re

da deo, Cd

diffate0). S, Quariaann of

tative parts (trunk and branches), genera and species with different life forms (trees, shrubs, herbaceous

2008 Elsevier Ltd. All rights reserved.

the Neologixico dof 20 kcity unee its

terns of exploitationof diverse species andconsequent impacton theenvironment (Salisbury and Jane, 1940; Godwin and Tansley, 1941;Minnis, 1978; Vernet, 1980; Kohler et al., 1984; Miller, 1985; Smartand Hoffman, 1988; Johannessen and Hastorf, 1990; Heinz, 1991;Newsom, 1993; Asouti and Hather, 2001; Scheel-Ybert et al., 2003;Rodrguez, 2004). However, this type of botanical evidence has re-ceived little attention in paleoethnobotanical research in Mexico,

was to contribute evidence for reconstructing the vegetation of theTeotihuacan Valley during the Classic period, when the landscapewas dominated by the powerful city of Teotihuacan. Identied taxawere associated with temperate forest (Pinus, Quercus, Alnus, Bac-charis, Buddleia), xerophytic scrub (Agave) and riparian vegetation(Fraxinus, Salix, Cornus). These genera suggested the existence ofdistinct vegetation types, of which a mixture of pine and oakcomprised the main component of the valleys landscape prior tohuman settlement. This evidence led Adriano-Moran to propose

Contents lists availab

e

ww

Journal of Archaeological Science 35 (2008) 29272936* Corresponding author. Tel.: 52 5556229503; fax: 52 5556229651.Tapia, 1990). In addition to an external supply network, the citydepended on the management of local resources to cover the basicneeds of its inhabitants. Wood resources must have played an im-portant role in the citys growth. However, little is known about theuse and management of wood and the impact this had on the en-vironment and human population.

Charcoal recovered from archaeological excavations providesdata for the reconstruction of past vegetation as well as the use offorest resources, including aspects of management, changes in pat-

studies where the identication of seeds was the main focus and isonly provided as complementary information. In such publications,only identied taxa are listed and inferences and interpretationsabout their uses are very general (Kovar,1970; Ford,1976; Gonzalez,1982; Alvarez del Castillo, 1984; Manzanilla, 1985; Flannery, 1986;Gonzalez et al., 1993; Popper, 1995; Montufar, 1996; Montufar,1999).

The rst systematic effort to explore the potential of charcoal asevidence was undertaken by Adriano-Moran (2000). The objectiveTreesShrubs

1. Introduction

The Teotihuacan Valley, located inMexico, is best known for the archathe most important city in central Me(ca. A.D. 1650). With a surface areainhabitants (Millon, 1970, 1973), theefcient supply network to guarantE-mail address: adrianoc65@gmail.com (C.C. Adria

0305-4403/$ see front matter 2008 Elsevier Ltd.doi:10.1016/j.jas.2008.06.001E sector of the Basin ofcal site of Teotihuacan,uring the Classic periodm2 and around 100,000doubtedly relied on anoperation (McClung de

particularly in the Teotihuacan Valley. As a result there is a consid-erable shortage of studies that systematize and analyze informationnecessary to propose hypotheses concerning relationships betweenhumans and their environment, particularly in forest ecosystems.

Until the end of the 1990s, studies did not concentrate on thistype of material and the little evidence available was either un-published or briey incorporated into analyses of seed remains.Published information on charcoal analyses comes mainly fromManagementExploitationthe inhabitants of the Teotihuacan Valley implemented practices to manage vegetation in order to assureresource availability in the region.WoodVegetation plants) and primary (Pinus and Quercus) as well as secondary taxa (Prunus and Arbutus). We propose thatTrees and shrubs: the use of wood in p

Carmen Cristina Adriano-Moran a,*, Emily McClunga Posgrado en Ciencias Biologicas, Facultad de Ciencias, Universidad Nacional Autonomb Instituto de Investigaciones Antropologicas, Universidad Nacional Autonoma de Mexic

a r t i c l e i n f o

Article history:Received 21 September 2007Received in revised form 2 June 2008Accepted 6 June 2008

Keywords:Teotihuacan regionCharcoal

a b s t r a c t

Charcoal recovered fromMexico was used to evalu(approx. 400 B.C.A.D. 150identied (including PinusBuddleia, among others). Vtensity of use rather thancontinuity in the utilizatio

Journal of Archa

journal homepage: ht tp: / /no-Moran).

All rights reserved.hispanic Teotihuacan

e Tapia b

Mexico, Cd. Universitaria, 04510 Mexico D.F., Mexico. Universitaria, 04510 Mexico D.F., Mexico

erent archaeological excavations carried out in the Teotihuacan Valley,whether signicant changes in the use of wood took place through timeixteen taxa of trees and shrubs characteristic of the Basin of Mexico wereercus, Arbutus, Cupressaceae, Alnus, Prunus, Taxodium, Salix, Baccharis andtions in the proportions of taxa are interpreted as a reection of the in-indicator of deforestation. The analysis of the data shows a pattern oftaxa throughout the occupation. This includes the use of different vege-

le at ScienceDirect

ological Science

w.elsevier .com/locate/ jasthat the utilization of different types of wood could be a function of

springs in the SW sector of the valley (Lorenzo, 1968; Mooser,1968). Principal soil types in the region include phaeozems (40%),

1998; Scheel-Ybert, 2000; Asouti, 2003). In other cases, contin-gency tables and four-way log-linear analysis have been employed

al overtisols (16%), cambisols (13%) and leptosols (13.5%), many ofwhich present an advanced state of erosion (McClung de Tapia andTapia-Recillas, 1997).

Six main types of vegetation currently predominate in the re-gion (Castilla and Tejero, 1987). The oak forest (Quercus crassipes,Quercus greggii and Quercus mexicana) community covers a smallarea restricted to the north slope of Cerro Gordo (28003050 masl)but may have extended over a broader area in the past, at least asfar as the area covered today by oak scrub. Oak scrub (Quercusmicrophylla) is situated between the oak forest and xerophyticscrub (28503000 masl). It has been suggested that the oak scrubcommunity replaces the pineoak component of the temperateforests in the Basin of Mexico as a result of repeated res (Rze-dowski et al., 1964). However, Castilla and Tejero (1987) considerthat its presence in the Teotihuacan region may be a consequenceof cutting the oak forest. Xerophytic scrub (Opuntia streptacantha,Zaluzania augusta, Mimosa biuncifera) is the most representativevegetation type of the region and it occupies the largest area(23002750 masl). Grasslands (Buchloe dactyloides, Hilaria cen-chroides, Bouteloua gracilis) cover a small part of the area and aremostly interdigitated with xerophytic scrub, oak scrub or scrubcomprised of Senecio salignus and Baccharis conferta (associationswithin the oak scrub community). Hydrophylous vegetation iscomprised of seasonal taxa associated with the rainy season ortheir availability in the environment and that, apparently, therewasa preference for certain genera over others. This study showed taxathat existed in the region and were used during the Teotihuacanperiod. Furthermore, it provided an important point of departure tocompare with present day environmental conditions in the region.

Based on this information, further investigation purported toreconstruct the patterns of use and exploitation of wood on the partof the valleys inhabitants from the Preclassic through the Post-classic periods (ca. 400 B.C.A.D. 1500). An understanding of theway inwhich these activities were carried out is important becausedeforestation, as a consequence of poor management of the land-scape and its resources, has been cited by some authors (Sanders,1965) as the cause of environmental deterioration in the region(soil loss and erosion, changes in hydrographic patterns) that couldhave been one of the factors leading to the fall of Teotihuacan.

In this study, statistical analyses are used to determine if sig-nicant differences are apparent in the proportions of taxa throughtime, in order to evaluate the evidence for changes in the use ofwood between 400 B.C.A.D. 1500. The long and continuous oc-cupation of the Teotihuacan region, added to the apparent changesin its demographic history and settlement patterns through thecourse of its prehispanic history, make this an ideal region for sucha study.

2. Study area

The Teotihuacan Valley is located in the NE sector of the Basin ofMexico (19

340 N, 99

400 W) between 2240 and 3100 masl (Fig. 1).

Climate is semi-arid on the plain and temperate subhumid in thesurrounding elevations. Because it is situated in a transition zonebetween dry and subhumid climate types, the region is highlysusceptible to climatic variations. Average annual temperature is14.9 C, and varies between 12 C and 18 C below 2800 masl andbetween 5 C and 12 C above 2800 masl; mean annual pre-cipitation for the region is 563.3 mm (Garca, 1968, 1988). The maindrainages are Rio San Juan, El Huixulco and Rio San Lorenzo, whichare largely seasonal and which inltrate the subsoil to reappear as

C.C. Adriano-Moran, E. McClung de Tapia / Journ2928articial systems such as dams, irrigation ditches and water de-posits. Natural stands occur in inundated areas. Anthropogenicto explore categorical variables (Kohler and Matthews, 1988).Studies in which signicance testing has been used include

techniques such as ANOVA and regression (Lepofsky et al., 1996;Johannessen, 1988). However, parametric statistical approachesrequire that data be normally distributed or close to it. Other sup-positions of the model include independence and homoscedastic-ity. Archaeobotanical data are not normally distributed, andfrequently a large number of samples contain a small number oftaxa or small numbers of samples contain numerous taxa. Signi-cance tests cannot be applied in this type of situation, and othermethods or techniques that do not require a normal distributionmust be employed.

Generalized Linear Models were developed to analyze bothdiscrete and continuous data that do not comply with the normalmodel, while recognizing that many of the properties of normaldistribution are found in the so-called exponential family, which, inaddition to the normal distribution, includes the binomial, negativevegetation includes secondary associations of agrestal and ruderalspecies that develop in areas of human impact.

The conformation of present day vegetation communities in theTeotihuacan Valley is undoubtedly a product of the combined ac-tion of natural and anthropogenic factors over time. Castilla andTejero (1983) suggest that vegetation types in the past were pos-sibly quite similar to the present although distributed morebroadly. This is based on the idea that if past climatic conditionswere similar to the present, climax communities (such as forest orscrub in temperate subhumid or semi-arid climates) should be thesame. Evidence from recent studies (McClung de Tapia and Tapia-Recillas, 1997; McClung de Tapia et al., 1998, 2003; Adriano-Moran,2000) supports the presence of these communities in the valleytogether with changes in their distribution as well as their oristiccomposition. The most noticeable change is the disappearance ofPinus from the present day ora in the region.

3. Numerical analysis of botanical materials

Archaeobotanical and paleoethnobotanical investigations com-monly use ubiquity to describe and analyze data. Ubiquity isa semiquantitative form of describing the data (Jones, 1991), de-ned as the number of samples of a collection in which a particulartaxon is found, expressed as a percentage (relative frequency).Ubiquities are compared in graphs or tables that allow viewers toobserve tendencies in the behavior of the taxa among contexts orthrough time. Often ubiquities vary noticeably from one context ortime period to another that frequently are interpreted as signicantdifferences, although without the application of statistical tests. Anumber of examples of this approach (Willcox, 1974; Minnis, 1978;Hastorf, 1988; February, 1992; Figueiral, 1993), where samples thathave been grouped by contexts or periods are compared numeri-cally without statistical testing, produce partial or even potentiallyerroneous conclusions. However, the paucity of statistical analysesin paleoethnobotanical studies, including signicance testing, re-ects the researchers awareness of random or nonrandom factorsthat affect the conservation of botanical remains so that they arenot populations representative of some larger universe and, assuch, are not considered appropriate subjects for formal statisticalanalysis.

In recent years, the use of multivariate techniques for data ex-ploration has increased, including correspondence analysis, prin-cipal components, cluster analysis and discriminant analysis (Priorand Price-Williams, 1985; Jones, 1987, 1991; Heinz and Thiebault,

f Archaeological Science 35 (2008) 29272936binomial, gamma and Poisson distributions. By means of thesemodels, equations can be developed that permit estimates of the

nal oC.C. Adriano-Moran, E. McClung de Tapia / Jourbehavior of the response variable as a result of one or more ex-planatory variables (Crawley, 1996).

The logit model is employed for the analysis of data expressed asproportions in a binomial distribution. Since ubiquity is repre-sented as a fraction, it should follow a binomial distribution. Themain advantage of this model is that it does not require that thedata be normally distributed or that the groups of samples to becompared be of the same size (McCullag and Nelder, 1989; Crawley,1996).

4. Method of analysis

A total of 2613 charcoal fragments, from 297 samples of charcoalrecovered fromvarious contexts in different excavations, were usedin this study. Based on their association with specic contexts(oors, ll, hearths and various), it is assumed that the charcoalfragments represent wood used as fuel. The forms and sizes of thefragments indicate that branches and twigs as well as trunks wereused.

The time frame represented by the samples is based on theceramic chronology previously established for the Teotihuacan re-gion by stratigraphic relationships and radiocarbon dating ofcharcoal associated with diagnostic ceramic types (Cowgill, 1996;Rattray, 2001). In this study the occupation phases dened by ar-chaeologists (Cowgill, 1996) have been grouped into periods toform broader more manageable and comparable time units. Thus,the Preclassic period (ca. 400100 B.C.) is considered as a unit, theClassic period is divided into Early (A.D. 0250), Middle (A.D. 250

Fig. 1. Location of the Teotihuacan Valley, including the Archaeological Zone (ancient city of Archaeological Science 35 (2008) 29272936 2929550) and Late (A.D. 550650), and the Postclassic period (A.D. 6501500) is also treated as a single unit. The approximate duration ofthese periods is based on Cowgill (1996). However, few excavationscorresponding to the Postclassic have been carried out in the studyregion and available charcoal was insufcient to permit a nerchronological division.

Sixteen taxa of trees and shrubs characteristic of the Basin ofMexico, together with monocotyledons (Agave and Zea mays)(Table 1, Fig. 3), were identied in the samples which come fromdiverse archaeological excavations undertaken in the region overthe past 30 years (Figs. 1 and 2). Ubiquity analysis shows importantuctuations in the percentages of different genera for contexts aswell as time periods, possibly indicating changes in their uses.

In order to build the database for the analyses, numerical ref-erences were assigned to four contexts, ve periods and 14 taxa.The context referred to as various includes samples that wererecovered from ll, but directly associated with burials, offeringsand storage features. Taxa that were present in a single sample orthat could not be identied to the genus form the category referredto as Others. Fabaceae and Cupressaceae are represented asfamilies due to the difculty in identifying the genus of some of thesamples. Z. mays was included because, although it is not a woodyplant it occurs frequently among the charcoal samples and is con-sidered an important fuel resource, as are other monocotyledons.

The statistical program GLIM 4.0 (Crawley, 1996) was employedfor the analysis. One of the characteristics of an interactive pro-cedure is that following the adjustment of a model that includes allof the variables and their interactions, the insignicant terms are

f Teotihuacan) and the sites of Cuanalan and Otumba, outside of the urban perimeter.

4

al oTable 1Identied taxa by excavation

Excavationa 1 2 3

Taxa

AgavaceaeAgave sp. AsteraceaeBaccharis spp. Cf. Senecio salignus DC. BetulaceaeAlnus spp. Cupressaceae Juniperus cf. deppeana Steud. Cupressus cf. lindleyi Klotzsch.EricaceaeArbutus cf. xalapensis H.B.K. Fabaceae Eysenhardtia polystachya (Ort.) Sarg. Mimosa cf. aculeaticarpa var. biuncifera (Benth) BarnaberyProsopis laevigata (Willd.) M.C. Johnst. FagaceaeQuercus spp. LoganiaceaeBuddleia cf. parviora H.B.K. OnagraceaeFuchsia cf. arborescens Sims. PinaceaePinus spp. PoaceaeZea mays L. Rosaceae

C.C. Adriano-Moran, E. McClung de Tapia / Journ2930eliminated as the model is simplied. This procedure allows thetesting of signicance of each of the variables in the presence of theothers, since the group of variables may explain patterns that in-dividual variables cannot. All of the contexts, periods and taxawerecompared with each other to establish which are signicantly dif-ferent in terms of the wood used. Because the fraction of samplesfrom each group that was analyzed showed the presence of eachtaxon, a logit transformation of the data, a binomial error and a c2

signicance test (a 95%) were employed.

5. Results

Comparison of all the contexts, time periods and taxa showedthat many of the components are signicantly different. However,only the values of total signicance of each variable and its in-teractions are presented in Table 2. In the section dealing with theinteractions themselves, the probabilities of signicant variablesare considered.

With respect to the archaeological contexts under consider-ation, c2 2.63 (Table 2), indicating that there is no signicantdifference among hearths, oors and ll and, as such, these may beconsidered as a single context. Thus, only two contexts differ fromeach other: (1) hearths/oors/lls and (2) various. The homoge-neity observed in the majority of the contexts is responsible for thefact that context as a variable explains only a small proportion ofvariation among the data.

With respect to the time periods considered, only Middle Classicand Late Classic showed no signicant difference. Therefore, thesewere combined into a single period, resulting in the following

Prunus serotina ssp. capuli (Cav.) McVaugh SalicaceaeSalix spp. TaxodiaceaeTaxodium mucronatum Ten. Monocotyledons Not identied

a Excavations: 1. Cuanalan (Manzanilla,1985), 2. Sun Pyramid (Millon,1973), 3. Moon PyOaxaca Barrio (Millon,1973), 7. Teopancazco (Millon, 1973; Manzanilla, 2003), 8. Tetitla (M1973), 12. TE 2 (Millon, 1973), 13. TE 11 (Millon, 1973), 14. Tunnels (Manzanilla et al., 1995 6 7 8 9 10 11 12 13 14 15

f Archaeological Science 35 (2008) 29272936temporal divisions: (1) Preclassic; (2) Early Classic; (3) MiddleLateClassic; and (4) Postclassic. Once this adjustment was carried out,a c2 of 63.36was obtained (Table 2). The time periods alone accountfor 5.44% of the total variation.

Insofar as the taxa are concerned, Pinus, Quercus, Cupressaceae,Z. mays, Monocotyledons and Others were distinct from eachother and from the rest of the taxa. The remaining taxa formed twogroups: Group 1 included Taxodium, Arbutus, Baccharis and Prunus,and Group 2 included Alnus, Buddleia, Fabaceae and Salix. Thus, taxawere reduced from 14 to 8: (1) Pinus; (2) Quercus; (3) Cupressaceae;(4) TaxodiumArbutusBaccharisPrunus; (5) AlnusSalixBuddleiaFabaceae; (6) Z. mays; (7) Monocotyledons; and (8) Others.Following this re-arrangement of taxa, all of the groups were sig-nicantly different from one another (c2 914.9).

Group 1 (TaxodiumArbutusBaccharisPrunus) is comprised ofgenera that are present during all of the periods (Taxodium andBaccharis) or during at least four of them (Arbutus and Prunus). Allshow similar behavior through time in that their proportions in-crease or decrease in the same period, thus they behave as a unit. Inthe second group (AlnusSalixBuddleiaFabaceae), most of thetaxa appear in three of the time periods (Alnus, Salix and Fabaceae)or in at least two (Buddleia), and all have low presence.

Taxa that are signicantly different are present in all of the timeperiods and generally show the highest percentages of ubiquity.Probably for this reason, taxa alone account for 78.66% of the var-iation evident in the data.

Based on the previous results, the interactions taxa vs. contextand taxa vs. period were evaluated, resulting in highly signicant c2

values (Table 2).

ramid (Sugiyama, 2004), 4. TE 5 (Millon,1973), 5. Oztoyahualco (Manzanilla,1993), 6.illon, 1973), 9. Tepantitla (Millon, 1973), 10. Yayahuala (Millon, 1973), 11. TE 1 (Millon,6), 15. Otumba (Charlton et al., 2000). See Figs. 1 and 2.

n fro

nal oFig. 2. Excavations located within the perimeter of the ancient city of Teotihuaca

C.C. Adriano-Moran, E. McClung de Tapia / Jour5.1. Variation of taxa among contexts

The next step in the procedure was to evaluate the behavior ofthe taxa in Contexts 1 and 2. The genera that show signicantdifferences between contexts are: Pinus (c27.931, p 0.0049),Quercus (c213.54, p 0.0002) and Group 1 (c2 7.62,p 0.0058). The observed differences are due to the fact that inContext 1, these genera appear constantly in high proportionsthroughout the chronological sequence. In Context 2, Pinus occursin a lower proportion of the samples and Quercus in a higher pro-portion, whereas the taxa that comprise Group 1 are practicallyabsent. In this case, only Prunuswas present (in low proportion andin a single period) (Fig. 4).

5.2. Variation among taxa through time

The last step of the procedure involved exploring variationamong taxa during each of the time periods. Cupressaceae, Group 2(AlnusSalixBuddleiaFabaceae) and Monocotyledons showed nosignicant change between periods, whereas Pinus, Quercus, Group1 (TaxodiumArbutusBaccharisPrunus), Z. mays and Others didchange signicantly.

Once it was established which taxa varied, the periods duringwhich variation took place were determined (Fig. 5). Pinus changedin periods 1, 2 (c2 3.951, p 0.0468) and 3 (c210.21,p 0.0014). For Quercus, on the other hand, variation was observedbetween periods 3 and 4 (c2 5.611, p 0.0178). For Group 1, allfour periods are different (periods 12, c2 28.26, p 0.0001;periods 23, c2 49.67, p 0.0001; periods 34, c2 6.199,p 0.0128). Z. mays varied between periods 2 and 3 (c213.54,p 0.0002) and extending to period 4 (c215.8, p 0.0001). Thechange in Others took place between periods 2 and 3 (c218.83,p 0.0001).m which charcoal samples were obtained for this study (based on Millon, 1973).

f Archaeological Science 35 (2008) 29272936 29316. Discussion

6.1. Variations in contexts, time periods and taxa

The fact that oors, lls and hearths were not signicantly dif-ferent from each other was evident from the ubiquity analysis (notshown), in which the composition of taxa in these different con-texts is essentially the same although their proportions differ. Thisconrms the idea that the charcoal present in these contexts rep-resents the remains of wood used as fuel.

The context referred to as Various differs from the others inthe number of taxa present (six as opposed to 14 for the othercontexts prior to grouping), the fact that those taxa do not occurconsistently throughout the sequence of periods, and that therewas considerably more variation in their ubiquity proportions. Theresults of the analysis may be pointing to the low diversity in taxarather than to differential or particular use of wood in the domesticand ritual activities represented by this context.

One possible explanation for the lack of any statistical differencebetween the Middle and Late Classic periods is that the proportionsof most of the taxa during these periods were very similar and inone case, Arbutus, the same. It may be that the model is unable todistinguish between such ne differences, and therefore, cannotdistinguish between the periods, even though at least one taxon(Pinus) shows dissimilar values in these periods. The remainingtime periods differ from each other based on the distinct pro-portions of taxa in each.

The variations in the proportions of Pinus during the Pre-classic, Early Classic, and MiddleLate Classic periods may bedue to more intense use and preference by the Teotihuacanosfor this genus, particularly if we consider that these periodsmarked the development and orescence of Teotihuacan cul-ture. The absence of change during the Late ClassicPostclassic

C.C. Adriano-Moran, E. McClung de Tapia / Journal o2932transition probably indicates a reduction in the intensity withwhich this resource was exploited as a consequence of pop-ulation decline in the region, the exhaustion of the resource orthe loss of important trade routes as Teotihuacan came to itsend.

In the case of Quercus, the proportions are stable during thePreclassic and the entire Classic, suggesting a constant but lessintensive use than that of Pinus. The most noticeable changeclearly takes place at the beginning of the Postclassic period,consisting of a drastic reduction in its proportion. This may in-dicate a reduction in the availability of oaks in the region,

Fig. 3. Some genera identied in wood charcoal from the Teotihuacan Valley. (a, b) Cupresaculeaticarpa var. biuncifera. (a, b, c, and e) Transverse section; (d, f) tangential longitudina

Table 2Signicance values obtained from logistic analysis

Variable c2 df Signicance % of variation explained

Taxa 914.9 7

nal o6.2. Use and management of vegetation: a hypothesis

Taxa present in the samples recovered from numerous archae-ological excavations in the Teotihuacan Valley represent differenttypes of vegetation: pineoak forest (Pinus, Quercus, Arbutus), xe-rophytic scrub (Fabaceae, Agave) and riparian vegetation (Taxodium,Alnus and Salix). Because these taxa are evident since the Preclassicperiod, this mosaic of vegetation types was present in the valley atthis time and, as a group, probably represents the vegetation of theregion before human settlement. This agrees with Castilla andTejero (1983), the results of Adriano-Morans initial study of ar-chaeological specimens (2000), and studies of other materials suchas pollen, phytoliths and soils (McClung de Tapia,1996; McClung deTapia et al., 1998, 2003).

The constant presence of wood from Pinus, Quercus andCupressaceae indicates a preference for these taxa used as fuel. Thevariations in proportions of identied taxa do not indicate a changein the composition of genera utilized throughout the time of oc-cupation [Asouti (2003) found similar evidence from a Neolithicsite in Turkey] nor do they indicate that arboreal species utilizedduring the earlier occupation stages were replaced by shrub speciesin later stages, as would be expected if the main resource (trees)had been depleted by overexploitation (deforestation) as suggestedby other studies (Newsom, 1993). The presence of Groups 1 and 2

Fig. 4. Differences in ubiquity among taxa by context. Floors, lls and hearths (black)differed signicantly from various (white). Gray indicates ubiquity of taxa that showedno signicant variation between contexts. Intervals correspond to standard errors.

C.C. Adriano-Moran, E. McClung de Tapia / Jour(comprised of trees and shrubs) from the Preclassic suggests thatboth were being utilized from the initial occupation of the valley.Their constancy over time, especially in the case of Group 1, mayindicate that they were preferred as fuel. It is probable that Pinus,Quercus, Cupressaceae and Groups 1 and 2were utilized together ascomplementary resources rather than as substitutes for one oranother.

The preference for the taxa mentioned above suggests that theywere selected intentionally because they were considered to begood materials for fuel. If this were the case, we can suppose thatthe prehispanic inhabitants of the Teotihuacan Valley were familiarwith the characteristics of the woods and, as a function of this fa-miliarity, chose themost suitable for their purposes. In otherwords,they maintained a sort of quality control over the taxa chosen forspecic purposes. Today in some of the rural communities of cen-tral Mexico in which wood remains the primary source of fuel, thegenera employed include Pinus, Quercus, Prunus, Alnus, Arbutus,Baccharis, Buddleia and Salix because these are considered to be ofgood to fair quality based on the selection criteria: easy ignition,slow burning, intense ame, little smoke, production of embers andavailability of the species (Camacho, 1985; Almeida, 1990; Estrada,1996).f Archaeological Science 35 (2008) 29272936 2933The behavior of Z. mays and other taxa such as monocotyledonsindicate that they were also used in combinationwith other woodyspecies rather than as a substitute due to a shortage of wood. Thecombinations of different woods (or other fuel materials) may bea function of the purpose and duration of the re (for example,ignition, illumination, heating, cooking, ring ceramics) and wouldalso include the use of different parts of the tree such as trunks,branches and twigs. This practice is reported among present dayrural communities where the species are utilized in different pro-portions according to the activity to be undertaken (Camacho,1985; Martorell, 1995). For example, in an Otom community in theState of Mexico, the rst step of the ring process in the productionof roof tiles is carried out with a combination of madrono (Arbutus)

Fig. 5. Differences in ubiquity among taxa by period. Periods during which the ubiq-uity of a taxon showed no signicant variation are indicated by the same bar. Intervalscorrespond to standard errors. Taxa in which ubiquity did not differ in any given periodare not shown.

al oand alder (Alnus), followed by the addition of broom (Baccharis)when the heat is intense, in order to produce the characteristicreddish tone. In this case, the intensity of caloric output and rapidcombustion of the wood is important. Other genera, such as oak(Quercus) which also has a high caloric output but burns moreslowly are not used because the tiles would break during ring.On the other hand, species of oak (and alder and madrono toa lesser extent) are preferred for domestic consumption becauseintense, longer lasting heat is required. In some cases, in order toget better heat yield from logs, branches and twigs from trees aswell as shrubs (such as Prunus, Baccharis and Buddleia) are added ascomplements. In addition, other woody or combustible materialssuch as maize cobs and stalks as well as Agave leaves (pencas) arealso used (Camacho, 1985).

Genera and families identied in the archaeological samplesrepresent components of primary as well as secondary vegetation.Genera such as Pinus, Quercus, Taxodium and Cupressus are repre-sentative of temperate forests characteristic of higher elevations inMexico and constitute primary vegetation or climax communities.Alnus, Salix, Juniperus, Prunus, Arbutus, Buddleia and Baccharis arecommonly present as companion species in stands of primaryforest but also play an important role as secondary taxa (Vazquez-Yanes et al., 1999; Rzedowski and Rzedowski, 2001). When dis-turbance occurs, either natural or anthropogenic, species of thesegenera are pioneers in the succession. They establish in disturbedand open spaces because they are mainly heliophilous and, there-fore, of considerable importance in the regeneration of the forest.Also, some of these species are able to reproduce vegetatively fromshoots (such as Alnus, Prunus and Juniperus), which allows morerapid recuperation of the vegetation. Because of these character-istics, Baccharis and Buddleia are considered indicators of secondaryvegetation.

The presence of these genera among the samples from archae-ological contexts in the Teotihuacan Valley may indicate pertur-bation of the vegetation from very early times and the possibilitythat vegetation types present in the region prior to human settle-ment persisted through time in different stages of regeneration(secondary vegetation). From this perspective, the inhabitants ofthe valley appear to have been intentionally using primary as wellas secondary species. The extension of secondary communitiesmust have been quite broad considering that these rapidly growingplants could establish in abandoned lands or fallow agriculturalplots following the completion of their planting/harvesting cycle.The use andmaintenance of secondary vegetationwould guaranteea constant wood supply and, particularly in the valley, a largeamount of wood suitable for domestic consumption, which wouldbe advantageous for the human population by reducing costs oftransportation and recollection (time and distance) of the resource.Independently, the acquisition of wood such as Pinus and Quercusfrom other parts of the Basin of Mexico was probably establishedearly in the occupation of the region, based on estimates suggestingthat only around 12% of the area was suitable to support primarypineoak forest (McClung de Tapia and Tapia-Recillas, 1997).Therefore, it is likely that local availability was not sufcient tocover the needs of the valleys inhabitants during periods of in-creased population.

The use of secondary vegetation among indigenousgroups (Huastecs, Totonacs and Maya) is well documented for thewarmhumid tropics in Mexico as well as other parts of LatinAmerica and represents one of the strategies included in a di-versied management system or multiple-use strategy (Alcorn,1983; Toledo et al., 1995, 2003). Such indigenous strategies purporttomaximize diversity and the range of available options in resourceexploitation to meet subsistence needs and minimize risks. They

C.C. Adriano-Moran, E. McClung de Tapia / Journ2934include the use of different families or species, life forms, parts ofplants, habitat and its manipulation, all of which imply a broadfamiliarity not only with the plants themselves but also with eco-logical processes. Indigenous communities are able to adapt andintegrate new practices into the original multiple managementstrategy in response to recent economic changes (Toledo et al.,2003). However, it is also possible that some of these strategieshave been employed since prehispanic times.

Several authors have suggested that Maya communities duringthe Classic period (ca. A.D. 250900) used andmanaged the tropicalforest (Barrera et al., 1977; Alcorn, 1981), resulting in the presentday oristic composition and distribution visible in associationwith ancient sites. Sylviculture formed part of the managementstrategies that included the use of multiple resources from pre-hispanic times. In the temperate zones of the central highlands ofMexico indigenous agricultural practices, including shifting culti-vationwith fallow periods, are mentioned in 16th and 17th centurydocuments. In shifting cultivation, the fallow periods and rotationof plots were employed as a means of regenerating vegetation aswell as soil fertility (Rojas, 1985). In addition, there are indicationsthat at least the Aztecs recognized that the vegetation growing onfallowed plots was different from the original. Based on Molinas(1992) 16th century vocabulary, the term quauhtapaolli refers tobrambles or upland weeds and can be understood to refer to sec-ondary vegetation, whereas quauhtla, meaningmountain, woods orforest and quauhyouacatlameaning forest or dense woods, suggestsprimary vegetation.

7. Conclusions

The use of wood by the prehispanic occupants of the Teoti-huacan Valley was a complex activity that required considerableknowledge on the part of the inhabitants of the resource and theprocesses that controlled its availability.

The taxa show a pattern of continuity with respect to the ex-ploitation of particular genera from the Preclassic through thePostclassic period. This pattern is characterized by the utilization ofdistinct genera including trees and shrubs, as well as herbaceous(Monocotyledons) and cultivated taxa (Z. mays); in addition itcomprises the use of different parts of trees and shrubs such asbranches and trunks. Both the taxa employed and their specicvegetative parts were used in diverse combinations and quantities,depending upon the energetic requirements. Changes in the pro-portions of different genera are interpreted as reecting preferenceand intensity of usage and not as indicators of regionaldeforestation.

Deforestation, as a consequence of the use of wood, could havebeen of two types: (1) partial, in which isolated patches of naturalvegetation were maintained throughout the region through time;or (2) total, in which the resource was rapidly extinguished,resulting in the disappearance of the arboreal community early onin the history of human occupation. The results of this study sup-port the rst option; however, it is also possible that the alternativescenario occurred at some point in the past but is unobservable inthe archaeological record.

The population of the Teotihuacan region apparently selectedrewood based on its quality. The data suggest that the preferencefor certain generawas diversied instead of limited; in other words,various genera with similar characteristics (e.g. easy cutting, quickignition, slow burning and intense heat production) were exploitedrather than a single taxon. This practice contributed to maintainingthe availability of preferred taxa (such as Pinus, Quercus, Cupres-saceae, Group 1) in the region throughout the period of prehispanicoccupation.

The exploitation of taxa from primary as well as secondarycommunities supports the idea that the inhabitants of the region

f Archaeological Science 35 (2008) 29272936employed management strategies oriented towards maintaininglocal vegetation zones that provided wood for daily use. These

References

nal oAdriano-Moran, C.C., 2000. Estudio del carbon arqueologico como indicador de loscambios en la vegetacion, en el Valle de Teotihuacan, Edo. de Mexico. MScThesis. Facultad de Ciencias, UNAM. Mexico.

Alcorn, J.B., 1981. Huastec non-crop resource management: implications for pre-historic rain forest management. Human Ecology 9, 395417.

Alcorn, J.B., 1983. El Telom Huasteco: Presente, pasado y futuro de un sistema desilvicultura indgena. Biotica 8, 315325.

Almeida, R., 1990. Analisis calorimetrico de especies vegetales que se utilizan comolena. Undergraduate thesis. Facultad de Ciencias, UNAM. Mexico.

Alvarez del Castillo, C., 1984. Resultados del analisis botanico de la excavacion ar-queologica realizada en Cuanalan, Edo. de Mexico. In: Dos estudios paleo-botanicos en la Cuenca de Mexico. Cuadernos de trabajo 20. INAH, Mexico D.F.,pp. 325.

Asouti, E., 2003. Woodland vegetation and fuel exploitation at the prehistoriccampsite of Pinarbasi, south-central Anatolia, Turkey: the evidence from thewood charcoal macro-remains. Journal of Archaeological Science 30, 11851201.

Asouti, E., Hather, J., 2001. Charcoal analysis and reconstruction of ancient wood-land vegetation in the Konya Basin, south-central Anatolia, Turkey: results fromthe Neolithic site of atalhoyuk East. Vegetation History and Archaeobotany 10,2332.

Barrera, A., Gomez-Pompa, A., Vazquez-Yanes, C., 1977. El manejo de las selvas porlos Mayas: sus implicaciones agrcolas y silvcolas. Biotica 2, 4760.

Camacho, J., 1985. Estudio del uso del bosque para extraccion de lena, madera paraconstruccion y fabricacion de herramientas en una comunidad Otom San An-dres Timilpan, Edo. de Mexico. Undergraduate Thesis. ENEP Iztacala, UNAM.Edo. de Mexico.

Castilla, M., Tejero, D., 1983. Estudio orstico del Cerro Gordo (proximo a San JuanTeotihuacan) y regiones aledanas. Undergraduate Thesis. ENEP Iztacala, UNAM.Edo. de Mexico.

Castilla, M., Tejero, D., 1987. Flora y vegetacion del Cerro Gordo (San Juan Teoti-huacan) y regiones aledanas, Valle de Mexico, Mexico. Biotica 12, 231255.

Charlton, T.H., Nichols, D.L., Otis Charlton, C.L., 2000. Otumba and its neighbors. Exoriente lux. Ancient Mesoamerica 11, 247265.

Cowgill, G.L., 1996. Discussion. Ancient Mesoamerica 7, 325331.Crawley, M.J., 1996. Glim for Ecologists. Blackwell Scientic Publications, Oxford.Estrada, E., 1996., Etnobotanica forestal en Sta. Isabel Chalma, Amecameca, Mexico.

MSc thesis. Colegio de Posgraduados. Chapingo, Edo. de Mexico.February, E.C., 1992. Archaeological charcoals as indicators of vegetation change and

human fuel choice in the late Holocene at Elands Bay, western Cape Province,zones played an important role as seed reservoirs contributing tothe partial regeneration of local vegetation. Based on McClung deTapia et al. (2005), this restoration likely occurred during periodsfollowing important declines in humanpopulation (i.e. after ca. A.D.650). From this perspective, the development of these zones wouldcontribute to erosion control and overall landscape stability. Theimplementation of different management strategies would be animportant contributing factor to the duration of the prehispanicpopulation for over 1000 years in an otherwise heavily impactedregion.

Acknowledgements

This research forms part of the doctoral dissertation of CarmenCristina Adriano-Moran and was supported by Consejo Nacional deCiencia y Tecnologa (CONACYT) and the Direccion General dePersonal Academico, Universidad Nacional Autonoma de Mexico(DGAPA UNAM). We would like to thank Linda Manzanilla Naimfor making available materials from her investigations at Cuanalan,Oztoyahualco, tunnels East of the Pyramid of the Sun and Teo-pancazco. We also thank Saburo Sugiyama for permission to studythe charcoal remains from the Moon Pyramid and Thomas H.Charlton for the materials from Otumba. Access to the botanicalmaterials from the Teotihuacan Mapping Project was granted byRene Millon (University of Rochester). Additional thanks go toCarlos Martorell (Facultad de Ciencias, UNAM) for collaboratingwith the statistical analysis and Silvia Espinosa (Facultad de Cien-cias, UNAM) for assistance with the Scanning Electron Microscope.Rodrigo Tapia designed Fig. 1. Finally, we thank the reviewers fortheir recommendations, which substantially improved the text.

C.C. Adriano-Moran, E. McClung de Tapia / JourSouth Africa. Journal of Archaeological Science 19, 347354.Figueiral, I., 1993. Charcoal analysis and the vegetational evolution of North-West

Portugal. Oxford Journal of Archaeology 12, 209222.Flannery, K., 1986. Wood used as fuels. In: Flannery, K. (Ed.), Guila Naquitz. ArchaicForaging and Early Agriculture in Oaxaca, Mexico. Academic Press, New York,pp. 169170.

Ford, R., 1976. Carbonized plant remains. In: Drennan, R. (Ed.), Fabrica San Jose andMiddle Formative Society in the Valley of Oaxaca, Prehistory and HumanEcology of the Valley of Oaxaca, Vol. 4, No. 8. Memoirs of the Museum of An-thropology. University of Michigan, Michigan, pp. 261268.

Garca, E., 1968. Clima actual de Teotihuacan. In: Lorenzo, J.L. (Ed.), Materiales parala arqueologa de Teotihuacan. Cuadernos de Trabajo. Serie Investigaciones 17.INAH, Mexico D.F., pp. 928.

Garca, E., 1988. Modicaciones al sistema de clasicacion climatica de Koppen.Instituto de Geografa, UNAM, Mexico D.F.

Godwin, H., Tansley, A.G., 1941. Prehistoric charcoals as evidence of former vege-tation, soil and climate. Journal of Ecology 29, 117126.

Gonzalez, J., Ibarra, E., Zurita, J., McClung, E., Tapia, H., 1993. Macrofosiles botanicos,tolitos y polen. In: Manzanilla, L. (Ed.), Anatoma de un conjunto residencialteotihuacano en Oztoyahualco. Instituto de Investigaciones Antropologicas,UNAM, Mexico D.F., pp. 661728.

Gonzalez, L., 1982. Informe preliminar sobre el estudio de restos organicos deTeotihuacan. In: Cabrera, R., Rodrguez, I., Morelos, N. (Eds.), Memoria delproyecto arqueologico Teotihuacan 8082. Coleccion cientca 132. INAH,Mexico D.F., pp. 463465.

Hastorf, C., 1988. The use of paleoethnobotanical data in prehistoric studies of cropproduction. In: Hastorf, C., Popper, V. (Eds.), Current Paleoethnobotany. Ana-lytical Methods and Cultural Interpretations of Archaeological Plant Remains.The University of Chicago Press, Chicago, pp. 119144.

Heinz, C., 1991. Upper Pleistocene and Holocene vegetation in the south of Franceand Andorra. Adaptations and rst ruptures: new charcoal analysis data. Re-view of Paleobotany and Palynology 69, 299324.

Heinz, C., Thiebault, S., 1998. Characterization and paleoecological signicance ofarchaeological charcoal assemblages during late and post-Glacial phases insouthern France. Quaternary Research 50, 5668.

Johannessen, S., 1988. Plant remains and culture change: are paleoethnobotanicaldata better than we think? In: Hastorf, C., Popper, V. (Eds.), Current Paleo-ethnobotany. Analytical Methods and Cultural Interpretations of ArchaeologicalPlant Remains. The University of Chicago Press, Chicago, pp. 145166.

Johannessen, S., Hastorf, C., 1990. A history of fuel management (A.D. 500 topresent) in the Manataro Valley, Peru. Journal of Ethnobiology 10, 6190.

Jones, G., 1987. A statistical approach to the archaeological identication of cropprocessing. Journal of Archaeological Science 14, 311323.

Jones, G., 1991. Numerical analysis in archaeobotany. In: Van Zeist, W.,Wasilikowa, K., Behre, K.E. (Eds.), Progress in Old World Palaeoethnobotany.Balkema, Rotterdam, pp. 6380.

Kohler, T.A., Lipe, W.D., Floyd, M.E., Bye Jr., R.A., 1984. Modeling wood resourcedepletion in the Grass Mesa locality. In: Breternitz, D.A. (Ed.), Dolores Archae-ological Program: Synthetic Report 19781981. U.S. Department of the Interior,Bureau of Reclamation, Engineering and Research Center, Denver, pp. 99105.

Kohler, T.A., Matthews, M.H., 1988. Long-term Anasazi land use and forest reduction:a case study from southwest Colorado. American Antiquity 53, 537564.

Kovar, A., 1970. The physical and biological environment of the basin of Mexico. In:The Teotihuacan Valley Project, Final Report. Vol. 1, Occasional Papers in An-thropology No. 3. Department of Anthropology, Pennsylvania State University,Pennsylvania, pp. 2124.

Lepofsky, D., Kusmer, K.D., Hayden, B., Lertzman, K.P., 1996. Reconstructing pre-historic socioeconomic from paleoethnobotanical and zooarchaeological data:an example the British Columbian Plateau. Journal of Ethnobiology 16, 3162.

Lorenzo, J.L., 1968. Clima y agricultura en Teotihuacan. In: Lorenzo, J.L. (Ed.), Ma-teriales para la Arqueologa de Teotihuacan. Cuadernos de Trabajo. Serie In-vestigaciones 17. INAH, Mexico D.F., pp. 5172.

Manzanilla, L., 1985. El sitio de Cuanalan en el marco de las comunidades pre-ur-banas del valle de Teotihuacan. In: Monjaras-Ruiz, J., Perez, E., Brambila, R. (Eds.),Mesoamerica y el centro de Mexico. Una antologa. INAH, Mexico D.F., pp. 133178.

Manzanilla, L., 1993. Anatoma de un conjunto habitacional teotihuacano en Oz-toyahualco, v. 1. Instituto de Investigaciones Antropologicas, UNAM, Mexico D.F.

Manzanilla, L., 2003. Teopancazco: Un conjunto residencial teotihuacano. Arqueo-loga Mexicana 11, 5053.

Manzanilla, L., Lopez, C., Freter, A.C., 1996. Dating results from excavations in quarrytunnels behind the Pyramid of the Sun at Teotihuacan. Ancient Mesoamerica 7,245266.

Martorell, C.,1995. Consecuencias ecologicas y alternativas de uso de la lena en lacomunidad de los Reyes Metzantla. Undergraduate thesis. Facultad de Ciencias,UNAM. Mexico.

McClung de Tapia, E., 1990. Modelos para la reconstruccion de actividades desubsistencia en la Cuenca de Mexico. In: Cardos de Mendez, A. (Ed.), La epocaClasica: nuevos hallazgos, nuevas ideas. Seminario de Arqueologa, Museo Na-cional de Antropologa, INAH, Mexico D.F., pp. 107114.

McClung de Tapia, E., 1996. Informe tecnico del proyecto: Cambios paleo-ambientales y sus efectos sociales en Teotihuacan 199596, Primera parte,Subproyecto: El paleoambiente de la region de Teotihuacan, Informe Tecnico 4.Laboratorio de Paleoetnobotanica y Paleoambiente, Instituto de InvestigacionesAntropologicas, UNAM, Mexico D.F.

McClung de Tapia, E., Cervantes, J., Meza, M., Zurita, J., Ibarra, E., 1998. Cronologa de

f Archaeological Science 35 (2008) 29272936 2935los procesos geomorfologicos en el Valle de Teotihuacan. In: Brambila, R.,Cabrera, R. (Eds.), Los ritmos de cambio enTeotihuacan: reexiones y discusionesde su cronologa. Coleccion Cientca 366. INAH, Mexico D.F., pp. 503518.

McClung de Tapia, E., Domnguez Rubio, I., Gama-Castro, J., Solleiro-Rebolledo, E.,Sedov, S., 2005. Radiocarbon dates from soil proles in the TeotihuacanValley, Mexico: indicators of geomorphological processes. Radiocarbon 47,159175.

McClung de Tapia, E., Gama-Castro, J., Solleiro-Rebolledo, E., Villalpando, J., Sedov, S., 2003. Paleosols in the Teotihuacan Valley, Mexico, evidence for paleo-environment and human impact. Revista Mexicana de Ciencias Geologicas 20,270282.

McClung de Tapia, E., Tapia-Recillas, H., 1997. Sistemas de informacion geograca enarqueologa. Ciencia y Desarrollo 22, 4856.

McCullag, P., Nelder, J.A., 1989. Generalized Linear Models. Chapman and Hall,London.

Miller, N., 1985. Paleoetnobotanical evidence for deforestation in ancient Iran:a case study of urban Malyan. Journal of Ethnobiology 5, 119.

Millon, R., 1970. Teotihuacan: completion of map of giant city in the Valley ofMexico. Science 170, 10771082.

Millon, R., 1973. Urbanization at Teotihuacan, Mexico, v. 1, Part 1, The TeotihuacanMap. University of Texas Press, Austin.

Minnis, P.E., 1978. Paleoethnobotanical indicators of prehistoric environmentaldisturbance: a case study. In: Ford, R.I., Brown, M.F., Hodge, M., Merrill, W.L.(Eds.), The Nature and Status of EthnobotanyAnthropological Papers No.67.University of Michigan, Ann Arbor, pp. 347366.

Molina, Fray A. de, 1992. Vocabulario en Lengua castellana y mexicana y mexicana ycastellana. Edicion facsimilar, Porrua, Mexico D.F.

Montufar, A., 1996. Estudio arqueobotanico en vasijas de la Ventilla, Teotihuacan,Mexico. Revista Mexicana de Estudios Antropologicos 42, 7176.

Montufar, A., 1999. Archaeobotanical study of the caves II and III, Teotihuacan.Perspectives in Human Biology 4, 101114.

Mooser, F., 1968. Geologa, naturaleza y desarrollo del Valle de Teotihuacan. In:Lorenzo, J.L. (Ed.), Materiales para la Arqueologa de Teotihuacan. Cuadernos deTrabajo, Serie Investigaciones 17. INAH, Mexico D.F., pp. 3137.

Newsom, L., 1993. Plants and people: cultural, biological and ecological responses towood exploitation. In: Scarry, C.M. (Ed.), Foraging and Farming in EasternWoodlands. University Press of Florida, Gainesville, pp. 115137.

Popper, V., 1995. Nahua plant knowledge and chinampa farming in the basin ofMexico: a middle postclassic case study. PhD. Thesis. University of Michigan.

Prior, J., Price-Williams, D., 1985. An investigation of climatic change in the Holo-cene epoch using archaeological charcoal from Swaziland, Southern Africa.Journal of Archaeological Science 12, 457475.

Rojas, T., 1985. La tecnologa agrcola mesoamericana en el siglo XVI. In: Rojas, T.,Sanders, W.T. (Eds.), Historia de la agricultura. Epoca prehispanica-siglo XVI.Coleccion Biblioteca, INAH, Mexico D.F., pp. 129231.

Rzedowski, G.C.de, Rzedowski, J. (Eds.), 2001. Flora fanerogamica del valle deMexico. Instituto de Ecologa, A.C., Centro Regional del Bajo y Comision Na-cional para el Conocimiento y Uso de la Biodiversidad, Patzcuaro (Michoacan).

Rzedowski, J., Guzman, G., Hernandez, A., Muniz, R., 1964. Cartografa de los prin-cipales tipos de vegetacion de la mitad septentrional del Valle de Mexico.Anales de la Escuela Nacional de Ciencias Biologicas 13, 3157.

Salisbury, E.J., Jane, F.W., 1940. Charcoals from Maiden Castle and their signicancein relation to the vegetation and climate conditions in prehistoric times. Journalof Ecology 28, 310325.

Sanders, W., 1965. The Cultural Ecology of Teotihuacan Valley. Department of An-thropology, Pennsylvania State University, Pennsylvania.

Scheel-Ybert, R., 2000. Vegetation stability in the southeastern Brazilian coastalarea from 5500 to 1400 14C yr BP deduced from charcoal analysis. Review ofPaleobotany and Palynology 110, 111138.

Scheel-Ybert, R., Gouveia, S.E.M., Pessenda, L.C.R., Aravena, R., Coutinho, L.M.,Boulet, R., 2003. Holocene palaeoenvironmental evolution in the Sao PauloState (Brazil), based on anthracology and soil d13C analysis. The Holocene 13,7381.

Smart, L.T., Hoffman, E.S., 1988. Environmental interpretation of archaeologicalcharcoal. In: Hastorf, C., Popper, V. (Eds.), Current Paleoethnobotany. AnalyticalMethods and Cultural Interpretations of Archaeological Plant Remains. TheUniversity of Chicago Press, Chicago, pp. 167205.

Sugiyama, S., 2004. Governance and polity at Classic Teotihuacan. In: Hendon, J.A.,Joyce, R.A. (Eds.), Mesoamerican Archaeology. Blackwell Publishing, Ltd., pp.97123.

Toledo, V., Batis, A.I., Becerra, R., Martnez, E., Ramos, C.H., 1995. La selva util: et-nobotanica cuantitativa de los grupos indgenas del tropico humedo de Mexico.Interciencia 20, 177187.

Toledo, V., Ortiz-Espejel, B., Cortes, L., Moguel, P., Odonez, M.J., 2003. The multipleuse of tropical forest by indigenous peoples in Mexico: a case of adaptativemanagement. Conservation Ecology 7 Available from: http://www.consecol.org/vol7/iss3/art9.

Vazquez-Yanes, C., Batis Munoz, A.I., Alcocer Silva, M.I., Gual Daz, M., SanchezDirzo, C., 1999. Arboles y arbustos potencialmente valiosos para la restauracionecologica y la reforestacion. Reporte tecnico del proyecto J084. CONABIO-In-stituto de Ecologa, UNAM, Mexico D.F.

C.C. Adriano-Moran, E. McClung de Tapia / Journal of Archaeological Science 35 (2008) 292729362936Rattray, E., 2001. Teotihuacan: Ceramics, Chronology and Cultural Trends. INAH,University of Pittsburgh, Mexico D.F., Pittsburgh.

Rodrguez, M.F., 2004. Woody plant resources in the southern argentine puna:Punta de la Pena 9 archaeological site. Journal of Archaeological Science 31,13611372.Vernet, J.L., 1980. La vegetation du bassin de lAude, entre Pyrenees et MassifCentral, Au tardiglaciaire et postglaciaire dapres lanalyse anthracologique.Review of Paleobotany and Palynology 30, 3355.

Willcox, G.H., 1974. A history of deforestation as indicated by charcoal analysis offour sites in eastern Anatolia. Anatolian Studies 24, 117133.

Trees and shrubs: the use of wood in prehispanic TeotihuacanIntroductionStudy areaNumerical analysis of botanical materialsMethod of analysisResultsVariation of taxa among contextsVariation among taxa through time

DiscussionVariations in contexts, time periods and taxaUse and management of vegetation: a hypothesis

ConclusionsAcknowledgementsReferences