Differentiating Aspen and Cottonwood in Prehistoric ?· Differentiating Aspen and Cottonwood in Prehistoric…

  • Published on

  • View

  • Download

Embed Size (px)


<ul><li><p>Journal of Archaeological Science (2002) 29, 521527doi:10.1006/jasc.2001.0746, available online at http://www.idealibrary.com on</p><p>Differentiating Aspen and Cottonwood in Prehistoric Woodfrom Chacoan Great House Ruins</p><p>David Tennessen</p><p>Department of Anthropology, University of Minnesota, Minneapolis, Minnesota 55455, U.S.A.</p><p>Robert A. Blanchette</p><p>Department of Plant Pathology, University of Minnesota, St Paul, Minnesota 55108-6030, U.S.A.</p><p>Thomas C. Windes</p><p>National Park Service, U.S. Department of Interior, Santa Fe, New Mexico 87504-0728, U.S.A.</p><p>(Received 8 June 2001, revised manuscript accepted 12 September 2001)</p><p>Accurate taxonomic identification is an essential part of archaeological wood analysis. However, making identificationsmore precise than the genus level is usually not possible since species within the same genus typically possess verysimilar cellular morphology. This paper describes a method for distinguishing aspen (Populus tremuloides) fromcottonwood (Populus fremontii, Populus angustifolia, Populus acuminata) in samples of wood collected from the SanJuan Basin in northwestern New Mexico and southwestern Colorado. This method is then applied to archaeologicalwood samples from the Anasazi great house at Aztec Ruins National Monument in Aztec, New Mexico. The results ofthis study demonstrate that quantifiable differences do exist between aspen and cottonwood species and that thetechnique can be used to separate archaeological specimens of Populus wood.</p><p> 2002 Elsevier Science Ltd. All rights reserved.</p><p>Keywords: ARCHAEOLOGY, ARCHAEOLOGICAL WOOD, WOOD IDENTIFICATION,WOOD ANATOMY, CHACO CANYON, AZTEC RUINS, CHACOAN GREAT HOUSES.</p><p>Introduction</p><p>T he analysis of wood from archaeological sitesholds great potential for providing informationabout the past. Dendrochronology has beenused successfully for decades to provide precisedating of archaeological sites (Douglass, 1935;Schweingruber, 1993). More recently, analysis ofarchaeological wood has been used to provide import-ant information about raw material production(Windes &amp; Ford, 1996; Windes &amp; McKenna, 2001),resource availability (Fitzhugh, 1996) and land useand deforestation (Kohler &amp; Matthews, 1988). Eveninvestigations on deteriorated wood have provideduseful information on wood species used, degree ofpreservation, and past environmental conditions(Blanchette, 2000; Blanchette et al., 1994). Regardlessof the kind of information being sought, an essentialstep in the analysis of wood from an archaeologicalcontext is identification.</p><p>52103054403/02/$-see front matter</p><p>The taxonomic level to which identifications arepossible depends on a number of factors including thespecies being examined and the size and condition ofthe specimen. Typically, wood identifications are madeto the genus level. Identifications to species are verydifficult if not impossible since the different specieswithin a genus usually display very similar cell mor-phology. In order to address some anthropologicalquestions, however, it may be necessary to determinefrom which tree species a sample of archaeologicalwood originated.</p><p>An example of a case where precise, species levelidentifications are important is the analysis of woodfrom Anasazi sites in the San Juan Basin of north-western New Mexico. Large amounts of wood from avariety of species were used in the construction ofAnasazi great houses (Lekson et al., 1988), includingmembers of the genus Populus. In New Mexico,Populus is represented by a number of species in differ-ent sections of the genus. These include one species of</p><p> 2002 Elsevier Science Ltd. All rights reserved.</p></li><li><p>522 D. Tennessen et al.</p><p>Figure 1. Cross-section of cottonwood comparative sample at 100magnification.</p><p>Figure 2. Cross-section of aspen comparative sample at 100magnification.</p><p>aspen, P. tremuloides from section Leuce, and twospecies of cottonwood, P. fremontii from sectionAigeros and P. angustifolia from section Tacamahaca.In addition, a hybrid species of cottonwood,P. acuminata (P. angustifoliaP. fremontii) alsooccurs.</p><p>Although their ranges overlap, aspen and cotton-wood may occur in very different habitats (Harlowet al., 1996). Aspen does not currently grow in theimmediate vicinity of great houses in the San JuanBasin, but species of cottonwood are common alongthe rivers, tributaries, and washes adjacent to wheremany great houses are located. It has been suggestedthat the regional vegetation has remained approxi-mately the same over the past 2000 years (Betancourt,1984), and certainly ethnobotanical remains recoveredfrom Anasazi sites reveal that local resources are littlechanged from today.</p><p>Differentiating aspen from cottonwood in samplesof wood from Anasazi great houses could providevaluable information about environmental changeand shifts in socio-economic strategies related towood harvesting and construction (Windes &amp; Ford,1996; Windes &amp; McKenna, 2001). Aztec Ruins, acluster of Chacoan great houses built in the early 1100s, has one of the largest samples of pre-historic wood left of any site in the AmericanSouthwest. The West Ruin, which provided the vastmajority of materials for this study, still containsabout 6000 pieces of visible wood. The majority ofthis is Populus sp. used for door and ventilator lintelsand for secondary roof support beams. Builders ofgreat houses in Chaco Canyon, 80 km due south ofAztec, procured 10s of 1000s of beams for construc-tion from around the periphery of the San Juan Basin6090 km away (e.g., Betancourt, Dean &amp; Hull, 1986;Dean &amp; Warren, 1983; Windes &amp; McKenna, 2001).The availability of cottonwood stands along thenearby Animas River, a mere 300 m away, however,would have provided the logical source for the pro-lific use of Populus sp. in the West Ruin. Yet, somenon-local species of ponderosa pine and spruce or firswere also used in construction at the West Ruin, withthe nearest present stands about 1824 km away.Thus, we cannot be certain that some or all of thePopulus sp. in the West Ruin is cottonwood withoutspecies-level identifications. The differences are pro-found, with cottonwood being locally abundant butaspen stands located some 50 km or more away.Aside from a shift in labour investments and harveststrategies, the differences in procurement could markdifferent organizational capabilities. In addition, theunique cluster of conifers constructed in the initialroom suites in the West Ruin suggests that Populussp. might also be informative as to room use anddifferences in social status or groupings.</p><p>The anatomy of wood tissue in both aspen andcottonwood is very similar (Hoadley, 1990; Panshin &amp;DeZeeuw, 1970) which makes species level identifica-</p><p>tion in archaeological samples problematic (Figures 1&amp; 2). Both woods are characterized by a diffuse porousto semi-ring porous distribution of vessels, simpleperforation plates within vessel elements, alternateintervessel pitting and homogenous uniseriate rays.Previous anatomical studies have suggested that somespecies of cottonwood (P. balsamifera, P. deltoides,P. trichocarpa) from both the Aigeros and Tacamahacasections tend to have larger vessels than aspen(Hoadley, 1990; Panshin &amp; DeZeeuw, 1970). To deter-mine if there is a difference in vessel size between aspenand cottonwood from the San Juan Basin, modernsamples of aspen and cottonwood were obtained fromnorthwestern New Mexico and southwestern Coloradoand detailed measurements of the first-formed early-wood vessels were made. Data obtained from thesemodern woods were then compared to vessel sizes ofprehistoric Populus wood samples obtained from dif-ferent rooms in the Aztec West Ruin great house inAztec, New Mexico.</p></li><li><p>Prehistoric Wood from Chacoan Great House Ruins 523</p><p>Methods</p><p>Comparative samples</p><p>Thirty samples of modern aspen and 16 samples ofmodern cottonwood were obtained for comparativepurposes from a variety of locations in New Mexico,Colorado and Arizona (Table 1). Aspen samples werecollected from Tafoya Canyon, New Mexico (n=8),the Sandia Mountains, New Mexico (n=3), Dolores(n=10) and La Plata (n=6) counties, Colorado andthe White Mountains, Arizona (n=3). Cottonwoodsamples were collected from along the Chaco Wash inChaco Canyon adjacent to Pueblo Bonito (n=6) andalong the Animas River adjacent to the Aztec Ruins(n=10). Wood was obtained from the trunk orbranches of trees with approximately the samediameter as the archaeological wood found in theAztec Ruin great house.</p><p>Archaeological samplesFifty-one archaeological samples were recovered fromlatillas (secondary roof beams) and door and ventilatorlintels at Aztec West Ruin great house in Aztec,New Mexico (Table 2). Archaeological specimens werecollected during sampling carried out for the ChacoWood Project (Windes &amp; McKenna, 2001) and wereobtained by sawing pieces from beams or by coring. Tofaciliate analysis, samples were organized by their fieldidentification numbers into seven groups; 900s, 1000s,1100s, 1200s, 1400s, 3000s and 4000s. All specimens inthe 900s and 1400s groups were recovered by coring,while the remainder of the archaeological specimenswas obtained by removing small sections from latillas.</p><p>Table 1. Modern aspen and cottonwood samples used for determiningdifferences in anatomical structure</p><p>Sample location SpeciesNumber of</p><p>samples</p><p>Mean porearea(2)</p><p>Tafoya Canyon Aspen 8 184003Sandia Mtns Aspen 3 146151Colorado Aspen 10 211478White Mtns Aspen 3 146254La Plata Aspen 6 141803Aspen mean 183480Chaco Canyon Cottonwood 6 289498Animas River Cottonwood 10 261052Cottonwood mean 271719</p><p>Table 2. Samples of wood from latillas (secondary roof beams)and door/ventilator lintels at West Ruin Great House, Aztec RuinsNational Monument, Aztec, New Mexico used for analyses</p><p>NumberSize(2)</p><p>Probability thatunknown specimen</p><p>is aspen</p><p>1471 89636 09916952521473 95274 09901446031436 108957 09850879911472 112556 09833783081435 121195 09784495691011 134747 09677145281006 137075 09654099381442 141670 0960387371474 142155 09598181161470 145528 09556370611007 149498 095018436</p><p>981 150986 09479813481231 151805 0946729822</p><p>980 156285 0939366785998 156884 0938312916992 170827 0908443338</p><p>1476 177860 08888722234813 178066 08882471871232 178578 08866803761230 182252 087486707</p><p>996 188840 08510460623540 196360 0819417363</p><p>978 196689 08179208381177 198805 08080651341061 200272 08009967151146 204827 07778124323359 206911 07665812411090 211260 07418960471475 213489 0728603098</p><p>994 213883 07262092481155 214402 07230359273596 236832 05676531133337 243271 05187366363338 243848 05143214593336 247837 04837738411097 254737 0431347923594 261700 03799617573340 269423 03259951393715 309266 01248585393713 310396 01211239973711 314757 01076035723714 316178 01034933653592 322718 0086322013716 333347 00638593873341 337601 00564959853591 373350 00196307123593 387125 00129590053539 392189 00111171243335 519913 00002244253712 556359 0000073474</p><p>Sample preparationThin sections of the transverse plane were made fromeach modern and archaeological specimen. These sec-tions were mounted on a glass slide for observationusing a compound light microscope. In addition, radialsections were made for each archaeological specimen</p><p>to confirm that the wood was Populus. One archaeo-logical specimen, 1128, was identified as Salix sp. basedon the presence of upright ray parenchyma cells andwas removed from further analysis.</p><p>In order to prepare thin sections, a small piece wascut from each comparative and archaeological speci-men. The segments cut from each specimen were inturn cut into smaller pieces, which typically measured</p></li><li><p>524 D. Tennessen et al.</p><p>151505 cm. Before sectioning, the small woodsegments were softened by boiling for up to 2 hdepending upon the condition of the specimen.Modern samples required boiling for the longest time,while decayed or degraded archaeological sampleswere only boiled for approximately 05 h.</p><p>After the specimens were softened, transversesections were made using a razor blade. The cuttingsurface was lubricated with a 50% alcohol:50% glycerinsolution. These sections were placed on glass slides,mounted in a few drops of the same alcohol/glycerinsolution and covered with a cover slip. Multiple slidesof each specimen were made, with each slide containingsections (usually four) from a single (151505 cm) piece. Slides were then heated to drive off airbubbles, and the edges of the cover slip were sealedwith Permount mounting medium (Fisher Chemicals,Fair Lawn, New Jersey).</p><p>Measurement of vessel areaPrevious anatomical investigations have noted that thevessels of some species of cottonwood tend to be largerthan those of aspen (Hoadley, 1990; Panshin &amp;DeZeeuw, 1970). The method used here to differentiateaspen from cottonwood is based on differencesbetween the two groups of species in mean area of thefirst-formed vessel elements in the earlywood. For thisstudy, pore size was represented as area in cross-section, measured in square microns, rather thanlength and width. Taking a single measurement savedtime and it was reasoned that area measurementsshould stay approximately the same even if the actualshape of the vessel was slightly distorted during themaking of thin sections.</p><p>The area of earlywood vessels was measured usingNIH Image, a program for the analysis and processingof images developed by the National Institutes ofHealth and available on the Internet (http://rsb.info.nih.gov/nih-image/). Using a Leitz Labor luxS compound transmitted light microscope with aPanasonic GP-KR222 digital video camera mountedon a 05 camera tube, images of earlywood vesselswere captured at 400 and displayed on a Macintoshcomputer. Once images were captured and displayed,NIH Image was used to obtain vessel area. In orderto record vessel area in square micrometers at thismagnification, the scale was set at 2058 pixels permicron. Manual area measurements were made byoutlining the interior surface of selected pores using thefreehand selection tool, then selecting the measurementcommand with the area measurement enabled.</p><p>Measurement protocolIn order to consistently measure the same featuresamong different specimens, the following measurementprotocol was used for both archaeological and modernwood specimens. Area measurements were made only</p><p>on the first-formed earlywood vessels of an annualring. These were defined as vessels in direct contactwith the radially flattened cells that demarcate the edgeof a growth ring. Vessels themselves were usually easilyidentified on the basis of their relatively large size.However, in some specimens, earlywood vesselsappeared to be quite small. In these cases, the presenceof intervessel pitting in vessel walls was helpful indistinguishing pores from the lumen of other cells. Ifcell identification could not be determined or if thepore was excessively distorted or damaged it was notmeasured.</p><p>In order to prevent measuring the same vessel morethan once, measurements were only made on one thinsection per slide. One hundred vessels were typicallymeasured on each comparative specimen. Fifty vesselswere typically measured on each arch...</p></li></ul>


View more >