Journal of Archaeological Science: Reports 4 (2015) 174–181

Contents lists available at ScienceDirect

Journal of Archaeological Science: Reports

j ourna l homepage: ht tp : / /ees.e lsev ie r .com/ jas rep

Local water source variation and experimental Chicha de Maíz brewing:Implications for interpreting human hydroxyapatite δ18O values inthe Andes

Celeste Marie Gagnon a, C. Fred T. Andrus b, Jennifer Ida c, Nicholas Richardson d

a Department of Anthropology, Wagner College, Staten Island, NY 10301, United Statesb Department of Geological Sciences, University of Alabama, Tuscaloosa, AL 35487, United Statesc Department of Anthropology, University of Colorado Boulder, Boulder, CO 80309, United Statesd Department of Physical Sciences, Wagner College, Staten Island, NY 10301, United States

E-mail address: celeste.gagnon@wagner.edu (C.M. Gag

http://dx.doi.org/10.1016/j.jasrep.2015.09.0082352-409X/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 25 June 2015Received in revised form 8 September 2015Accepted 9 September 2015Available online xxxx

Keywords:Stable oxygen isotopesFermented beveragesWater sourcingMobilityChichaAndes

Oxygen isotope data (expressed as δ18O values) recovered from human skeletal remains have been the centralfocus of a number of archeological analyses tracking humanmigration. However, in the Andes the use of oxygenisotopes to investigate residential mobility is subject to two issues addressed in this study: 1) local variation inwater sources and 2) pre-consumption processing of water. To explore these factors we sampled a wide varietyof water sources in the Moche watershed of north coastal Peru and experimentally and ethnographically pro-duced chicha de maíz, a traditional brewed beverage. Our data indicate an unexpected spatial pattern in Mochevalley water source δ18O values, and identify springs as an important influence on river water. Using chicha demaíz as an example of pre-consumption processing of water, we find that brewed beverages may impacthuman δ18O values. Together these data indicate that in the Andes the consumption of chicha demaíz has the po-tential to swamp spatial variation in water source δ18O values. Therefore, we cannot assume that the identifica-tion of different δ18O values in the remains of Andean groups necessarily indicates the presence of migrants.

© 2015 Elsevier Ltd. All rights reserved.

Oxygen isotope analysis is a well-established approach to assess anumber of biological, environmental, and behavioral variables relevantto archeology (e.g., Andrus, 2011; Katzenberg, 2008; Makarewicz andSealy, 2015; Scherer et al., 2014; Sponheimer and Lee-Thorp, 1999). Ox-ygen isotope data (expressed as δ18O values) from humans bones andteeth can be interpreted to determine the isotopic composition ofwater or other liquids humans drink. Rainfall and continental surfaceand groundwater δ18O values vary geographically in a systematic way.This is the result of fractionation during phase transitions and mixingof water from different sources. The lightest oxygen isotope, 16O, evap-orates more readily than the heavier isotopes. Conversely, the heaviestisotope, 18O, condenses more readily. Consequently there is a regularand predictable pattern inwhich precipitation tends to become isotopi-cally lighter with increasing distance from the evaporation point, typi-cally resulting in more negative δ18Omw (meteoric water) values withincreasing latitude and/or altitude (see Kendall and Coplen, 2001 forNorth American rivers; http://www-naweb.iaea.org/napc/ih/IHS_resources_gnip.html; http://wateriso.utah.edu/waterisotopes/pages/data_access/ArcGrids.html). When imbibed, δ18Omw is transferred toδ18Obw (body water) and the local δ18O values are reflected in the car-bonate and phosphate of hydroxyapatite in human bones and teeth,

non).

transferring their signature to human remains (Longinelli, 1984). Thishas enabled researchers to use δ18Op (phosphate) and δ18Oc (carbonate)values to reconstruct both paleoclimates (e.g., Fricke et al., 1995) andmigration (e.g., White et al., 1998).

A variety of factors beyond climate change and residential mobilitymay affect human δ18Op & c values. Oxygen from inhalation and foodconsumption is incorporated into the body and so the levels of 18O inair and solid food could affect human δ18Op & c values. Research suggestshowever, that the effects of these sources is minimal, thus the value ofδ18Obw primarily reflects the δ18O value of imbibed water (Longinelli,1984; Longinelli and Peretti Paladino, 1980). Hydroxyapatite drawnfrom different skeletal elements of the same individual can be charac-terized by varying δ18O values. For example, different teeth form at dif-ferent ages but once formed are not remodeled, thus the δ18O values ofindividual teeth reflect the water imbibed during formation. Boneshowever, are continuously remodeled until death, and so their δ18Ovalues represent water imbibed during the last ten years of an individ-ual's life (Hedges et al., 2007). Because different elements of a single in-dividual capture δ18O values of water imbibed at different ages, oxygenisotopes have been used to track residential mobility. Tracking suchsources of variation has been central to a number of recent analysesthat focus on the δ18O values of human skeletal remains in the Andes(e.g., Knudson et al., 2009; Buzon et al., 2011; Turner et al., 2009;Toyne et al., 2014). However the use of oxygen isotopes to explore

175C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

residential mobility in the region is subject to two issues addressed inthis study: local variation in water sources and pre-consumption pro-cessing of water.

The general arid climate and complex hydrology of the Andes canlead to a wide range of δ18Omw values in local water sources in somebut not all areas. A number of recent local water source studies havebegun to examine such local variation in the southern Andes (Buzonet al., 2011; Knudson, 2009; Webb et al., 2013), however the northernAndes remain under-characterized. To begin to characterize northcoast water sources, we present δ18Omw values from several dozenriver locations, as well as canal and spring sources in the north coastalregion.

The δ18O value of imbibed water may differ from local δ18Omw valuebecause imbibedwatermay be subject to evaporation during food prep-aration. Such enrichment from processing has been documentedthrough isotopic analysis of commercially available beer (Carter et al.,2015) and experimentally produced, Medieval-style foods and drinks(Brettell et al., 2012). Andean food and drink preparation, while notedby researchers as being a potentially important source of variation inhuman δ18Op & c values, has not yet been experimentally investigated.Such work is especially important as the consumption of chicha demaíz, a traditional, Andean brewed beverage, has been well docu-mented in the archeological record. This study examines this secondissue through stable isotopic analysis of both experimentally and ethno-graphically brewed chicha de maíz.

1. Hydrology of the Andes

Precipitation in the Andes is derived primarily from moisturesources to the east (Aravena et al., 1999). The western slope of theAndes is a rain shadow, consequently much of coastal Peru and north-ern Chile receive almost no precipitation except during El Niño events,which cause spatially and temporally variable rains (Bourrel et al.,2015). As onewould expect in high altitude locations, meteoric, surface,and ground waters in the Andes are depleted in the heavy isotopes ofoxygen (Aravena et al., 1999; Ohlanders et al., 2013). The rivers thatcross the coastal desert are sourced in the western slope of the Andesand fed by high altitude precipitation and glaciermelt, resulting in com-paratively negative δ18O values similar to those measured in high alti-tudes (see Knudson, 2009 for a more complete explanation of thedistribution of meteoric δ18O values across altitude zones). Groundwa-ter aquifers in this coastal desert are fed by contributions from these riv-ers, high altitude precipitation, and to a lesser extent, from percolationby irrigation waters in agricultural fields (Gilboa, 1971). As a result,groundwater δ18O values in this region are similar to surface water,but could be variably heavier if the source waters underwent evapora-tion prior to percolation. Anthropogenic change to Andean surface andgroundwater hydrology over thepast fewmillennia is likelywidespreadconsidering the extensive irrigation practices used by pre-Hispanic cul-tures, though the impacts to δ18O values are not known.

2. Chicha de maíz

Broadly, the term chicha refers to any locally produced, fermentedbeverage (Goldstein et al., 2009). Most commonly, the term chicha isused in the Andes as a short hand for chicha de maíz, which is primarilymade from maize. Chicha, was and continues to be socially, politically,and economically important in the Andes and Amazon Basin (Allen,1988; Jennings, 2005; Jennings and Bowser, 2009; Weismantel, 1988).The use of chicha de maíz in the Andes has been well documentedethnohistorically, archeologically and ethnographicallywith the earliestevidence of consumption being identified in Southern Highlands anddating to as early as 800 B.C. (Logan et al., 2012).

In Gillin's (1947:46) mid-twentieth century ethnographic study ofthe Moche valley of Peru, he estimated adult chicha de maíz consump-tion of approximately two liters daily, with substantially more

consumption occurring during feasting events. Incan chroniclers deBetanzos (1996:56) and Cobo (1979:28) both note that people drankprodigious amounts of chicha at social events with Cobo (1979:27) sug-gesting that “there is no worse torment for them than being compelledto drink water (a punishment which the Spaniards sometimes givethem).” In more recent ethnographic investigations Peruvian work-party and fiesta consumption has been estimated between nine and fif-teen liters per person (Jennings, 2005).

Ethnohistoric and archeological research has found that chicha demaíz played an important role in Inca state politics. It was consumedin large quantities at political feasting events and its consumption wascentral to the idea of reciprocity that supported the social structure ofthe state (Bray, 2009; Dillehay, 2003; Morris, 1979). Chicha was used,not just in the heartland of the southern Andes, but throughout thestate to “pay” men for their work on large-scale, state-sponsored pro-jects and the production of chicha was a site of power negotiations atthe local level (Costin and Earle, 1989; Hastorf, 1990, Hastorf andJohannessen, 1993; Morris, 1979; Murra, 1960). Bray (2003) identifiedthree ceramic forms associated with three different elite foods: chicha,meat, and maize stew. Of these, the vessels used to store and transportchichawere the most common forms found outside the state heartland,indicating that chicha was an important tool of state expansion.

The importance of chicha de maíz in the Andean highlands pre-datesthe Inca. Goldstein (2003) discusses the role chicha played in earlierTiwanaku state politics in the Southern Highlands. Unlike among thelater Inca, it appears that chicha de maíz was primarily consumed atthe household and corporate group level, rather than as part of templerituals. However, Goldstein (2003) notes that corporate competitivefeasting with chicha de maíz may have played an important role in theTiwanaku expansion, particularly into lower elevation,maize producingregions. Among the contemporaneous Wari state of the central high-lands, archeological evidence also argues for the importance of maizeand chicha production (Valdez, 2006; Valdez et al., 2010), butGoldstein et al. (2009) and Sayre et al. (2012) suggest that chicha demolle (made with fruit of the Peruvian pepper tree) was the Waridrink of choice. Goldstein et al. (2009) identified several scales of chichade molle production at oneWari administrative center, but Laffe (2015)suggests that theWarimay not have centralized chicha demolleproduc-tion everywhere, allowing for substantial amounts at the householdlevel.

On the north coast of Peru, ethnohistoric research suggests the usesof chicha de maíz mirror those found in the Southern Highlands(Netherly, 1977; Rostworowski de Diez Canseco, 1977). Productionhas been identified archeologically based on the presence of fermenta-tion vessels or preserved jora (sprouted maize) at Huacas de Moche,the capital of the first regional polity in the Moche valley (UcedeCastillo, 2010), in the Lambayeque valley at Pampa Grande (Shimada,1994), at Chan Chan, the Chimu capital (Topic, 1990) and at Manchan,a Chimu region center in the Casma valley (Morre, 1989).

Stable isotopic analysis of human skeletal remains has been used toidentify the prehistoric use of chicha de maíz. In her analysis of pre-Incaand Inca period elites and non-elites from the Southern Highlands ofPeru, Hastorf (1990, 1991) identified increasing production of maizeand changing patterns of δ13C values as a result of Inca imperialism.She suggested these changes resulted from Inca elites re-orienting agri-cultural production toward maize for chicha, which they then suppliedto men involved in mit'a labor. In their analysis of elite and non-eliteprehistoric period individuals from highland Ecuador, Ubelaker andKatzenberg (1995) identified status differences in δ13C values. Becauseδ13Ccollagen and δ13Ccarbonate are tightly correlated, they argue that lessnegative δ13C values among elites result from the consumption ofmaize, the only C4 plant in the area, not from variations in proteinsources. This assessment was further supported by a lack in significantdifference between elites and non-elites in δ15N values. These re-searchers have interpreted changes in δ13C values as indicative of chichade maíz consumption based on ethnohistoric models. However, it is

176 C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

important to note that while δ13C values are indicative of maize con-sumption, they do not distinguish among possible preparations. As ofyet, δ18O has not been used to study the consumption of chicha demaíz in either archeological or ethnographic contexts.

Ethnographic studies of chicha de maíz provide us with models forreconstructing chicha production and consumption in the past. Tobrew chicha de maíz the complex carbohydrates in dried, hulled maizemust first be transformed into simple sugars, which can be accom-plished either through sprouting or through the introduction of salivaryamylase. In the first method, the sprouted maize or jora is then dried,ground, mixed with water, cooked and fermented (Bonavia, 2013;Culter and Cardenas, 1947; Hayashida, 2008; McCool and Parker,2014; Morris, 1979; Nicholson, 1960). The second method requiresdried maize to be ground, mixed with water and made into a smallball that is worked in the mouth to introduce amylase. These salivatedballs are mixed with water and sometimes unsalivated ground maize,cooked and fermented (Bonavia, 2013; Culter and Cardenas, 1947).

3. Methods

3.1. Water sampling

Water sampleswere collected from a variety of sources in theMochewatershed of north coastal Peru. The Moche watershed is divided intolower, middle and upper sections (Fig. 1). The lower valley, which ischaracterized by a wide alluvial plain, stretches from the mouth at thePacific to the neck located at approximately 160 m.a.l.s. In the middlevalley, the alluvial plain becomes constricted as the river cuts into thefoothills of the Andes. The upper valley, at elevations greater than 320m.a.l.s., lies above the confluence of the upper Moche with the Sinsicapand Cuesta rivers. In both the middle and upper valleys, springs dis-charge into the rivers. In each section, canals have been constructed todraw river water to farmland.

Water samples were collected during the austral winters of 2011and 2012. To insure the quality of each sample, a 100-ml plastic bottlewas first thoroughly rinsed in the water to be sampled before fillingand capping. Bottles were not opened again until they were processed.Location elevation, northing and easting were recorded using a hand-held Garmin GPS. Each location was characterized as lower, middle orupper valley. In addition, samples were identified by date collectedand as being drawn from river, spring, or canal. Only small canalswere sampled, as they more closely approximate prehistoric systemsthan do large, industrial canals.

Fig. 1.Map of Moche Watershed.

3.2. Experimental brewing

The consumption of chicha demaíz has the potential to affect humanδ18O values because of the substantial boiling it undergoes before fer-mentation. In this regard chicha is fundamentally different from barleyor wheat beers, in which the wort is heated but not boiled. Because ofits long boil time, chicha likely impacted the δ18Op & c values of peoplewho consumed significant amounts of the beverage. In order to exam-ine the amount of enrichment in chicha de maíz, we developed the fol-lowing basic recipe inspired by the jora recipes published byNicholson (1960) and Hayashida (2008).

Fill 1/3 of boiling vessel with ground jora.Add measured volumes of water to fill within 2 in. of vessel rim.Boil for 6 h, adding measured volumes of water to maintain waterlevel.Cool overnight.Next day, boil for 3 h and then strain to remove solids.Decant upi to fermentation vessel and allow to cool.When room temperature, add brewer's yeast and cover but do notseal.Ferment for 3 days.

We produced our chicha de maíz in a chemistry lab atWagner Col-lege in Staten Island, New York, using tap water, as we were not in-terested in the specific δ18Omw value of our source but rather thescale of enrichment resulting from brewing.We filled a sealed carboywith tap water to use throughout the brewing of each batch, takingthree pre-process samples to provide us with a baseline δ18Omw

value to compare to our three post-process samples. So as to under-stand the effects of vessel size on oxygen enrichment, we brewed inboth small and large batches. Our small batches used a 1-l beaker asour boiling vessel and a l-liter Erlenmeyer flask as our fermentationvessel. Large samples were boiled in a 30-l stock pot and fermentedin a 25-l carboy.

Our first small vessel chicha demaíz preparations were unsuccessful,as our product did not sufficiently ferment, resulting in 1% and 1.3% al-cohol, respectively. We hypothesized that the maize we purchased lo-cally was not sufficiently sprouted to allow for full conversion ofcarbohydrates to sugars. Insufficient sproutingwas indicated by the sig-nificantly shorter sprouts visible on our maize than on that of jora avail-able in Peruvianmarkets. To aid in fermentationwemodified the recipeby adding brewer's barley malt to the upi. Our subsequent preparationsproduced adequate fermentation, creating chicha de maíz between 3%and 4% alcohol. This modified recipe was followed for all batches sub-jected to stable isotopic analysis.

3.3. Ethnographic brewing

Our chicha de maíz was produced under experimental conditionswith resources available in NewYork City.Wewished to be able to com-pare the oxygen enrichment of our laboratory brew to a chicha de maízproduced in less artificial conditions. For this purpose, we worked witha chicheras (as women who brew chicha are known) who runs a café inthe Campiña de Moche, a small farming village located adjacent to theHuacas de Moche archeological site. Her recipe differed slightly fromours.

Fill boiling vessel with 2 kg jora entera, 1 kg each of whole barely andfava beans.Add water to fill vessel.Boil for 9–10 h, adding water to maintain water level.Cool overnight.Next day, strain upi to remove solids.Decant upi to fermentation vessel, add panela and cover but do notseal.

Fig. 3. Large vessel, experimental chicha de maize.

177C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

Ferment for 8 days.

Two of these differences could have affected enrichmentwith heavyoxygen. First, the chichera used jora entera rather than ground jora. As aresult of using unground jora (as well as whole barely and fava beans),the consistency of her pre-fermented chicha de maíz was that of a par-tially pureed soup, while ours was the consistency of a thick oatmeal(Figs. 2 and 3). Second, the chichera's boil vessel was larger, with a vol-ume of approximately 89 l (Fig. 4). Other differences in her recipe areless likely to impact δ18O, including the use of panela (unrefinedsugar) rather than barley malt to aid fermentation, and the lack ofadded yeast. No yeastwas added to the chichera's upi because fermenta-tion occurred in a ceramic vessel (Fig. 5) that had been inoculated withnative yeast during previous fermentations (Jennings, 2005; McCooland Parker, 2014). As with our experimentally produced chicha demaíz, we took pre-process water samples to compare to post-processsamples. In this case, however, we took three post-process samples be-fore fermentation and three samples after fermentation to test if herlonger fermentation time affected heavy oxygen enrichment.

3.4. Isotopic analysis

All water and chicha samples were measured for oxygen isotopes inthe University of Alabama's Stable Isotope Laboratory (ASIL). Prior toanalysis all samples were sealed in airtight bottles for transportationand brief storage (weeks). The samples were measured using the CO2

equilibrationmethod on a Thermo GasBench II peripheral device linkedvia continuous helium flow to either a ThermoDelta Plus or Delta V iso-tope ratio mass spectrometer (IRMS). Samples were equilibrated at25 °C. The precision (1σ) of the values (±0.1‰) was obtained frommultiple standards analyzed during each sample run. Each run alsocontained blanks and replicates. The standards used includedGreenlandIce Sheet Precipitation (GISP), Standard Light Antarctic Precipitation(SLAP), Vienna Standard Mean OceanWater (VSMOW) and/or the lab-oratory working standard AlabamaWater Isotope Standard (AWIS). Alldatawere normalized using both VSMOWand SLAP to bracket the sam-ples' δ18O values and the resulting 18Ow values were reported in partsper mil (‰) relative to VSMOW.

4. Results

4.1. Moche valley water sources

Moche watershed sample δ18O values are provided in Table 1. Thesedata diverge from the expected pattern of increasing enrichment (thusmore negative values) with increasing altitude, with a mean δ18O value

Fig. 2. Peruvian chicha de maize.

of−11.4‰ (SD= .2) in the lower valley,−10.4‰ (SD= .6) in themid-dle valley, and−9.9‰ (SD= .3) in the upper valley. Thus river water ischaracterized by aΔ18O of 1.5‰ across the 800m elevation change doc-umented in our samples. Springs sampled in the middle and upper val-ley have more variable δ18O values than do river samples, with a meanof−11.3‰ (SD= .5) in themiddle valley and amean of−9.2‰ (SD=1.0) in the upper valley. Canal δ18O values are quite similar to the valuesthat characterize the river that feeds them in the lower and middle val-ley (−11.6‰ and −10.2‰, respectively). But in the upper valley thedifference between mean δ18O river (−9.9‰) and canal (−9.4‰)values is greater.

4.2. Chicha de Maíz

Three pre and post-boil samples were taken from each of 10 differ-ent preparations. These preparations included two small vessel water-only runs, three large vessel water-only runs, two each of small andlarge vessel experimental chicha runs, and one Peruvian chicha run(Table 2). As these data show, boiling significantly enriched each runwith heavy oxygen (T-test p b .05). In our two small vessel, water-only runs, δ18O values increased from−8.3‰ to 0.5‰ and 0.9‰ respec-tively, for a mean Δ18Opre-post (enrichment) of 9‰ (Table 3). We con-ducted three large vessel, water-only preparations as we had difficultykeeping the water at a constant boil. The data reflect this challenge.The three large vessel preparations were characterized by beginningδ18O of −8.3‰, −8.1‰ and −8.6‰, while post-boil values rangefrom 8.1‰ to −2.8‰. Although the three pre-boil, mean δ18O valuesdo not differ significantly (either from one another or from any other

Fig. 4. Peruvian boil vessel.

Fig. 5. Peruvian fermentation vessel.

178 C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

pre-boil Staten Island δ18Omw), ANOVA identified the large vessel, post-boilmeans to be significantly different (p b .05). The Tukeypost-hoc testfound all three means to be significantly different from one another,thus there was not a clear outlier that we could exclude. Therefore weused the averaged values from all three runs to calculate the large vesselΔ18Opre-post of 7.9‰.

Our experimentally produced chicha de maíz also became enrichedin heavy oxygen during the brewing process (Table 2) with meanΔ18Opre-post of 7.4‰ in small vessels and 4.3‰ in large vessels(Table 3). Our difficulties with large vessel, water-only runs were not

Table 1Moche watershed δ18O values.

Sample location date Elevation (m.a.l.s.) Zone N

Lower Moche river7/14/2011 17 17L 77/25/2012 17 17L 77/14/2011 29 17L 77/25/2012 29 17L 7

Lower Moche canal8/1/2011 6 17L 77/14/2011 30 17L 7

Middle Moche river7/29/2011 172 17L 76/20/2012 309 17L 7

Middle Moche canal8/12/2012 174 17L 77/30/2011 545 17L 7

Middle Moche spring7/17/2012 252 17L 77/17/2012 258 17L 77/11/2012 315 17L 7

Upper Moche-Sinsicap-Cuesta river7/29/2011 691 17L 77/12/2012 886 17M 7

Upper Moche-Sinsicap-Cuesta canal7/29/2011 500 17M 77/22/2012 615 17M 77/29/2011 627 17L 77/29/2011 709 17L 77/28/2011 861 17M 7

Upper Moche-Sinsicap-Cuesta spring7/29/2011 620 17M 77/2/201 886 17M 77/28/2011 929 17M 7

repeated with our large vessel chicha brews, thus there was no signifi-cant difference between post-boil mean δ18O values of the two largevessel chichas (Table 2).

The ethnographically produced chicha de maíz was also enriched.The Campiña de Moche ground water used by our chichera in herbrew process was characterized by a mean δ18O of −11.9‰ (Table 3).This value is similar to other δ18Omwvalues from the lowerMoche valleyreported here (Table 1) and elsewhere (IAEA/WMO, 2015; Toyne et al.,2014). After the boiling process, the Peruvian produced chicha de maízbecame enriched by 5.7‰. Our Peruvian chicha was sampled both be-fore and after fermentation, but no significant difference in mean δ18Ovalues was identified (Table 2).

5. Discussion

The goals of this studywere tomore fully characterize δ18O values ofwater sources in the Moche valley and to assess the potential impact ofthe consumption of brewed beverages on δ18Op & c values of human hy-droxyapatite by focusing on chicha de maíz, an economically, politically,and socially important native brew of the Andes (Allen, 1988; Jennings,2005; Jennings and Bowser, 2009; Weismantel, 1988). To address ourfirst goal, 23 samples were collected from rivers, springs and canals intheMochewatershed.We identified a pattern of decreasing δ18O valueswith decreasing altitude, opposite of the pattern of enrichment thatgenerally characterizes rainfall, and opposite the pattern typically seenwith progressive evaporation. Given that a similar pattern is seen inspring water δ18O values, it seems likely that this unexpected spatialchange is the result of enriched spring water mixing with meteoricwater at higher elevations. As elevation falls springswithmore negativevalues feed the river and thus riverwater becomes increasingly negativeas it nears the ocean. The variability introduced by springwater is visiblein the increased variability in δ18O values (and thus higher standard de-viations) among samples collected in themiddle and upper valleys. Thesecond pattern seen in the Moche watershed samples is a close corre-spondence between river, spring and canal mean δ18O values within

orthing Easting δ18O (‰, VSMOW) Mean SD

−11.4 .218971 9099382 −11.118961 9099364 −11.620722 9101080 −11.520776 9101069 −11.4

−11.6 .210017 9102722 −11.420956 9100770 −11.7

−10.4 .634351 9106035 −9.937620 9112291 −10.8

−10.2 .434401 9106129 −10.435160 9108445 −9.9

−11.3 .537770 9109810 −11.837796 9110200 −11.037611 9112390 −11.0

−9.9 .347732 9113704 −9.742403 9123235 −10.1

−9.4 .94045 9116330 −9.141458 9117632 −8.245758 9113860 −10.747779 9114044 −9.741826 9122997 −9.2

−9.2 1.041503 9117690 −8.042401 9123242 −10.042407 9123253 −9.5

Table 2Results of stable isotopic analysis.

Sample type Run N Mean δ18O (‰, VSMOW) SD Mean Δ18Opre-post (‰, VSMOW)

Water testSmall pre-boil 1 3 −8.3 .06Small boiled 1 3 0.5 .12 8.8a

Small pre-boil 2 3 −8.3 .06Small boiled 2 3 0.9 .12 9.2a

Large pre-boil 3 3 −8.3 .49Large boiledb 3 3 8.1 .10 16.3a

Large pre-boil 4 3 −8.1 .49Large boiledb 4 3 −6.4 .25 5.8a

Large pre-boil 5 3 −8.6 .17Large boiledb 5 3 −2.8 .25 5.8a

Experimental ChichaSmall pre-boil 6 3 −8.4 .25Small chicha 6 3 −1.2 .55 7.2a

Small pre-boil 7 3 −8.6 .20Small chicha 7 3 −1.0 .06 7.6a

Large pre-boil 8 3 −8.6 .06Large chicha 8 3 −4.2 .21 4.4a

Large pre-boil 9 3 −8.4 .17Large chicha 9 5 −3.7 .15 4.7a

Ethnographic chichaMoche pre-boil 10 3 −11.9 .21Chicha pre-fermentation 10 3 −6.2 .12 5.7a

Chicha post-fermentation 10 3 −6.2 .26 5.7a

a Two-tailed T-test identifies the pre-boil and post-boil δ18O values are significantly different (p b .05).b ANOVA with Tukey's post hoc test identified significantly different means among the three large water boiled samples.

179C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

each section of the valley. The small canals sampled in theMochewater-shed are gravity fed, thus water moves quickly, leaving little opportu-nity for additional evaporation and thus limiting the enrichment ofcanal water.

To address our second goal,we prepared 10 experimental samples ofvarying recipes, volumes, and authenticity. We found that both smalland large vessel water-only runs became most enriched in heavy oxy-gen, but both experimentally and ethnographically produced chicha demaízwere also enriched. This is expected because of the greater affinityformolecules containing lighter isotopes to evaporate. Over time the re-sidual liquid will become comparatively enriched in heavy isotopes.With more evaporation, residual liquids will display more positiveδ18O values. Our experimentally produced, large vessel chicha was theleast enrichedwithΔ18Opre-post 4.3‰ (Table 2). This level of enrichmentis greater than the Δ18Opre-post 1.3‰ identified in traditionally brewedales (Brettell et al., 2012). And as noted, the δ18O value of meteoricwater sampled in the Moche watershed is spatially patterned with amean Δ18O of 1.4‰ from the upper to lower end of the valley(Table 1). A characterization of all sampled water sources in the water-shed identified a maximum spatial difference of 18O of 3.6‰, from themost negative middle valley spring (−11.8‰) to the most enrichedupper valley spring (−8.2‰). Thus, Δ18O of 4.3‰ in experimentallyproduced chicha de maíz suggests that if sufficient amounts were

Table 3Results of stable isotopic analysis, data pooled by sample type.

Sample type N Mean δ18O (‰, VS

Water TestsSmall pre-boil 6 −8.3Small boiled 6 0.7Large pre-boil 6 −8.3Large boiled 6 −0.4

Experimental ChichaSmall pre-boil 6 −8.5Small chicha 6 −1.1Large pre-boil 6 −8.4Large chicha 8 −4.1

Ethnographic chichaMoche pre-boil 3 −11.9Chicha post-fermentation 6 −6.2

consumed, it would be possible for its δ18O signature to obfuscate vari-ation in human δ18Obw values resulting from variability in watersources. Although controlled feeding experiments are required to esti-mate approximately what proportion of liquid must be consumed inthe form of chicha de maíz to substantially raise human δ18Op & c values,the ethnographic, ethnohistoric and archeological studies previouslydiscussed indicate that consumption was substantial among Andeanpopulations.

Two other patterns emerged from our data that potentially impactthe interpretation of human δ18Op & c values in bioarchaeologicalsamples. First, we observed that vessel size impacts the level ofheavy oxygen enrichment. In the smaller vessels there is a greatersurface area to volume ratio, which allows for an increased rate ofevaporation (and hence increased enrichment) compared to largervessels, which have a smaller surface area to volume ratio (Fig. 6).This suggests that the remains of individuals who consumed chichade maíz produced in small batches should be characterized by lessnegative δ18O values than the remains of individuals who consumedchicha de maíz produced in large batches. It may be that small scaleproduction of fermented beverages, such as occurs at the householdlevel, would have had a greater impact on human δ18Op & c valuesthan large-scale production, if the volume of chicha consumed wassimilar.

MOW) SD Mean Δ18Opre-post (‰, VSMOW)

.05

.21 9.0

.256.51 7.9

.22

.34 7.4

.15

.36 4.3

.21

.18 5.7

Fig. 6. Plot of average enrichment by surface area to volume ratio (cm2/l).

180 C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

Second, the presence of solublematerial (in this case jora) in the boilclearly affected the amount of evaporation. This can been seenwhenwecomparewater runs to chicha runs, and experimentally produced chichawith the ethnographically produced chicha. The experimental brewused ground jora which introduced greater amounts of solute to thebrew than did the jora entera used in the ethnographic chicha (Fig. 6).The addition of a solute to create a solution leads to a decrease invapor pressure compared to the pure solvent, and so fewer moleculesenter the vapor phase. This results in a reduced rate of evaporation forthe solution compared to the pure solvent (Atkins and de Paula,2010:164–165) and therefore reduced enrichment in heavy oxygen.The magnitude of the effect of solids is a product of the degree of starchgelatinization (the breakdown of a starch that allows it to be dissolved),which increases the viscosity of the solution and results in a decreasedrate of evaporation (Jenkins and Donald, 1998; Li et al., 2015). Thelevel of gelatinization depends on the type of starch, amount of water,solution pH, and other features of the recipe (Hans-Dieter Belitz et al.,2004). It is important to note that insoluble materials present duringthe boil will have negligible effects on rates of evaporation.

Only soluble material leads to gelatinization and once the boil iscomplete, un-dissolved solids and fermentation have no further effecton δ18O values. Therefore chichas that have been subject to differentlevels of upi straining and so vary in their thickness, or that are“ready” for consumption immediately or after many days of fermenta-tion and so vary in their alcohol content, will have the same δ18O signa-ture if their pre-boil recipe, boil vessel size and boil timewere the same.Conversely, δ18O signatures will vary among chichas that have differentpre-boil recipes, boil vessel sizes or boil times, even if post-boil strainingand fermentation practices create chichas of similar thicknesses and al-cohol contents. The complexity of these interacting variables can beseenwhen the surface area to volume ratio is compared to themean en-richment in our five different sample types (Fig. 6). As we did not antic-ipate the recipe effect, we did not measure changes in chicha viscosityduring our experimental or ethnographic brewing processes. An exper-imental protocol in which both vessel size and viscosity weremeasuredfor various brews would make it possible to distinguish the greater en-richment caused by small vessels sizes, greater boil times, or lessgelatinized brews. These issues are important to consider as diversechicha demaíz recipes and production vessel sizeswere specifically cho-sen in support (or contestation of) particular sociopolitical purposes, assuggested by contrasts in both chicha recipes and size of consumptiongroups (and thus presumably production vessels) identified amongpre-Incan (Goldstein, 2003; Laffe, 2015) and the Inca (Bray, 2009;Dillehay, 2003; Morris, 1979). Tracing the consumption of particular

chichas could provide us with interesting insights into the pre-Columbian use of brewed beverages.

6. Conclusions

The goals of this study were to more fully characterize the δ18Ovalues of Moche watershed water sources and to examine the possibleimpacts that consumption of brewed beverages could have on humanhydroxyapatite δ18O values. Water sampling provided a more completepicture ofMochewater sources and identified an important influence ofenriched spring water on meteoric water δ18O values. Using chicha demaíz as an example we have found that brewed beverages may signifi-cantly impact δ18O p & c values. We have also identified vessel size andrecipe as important factors that affect the amount of heavy oxygen en-richment that occurs during the brewing process. This research projectraises a number of issues that are important to consider wheninterpreting δ18O p & c values calculated from remains recovered fromAndean contexts, and others in which the consumption of brewed bev-erages may have been substantial. The data show that chicha de maízconsumption in the Andes has the potential to swamp spatial variationinwater source δ18O values, sowe cannot assume that the identificationof different δ18O values in the remains of Andean groups necessarily in-dicates the presence of migrants. Recognizing the complexities of track-ing population movement through δ18O values, many researchers havesuccessfully combined oxygen and strontium analyses (Buzon et al.,2011; Evans et al., 2006; Knudson 2008; Knudson et al., 2014; Turneret al., 2009). In addition, both the size of the production vessel and theingredients used will affect the final δ18O of chicha de maíz. Therefore,if we wish to model consumption patterns, bioarchaeological data arenot sufficient. Bringing together multiple indicators of mobility andchicha de maíz production and consumption (e.g., δ18O, δ13C, 87Sr/86Sr,oral health, paleoethnobotanical, and assemblage analyses) may allowus to better trace consumption patterns of chicha de maíz and thus bet-ter reconstruct its ritual, economic, and sociopolitical importance in theAndes.

Acknowledgments

Support for this research was provided by Wagner College Anony-mous Donor Grant and Student Support and Faculty Research Grantand Wagner College Faculty Research Grant. We are grateful to thechichera of La Jarrita for graciously talking to us and allowing us to ob-serve her brewing process and sample her ingredients. We would liketo thank Brandi Adduce for her help brewing in the Wagner chemistrylab and Alicia Boswell, Violeta Capric, Sophia Fox-Sowell, NicholasGibaldi, and Rose Tobiassen for their help in collecting and transportingMoche valley water samples and Joe Lambert in the Alabama Stable Iso-tope Laboratory for isotope analyses. Finally, we recognize our peer re-viewers for their thoughtful comments.

References

Allen, C.J., 1988. The Hold Life Has: Coca and Cultural Identity in an Andean Community.Smithsonian Institution Press, Washington.

Andrus, C. Fred T. 2011. Shell midden sclerochronology. Quat. Sci. Rev. 30:2892–2905.Aravena, R., Suzuki, O., Peña, H., Pollastri, A., Fuenzalida, H., Grilli, A., 1999. Isotopic com-

position and origin of the precipitation in Northern Chile. Appl. Geochem. 14,411–422.

Atkins, P., de Paula, J., 2010. Physical Chemistry. Ninth Edition. W.H. Freeman and Com-pany, New York.

Bonavia, D., 2013. Maize: Origin, Domestication, and its Role in the Development of Cul-ture. Cambridge University Press, New York.

Bourrel, L., Rau, P., Dewitte, B., Labat, D., Lavado, W., Coutaud, A., Vera, A., Alvarado, A.,Ordoñez, J., 2015. Low-frequency modulation and trend of the relationship betweenENSO and precipitation along the northern to centre Peruvian Pacific coast. Hydrol.Process. 29, 1252–1266. http://dx.doi.org/10.1002/hyp10247.

Bray, T.L., 2003. To dine splendidly: imperial pottery, commensal politics, and the Incastate. In: Bray, T.L. (Ed.), The Archaeology and politics of Food and Feasting in EarlyStates and Empires. Kluwer Academic/Plenum Publisher, New York, pp. 93–142.

181C.M. Gagnon et al. / Journal of Archaeological Science: Reports 4 (2015) 174–181

Bray, T.L., 2009. The role of chicha in Inca state expansion: a distributional study of IncaAríbalos. Drink, Power, and Society in the Andes, Jennings, Justin and Brenda J.Bower, eds. University of Florida Press, Tallahassee, pp. 108–132.

Brettell, R., Montgomery, J., Evans, J., 2012. Brewing and stewing: the effect of culturallymediated behaviour on the oxygen isotope composition of ingested fluids and theimplication for human provenance studies. J. Anal. At. Spectrom. 27, 778–785.

Buzon, M.R., Conlee, C.A., Bowen, G.J., 2011. Refining oxygen isotope analysis in the Nascaregion of Peru: an investigation of water sources and archaeological samples. Int.J. Osteoarchaeol 21, 446–455.

Carter, J.F., Yates, H.S.A., Tinggi, U., 2015. A global survey of the stable isotope and chem-ical compositions of bottled and canned beers as a guide to authenticity. Sci. Justice55, 18–26.

Cobo, Bernabe. 1979. History of the Inca Empire. Roland Hamilton translator and ed. Aus-tin: University of Texas Press [1653].

Costin, C.L., Earle, T., 1989. Status distinction and legitimation of power as reflected inchanging patterns of consumption in late Prehispanic Peru. Am. Antiq. 54 (4),691–714.

Culter, H.C., Cardenas, M., 1947. Chicha, a native South American beer. Bot. Mus. Leafl.Harv. Univ. 13, 33–60.

de Betanzos, Juan. 1996. Narrative of the Incas. Roland Hamilton and Dana Buchanantranslator and ed. Austin: University of Texas Press.

Dillehay, T.D., 2003. El conlonialismo Inka, el consumo de chicha y los festines desde unaperspectiva de banquetes políticos. Boletín de Arqueología PUCP 7, 355–363 [1557].

Evans, J.A., Chenery, C.A., Fitzxpatrick, A.P., 2006. Bronze Age childhood migration of indi-viduals near Stonehenge, revealed by strontium and oxygen isotope tooth enamelanalysis. Archaeometry 48, 309–321.

Fricke, H.C., O'Neil, J.R., Lynnerup, N., 1995. Oxygen isotope composition of human toothenamel from medieval Greenland: linking climate and society. Geology 23, 869–872.

Gilboa, Y., 1971. Replenishment sources of the alluvial aquifers of the Peruvian coast.Groundwater 9, 39–46.

Gillin, J.R., 1947. Moche, A Peruvian Coastal Community. Smithsonian Institution, Instituteof Social Anthropology, Washington.

Goldstein, D.J., 2003. From stew-eaters to maize-drinkers: the Chicha economy and theTiwanaku expansion. In: Bray, T.L. (Ed.), The Archaeology and Politics of Food andFeasting in Early States and Empires,. Kluwer Academic/Plenum Publisher, NewYork, pp. 143–172.

Goldstein, D.J., Coleman Goldstein, R.C., Williams, P.R., 2009. You arewhat you drink: a so-ciocultural reconstruction of Pre-Hispanic fermented beverage use at Cerro Baúl,Moquegua, Peru. In: Jennings, J., Bowser, B.J. (Eds.), Drink, Power, and Society in theAndes. University of Florida Press, Tallahassee, pp. 133–166.

Belitz, H.-D., Grosch, W., Schieberle, P., 2004. Food Chemistry. Springer, New York.Hastorf, C.A., 1990. The effect of the Inka state on Sausa agricultural production and crop

consumption. Am. Antiq. 55 (2), 262–290.Hastorf, C.A., 1991. Gender, space, and food in prehistory. In: Gero, J.M., Conkey, M.W.

(Eds.), Engendering Archaeology. Blackwell, Oxford, pp. 132–159.Hastorf, C.A., Johannessen, S., 1993. Pre-Hispanic political change and the role of maize in

the central Andes of Peru. Am. Anthropol. 95 (10), 115–138.Hayashida, F.M., 2008. Ancient beer and modern brewers: ethnoarchaeological observa-

tions of chicha production in two regions of the North Coast of Peru. J. Anthropol.Archaeol. 27, 161–174.

Hedges, R.E.M., Clement, J.G., David, C., Thomas, L., O'Connell, T.C., 2007. Collagen turnoverin the adult femoral mid-shaft: modeled from anthropogenic radiocarbon tracermeasurements. Am. J. Phys. Anthropol. 133, 808–816.

IAEA/WMO. 2015. Global Network of Isotopes in Precipitation. The GNIP Database. Acces-sible at: http://www.iaea.org/water.

Jenkins, P.J., Donald, A.M., 1998. Gelatinisation of starch: a combined SAXS/WAXS/DSCand SANS study. Carbohydr. Res. 308, 133–147.

Jennings, J., 2005. La chichera y el patrón: chicha and the energetics of feasting in the pre-historic Andes. In: Conlee, C.A., Ogburn, D., Vaughn, K. (Eds.), Foundations of Power inthe Prehispanic Andes, pp. 241–259.

Jennings, J., Bowser, B.J. (Eds.), 2009. Drink, Power, and Society in the Andes. University ofFlorida Press, Tallahassee.

Katzenberg, M. Anne. 2008. Stable isotope analysis: a tool for studying past diet, demog-raphy, and life history. In Biological Anthropology of the Human Skeleton, SecondEdition, M. Anne Katzenberg and Shelley R. Saunders, eds., 411–441. Hoboken:Wiley-Liss.

Kendall, C., Coplen, T.B., 2001. Distribution of oxygen-18 and deuterium in river watersacross the United States. Hydrol. Process. 15 (7), 1363–1393.

Knudson, K.J., 2009. Oxygen isotope analysis in a land of environmental extremes: thecomplexities of isotopic work in the Andes. Int. J. Osteoarchaeol. 19, 171–191.

Knudson, K.J., Williams, S.R., Osborn, R., Forgey, K., Williams, P.R., 2009. The geographicorigins of Nasca trophy heads using strontium, oxygen, and carbon isotope data.J. Anthropol. Archaeol. 28, 244–257.

Knudson, K.J., Goldstein, P.S., Dahlstedt, A., Somerville, A., Schoeninger, M.J., 2014.Paleomobility in the Tiwanaku diaspora: biogeochemical analyses at Rio Muerto,Moquegua, Peru. Am. J. Phys. Anthropol. 155, 405–421.

Laffe, A.O., 2015. The preliminary residues of Power: women, chicha, and agency in theMiddle Horizon Andes c. AD 600–1100. In: Druc, I.C. (Ed.), Ceramic Analysis in theAndes. Deep University Press, Wisconsin, pp. 83–101.

Logan, A.L., Hastorf, C.A., Pearsall, D.H., 2012. “Let's drink together”: early ceremonial useof maize in the Titicaca Basin. Lat. Am. Antiq. 23 (3), 235–258.

Longinelli, A., 1984. Oxygen isotopes in mammal bone phosphate: a new tool forpaleohydrological and paleoclimatological research? Geochim. Cosmochim. Acta 48,385–390.

Longinelli, A., Peretti Paladino, A., 1980. Oxygen isotope composition of water frommam-mal blood: first results. Mass Spectrometry in Biochemical Medical and Environmen-tal Research 1, 135–139.

Li, Z., W. Liu, Z. Gu, C. Li, Y. Hong and L. Cheng, L. 2015. The effect of starch concentrationon the gelatinization and liquefaction of corn starch. Food Hydrocoll., 48:189–196.

Makarewicz, C.A., Sealy, J., 2015. Dietary reconstruction, mobility, and the analysis of an-cient skeletal tissues: expanding the prospects of stable isotope research in archaeol-ogy. J. Archaeol. Sci. 56, 146–158.

McCool, W., Parker, B., 2014. Household maize beer production in the Andes: anethnoarchaeological investigation. Paper Presented at the Society for American Ar-chaeology 79th Annual Meeting, Austin, Texas.

Morre, J.D., 1989. Pre-Hispanic beer in coastal Peru: technology and social context of pre-historic production. Am. Anthropol. 91, 682–695.

Morris, C., 1979. Maize beer in the economics, politics, and religion of the Inca Empire. In:Gastineau, C.F., Darby, W.J., Turner, T.B. (Eds.), Fermented Food Beverages ion Nutri-tion. Academic Press, New York, pp. 21–34.

Murra, 1960. Rite and crop in the Inca state. In: Diamond, S. (Ed.), Culture in History. Co-lumbia University Press, New York, pp. 393–407.

Netherly, P., 1977. Local Level Lords on the North Coast of Peru. University Microfilms,Ann Arbor.

Nicholson, E.G., 1960. Chicha maize types and chicha manufacture in Peru. Econ. Bot. 14,290–299.

Ohlanders, N., Rodriguez, M., McPhee, J., 2013. Stable water isotope variation in a CentralAndean watershed dominated by glacier and snowmelt. Hydrol. Earth Syst. Sci. 17,1035–1050.

Rostworowski de Diez Canseco, M., 1977. Etnia y Sociedad: Costa Peruana Prehispanica.Instituto de Estudios Peruanos, Lima.

Sayre, M., Goldstein, D., Whitehead,W.,Williams, P.R., 2012. Amarked preference: Chichade molle and Huari state consumption practices. Ñawpa Pacha 32 (2), 231–258.

Scherer, A.K., de Carteret, A., Newman, S., 2014. Local water resource variability and oxy-gen isotopic reconstructions of mobility: a case study from the Maya area. Journal ofArchaeological Sciences (Reports online).

Shimada, I., 1994. Pampa Grande and the Mochica Culture. University of Texas Press,Austin.

Sponheimer, M., Lee-Thorp, J.A., 1999. Oxygen isotopes in enamel carbonate and theirecological significance. J. Archaeol. Sci. 26, 723–728.

Topic, J.R., 1990. Craft production in the Kingdom of Chimor. In: Mosely, M.E., Cordy-Collins, A. (Eds.), The Northern Dynasties: Kingship and Statecraft in Chimor. Dum-barton Oaks, Washington, pp. 145–176.

Toyne, M.J., White, C.D., Verano, J.W., Castillo, S.U., Longstaffe, F.J., Millaire, J.F., 2014. Res-idential histories of elites and sacrificial victims at Huacas de Moche, Peru, as recon-structed from oxygen isotopes. J. Archaeol. Sci. 42, 15–28.

Turner, B.L., Kamenov, G.D., Kingston, J.D., Armelagos, G.J., 2009. J. Archaeol. Sci. 36,317–332.

Ubelaker, D.H., Katzenberg, M.A., 1995. Status and diet in precontact highland Ecuador.Am. J. Phys. Anthropol. 97, 403–411.

Ucede Castillo, S., 2010. Los contextos urbanos de producción artesanal en el complejoarqueológico de las huacas del Sol y de la Luna. Bulletin de I'Institut Français d'ÉtudesAndines 39 (2), 243–297.

Valdez, L.M., 2006. Maize beer production in Middle Horizon Peru. J. Anthropol. Res. 62(1), 53–80.

Valdez, L.M., Bettcher, K.J., Ernesto Valdez, J., 2010. Production of maize beer at aWari sitein the Ayachucho valley, Peru. Arqueologia Iberoamericana 5, 23–35.

Webb, E.C., White, C.D., Longstaffe, F.J., 2013. Exploring geographic origins at Cahuachiusing stable isotopic analysis of archaeological human tissues and modern environ-mental waters. Int. J. Osteoarchaeol. 23, 698–715.

Weismantel, M.J., 1988. Food, Gender, and Poverty in the Ecuadorian Andes. WavelandPress, Prospect Heights.

White, C.D., Spence, M.W., Stuart-Williams, H.L.Q., Schwarcz, H.P., 1998. Oxygen isotopesand the identification of geographic origins: the valley of Oaxaca verse the valley ofMexico. J. Archaeol. Sci. 25, 643–655.