirms the importance of dairying as a subsistence strategy on the site (Table 3; Appendix 4).This suggests that the site was part of a cultural landscape of pastoral societies on the Atlantic coast, established at least by the Early Neolithic (Cubas et al. 2020;Evershed et al. 2022) and continuing into the Bronze Age, as evident from sites across the channel (Copley et al. 2003;Šoberl 2011).

Was pottery used for processing aquatic products?
Zooarchaeological analyses have demonstrated that the populations of Beg ar Loued exploited a wide range of coastal resources, especially during summer for fishing gilthead bream, as well as the collection of limpets close to the site (Pailler & Nicolas 2019a).It has been suggested that fish was stored for long-term consumption by drying, smoking and/or salting (Dréano 2019).
However, lipid and isotopic data do not provide a clear indication of the exploitation of aquatic resources in ceramics.Among the 31 acid-extracted samples, some of the vessels have partial biomarkers for aquatic products.TMTD, the isoprenoid marker specific for aquatic products when associated with PRI and PHY, co-eluted with another compound, but may be present in low amounts in only eight vessels.Fatty acid isotopic values for only one of these (MR6545T) are available, which demonstrates slightly enriched 13 C values for stearic acid, suggesting a mixture of mainly ruminant meat and marine products (as suggested in Identification of dairy products and ruminant meat).The SRR%>70% for 4 vessels may indicate the aquatic origin of phytanic acid and therefore an input of fish into the ceramics.None of the samples contained APAAs in significant quantities and they were restricted to the non-specific APAAs C18 (see Evidence of plant products and contamination from sediment; Cramp & Evershed 2014;Bondetti et al. 2021), which suggests that aquatic products were not heated in vessels.
Pending further detection of aquatic products, the results of this study suggest minimal use of ceramics to process aquatic products at Beg ar Loued.These findings support that the settlement at Beg ar Loued was linked to seasonal fishing activities, for the stocking of dried/smoked/salted fish (aceramic process), while shellfish could be eaten fresh or cooked on stone on-site (Dupont 2019; Dréano 2019; Mougne 2019; Pailler & Nicolas 2019a).The pattern observed at Beg ar Loued appears similar to those studied at Bronze Age sites in Britain and Scotland, where fishing did not play a central role in the subsistence economy (Craig et al. 2000;Copley et al. 2003;Richards et al. 2003;Šoberl 2011;Baeten et al. 2013).Such a subsistence strategy is unique when compared to other areas of the Baltic coast, where lipids studied in ceramics reveal the integration of pottery in the processing and consumption of aquatic resources at sub-Neolithic and Bronze Age sites (Craig et al. 2011;Cramp & Evershed 2014;Cramp et al. 2014a, b;Heron et al. 2015;Robson et al. 2019).

vesseL use and technoLogy
Given the fragmentary nature of the ceramic assemblage at Beg ar Loued, the reconstruction of the forms of vessels has been limited.Available evidence from the location of the sherd on the vessel (rim, base, body; see Appendix 7) did not allow for a reconstruction of the gradient of lipid concentration in order to make inferences about specific culinary activities such as boiling or roasting (Drieu et al. 2022).
The presence of mid-chain ketones (K 29 -K 35 ) formed by the condensation of free fatty acids provides an indication of vessel use for heating (Raven et al. 1997).Mid-chain The colour scale provides an indication of the Δ 13 C value: we use accepted threshold values to distinguish dairy (-3.3‰<Δ 13 C), ruminant (3.3‰<Δ 13 C<-1.0‰)and non-ruminant fats (Δ 13 C>-1.0‰)(Dudd 1999;Outram et al. 2009;Craig et al. 2011;Cramp et al. 2014b;Taché & Craig 2015;Colonese et al. 2015;Horiuchi et al. 2015;Choy et al. 2016;Lucquin et al. 2016b;Liu et al. 2017;Courel et al. 2020b;Pääkkönen et al. 2020).R Markdown file in Appendix 13.Credits: R code and CAD, C. Prévost.ketones were mostly found absorbed in ceramic matrices (see Appendices 5; 6), and for some of the samples, as the sherds do not have charred visible encrustations, they probably result from repeated use of vessels rather than long heating periods (ibid.).Moreover, they are found rather at the level of the body and absent from the bases (see Appendix 7).Unlike other studies on cooking pots, where mid-chain ketones are generally found at the bottom or on the rims of containers (Duplaix-Rata 1997; Evershed 2008); here the pattern is slightly more consistent with the boiling of half-filled containers.Visible charred residues are more often found on body fragments of pots (21 of 27 samples; 78%) with walls with medium thickness (median of the average thickness c. 7 cm; Appendices 5; 6).These observations suggest that vessels may have been heated laterally rather than centrally or suspended over a fire, as it was reported in a recent study of ethnological vessels used for indirect/steam cooking (Drieu et al. 2022).
Investigations into surface treatments of pottery on the Atlantic coast of France are mostly limited to the examination of ceramic thin sections or vessel surface traces (Convertini & Querré 1998;Convertini 2019).Very little literature exists on the chemical determination of surface treatments in general for European prehistoric contexts (Drieu et al. 2020).Our preliminary results on six potsherds present one of the first investigations of the organic component of vessels coating layers (engobe) in the region that we know of.
The gradient of lipid accumulation and composition indicates that organic components in the external layers are often not related to the contents of the vessel, and they are not analogous to the surrounding sediments (see Fig. 6; Appendices 4C-6; 12).
One can note that coating layers are mainly identified on vessels with thin walls (c. 5 cm; see Appendix 5).Heat markers (charring, visible residues and ketones) are not detected on these fragments, indicating that these might constitute an assemblage distinct from cooking pots.In most of the surface/coating layers investigated, wax/plant biomarkers (esters, n-alkanes, n-alcohols) were identified together with the typical composition of animal fats (see Appendix 4).
As an example, the external layer, internal layer and absorbed residue of the sample (MR6517) had similar lipid concentrations (c.200 µg.g -1 ; see Fig. 6).Degraded wax and potential dairy products were found in the external red lustrous layer (MR6517SE), while the internal layer (MR6517SI) only supports evidence for potential dairy products.The absorbed residue (MR6517T) showed free fatty acids, presumably from animal fats.This suggests that the products in the outer layer are not related to the contents of the vessel.Other examples suggest the presence of ruminant carcass fats and waxy products in red lustrous external coating layers (e.g.MR6507SE, MR6520SE).Overall, as these pottery fragments have no heat markers (visible residues or ketones); show particular investment in the finishing treatment; and a specific gradient of lipid impregnation, this evidence suggests that waxes and/or animal grease were used for vessel finishing treatment, or constituted the organic component of vessel coating layers, particularly for those with red lustrous external layers.In these cases, the adding of wax may have been used to obtain the lustrous finish rather than to act as a vessel sealant or as culinary input.The esters detected do not make it possible to clearly determine wax sources.Plant waxes were probably used, but still need further investigation.Meanwhile, knowledge about the use of beehive products in Europe was known long before the occupation of Beg ar Loued (Regert et al. 2001;Evershed et al. 2008;Salque et al. 2013;Roffet-Salque et al. 2015, 2016;Drieu et al. 2021;Tarifa-Mateo et al. 2021).The use of beeswax is well attested in southern England in vessels from the Collared Urns (c.2200-1780 BCE) at the site of Gayhurst Barro (Šoberl 2011), but as far as we know, there are no contemporary published examples on the Atlantic coast.It is possible that inhabitants of Molène may have used beeswax mixed with other plant waxes while finishing vessels so as to improve the mechanical strength of the induction, or to make up for a lack of sufficient quantities during the manufacture of pottery.

CONCLUSION AND OUTLOOK: FROM POTTERY CONTENT TO COOKING AND CULINARY PRACTICES
The lipid residue results presented here facilitate a new understanding of resource exploitation, vessel use and culinary practices at Beg ar Loued, a unique island settlement off the northwest French Atlantic coast.Our study investigates lipids in ceramics from the Final Neolithic and Early Bronze Age, filling a gap in lipid residue research in this region, which has mostly focused on the Early Neolithic period.
The exceptional lipid preservation in the ceramics, evidenced by the presence of TAGs and esters in several samples, indicates that burial environment conditions (leached and fairly acidic soils) in Brittany likely enable the survival of lipids in pottery.
Domestic ruminants make up an essential component of the products being processed in vessels.The strong evidence of dairy products in vessels highlights that pastoral lifeways were adopted by the inhabitants of the island, much like other populations along the Atlantic coast (Cubas et al. 2020;Evershed et al. 2022).Plant biomarkers are also occasionally identified but their origin is often difficult to establish with certainty due to their low amount in ceramic vessels and their presence in the sediment control sample.
Despite the site's island context and close relationship with the sea, minimal evidence for the processing of aquatic products in the ceramics is detected, and the extent of their use in vessels remains an open question.This pattern fits with the general lack of evidence of fish in vessels observed along the Atlantic coast in the Neolithic period (Cubas et al. 2020).This perhaps indicates the continuity of specific culinary preferences over time and specific local choices for the processing of fish without the use of ceramics, as demonstrated by the diversity of marine fauna present at the site.Together with the material evidence and the ceramic lipid residue results, it is clear that populations at Beg ar Loued maintained an integrated land-and coastal-based subsistence economy, connecting both the mainland and the ocean.
Finally, the analyses reveal at least two categories of vessels: cooking pots with molecular evidence of heating, and vessels with coating layers or special finishing treatment, possibly linked to consumption or service of food commodities without any heating operation.The latter category has evidence of organic components such as mixtures of waxes and animal fats used for surface treatments.
Our study combines the data obtained from lipid residue results, characteristics of the ceramics, environmental data and faunal remains to contribute new aspects about culinary practices of prehistoric Atlantic coastal communities.Developing other studies using such multiproxy approaches in the future will be useful for studying natural resource management and culinary practices in greater depth.A focus on ceramics and their use will enable a better understanding of the role of pottery in food production and variabilities or links between different settlements across time.

LIPID EXTRACTION
soLvent extraction (dcM/Meoh) A total of 174 archaeological samples including sherds, adhering visible charred residues, surface/coating layers and a topsoil sample were extracted with this method in the first place.Sherd fragments (c. 2 g) were cleaned with a modeling drill and ground to a powder in a solvent-cleaned mortar and pestle, while charred internal or external surface visible residues (between 5-50 mg) adhering to pottery were scraped with a solvent-cleaned scalpel and ground in a solvent-cleaned mortar and pestle.External/internal surface/coating layers were gently cleaned with a modeling drill to remove the top-most contaminating surface, and then drilled to obtain a powder.The control topsoil sample (5 g) was sieved (2 mm grid) and ground.
Then, lipids were extracted via conventional solvent extraction (Evershed et al. 1990).An internal standard was added in the archaeological samples (20 µg of n-C 34 , 1 mg.mL -1 in n-hexane) and control topsoil sample (50 µg of n-C 34 , 1 mg.mL -1 in n-hexane).A solution of dichloromethane/ methanol (DCM/MeOH) 2:1 v/v was added to the samples (10 mL for crushed sherds, charred visible residues, and surface/coating layers, and 20 mL for the control topsoil sample) and sonicated (2 × 15 min).After centrifugation (12 min; 3000 rpm), the supernatant was evaporated to dryness and dissolved in 500 µl of DCM/MeOH to obtain the total lipid extract (TLE).Aliquots of the TLE were trimethylsilylated (N,O-is(trimethylsilyl)trifluoroacetamide, 70 µL, 70°C, 60 min), and submitted to analysis by GC and GC-MS.A method blank was extracted alongside each batch of samples to identify any contaminant introduced during the extraction.
A sample-set (n = 24) was selected for compound-specific isotopic analysis.Samples with poor preservation only with fatty acids, or those with traces of triacylglycerols (TAGs) which could be indicative of ruminant carcass/dairy products were chosen.
acid extraction (Meoh/h 2 so 4 ) Archaeological samples with good preservation of lipids, or with the presence of polyunsaturated fats (PUFAs), isoprenoid fatty acids (IFAs) and/or ketones were selected for further extraction to detect aquatic resources and improve the characterisation of ruminant fats.
The acid extraction protocol was adapted from Craig et al., 2007Craig et al., , 2013;;Correa-Ascencio & Evershed 2014.Samples were ground (c.0.8 g) and an internal standard was added (triacontane, 1 mg.mL -1 , 10 µL), after which methanol was added (4 mL).The samples were sonicated (15 min), then after cooling, concentrated sulphuric acid was added (c. 800 µL).The tubes were heated for 4 hours at 70°C.Once cooled, they were centrifuged (3000 rpm; 5 min) to facilitate separation of the solid fraction.The liquid fraction was recovered.The acid phase was then extracted by adding 2 mL of hexane and rinsing the solid fraction with 2 mL of hexane, again after centrifugation (3000 rpm; 15 min).The acid phase (6 mL) was evaporated under a stream of nitrogen (40°C), and rediluted in 500 µL of cyclohexane.Finally, 1 µL was used for each analysis, while the rest was kept in the freezer (-40°C).
ANALYSIS VIA GC-FID AND GC-MS archaeoLogicaL saMpLes GC-FID analyses (gas chromatography-flame ionization detection) were performed on an Agilent Technologies 7890A device.A volume of 1 µL of sample was introduced via an on-column injector into a 15 m × 0.32 mm i.d.fused silica capillary (DB5-MS, 0.1 µL film thickness, Agilent J&W), with helium used as carrier gas.The GC temperature program was as follows: increased from 50°C to 100°C at 15°C.min −1 , then from 100°C to 375°C at 10°C.min −1 .
For GC-MS analysis (gas chromatography-mass spectrometry), the instrument used was a Shimadzu GC 2010 PLUS chromatograph coupled to a Shimadzu QP 2010 ULTRA mass spectrometer, fitted with a high temperature non-polar column (DB5-HT, 15 m × 0.322 mm i.d., 0.1 µm film thickness, Agilent J&W).The injection was performed using a splitless injector.The temperature programme consisted of a 1 min isothermal hold at 50°C followed by an increase to 150°C at 20°C.min −1 , then to 250°C at 10°C.min −1 and to 350°C.min−1 and a final isothermal hold for 10 min.The GC-MS interface was maintained at a temperature of 300°C and the mass spectrometer was run in electron ionization mode (EI, 70 eV).Mass spectra were acquired over the range m/z 50-950.
The instrument was also used on SIM mode to track specific ions for the detection of aquatic resources and ruminant fat, such as isoprenoid fatty acids (IFAs) and ω-(o-alkylphenyl) alkanoic acids (APAAs).

Faire
fig. 2. -A, B, C, Examples of pottery fragments analysed in the framework of this study (internal face on left); D, diagram showing different sampling options and abbreviations used.Abbreviations: RE, external visible charred residues; RI, internal visible charred residues; SE, external surface/coating layer; SI, internal surface/coating layer.Scale bars: 1 cm.Credits: photos, J.-D.Strich; CAD, C. Prévost.

Faire
gras à Molène: dairy products and ruminant fats in pottery from Beg ar Loued COMPTES RENDUS PALEVOL • 2024 • 23 (1) fig. 3. -Chromatograms of TMS lipid extracts (obtained by conventional solvent extraction) of vessels from Beg ar Loued that demonstrate: A, the presence of animal fats with the survival of TAGs indicating the possible presence of dairy products; B, mixtures of substances including animal fats (possibly ruminant carcass) and degraded unspecific wax; C, the presence of an unspecific wax with evidence of heat treatment (mid-chain ketones); D, E, are chromatograms of samples MR6506T and MR6506Re, an absorbed residue and external visible surface residue that indicate the presence of wax (possible mixture of plant wax and beeswax) on the external surface: D includes (inset) a partial fragmentogram for the m/z 257 and m/z 285 ions showing the relative proportion of palmitate vs stearate esters characterising the wax; F, is a compilation of the chromatogram and the partial fragmentograms for the m/z 257, m/z 285 and m/z 117 ions of sample MR6558T, which shows the most significant evidence concerning potential beeswax.Abbreviations: AL, n-alkanes (blue squares); Cn:x, fatty acids with n carbon atoms and x double bonds (red star); D, diacylglycerols; E, wax esters (yellow triangles); K, ketones; M, monoacylglycerols (inverted purple triangles); OH, n-alcohols (green dots); T, triacylglycerols.Scale bars: 1 cm.Credits: sherds photographed by C. Prévost; CAD, C. Prévost.
pottery used for processing ruMinant carcass and dairy products Domestic cattle, pig and sheep/goat are first reported on the islands of the Iroise Sea off the coast of Brittany in the 4th millennium BC(Hanot & Tresset 2019).At Beg ar Loued, zooarchaeological data suggest breeding of sheep/ goat at the site(Hanot & Tresset 2019).Slaughter curves show that sheep/goat were likely exploited for their milk;
fig.6.-A, B, Histograms of the total lipid extract (%) from selected samples with separated brownish surface layer (MR6510) and red/white coating layers (MR6507, MR6520, MR6516, MR6517); C, scheme presenting the usual pattern of absorption of lipids in vessels.A, Pattern A fits the distribution of lipid impregnation through the vessel profile generally observed in archaeological pottery; B, pattern B shows a unique distribution of lipid impregnation supporting the use of fatty substances in ceramic finishing processes for samples from Beg at Loued.R Markdown file available in Appendix 13.Abbreviations: SE, external surface/ coating layer; SI, internal surface/coating layer; T, sherd.Credits: R code and CAD, C. Prévost and A. Suryanarayan.

Faire
gras à Molène: dairy products and ruminant fats in pottery from Beg ar Loued COMPTES RENDUS PALEVOL • 2024 • 23 (1) appenDix 3. -Chromatogram of lipid extract of control sediment and mass spectrum of unknown compounds (Beg ar Loued): A, chromatogram of lipid extract of control sediment; B, C, mass spectrum of unknown compounds.appenDix 4. -Tables of all molecular and isotopic data: A, acronyms used; B, general archaeological data; C, general analysis data; D, table comparing the total lipid extract (TLE) of samples that have external and internal charred residues, external and internal coating/surfaces and absorbed residues; E, table with esters data: https://doi.org/10.5852/cr-palevol2024v23a1_s1Molène: dairy products and ruminant fats in pottery from Beg ar Loued COMPTES RENDUS PALEVOL • 2024 • 23 (1) appenDix 6. -Boxplots comparing different sample residue types with: A, total lipid extract (TLE); B, triacylglycerol preservation (TAG%); C, average pot wall thickness; D, identification of molecular heating markers.C, D, do not include samples from 2019 study.Black diamonds indicate mean value.R Markdown file for plots available in Appendix 14. Abbreviations: TAG, triacylglycerol(s); TLE, total lipid extract.et al. appenDix 7. -Boxplots comparing vessel fragment types with: A, total lipid extract (TLE); B, triacylglycerol preservation (TAG%); C, average pot wall thickness; D, identification of molecular heating markers.C, D, do not include samples from 2019 study.Black diamonds indicate mean value.R Markdown file for plots available in Appendix 14. Abbreviations: TAG, triacylglycerol(s); TLE, total lipid extract.Molène: dairy products and ruminant fats in pottery from Beg ar Loued COMPTES RENDUS PALEVOL • 2024 • 23 (1)

table 2 .
-Details of the total lipid extract (TLE) obtained by solvent extraction from absorbed lipids in sherd fragments 'T', charred surface residues adhering to either the internal 'RI' or external surface of the sherd 'RE', and surface/coating layers (either internal 'SI' or external 'SE'), and individual vessels (combining different samples linked to one vessel) analysed in this study.In bold, total values for the whole sample-set.

table 3 .
-Overview of the molecular characterisation and products/biomarkers identified in vessels from Beg ar Loued.
appenDix 11. -Plots with published examples of ester profiles of beeswax (extracted from Mirabaud et al. 2007).R Markdown file for figure available in Appendix 14. appenDix 10. -Plots with ester profiles (Beg ar Loued).R Markdown file for figure available in Appendix 14. Prévost C. et al. appenDix 12. -Plots with percent composition of wax ester profiles (Beg ar Loued).R Markdown file for figure available in Appendix 14.appenDix 13. -R Markdown file with code used to make all figures for the article: https://doi.org/10.5852/cr-palevol2024v23a1_s2appenDix 14. -R Markdown file with code used to make all figures for Appendices 5-12: https://doi.org/10.5852/cr-palevol2024v23a1_s3