Early Holocene crop cultivation and landscape modification in Amazonia.

The onset of plant cultivation is one of the most important cultural transitions in human history1-4. Southwestern Amazonia has previously been proposed as an early centre of plant domestication, on the basis of molecular markers that show genetic similarities between domesticated plants and wild relatives4-6. However, the nature of the early human occupation of southwestern Amazonia, and the history of plant cultivation in this region, are poorly understood. Here we document the cultivation of squash (Cucurbita sp.) at about 10,250 calibrated years before present (cal. yr BP), manioc (Manihot sp.) at about 10,350 cal. yr BP and maize (Zea mays) at about 6,850 cal. yr BP, in the Llanos de Moxos (Bolivia). We show that, starting at around 10,850 cal. yr BP, inhabitants of this region began to create a landscape that ultimately comprised approximately 4,700 artificial forest islands within a treeless, seasonally flooded savannah. Our results confirm that the Llanos de Moxos is a hotspot for early plant cultivation and demonstrate that-ever since their arrival in Amazonia-humans have markedly altered the landscape, with lasting repercussions for habitat heterogeneity and species conservation.

Recent genetic and archaeological evidence suggests the existence of at least four independent centres of early Holocene domestication, two in the Old World and two in the New World (Mesoamerica and northern South America)[1]. However, the closest wild ancestors of several globally important domesticated cultigens occur in south-western (SW) Amazonia. These include the wild ancestors of manioc (Manihot esculenta) M. e. subsp. flabellifolia[7]; squash (Cucurbita maxima spp. maxima) C. m. spp. andreana[8]; peach palm (Bactris gasipaes)[9]; jack bean (Canavalia plagiosperma) C. piperi[4] and chili peppers’ (Capsicum baccatum var. pendulum) C. baccatum var. baccatum[10]. This suggests that SW Amazonia could be the fifth global early Holocene centre of domestication. However, with the exception of Calathea sp. phytoliths, possibly representing lerén (C. alluoia), recently documented in the upper Madeira[11], archaeological evidence has not been found for early plant cultivation in SW Amazonia. Our research fills this gap with new data from 61 early and mid-Holocene archaeological sites that we refer to as forest islands (FIs)[12-14] because they now occur as patches of forest surrounded by savannah.

Mapping of forest islands

Using remote sensing data, we mapped 6643 FIs in the Llanos de Moxos (LM). The average size of FIs is 0.5 ha (min: 0.05 ha; max: 16 ha; SD 0.65 ha). We surveyed 82 of these, which represents ca. 1.2% of all sites. We took column sediment samples in all of them and carried out archaeological excavations in four. Out of the total 82 sites sampled, 83 including Monte Castelo in Brazil[14], we classified 64 sites as anthropic, based on the presence of deep dark sediments rich in organic matter, charcoal, and burned earth frequently associated with shell and bone fragments (Fig. 1). The FIs surveyed are between ~0.5 m and 3 m high. The proportion of anthropic vs. natural sites suggests the existence of at least 4700 anthropic FIs in the LM (Fig. ED1). This is probably far less than the original amount built in the early and mid-Holocene, as during the transition to the late Holocene, most of the rivers in the SW part of the LM became very active and many of the pre-existing soils and potential archaeological sites were covered by alluvial deposits, sometimes up to 5 m thick[15]. This explains the modern distribution of FIs and why 48% of the 6643 FIs we mapped are concentrated in a relatively small area in the north-western LM (Fig. ED1), where the landscape did not change notably during this period. Anthropic FIs are mostly located in interfluvial settings covered by seasonally flooded savannahs, and they account for an estimated 24 km2 of forested area and 1,000 km of forest/savannah ecotone.

Figure 1

Forest islands mapped in the LM. Numbers associated with middens are ages expressed in median cal yr BP from the deepest anthropic datable layer at each site (Table ED1). Insets a, b, d and e identify the four areas that were surveyed to estimate the total number of anthropic FIs in the LM (Fig. ED1). The square in inset c identifies the study area; the shaded area in inset c depicts Greater Amazonia.

Fig. ED1

Estimated number of anthropic FIs. The number was estimated by extending the proportion of anthropic FIs in the surveyed areas (in green) to the portion of the LM with similar physical geography and land cover (polygons). Background image represents the density of FIs, calculated using the Kernel Density tool in Esri ArcGIS.

Sixty-six AMS 14C dates from 31 archaeological sites (Table S1) bracket the human occupation of FIs throughout the Holocene between ca.10,850 and 2,300 cal yr BP. Except for three sites in the north-eastern LM dated at ca. 2,350, 2,350 and 4,100 cal yr BP, the remaining dated sites were established between the early and mid-Holocene.

Evidence of early plant cultivation

We analysed phytoliths from the radiocarbon-dated profiles of 30 FIs (Figs. 2, 3). The earliest evidence of Manihot is a heart-shaped phytolith[16] documented at the Isla Manechi dated ca. 10,350 cal yr BP and in Isla del Tesoro dated ca. 8,250 cal yr BP (Fig. 1). Scalloped sphere phytoliths derived from the rind of Cucurbita sp. were identified in layers dated to ca. 10,250 cal yr BP in Isla Manechi and ca. 9,850 cal yr BP in Site 575. We have identified wavy-top rondel phytoliths, produced in the cob of maize[17], dated ca. 6,850 cal yr BP (Site 570) and ca. 6,700 cal yr BP (Site 421). We detected the early presence of Marantaceae Calathea sp. rhizome phytoliths at ca. 8,275 cal yr BP (Isla del Tesoro), before ca. 7,800 cal yr BP (Site 433) and at ca. 7,400 cal yr BP (Site FIN 14). Other Marantaceae, Cyperaceae, Phenakospermum guianensis, and Heliconia phytoliths have been found in almost all the samples starting from ca. 10,400 cal yr BP (Fig. 2, S2). We identified Oryza seed phytoliths dated at ca. 6,250 cal yr BP (San Pablo); and phytoliths from the epidermis of Celtis seeds from contexts dated to ca. 9,600 cal yr BP (Site FIN8). Hat-shaped and globular echinate phytoliths diagnostic of subfamily Arecoideae[17] of the palm family (Arecaceae) are present at ca. 9,975 cal yr BP (site 493). Peach palm (Bactris sp.), a member of this subfamily, is the only palm domesticated in South America, and its domestication likely took place in SW Amazonia[6]. Bactris produces these hat-shaped phytoliths, but does not produce diagnostic phytoliths at the species level[18]. Bactris and other arecoid genera growing today in the LM (Astrocaryum, Desmoncus, Geonoma and Socratea), are used for food, building materials and medicine in present Amazonia[19]. The size of squash phytoliths is well in the range of the domesticated species (table ED2), however, they do not show increase in size over time as would be expected under domestication pressure (Table. ED2)[17].The presence of domesticated Cucurbita sp. beginning at ~10,250 cal yr BP in Isla Manechi is the oldest evidence of Cucurbita sp. in Amazonia and coincides with the domestication of several species of Cucurbita across Central [20] and South America[21,22] at the very beginning of the Holocene. Further studies analysing larger sample sizes are needed to determine whether the domesticated squash cultivated in the early Holocene was adopted in the LM from other regions or was domesticated in situ. Maize cob phytoliths were documented at site 570 ca. 6,850 cal yr BP representing the oldest evidence, by a few centuries, of maize cultivation in the Amazon basin. As hypothesized by Kistler et al.[23], early maize in this and other areas probably represented a partially domesticated variety that later diverged into two South American groups of fully domesticated maize varieties. This early evidence of maize phytoliths is consistent with a temporal gradient of maize dispersal that began in western Amazonia and reached the eastern Amazon by ~4,300 cal yr BP. The early use of Manihot in Isla Manechi in the LM began more than 10,000 years ago, which coincides with the time of estimated molecular divergence from the wild ancestor and with current biogeography of manioc’s closest wild ancestor[7,24]. It possibly spread later to northern Peru (8,500 cal yr BP), Colombia (7,000 cal yr BP) and Panama (7,600 cal yr BP)[5]suggesting the bi-directional exchange of cultivars between Amazonia and the Andes beginning since the early Holocene. Our study shows that, as in other regions of Amazonia and Central America, also in the LM the development or arrival of full-blown agricultural societies was a very late phenomenon[9], as there is no evidence of land prepared for agriculture in the LM until raised fields and drainage canals were built around 1,500-1,000 years ago[25].

Figure 2

Diagram showing the percentage of most relevant phytolith groups from anthropic forest islands. Dots indicate presence ≤ 1% (for more details see Fig. ED2 and table ED1). *Non-domesticated, edible plants (used as food resources)[26]. Phytolith percentages are based on a minimum of 200 diagnostic phytoliths counted per slide.

Figure 3

Photomicrographs of phytolith morphotypes recovered from Isla del Tesoro, La Chacra and Isla Manechi: (a) wavy-top rondel from the cob of maize (Zea mays) (sampleIT190-200); (b) heart-shaped phytolith from the secretory cells of manioc (Manihot) (sample BANR17-UE1-57); (c) scalloped sphere from the rind of squash (Cucurbita sp.) (sample BANR17-UE1-31); (d) double-peaked glume from the seed of rice (Oryza sp.) (sample SM3-116s); (e) flat domed cylinder from the rhizome of Calathea sp. (sample IT30-40); (f) short trough body from the rhizome of Heliconia sp. (sample IT130-140); (g) stippled polygonal body from the seed of Cyperaceae (sample IT150-170); (h) nodular projections with a pointed apex phytolith from the seed of Marantaceae (sampleIT90-100) (i) stippled plate from the seed/fruit of hackberries (Celtis sp.) (sample SM3-69-74). Scale bars = 20μm.

Table ED2

Length and thickness range and average size of scalloped-sphere phytoliths identified in this study. Based on Piperno et al. (2000), we consider scalloped spheres longer than 72 μm or thicker than 59 μm as coming from domestic varieties of Cucurbita sp.

Site	Date cal. yr.B.P.	Length(μm)	Thickness(μm)	Domesticated	Depth
Isla del Tesoro	7839-7020	80,361	59,804	Yes	170-180 cm
575	10155-9632	82,524	58,318	Yes	270-280 cm
		72,804	65,178	Yes
		72,775	54,214	Yes
519	6443-6380	77,884	55,212	Yes	100 cm
		82,418	65,752	Yes
		75,938	54,452	Yes
		75,296	57,433	Yes
FIN-12	4785-4424	81.19	55,162	Yes	90-95 cm
		80,653	60,988	Yes
		61,005	45,635	No
		77,871	67,843	Yes
La Chacra	6707-5826	83,697	59,724	Yes	80-84 cm
		78,895	52,993	Yes
	7350-6945	75,789	56,754	Yes	108-113 cm
	8530-7965	63,119	41,238	No	140-145 cm
	9212-8665	75,419	65,484	Yes	160-165 cm
Isla Manechi	3319-3007	74.28	52.33	Yes	25 cm
		72.7	57.57	Yes
		78.49	50.79	Yes
		81.34	48.89	Yes
		70.1	45.38	Yes
		78.12	64.79	Yes
		61.29	51	No
		69.22	56.6	No
		61.78	46.56	No
		59.96	49.13	No
	4957-4728	68.67	54.95	No	55 cm
		57.22	40.23	No
		70.53	49.5	No
		63.21	49.51	No
		70.59	44.29	No
		81.74	66.85	Yes
		72.76	58.61	Yes
	5288-4965	74.27	57.12	Yes	85 cm
		74.53	45.13	Yes
		58.98	44.15	No
		81.95	56.8	Yes
		68.1	59.06	No
		66.84	55.37	No
		76.91	49.26	Yes
		74.69	47.15	Yes
		84.72	60.19	Yes
	10284-10196	78,173	66,108	Yes	105 cm
		84,996	67,836	Yes
		72,601	54,422	Yes
		64,201	50,158	No
		84,858	64,479	Yes
		68.44	57.37	No
		81.82	60.62	Yes
		86.81	63.17	Yes
	10664-10380	60.46	51.16	No	145 cm

The importance of starch-based foods

Paleoecological studies indicate that the LM was covered by cerrado-like savannah during the early and mid-Holocene[15] indicating that plant cultivation in the Neotropics also started in shrub savannahs along with seasonal tropical forest environments[4]. It does not come as a surprise that phytoliths derived from plants producing underground storage organs constitute an important part of the total phytolith assemblages from the LM’s FIs including Manihot, Calathea and other Marantaceae, Heliconia, Cyperaceae and Phenakospermum. These plants, which are abundant in savannahs, produce carbohydrate-rich foods which, with the exception of some varieties of manioc[27], are easy to process and cook. They are nowadays consumed by indigenous groups[19] and probably provided a considerable part of the calories consumed by the first inhabitants of the LM. Large herbivores and fish available in the savannahs[12,13] would have complemented a mixed economy. The fertile FIs were probably the home gardens where these crops were cultivated. Our data are in agreement with the hypothesis that plants producing underground storage organs were a fundamental part of the diet of human populations colonizing new territories[28,29].

Implication for biodiversity

Our results show that inland savannahs were a key region for the early occupation of Neotropics and that they began to be transformed by the arrival of very early human settlers. FIs are entirely anthropic. However, their formation is not only an incidental effect of food-waste dumping, but it can also be seen as an active process of niche construction[30]. These accumulative middens also constitute fertility hotspots amid poor savannah soils, since: i) they became loci for the accumulation of nutrients that come from gathering activities in the surrounding savannah, and ii) they remained above the water level during the wet season[12]. It is only after 4,000 years BP, when the old and infertile soils of the south of the LM were covered with fertile alluvium deposited by the Río Grande, that agriculture in the savannahs was facilitated[31]. Overall, the construction of pre-Columbian FIs, which became key structures in the landscape[32,33], increased forest patchiness (Fig. ED3, ED4a) and probably contributed to maintaining landscape-scale species richness in this Ramsar threatened biome. Nowadays, these anthropic FIs are preferential feeding and roosting sites for many species of birds, including the endemic and critically endangered blue-throated macaw (Ara glaucogularis)[34]. Taken together, our data show that the earliest inhabitants of the LM were not just tropical hunter-gatherers, but colonizers who had engaged in plant cultivation since the early Holocene, thus opening up the possibility that they already had a mixed economy when they arrived in the region. The thousands of keystone structures represented by FIs show that the human footprint on Amazonia is not just restricted to large-scale transformations by late Holocene farming groups[9,35], but is rooted in the earliest human dispersal into this region and has lasting implications for habitat heterogeneity and biodiversity conservation.

Fig. ED3

Map of all the FIs and platform fields in a north-western subset of the LM. Platform fields are mostly built along paleochannels. Rivers Omi and Yruyañez flow inside old channels of the Beni River. FIs are mostly located in interfluvial areas. The region contains a total of 2428 patches of forest, 955 of which are FIs. Once all the patches of forest within a 2-km buffer of a river, paleoriver and lake are removed, FIs account for 60% of the remaining 1191 patches.

Fig. ED4

Characteristics of FIs. a) Photo of the anthropic landscape dotted with forest islands in the Barba Azul Nature Reserve; b) Histogram showing the distribution of the diameter of FIs. The left side of the distribution is truncated at 25 m because smaller FIs have not been mapped; c) Photo taken at Site 579, natural FI. Samples are taken every 10 cm, from top left to down right. Material is silt with no organic matter; d) Photo taken at Site 425, anthropogenic FI. Depth of the sample 140 cm; e) Photo taken at Site 430, anthropogenic FI. Depth of the sample 160 cm. Samples in d) and e) are representative of the whole profile in site 425 and 430 respectively (see also Fig. ED7). We performed one core for each FI we visited.

Methods

Mapping of FIs

FIs were mapped by visual scanning of high-resolution satellite imagery of Esri ArcGIS basemaps (Fig. ED5). When the identification of FIs was not straightforward due to cloud cover or poor resolution, TanDemX and SRTM digital elevation models were used as complementary resources. Patches of forest were classified as FIs when they had a round shape and were completely or partially surrounded by savannah (n = 4,341); or they had an irregular shape but were relatively small (<400 m in diameter) and completely surrounded by savannah (n = 2,304). For each FI, the following attributes were recorded: diameter; shape (perfectly round, almost round, elongated or irregular); location (along paleochannel, along modern river, seasonally flooded savannah, border between seasonally flooded savannah and upland, along a drainage stream, upland surrounded by bushes); presence of other earthworks within ~500 meters; and whether or not FIs were established over fluvial deposits or uplands. The latter attribute is partly redundant with “Location” but sometimes fluvial deposits are not connected to paleochannels (as in the case of old crevasse splays or old meander belts where paleochannels have been infilled) or the FIs are located along a paleochannel with completely eroded levees and the FIs have clearly been built after the erosion of the levees.

Fig. ED5

Examples of surveyed FIs as seen in high-resolution satellite imagery of Esri ArcGIS basemap. a-f: FIs classified as anthropic (a, Sm4; b, Isla Manechi; c, Site 575; d, SM3; e, FIN 12; f, Isla del Tesoro); g-i: FIs classified as natural (g, FIN2; h, FIN11; I, Sire 529). Source: ESRI, DigitalGlobe, GeoEye, Earthstar Georaphics,CNES/Airbus DS.

Selection of survey areas

Four survey areas were selected in different regions of the LM to ground-truth FIs and evaluate their natural or anthropic origin. The four areas (Fig. 1) were selected based on differences in soil, landcover, hydrology, and accessibility by car. These four areas cover all the different eco-regions identified in the LM[36,37]. These areas belonged to organizations (area a -see inset in Fig. 1- in the Barba Azul Nature Reserve) or ranchers (b, c, and d) who granted us permission to conduct surveys. In total, we surveyed 21 FIs in area a, 22 in b, 13 in d, and 17 in e. Nine other FIs were surveyed outside of these four areas.

Criteria for identification of anthropic FIs

Several anthropic FIs in the LM have been excavated [12-14]. These excavations have revealed thick strata of sediments rich in organic matter, charcoal, burnt earth, and fragmented animal bones and shells; they also have revealed human burials. The clear difference between the sediments found in the anthropic FIs and the soil types found in the LM[38-41] makes the field identification of FIs relatively straightforward. In the present work, the FIs surveyed have been classified as anthropic when thick layers of organic-rich sediments contained at least two archaeological materials such as charcoal, burnt earth, animal bones, and shells.

Sampling of FIs

Sampling of undisturbed material was performed at regular intervals in the four sites where archaeological excavations were conducted: Isla del Tesoro (SM1), La Chacra (SM3), San Pablo (SM4), and Isla Manechi (Fig. ED6). The rest of the sites were sampled using an auger soil sampler. The stratigraphy of the recovered cores was described in the field and sampling was carried out only where stratigraphic changes were detected in the field (Fig. ED7). The deepest sample with evidence of charcoal was always sampled. After extraction, cores were inspected to avoid contamination and check that the soil section sampled show no evidence of soil mixing (i.e. root penetration and invertebrate burrowing were absent). The excess of material has been cut off with a knife and only the inner, uncontaminated part of the extracted samples has been stored in plastic bags. Samples have been air-dried in Bolivia before being shipped. Charcoal fragments for 14C have been collected in situ, enveloped in aluminium foil and stored in plastic bags.

Fig. ED6

Stratigraphic profile and sampling sites at Isla Manechi. Charcoal fragments for AMS 14C dating were collected during the excavation and yellow triangles indicate their depth. Yellow squares indicate the sampling locations of sediments analyzed for phytoliths, which have been sampled after the excavation from the vertical profile. The dashed line indicates the transition from the ceramic phase to the pre-ceramic one. This transition is characterized by a sharp increase in the compactness of sediments and amount of burnt earths. Values in parenthesis are median calibrated radiocarbon ages BP. A chronological gap of almost 4500 years exists between the two phases (Table ED1). All of these differences make it possible to exclude any contamination of phytoliths coming from the ceramic to the pre-ceramic contexts. For a description of archaeological excavations at Isla del Tesoro (SM1), La Chacra (SM3), and San Pablo (SM4) see Capriles et al.[13]

Fig. ED7

Stratigraphic descriptions of cored sites. Radiocarbon dates are included in ovals as median calibrated years before present.

Phytolith processing and identification

Phytoliths were extracted from sediments following Lombardo et al.[42]. Phytoliths were identified and counted using a Zeiss Axioscope 40 light microscope at 500× magnification. Phytolith identifications were made using published material for the Neotropics[17,43-46] and by direct comparison with the phytolith reference collection of the Archaeobotany and Palaeoecology Laboratory in the Department of Archaeology of the University of Exeter. A minimum of 200 diagnostic phytoliths were counted per slide. A full scan of the slides was performed to detect the presence of squash, manioc and maize. Phytolith assemblages in SW Amazonia have been studied in modern soils[46] and 29 paleosols from the early and late Holocene[15] in different natural environments and land covers. In none of these natural contexts have phytoliths of Manihot or Curcubita been found, strongly suggesting that phytoliths of these two genera found in FIs are the direct result of human activity and not of the chance occurrence of wild relatives on these FIs

Radiocarbon dates

The deepest recoverable sample of charcoal from 32 sites was dated in order to establish the minimum age of site foundation. For the four sites that were excavated, 35 samples from different depths were dated in order to establish periods of occupation and abandonment. The complete dataset and code used to calibrate all of radiocarbon dates are available respectively in Table ED1 and Supplementary Information. Radiocarbon dates from the studied sites were calibrated using SHCAL13[47]. For Isla del Tesoro, Isla Manechi and La Chacra, where stratigraphically ordered ages were available, we run a series of Bayesian age-depth models using the P_Sequence command in OxCal 4.3[48] with default settings. Each model was stratigraphically constrained by the youngest age in the profile and the deepest section reached in each site (Fig. ED8). Ages of undated samples were estimated using the command Date within the model.

Table ED1

Radiocarbon ages of all the dated sites cited in the text including provenance and calibrations. Stratigraphically ordered dated depths have been modelled using a Bayesian age-depth model (OxCal 4.3’s P_Sequence). The modelled ages for samples BANR17-UE1-53 and BANR17-UE1-64 have been calculated using the Date command. Code is available in Supplementary Information.

Lab _ Code	Site	14C BP	14C Age SD	Unmodelled (BP) 95.4%		Modelled (BP) 95.4%		Median	Material	Depth (cm)
				from	to	from	to
D-AMS-1737	SM4, San Pablo	5476	35	6306	6124			6239	Bulk	58
Poz-46397	SM4, San Pablo	5190	80	6178	5664			5898	Charcoal	65
D-AMS-1741	SM4, San Pablo	5490	32	6306	6186			6245	Bulk	93
PSUAMS-4659	SM4, San Pablo	6665	25	7571	7441			7512	Charcoal	150
D-AMS-1739	SM4, San Pablo	6910	30	7787	7621			7696	Charcoal	150
Poz-46396	SM4, San Pablo	7700	90	8636	8218			8463	Charcoal	197
PSUAMS-1450	SM3, La Chacra	6030	30	6935	6736	6925	6734	6825	Charcoal	100
Poz-38862	SM3, La Chacra	5140	40	5930	5733	5933	5733	5825	Shell	107
PSUAMS-1563	SM3, La Chacra	6650	30	7566	7437	7568	7441	7506	Charcoal	121
PSUAMS-1564	SM3, La Chacra	7930	30	8972	8590	8951	8586	8672	Charcoal	155
Poz-38865	SM3, La Chacra	7860	50	8847	8435	8682	8426	8557	Shell	236
Poz-38866	SM3, La Chacra	7790	80	8760	8383	8748	8456	8590	Charcoal	240
Poz-38853	SM2, San Francisco	4950	40	5736	5585			5638	Shell	85
Poz-38850	SM2, San Francisco	4770	60	5588	5320			5465	Charcoal	155
Poz-38851	SM2, San Francisco	5380	40	6272	5994			6117	Shell	205
Poz-38852	SM2, San Francisco	5500	40	6393	6128			6253	Charcoal	205
Poz-34228	SM1, Isla del Tesoro	345	25	452	304			391	Bone	15
Poz-34229	SM1, Isla del Tesoro	3895	35	4411	4153	4414	4155	4292	Shell	35
Poz-34230	SM1, Isla del Tesoro	4945	35	5722	5586	5721	5585	5629	Shell	45
Poz-28854	SM1, Isla del Tesoro	3830	50	4405	3986			4172	Shell	48
Poz-28855	SM1, Isla del Tesoro	4415	35	5210	4849			4937	Charcoal	77
Poz-22902	SM1, Isla del Tesoro	5520	40	6395	6190			6280	Charcoal	115
Poz-34231	SM1, Isla del Tesoro	5520	40	6395	6190	6295	6190	6234	Shell	115
Poz-24633	SM1, Isla del Tesoro	5360	40	6266	5950			6096	Shell	115
D-AMS-1740	SM1, Isla del Tesoro	5502	30	6310	6190	6309	6207	6275	Charcoal	117
PSUAMS-4658	SM1, Isla del Tesoro	5565	20	6398	6281	6393	6278	6305	Charcoal	120
Poz-24634	SM1, Isla del Tesoro	5505	35	6388	6184			6262	Charcoal	120
Poz-34232	SM1, Isla del Tesoro	5460	40	6304	6021	6302	6205	6254	Shell	125
Poz-28856	SM1, Isla del Tesoro	4480	40	5284	4872			5047	Charcoal	140
Poz-28850	SM1, Isla del Tesoro	4495	35	5288	4886			5109	Shell	140
Poz-36136	SM1, Isla del Tesoro	5800	35	6657	6453	6660	6461	6563	Charcoal	160
D-AMS 032885	SM1, Isla del Tesoro	7271	40	8162	7966	8151	7956	8019	Bulk orqanic	205
D-AMS 032884	SM1, Isla del Tesoro	7447	37	8348	8065	8347	8174	8274	Bulk orqanic	225
Poz-34301	SM1, Isla del Tesoro	9270	60	10556	10248	10573	10259	10434	Bulk orqanic	235
Poz-36135	SM1, Isla del Tesoro	9420	50	10743	10433	10715	10416	10574	Charcoal	245
BE-4254.1.1	FIN8_182-187	8681	22	9671	9537			9582	Charcoal	185
BE-4253.1.1	FIN5_92-97	6963	25	7828	7680			7745	Shell	95
BE-4250.1.1	FIN3_95-100	8572	48	9560	9438			9514	Charcoal	97
BE-4257.1.1	FIN15_145-150	7997	25	8985	8649			8834	Shell	147
BE-4256.1.1	FIN14_85-90	6552	25	7486	7325			7429	Shell	87
BE-4255.1.1	FIN12_90-95	4092	24	4785	4424			4522	Shell	92
BE-8256.1.1	BANR17-UE1-5	3017	21	3236	3005	3319	3007	3144	Charcoal	25
BE-8257.1.1	BANR17-UE1-20	4324	22	4959	4743	4957	4728	4848	Charcoal	55
BE-8258.1.1	BANR17-UE1-31	4491	23	5285	4888	5288	4965	5172	Charcoal	85
BE-8259.1.1	BANR17-UE1-38	9138	24	10367	10195	10284	10196	10239	Charcoal	105
	BANR17-UE1-53					10546	10231	10372		125
D-AMS 029221	BANR17-UE1-57	9346	41	10653	10298	10664	10380	10521	Charcoal	145
	BANR17-UE1-64					13918	10307	10883		155
BE-7663.1.1	575_270-280	8849	50	10155	9632			9848	Charcoal	275
BE-7671.1.1	574_80-90	5516	62	6409	6025			6273	Charcoal	85
BE-7667.1.1	570_60-70	6046	48	6984	6695			6844	Charcoal	65
BE-7675.1.1	569_40-50	407	19	496	328			453	Charcoal	45
BE-7672,1.1	569_130-140	5759	53	6659	6399			6512	Charcoal	135
BE-7664.1.1	536_110-120	4994	37	5857	5595			5678	Charcoal	115
BE-7668.2.1	535_190-200	6069	51	7145	6730			6871	Charcoal	195
BE-7673.2.1	530_160-170	2397	20	2465	2327			2362	Charcoal	165
BE-7661.1.1	527_200-210	5337	281	6730	5472			6076	Charcoal	205
BE-7662.1.1	526_250-260	6217	40	7239	6945			7076	Charcoal	255
BE-7674.1.1	521_90-100	2346	89	2701	2094			2331	Charcoal	95
BE-7666.1.1	519_100	5647	22	6443	6308			6368	Charcoal	100
BE-6164.1.1	502 65	4365	110	5295	4583			4927	Charcoal	65
BE-6166.1.1	493 155	8920	49	10186	9766			10006	char/sed	155
BE-6153.1.1	490 95-100	3796	33	4237	3985			4115	Shell	97
BE-6167.1.1	434 270	7811	57	8696	8413			8543	char/sed	270
BE-6163.1.1	433 175	7058	50	7954	7718			7852	char/sed	175
BE-6168.1.1	430 235	6397	130	7552	6951			7269	chat/sed	235
BE-6158.1.1	429 220	8028	24	8999	8717			8870	char/sed	220
BE-6157.1.1	423 170	4365	21	4965	4840			4877	char/sed	170
BE-6159.1.1	421 170	5875	127	6953	6322			6646	char/sed	170

Fig. ED8

Age-depth models of the modelled profiles. a) and b), Isla del Tesoro; c) Isla Manechi; d) and e) Isla La Chacra. The age-depth models have been produced using OxCal V4.3, code is available in Supplementary Information

Detailed percentage phytolith diagram for all the samples analyzed from anthropogenic FIs. *Non-domesticated, edible plants. Phytolith percentages are based on a minimum of 200 diagnostic phytoliths that were counted per slide

Supplementary Material

Supplementary Information is available for this paper.