ArticlesDOI 101038s41559-017-0322-4

copy 2017 Macmillan Publishers Limited part of Springer Nature All rights reserved

Evidence for mid-Holocene rice domestication in the AmericasLautaro Hilbert1 Eduardo Goacutees Neves2 Francisco Pugliese2 Bronwen S Whitney   3 Myrtle Shock4 Elizabeth Veasey5 Carlos Augusto Zimpel6 and Joseacute Iriarte   1

1 Department of Archaeology University of Exeter Exeter EX4 4QE UK 2 Museo de Arqueologia e Etnologia Universidade de Satildeo Paulo Satildeo Paulo-SP 05508-900 Brazil 3 Department of Geography Northumbria University Newcastle Newcastle upon Tyne NE1 8ST UK 4 Departamento de Arqueologia Universidade Federal do Oeste de Paraacute Santareacutem-PA 68035-110 Brazil 5 Departamento de Geneacutetica Escola Superior de Agricultura ldquoLuiz de Queirozrdquo Universidade de Satildeo Paulo Piracicaba-SP 13418-900 Brazil 6 Laboratoacuterio de Arqueologia dos Troacutepicos Universidade Federal de Rondocircnia Porto Velho-RO 76801-974 Brazil e-mail JIriarteexeteracuk

SUPPLEMENTARY INFORMATION

In the format provided by the authors and unedited

NAturE ECoLoGy amp EVoLutIoN | wwwnaturecomnatecolevol

Supplementary Note 1 Excavations at Monte Castelo

The excavation of the site in 2014 consisted of reopening the previous excavation of Eurico Miller1

cleaning the abandoned profile and expanding the excavation from the abandoned undisturbed

walls The provenience map in Supplementary Figure 2 illustrates the previous excavation of E Miller

(grey) and the new excavation carried out in 2014 (brown) A datum point with a Total Station was

established to have a consistent vertical measurement of the stratigraphy and excavation followed

10-cm artificial levels while paying close attention to the artificial strata Depth is reported in cm

below datum (cm bd) The stratigraphy shows a sequence of construction events evidenced by

unburnt entire Pomacea shell layers occupation floors marked by lenses of crushed shells primary

and incomplete burials and human-created dark soils (Supplementary Figure 3) Sample collection

was carried out by column sample in undisturbed contexts in each of the stratigraphic layers defined

during the excavation and targeted samples were collected from particular features such as burials

(Supplementary Table 2)

Below follows a summary description of the stratigraphy

Layer A ndash Superficial stratum (30-40 cm bd) This is a 10-20 cm deep layer composed of dark soils

(10yr 22 very dark brown) and containing a few ceramic sherds and lithics Based on its ceramics it

can be relatively dated to around AD 1200

Layer B ndash Later shell mound construction stratum (50-60 cm bd) This is a 30 cm deep layer

composed of dark silty clay loam soil (10yr 22 very dark brown) and containing about 50 of

Pomacea shells It also exhibits a high density of ceramic fragments including zoomorphic

appendages belonging to the Bacabal Phase2 It was preliminarily interpreted as a mound building

construction stratum built by the later occupants of the site represented in stratum A

Layer C ndash Dark earth upper stratum (70-80 cm bd) This is a 30 cm deep layer composed of dark clay-

silt soil (10yr 21 black) exhibiting a decrease in shell materials abundance when compared to the

above stratum but with a higher density of ceramic sherd corresponding to the Bababal phase

Layer D ndash Funerary Bacabal stratum (90-100 cm bd) This is a 30-60 cm deep layer composed of

thick clay-silty-loam soil matrix (10yr 22 very dark brown) and is characterised by the presence of

two human burials and clear features including hearths One is a primary burial of a juvenile

individual buried with an E-W orientation with a clear grave delimitation made by accompanying

gastropod shells The other was found near the end of this stratum where a deer skull with antlers

(likely Blastocerus dichotomus or Odocoileus virginianus given its large size) was found intentionally

buried over the left arm of an adult human burial surrounded by a circular shell structure The

interface between layers D and E is characterised by circular features that penetrate into the lower

layer which are likely postholes

Layer E - Burnt shell intermediate stratum (140-150 cm bd PN67) This is a 15 cm deep layer

composed of a thin layer of burnt and crushed shells of the Pomacea sp shells Soil matrix is

composed predominately of very fragmented (lt1 to 10mm) and burnt shells and ash (10yr 81

white to 10yr 71 white grey) mixed with areas of a light brown very fine clay-silt soil (10yr 64

yellowish brown) Fragments of chelonian bones and a well-preserved bivalve shell were also found

Lithic remains include some flakes and a rolled pebble hammer

Layer F ndash Intermediate shell mound construction stratum (200-210 cm bd PN111) This is thick

150 cm deep layer composed almost entirely of entire or slightly fragmented shells of apparently

unburnt Pomacea sp gastropods deposited in a clay-silt matrix with a coloration varying from 10yr

34 (dark yellowish brown) to 10yr 43 (brown) Interspersed with the gastropod shells this stratum

contains abundant charred faunal material lithics and charcoal

Layer G ndash Dark earth intermediate stratum (220-230 cm bd PN117) This is a 40-70 cm deep layer

composed of a very dark (10yr 21 black) clay soil matrix showing lenses of small fragments of

burnt and crushed snail shells similar to the ones in Layer E

Layer H ndash Former shell mound construction stratum (260-270 cm bd PN126) This is a 30 cm deep

layer mainly composed of unburnt Pomacea sp entire shells with abundant faunal remains

deposited in a dark clay matrix (10yr 22 very dark brown) In between the lenses of entire unburnt

Pomacea shells it exhibits lenses of ground shells that appear to be superimposed occupation floors

Layer I ndash Burnt shell intermediate stratum (350-360 cm bd) This layer has several textures and

colours showing a much more compact dark soil than stratum H and many lenses of ground shells

without visible associations of artefacts It is not clear yet whether soil colour results from anthropic

modification and enrichment or is just a representation of the adjacent dark soils of the Guaporeacute

floodplain It is not clear likewise whether this is a construction or occupation unit It follows

practically the same inclination as the inferior portion of layer H

Layer J ndashDark earth stratum (390-400 cm bd PN 112) This is a 100 cm deep layer composed of a

dark earth layer (10yr 21 black) with a highly compacted clay-sand texture It contains a high

density of charcoal (at least three different types of palm pericarp were identified) several flakes

and other lithic artefacts (including polished tools) as well as vertebrate and invertebrate remains

that are sometimes found associated with soil concretions

Supplementary Note 2 Multivariate statistics

Principal Component Analysis (PCA) of the five size measurements of Oryza phytoliths was

performed on the modern wild reference (botanical) dataset to determine the key variables that

define phytolith morphology (Supplementary Figure 6) Height (H1 H2) and Width (TW MW)

measurements are each shown to be highly correlated and comprise most of PCA axis 1 (60) PCA

of archaeological material (Supplementary Figure 7) and all specimens (archaeological and botanical)

(Supplementary Figure 8) confirm the findings that Height and Width measurements are key

defining variables All PCA analyses were performed in R package Community Ecology package

lsquoveganrsquo version (22-1)3

Supplementary Table 1 Radiocarbon dates from the Monte Castelo shell mound Rondonia State Brazil

Layer Depth (cm)

Phase Radiocarbon

date BP Calyr BP

Material

dated ID Reference

A 10-15

----

----

----

-Bac

abal

----

----

----

----

- 810plusmn70 805-625 (85) Charcoal B103185 Miller (2009)4

A 20-30 2475plusmn105 2756-2306 (945) Shell SI6843 Miller (2009)

A-D 10-110 2270plusmn105 2493-1992 (925) Shell SI6844 Miller (2009)

D 110-120 3160plusmn70 3482-3140 (94) Charcoal B66309 Miller (2009)

D 120-130 3700plusmn30 4085-3895 (95) Charcoal B408413 Pugliese et al (press)5

D 130-140 4810plusmn90 5662-5305 (95) Charcoal B66310 Miller (2009)

F 160-170 3945plusmn110 4628-3984 (94) Shell SI6845 Miller (2009)

F 210-220 3920plusmn85 4526-4078 (935) Shell SI6847 Miller (2009)

G 230-240 ____ 4455plusmn100 5318-4827 (954) Shell SI6852 Miller (2009)

H 230-240

----

----

--Si

nim

buacute

----

----

-

4395plusmn70 5285-4822 (927) Charcoal SI6848 Miller (2009)

J 430-440 4570 plusmn 30 5310-5210 (95) Charcoal B408414 Pugliese et al (press)

600-610 6316plusmn105 7424-6938 (954) Charcoal SI6850 Miller (2009)

600-610 5065plusmn85 5923-5599 (954) Charcoal SI6849 Miller (2009)

610-620 5605plusmn95 6570-6182 (933) Charcoal SI6853 Miller (2009)

610-620 5165plusmn80 6021-5655 (925) Shell SI6854 Miller (2009)

620-630

--C

up

im--

7010plusmn80 7959-7661 (954) Charcoal B118274 Miller (2009)

650-670 5970plusmn80 6970-6540 (954) Charcoal B118275 Miller (2009)

685-700 8350plusmn70 9478-9090 (954) Charcoal B103187 Miller (2009)

Supplementary Table 2 Provenience of wild rice modern reference material

Species Population identification Hydrographic basin

Ograndiglumis1 PA-1 Solimotildees

Ograndiglumis SO-23 Solimotildees

Ograndiglumis PU-1 Purus

Ograndiglumis JA-1 Japuraacute

Oalta RI-1 Ribeira

Oalta RI-2 Ribeira

Oglumaepatula SO-17 Solimotildees

Oglumaepatula JA-4 Japuraacute

Oglumaepatula RB-9 - Dauau Lake Negro

Oglumaepatula Pu-1 Purus

Olatifolia Arg-11 ParaguaiParanaacute

Olatifolia Arg-5 ParaguaiParanaacute

Olatifolia Arg-7 ParaguaiParanaacute

Olatifolia Arg-8 ParaguaiParanaacute

Olatifolia E-00814355 19264

Olatifolia E-00814354 19263

Olatifolia E-00814353 19262

Olatifolia E-00258728 19261

Olatifolia E-00258733 192610

Osativa E-00258732 19269

Osativa E-00258729 19266

Osativa E-00258730 19267 1 Except for the last eight samples wild rice seed samples were obtained from a collection of the

Genetics Dept at Escola Superior de Agricultura ldquoLuiz de Queirozrdquo University of Satildeo Paulo

Supplementary Table 3 Provenience of samples for phytolith analysis from Monte Castelo

Layer Depth Coordinates

A 30-40 cm E1010N1011

B 50-60 cm E1008N1011

C 70-80 cm E1008N1011

D PN75 90-100 cm E1010N1011

D PN64 130 cm E1008N1011

D PN1003 130-140 cm E1008N1011

DE PN 65 140 cm E1009N1011

E PN67 140-150 cm E1008N1011

F PN111 200-210 cm E1008N1011

G PN117 220-230 cm E1008N1012

H PN126 260-270 cm E1008N1012

I 350-360 cm E1009N1010

J PN 112 390-400 cm E1010N1010

J PN137 400-410 cm E1009N1010

J PN142 420-430 cm E1009N1010

J PN 159 450-460 cm E1010N1010

Supplementary Table 4 Results of pairwise comparison of overall phytolith size (mean Height mean Width) of archaeological and wild

reference specimens using a Bonferroni corrected p-value Highly significant differences among samples (plt0001) are shaded in dark grey

Significantly different specimens (p lt 005 ndash 001) are shown in light grey Phytoliths in the A layer are largest compared to all specimens and

phytoliths in the DE layers show a shift towards larger specimens

A B C D DE E F G H I J OALT OGG OGP

B 0537

C 1000 1000

D 0000 1000 0622

DE 0002 1000 1000 1000

E 0000 0495 0026 1000 1000

F 0000 1000 0141 1000 1000 1000

G 0000 1000 0238 1000 1000 1000 1000

H 0002 1000 1000 1000 1000 1000 1000 1000

I 0000 0056 0002 1000 1000 1000 1000 1000 1000

J 0000 0000 0000 0004 0522 1000 1000 1000 0719 1000

OALT 0000 0000 0000 0000 0006 1000 0348 0193 0009 1000 1000

OGG 0000 0000 0000 0000 0000 0002 0000 0000 0000 0059 0000 1000

OGP 0000 0000 0000 0000 0000 0001 0000 0000 0000 0041 0000 1000 1000

OLAT 0000 0000 0000 0000 0011 1000 0982 0526 0019 1000 1000 1000 0001 0000

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Layer F ndash Intermediate shell mound construction stratum (200-210 cm bd PN111) This is thick

150 cm deep layer composed almost entirely of entire or slightly fragmented shells of apparently

unburnt Pomacea sp gastropods deposited in a clay-silt matrix with a coloration varying from 10yr

34 (dark yellowish brown) to 10yr 43 (brown) Interspersed with the gastropod shells this stratum

contains abundant charred faunal material lithics and charcoal

Layer G ndash Dark earth intermediate stratum (220-230 cm bd PN117) This is a 40-70 cm deep layer

composed of a very dark (10yr 21 black) clay soil matrix showing lenses of small fragments of

burnt and crushed snail shells similar to the ones in Layer E

Layer H ndash Former shell mound construction stratum (260-270 cm bd PN126) This is a 30 cm deep

layer mainly composed of unburnt Pomacea sp entire shells with abundant faunal remains

deposited in a dark clay matrix (10yr 22 very dark brown) In between the lenses of entire unburnt

Pomacea shells it exhibits lenses of ground shells that appear to be superimposed occupation floors

Layer I ndash Burnt shell intermediate stratum (350-360 cm bd) This layer has several textures and

colours showing a much more compact dark soil than stratum H and many lenses of ground shells

without visible associations of artefacts It is not clear yet whether soil colour results from anthropic

modification and enrichment or is just a representation of the adjacent dark soils of the Guaporeacute

floodplain It is not clear likewise whether this is a construction or occupation unit It follows

practically the same inclination as the inferior portion of layer H

Layer J ndashDark earth stratum (390-400 cm bd PN 112) This is a 100 cm deep layer composed of a

dark earth layer (10yr 21 black) with a highly compacted clay-sand texture It contains a high

density of charcoal (at least three different types of palm pericarp were identified) several flakes

and other lithic artefacts (including polished tools) as well as vertebrate and invertebrate remains

that are sometimes found associated with soil concretions

Supplementary Note 2 Multivariate statistics

Principal Component Analysis (PCA) of the five size measurements of Oryza phytoliths was

performed on the modern wild reference (botanical) dataset to determine the key variables that

define phytolith morphology (Supplementary Figure 6) Height (H1 H2) and Width (TW MW)

measurements are each shown to be highly correlated and comprise most of PCA axis 1 (60) PCA

of archaeological material (Supplementary Figure 7) and all specimens (archaeological and botanical)

(Supplementary Figure 8) confirm the findings that Height and Width measurements are key

defining variables All PCA analyses were performed in R package Community Ecology package

lsquoveganrsquo version (22-1)3

Supplementary Table 1 Radiocarbon dates from the Monte Castelo shell mound Rondonia State Brazil

Layer Depth (cm)

Phase Radiocarbon

date BP Calyr BP

Material

dated ID Reference

A 10-15

----

----

----

-Bac

abal

----

----

----

----

- 810plusmn70 805-625 (85) Charcoal B103185 Miller (2009)4

A 20-30 2475plusmn105 2756-2306 (945) Shell SI6843 Miller (2009)

A-D 10-110 2270plusmn105 2493-1992 (925) Shell SI6844 Miller (2009)

D 110-120 3160plusmn70 3482-3140 (94) Charcoal B66309 Miller (2009)

D 120-130 3700plusmn30 4085-3895 (95) Charcoal B408413 Pugliese et al (press)5

D 130-140 4810plusmn90 5662-5305 (95) Charcoal B66310 Miller (2009)

F 160-170 3945plusmn110 4628-3984 (94) Shell SI6845 Miller (2009)

F 210-220 3920plusmn85 4526-4078 (935) Shell SI6847 Miller (2009)

G 230-240 ____ 4455plusmn100 5318-4827 (954) Shell SI6852 Miller (2009)

H 230-240

----

----

--Si

nim

buacute

----

----

-

4395plusmn70 5285-4822 (927) Charcoal SI6848 Miller (2009)

J 430-440 4570 plusmn 30 5310-5210 (95) Charcoal B408414 Pugliese et al (press)

600-610 6316plusmn105 7424-6938 (954) Charcoal SI6850 Miller (2009)

600-610 5065plusmn85 5923-5599 (954) Charcoal SI6849 Miller (2009)

610-620 5605plusmn95 6570-6182 (933) Charcoal SI6853 Miller (2009)

610-620 5165plusmn80 6021-5655 (925) Shell SI6854 Miller (2009)

620-630

--C

up

im--

7010plusmn80 7959-7661 (954) Charcoal B118274 Miller (2009)

650-670 5970plusmn80 6970-6540 (954) Charcoal B118275 Miller (2009)

685-700 8350plusmn70 9478-9090 (954) Charcoal B103187 Miller (2009)

Supplementary Table 2 Provenience of wild rice modern reference material

Species Population identification Hydrographic basin

Ograndiglumis1 PA-1 Solimotildees

Ograndiglumis SO-23 Solimotildees

Ograndiglumis PU-1 Purus

Ograndiglumis JA-1 Japuraacute

Oalta RI-1 Ribeira

Oalta RI-2 Ribeira

Oglumaepatula SO-17 Solimotildees

Oglumaepatula JA-4 Japuraacute

Oglumaepatula RB-9 - Dauau Lake Negro

Oglumaepatula Pu-1 Purus

Olatifolia Arg-11 ParaguaiParanaacute

Olatifolia Arg-5 ParaguaiParanaacute

Olatifolia Arg-7 ParaguaiParanaacute

Olatifolia Arg-8 ParaguaiParanaacute

Olatifolia E-00814355 19264

Olatifolia E-00814354 19263

Olatifolia E-00814353 19262

Olatifolia E-00258728 19261

Olatifolia E-00258733 192610

Osativa E-00258732 19269

Osativa E-00258729 19266

Osativa E-00258730 19267 1 Except for the last eight samples wild rice seed samples were obtained from a collection of the

Genetics Dept at Escola Superior de Agricultura ldquoLuiz de Queirozrdquo University of Satildeo Paulo

Supplementary Table 3 Provenience of samples for phytolith analysis from Monte Castelo

Layer Depth Coordinates

A 30-40 cm E1010N1011

B 50-60 cm E1008N1011

C 70-80 cm E1008N1011

D PN75 90-100 cm E1010N1011

D PN64 130 cm E1008N1011

D PN1003 130-140 cm E1008N1011

DE PN 65 140 cm E1009N1011

E PN67 140-150 cm E1008N1011

F PN111 200-210 cm E1008N1011

G PN117 220-230 cm E1008N1012

H PN126 260-270 cm E1008N1012

I 350-360 cm E1009N1010

J PN 112 390-400 cm E1010N1010

J PN137 400-410 cm E1009N1010

J PN142 420-430 cm E1009N1010

J PN 159 450-460 cm E1010N1010

Supplementary Table 4 Results of pairwise comparison of overall phytolith size (mean Height mean Width) of archaeological and wild

reference specimens using a Bonferroni corrected p-value Highly significant differences among samples (plt0001) are shaded in dark grey

Significantly different specimens (p lt 005 ndash 001) are shown in light grey Phytoliths in the A layer are largest compared to all specimens and

phytoliths in the DE layers show a shift towards larger specimens

A B C D DE E F G H I J OALT OGG OGP

B 0537

C 1000 1000

D 0000 1000 0622

DE 0002 1000 1000 1000

E 0000 0495 0026 1000 1000

F 0000 1000 0141 1000 1000 1000

G 0000 1000 0238 1000 1000 1000 1000

H 0002 1000 1000 1000 1000 1000 1000 1000

I 0000 0056 0002 1000 1000 1000 1000 1000 1000

J 0000 0000 0000 0004 0522 1000 1000 1000 0719 1000

OALT 0000 0000 0000 0000 0006 1000 0348 0193 0009 1000 1000

OGG 0000 0000 0000 0000 0000 0002 0000 0000 0000 0059 0000 1000

OGP 0000 0000 0000 0000 0000 0001 0000 0000 0000 0041 0000 1000 1000

OLAT 0000 0000 0000 0000 0011 1000 0982 0526 0019 1000 1000 1000 0001 0000

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Table 1 Radiocarbon dates from the Monte Castelo shell mound Rondonia State Brazil

Layer Depth (cm)

Phase Radiocarbon

date BP Calyr BP

Material

dated ID Reference

A 10-15

----

----

----

-Bac

abal

----

----

----

----

- 810plusmn70 805-625 (85) Charcoal B103185 Miller (2009)4

A 20-30 2475plusmn105 2756-2306 (945) Shell SI6843 Miller (2009)

A-D 10-110 2270plusmn105 2493-1992 (925) Shell SI6844 Miller (2009)

D 110-120 3160plusmn70 3482-3140 (94) Charcoal B66309 Miller (2009)

D 120-130 3700plusmn30 4085-3895 (95) Charcoal B408413 Pugliese et al (press)5

D 130-140 4810plusmn90 5662-5305 (95) Charcoal B66310 Miller (2009)

F 160-170 3945plusmn110 4628-3984 (94) Shell SI6845 Miller (2009)

F 210-220 3920plusmn85 4526-4078 (935) Shell SI6847 Miller (2009)

G 230-240 ____ 4455plusmn100 5318-4827 (954) Shell SI6852 Miller (2009)

H 230-240

----

----

--Si

nim

buacute

----

----

-

4395plusmn70 5285-4822 (927) Charcoal SI6848 Miller (2009)

J 430-440 4570 plusmn 30 5310-5210 (95) Charcoal B408414 Pugliese et al (press)

600-610 6316plusmn105 7424-6938 (954) Charcoal SI6850 Miller (2009)

600-610 5065plusmn85 5923-5599 (954) Charcoal SI6849 Miller (2009)

610-620 5605plusmn95 6570-6182 (933) Charcoal SI6853 Miller (2009)

610-620 5165plusmn80 6021-5655 (925) Shell SI6854 Miller (2009)

620-630

--C

up

im--

7010plusmn80 7959-7661 (954) Charcoal B118274 Miller (2009)

650-670 5970plusmn80 6970-6540 (954) Charcoal B118275 Miller (2009)

685-700 8350plusmn70 9478-9090 (954) Charcoal B103187 Miller (2009)

Supplementary Table 2 Provenience of wild rice modern reference material

Species Population identification Hydrographic basin

Ograndiglumis1 PA-1 Solimotildees

Ograndiglumis SO-23 Solimotildees

Ograndiglumis PU-1 Purus

Ograndiglumis JA-1 Japuraacute

Oalta RI-1 Ribeira

Oalta RI-2 Ribeira

Oglumaepatula SO-17 Solimotildees

Oglumaepatula JA-4 Japuraacute

Oglumaepatula RB-9 - Dauau Lake Negro

Oglumaepatula Pu-1 Purus

Olatifolia Arg-11 ParaguaiParanaacute

Olatifolia Arg-5 ParaguaiParanaacute

Olatifolia Arg-7 ParaguaiParanaacute

Olatifolia Arg-8 ParaguaiParanaacute

Olatifolia E-00814355 19264

Olatifolia E-00814354 19263

Olatifolia E-00814353 19262

Olatifolia E-00258728 19261

Olatifolia E-00258733 192610

Osativa E-00258732 19269

Osativa E-00258729 19266

Osativa E-00258730 19267 1 Except for the last eight samples wild rice seed samples were obtained from a collection of the

Genetics Dept at Escola Superior de Agricultura ldquoLuiz de Queirozrdquo University of Satildeo Paulo

Supplementary Table 3 Provenience of samples for phytolith analysis from Monte Castelo

Layer Depth Coordinates

A 30-40 cm E1010N1011

B 50-60 cm E1008N1011

C 70-80 cm E1008N1011

D PN75 90-100 cm E1010N1011

D PN64 130 cm E1008N1011

D PN1003 130-140 cm E1008N1011

DE PN 65 140 cm E1009N1011

E PN67 140-150 cm E1008N1011

F PN111 200-210 cm E1008N1011

G PN117 220-230 cm E1008N1012

H PN126 260-270 cm E1008N1012

I 350-360 cm E1009N1010

J PN 112 390-400 cm E1010N1010

J PN137 400-410 cm E1009N1010

J PN142 420-430 cm E1009N1010

J PN 159 450-460 cm E1010N1010

Supplementary Table 4 Results of pairwise comparison of overall phytolith size (mean Height mean Width) of archaeological and wild

reference specimens using a Bonferroni corrected p-value Highly significant differences among samples (plt0001) are shaded in dark grey

Significantly different specimens (p lt 005 ndash 001) are shown in light grey Phytoliths in the A layer are largest compared to all specimens and

phytoliths in the DE layers show a shift towards larger specimens

A B C D DE E F G H I J OALT OGG OGP

B 0537

C 1000 1000

D 0000 1000 0622

DE 0002 1000 1000 1000

E 0000 0495 0026 1000 1000

F 0000 1000 0141 1000 1000 1000

G 0000 1000 0238 1000 1000 1000 1000

H 0002 1000 1000 1000 1000 1000 1000 1000

I 0000 0056 0002 1000 1000 1000 1000 1000 1000

J 0000 0000 0000 0004 0522 1000 1000 1000 0719 1000

OALT 0000 0000 0000 0000 0006 1000 0348 0193 0009 1000 1000

OGG 0000 0000 0000 0000 0000 0002 0000 0000 0000 0059 0000 1000

OGP 0000 0000 0000 0000 0000 0001 0000 0000 0000 0041 0000 1000 1000

OLAT 0000 0000 0000 0000 0011 1000 0982 0526 0019 1000 1000 1000 0001 0000

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Table 2 Provenience of wild rice modern reference material

Species Population identification Hydrographic basin

Ograndiglumis1 PA-1 Solimotildees

Ograndiglumis SO-23 Solimotildees

Ograndiglumis PU-1 Purus

Ograndiglumis JA-1 Japuraacute

Oalta RI-1 Ribeira

Oalta RI-2 Ribeira

Oglumaepatula SO-17 Solimotildees

Oglumaepatula JA-4 Japuraacute

Oglumaepatula RB-9 - Dauau Lake Negro

Oglumaepatula Pu-1 Purus

Olatifolia Arg-11 ParaguaiParanaacute

Olatifolia Arg-5 ParaguaiParanaacute

Olatifolia Arg-7 ParaguaiParanaacute

Olatifolia Arg-8 ParaguaiParanaacute

Olatifolia E-00814355 19264

Olatifolia E-00814354 19263

Olatifolia E-00814353 19262

Olatifolia E-00258728 19261

Olatifolia E-00258733 192610

Osativa E-00258732 19269

Osativa E-00258729 19266

Osativa E-00258730 19267 1 Except for the last eight samples wild rice seed samples were obtained from a collection of the

Genetics Dept at Escola Superior de Agricultura ldquoLuiz de Queirozrdquo University of Satildeo Paulo

Supplementary Table 3 Provenience of samples for phytolith analysis from Monte Castelo

Layer Depth Coordinates

A 30-40 cm E1010N1011

B 50-60 cm E1008N1011

C 70-80 cm E1008N1011

D PN75 90-100 cm E1010N1011

D PN64 130 cm E1008N1011

D PN1003 130-140 cm E1008N1011

DE PN 65 140 cm E1009N1011

E PN67 140-150 cm E1008N1011

F PN111 200-210 cm E1008N1011

G PN117 220-230 cm E1008N1012

H PN126 260-270 cm E1008N1012

I 350-360 cm E1009N1010

J PN 112 390-400 cm E1010N1010

J PN137 400-410 cm E1009N1010

J PN142 420-430 cm E1009N1010

J PN 159 450-460 cm E1010N1010

Supplementary Table 4 Results of pairwise comparison of overall phytolith size (mean Height mean Width) of archaeological and wild

reference specimens using a Bonferroni corrected p-value Highly significant differences among samples (plt0001) are shaded in dark grey

Significantly different specimens (p lt 005 ndash 001) are shown in light grey Phytoliths in the A layer are largest compared to all specimens and

phytoliths in the DE layers show a shift towards larger specimens

A B C D DE E F G H I J OALT OGG OGP

B 0537

C 1000 1000

D 0000 1000 0622

DE 0002 1000 1000 1000

E 0000 0495 0026 1000 1000

F 0000 1000 0141 1000 1000 1000

G 0000 1000 0238 1000 1000 1000 1000

H 0002 1000 1000 1000 1000 1000 1000 1000

I 0000 0056 0002 1000 1000 1000 1000 1000 1000

J 0000 0000 0000 0004 0522 1000 1000 1000 0719 1000

OALT 0000 0000 0000 0000 0006 1000 0348 0193 0009 1000 1000

OGG 0000 0000 0000 0000 0000 0002 0000 0000 0000 0059 0000 1000

OGP 0000 0000 0000 0000 0000 0001 0000 0000 0000 0041 0000 1000 1000

OLAT 0000 0000 0000 0000 0011 1000 0982 0526 0019 1000 1000 1000 0001 0000

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Table 3 Provenience of samples for phytolith analysis from Monte Castelo

Layer Depth Coordinates

A 30-40 cm E1010N1011

B 50-60 cm E1008N1011

C 70-80 cm E1008N1011

D PN75 90-100 cm E1010N1011

D PN64 130 cm E1008N1011

D PN1003 130-140 cm E1008N1011

DE PN 65 140 cm E1009N1011

E PN67 140-150 cm E1008N1011

F PN111 200-210 cm E1008N1011

G PN117 220-230 cm E1008N1012

H PN126 260-270 cm E1008N1012

I 350-360 cm E1009N1010

J PN 112 390-400 cm E1010N1010

J PN137 400-410 cm E1009N1010

J PN142 420-430 cm E1009N1010

J PN 159 450-460 cm E1010N1010

Supplementary Table 4 Results of pairwise comparison of overall phytolith size (mean Height mean Width) of archaeological and wild

reference specimens using a Bonferroni corrected p-value Highly significant differences among samples (plt0001) are shaded in dark grey

Significantly different specimens (p lt 005 ndash 001) are shown in light grey Phytoliths in the A layer are largest compared to all specimens and

phytoliths in the DE layers show a shift towards larger specimens

A B C D DE E F G H I J OALT OGG OGP

B 0537

C 1000 1000

D 0000 1000 0622

DE 0002 1000 1000 1000

E 0000 0495 0026 1000 1000

F 0000 1000 0141 1000 1000 1000

G 0000 1000 0238 1000 1000 1000 1000

H 0002 1000 1000 1000 1000 1000 1000 1000

I 0000 0056 0002 1000 1000 1000 1000 1000 1000

J 0000 0000 0000 0004 0522 1000 1000 1000 0719 1000

OALT 0000 0000 0000 0000 0006 1000 0348 0193 0009 1000 1000

OGG 0000 0000 0000 0000 0000 0002 0000 0000 0000 0059 0000 1000

OGP 0000 0000 0000 0000 0000 0001 0000 0000 0000 0041 0000 1000 1000

OLAT 0000 0000 0000 0000 0011 1000 0982 0526 0019 1000 1000 1000 0001 0000

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Table 4 Results of pairwise comparison of overall phytolith size (mean Height mean Width) of archaeological and wild

reference specimens using a Bonferroni corrected p-value Highly significant differences among samples (plt0001) are shaded in dark grey

Significantly different specimens (p lt 005 ndash 001) are shown in light grey Phytoliths in the A layer are largest compared to all specimens and

phytoliths in the DE layers show a shift towards larger specimens

A B C D DE E F G H I J OALT OGG OGP

B 0537

C 1000 1000

D 0000 1000 0622

DE 0002 1000 1000 1000

E 0000 0495 0026 1000 1000

F 0000 1000 0141 1000 1000 1000

G 0000 1000 0238 1000 1000 1000 1000

H 0002 1000 1000 1000 1000 1000 1000 1000

I 0000 0056 0002 1000 1000 1000 1000 1000 1000

J 0000 0000 0000 0004 0522 1000 1000 1000 0719 1000

OALT 0000 0000 0000 0000 0006 1000 0348 0193 0009 1000 1000

OGG 0000 0000 0000 0000 0000 0002 0000 0000 0000 0059 0000 1000

OGP 0000 0000 0000 0000 0000 0001 0000 0000 0000 0041 0000 1000 1000

OLAT 0000 0000 0000 0000 0011 1000 0982 0526 0019 1000 1000 1000 0001 0000

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 1 A Oryza alta growing in the Guaporeacute River (courtesy C Zimpel)

B Wild rice growing on the Maniquiri lake during the dry season Amazonas state (courtesy

Carlos Augusto da Silva)

A

B

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 2 Provenience map of the Monte Castelo excavations

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 3 Stratigraphy of the Monte Castelo shell midden Unit N1010-11

North Face (drawn by F Pugliesse)

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 4 Relative frequencies of phytoliths recovered in the A and C fractions of samples from levels A to J of the Monte

Castelo shell mound Horizontal bars represent relative percentages circles correspond to the presence of plant taxa lower than 1 in

abundance Non-diagnostic Poaceae phytoliths include non-diagnostic bulliforms rondels and crosses Non-diagnostic Arboreal morphotypes

include globular granulates sclereids and various types of faceted tracheary elements

30-40 cm

50-60 cm

70-80 cm

90-100 cm

130 cm

130-140 cm

140 cm

140-150 cm

200-210 cm

220-230 cm

260-270 cm

350-360 cm

390-400 cm

400-410 cm

420-430 cm

450-460 cm

Layers

A

B

C

D

DE

E

F

G

H

I

J

00

Aris

tidoi

deae

0 5 10 15

Bam

buso

idea

e

00 40 80

Chlor

idoide

ae

0 5 10 15 20 25

Pan

icoide

ae

0 5 10 15

Non

-diagn

ostic

Poa

ceae

0 5 10 15 20 25

Ory

za d

ouble-

peak

edd

eeply

serra

ted

glum

es

0 5 10 15 20

Ory

za cun

eifo

rm k

eyston

e bu

llifor

m

00 20 40

Ory

zeae

sco

oped

end

billo

bate

00

Zea m

ays

wav

y-to

p ro

ndel

00

Zea m

ays

leaf

pre

senc

e

00 40 80

Cuc

urbita

sca

llope

d sp

here

0 5 10 15 20 25

Cyp

erac

eae

0 10 20

Mar

anta

ceae

00

Helicon

iace

ae

00 50 100

Aster

acea

e

0 5 10 15 20 25

Are

cace

ae g

lobu

lar e

chinat

e

00 30 60

Are

cace

ae c

onical to

hat

-sha

pe

00 25 50

Celtis

stip

pled

plate

0 25 50

Non

-diagn

ostic

Arb

orea

l

00

Ann

onac

eae

00

Tricho

man

es

00

Cys

tolith

Poaceae Cultigens Herbs Arboreal

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 5 Boxplots of Oryza double-peaked glume metric attributes from

modern reference species and the Monte Castelo stratigraphy

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 6 PCA of all botanical specimens showing that the two Width and

Height measurements are highly correlated and correspond to PCA1 explaining 60 of the

variance in the dataset Curvature Depth is correlated to PCA 2 which comprises 27 of the

variance

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 7 PCA of archaeological specimens only Height and width

measurements explain most variance in the data and Curvature Depth has a minor

contribution to Axis 1 (54) the species score however is low compared to height and

width Axis 2 is correlated to Top Width and Curvature Depth and explains 25 of the

variance

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 8 PCA for all specimens (botanical and total) confirming that Height

and Width explain most variance in the dataset PCA1 explains 58 of the variance PCA

axis 2 is correlated to Curvature Depth and Top Width

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

Supplementary Figure 9 Distribution of phytolith size among archaeological and wild

reference samples as determined by simple model (MWTW2)(H1H22) A Wilcoxon non-

parametric demonstrates the means of the two groups are significantly different (plt00001)

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13

References

1 Miller E T in Prehistoria Sudamericana Nuevas Perspectivas (ed B J Meggers) 219-232 (Taraxacum 1992)

2 Zimpel C A amp Pugliese F A in Ceracircmicas Arqueoloacutegicas da Amazocircnia Rumo a Uma Nova Siacutentese (eds C Barreto H P Lima amp C J Betancourt) 420-434 (Museu Paraense Emilio Goeldi 2016)

3 vegan Community Ecology Package R package v 21-1 (2015) 4 Miller E T A cultura ceracircmica do tronco Tupi no alto Ji-Paranaacute Rondocircnia Brasil algumas

reflexotildees teoacutericas hipoteacuteticas e conclusivas Revista Brasileira de Linguiacutestica Antropoloacutegica 1 35-136 (2009)

5 Pugliese Jr F A A Z N C amp Neves E G in Encyclopedia of Global Archaeology Vol Environmental Archaeology (ed Claire Smith) (Springer In press)

• Evidence for mid-Holocene rice domestication in the Americas13