389
Settlement
Prehispanic
Cuauhtemoc Soconusco, Mexico
Region
in the
of the
Chiapas,
Robert M. Rosenswig at Albany?SUNY University Albany, New York
This paper presents the results ofa 28 sq km, 100% coverage settlementsurveyundertaken on thePacific coast of southernChiapas, Mexico in a region known sinceAztec timesas the Soconusco. The Soconusco has always been one of the richestagricultural regions in Mesoamerica, and the surveyzone is now coveredhy industrial-scalefruit cultivation. This a trenchesand allows a high level of confidence in the agricultural practice results in grid of cultural materials. Detailed descriptions of the survey surface/subsurfacecorrespondenceof methods and the substantive results of80 new sites recorded from 860 systematicallycollect ed surfaceunits are presented. Three and a halfmillennia ofPrehispanic occupation were documented,from some of the earliest ceramic-usingBarra phase villages at approximately 1900 cal. b.c. to theLate Postclassic settlementsof the early 16th centuryad. The results zone are related towhat is knownfrom other surveysof settlement from this small survey the Pacific coast of southernChiapas and northernGuatemala. changes along
Introduction document most
regional-scale cultural surface. In patterns through survey of the ground as elsewhere, survey data have been em Mesoamerica, Archaeologists
ployed to establish many of the fundamental conclusions concerning the rise of social complexity and the dynamics of political evolution more generally (e.g., Billman and Feinman 1999; Blanton et al. 1993; Drennan et al. 1991; de Montmollin 1989). Survey data have also proved valu able to explore trade patterns (Hirth 1998), economic pro
duction (Stark and Garraty 2004; Willey 1989), and hu man 1994), systems (Sanders and Webster ecological other among things (Balkansky 2006; Nichols 1996). Sur face surveymethods, however, vary between regions and between researchers, and there is no single standard with which to evaluate all survey results (Kanter 2008). This is true even from within a area like relatively small (see Sandey 1990: 331-336), where, for ex extensive surveys have been conducted primarily in ample, the arid centralMexican highlands (e.g., Balkansky et al.
Mesoamerica
et al. 1982; Kowalewski et al. 1989; Sanders, Parsons, and Santley 1979). Aided by scant cover and considerable surface disturbance, such ground 2000;
Blanton
studies rely on surface artifact scatters to determine the ex tent of human occupation. In contrast, in tropical lowland surface indicators of subsurface regions of Mesoamerica, are to remains harder detect due to thick vegetation and/or sediment build-up to "mound
Further, buried counting" survey methodology, sites make estimates problematic lowland population (Johnston 2004; Pyburn 1989: 1-4). As a result, certain survey projects in the lowland regions of Mesoamerica have adopted intensive sampling methods (e.g., Killion et
due
(Balek 2002;
Johnston 2002).
al. 1989; Pool 2003; Stark 1991). Other survey projects have taken advantage of destructive agricultural practices (e.g., Ashmore 1984, 1991) or undertaken systematic au gur tests (e.g., Wendt 2003) to establish a more reliable surface-subsurface correspondence. In all regions, environ
mental conditions, soil deposition, and current land-use af fect how reliably surface survey data reflectpast population dynamics. Here, I present survey data collected by the Soconusco Formative Project (SFP) in the lowland Cuauhtemoc zone of the Soconusco
in southern Chiapas, Mexico (fig. i). In 2002, the SFP systematically surveyed 28 sq km and doc umented 80 new sites. Taking advantage of a system of drainage canals, the extent of Prehispanic settlement in
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
390
~
/
?
j7
/
W
^/AcapetahuaT Regiorw (W
^^^^Mh^~-\/\
^^^^b^^^^^^^^^^
n
o1 25 ^/
*
/
Mazatdn/ /
/ !zap?j Tapachula 7 -
Teotihuacan ^^J^^^^^p #San
,M
1.Map
Figure
of the Soconusco
~
g^^^~
"
'1 ^
'^^
and its location
each period was recorded as the number of hectares on artifactswere documented on the ground surface. Soconusco
is the name given to the coastal plain and the Sierra and piedmont between the Pacific Ocean Madre mountains of southeastern Chiapas and an adjacent section of Guatemala. The focus of this paper is the south The
ffeK
inMesoamerica.
which The
Ajfoer ^
^ / / ^ai)C^^S
^^^^^^^^^^^^
Lorenzo
/
/:
/
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^Madero
fc..
^rchaeologloalalte
Modern settlement
^^^^
HHHn^^llHl^^^W.
MonteAlbdn
"
.-"
/
~
Soconusco
eastern half of the Soconusco that is approximately 60 km long and 30 to 40 km wide (fig. i). This region is defined sw is the Pacific on all sides by geographic barriers. To the are to ne Sierra the the Madre mountains, and to Ocean,
both the nw and the se are two large swamp systems: the Cantilena and Guamuchal/Manchon (see Voorhies 2004:
area between the two relatively small contains the swamp systems greatest concentration of fer tile land on the Pacific coast ofMexico and Guatemala and to grow a variety of fruit crops. The area to today is used the nw contains poor gravelly soils with lower rainfall and less productive agricultural lands (Lowe 1977: 202). In fact, the area between Pijijiapan and Tonala was called the despoblado del Soconusco (uninhabited part of the Soconus co) in colonial times due to its extremely low population 6-13,
20-21).
The
(Orellana 1995: 11). Today the area is still sparsely popu lated and used mainly to graze cattle. To the se of the land between these swamps, the coastal plain becomes wider and the piedmont is furtherfrom the ocean. In this region, there are no swamp systems and the soils contain so much
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
Journal ofField Archaeology[Vol. 33, 2008
clay thatmaize is still not grown today (Neff et al. 2006: 397). It is thus not surprising that a disproportionate num ber of significant human developments on the Pacific coast were in the southeastern part of the So ofMesoamerica conusco
these two large swamp systems (see in press a). Rosenswig Four major rivers transect the land between these two swamps. From north to south these rivers are the Coatan, between
the Cahuacan, the Suchiate, and the Naranjo (fig. i). These rivers originate on the piedmont and drain thewa
tersheds of theTacana and Tajamulco volcanoes. A number of smaller streams originate on the coastal plain and form a series of mangrove estuary systems parallel to the ocean where water flow is insufficient to break through the bar rier beaches. These rivers create a dynamic set of microen vironments. There are a number of detailed descriptions of the Soconusco environment (e.g., Coe and Flannery 1967: 1982: 55-62). For our 9-15; Lowe, Lee, and Espinoza
purposes, five environmental zones were distinguished that run roughly parallel to each other between the Pacific Ocean and the SierraMadre mountains. These distinctions
reflect current land-use practices. As current land-use is the main determinant of soil disturbance and surface visibility, such divisions are relevant for designing archaeological sur
vey methods (e.g., Banning 2002: 39-74; Drennan et al. 1991: 304-305; Schiffer 1987: 235-262; Terrenato and Ammerman 1996). Land-use changes as one moves inland from the Pacific Ocean. Behind the active beach is the firstzone of interest here, a system of estuary mangrove swamps not seriously impacted by modern activity. In this zone, higher eleva tions contain Prehispanic remains and lower ground is sea inundated. The zone second land-use is the occa sonally a sionally-inundated landwith high salt content behind the swamp and estuary systems. These lands are not suitable for fruitproduction and are often used to graze cattle. This zone is usually covered in high grasses and surface visibili is low. ty The third land-use zone is the fertile coastal plain formed by alluvium washed down from the Sierra Madre mountains. This is a zone of intensive agriculture, em ployed mostly for fruit crops such as bananas, mangos, and papayas, with some cacao production. Previously, sesame seeds, sorghum, and cotton were widely cultivated, and are still produced in limited areas. This alluvial zone has been
extensively trenched with a system of irrigation canals and ground cover is kept to a minimum in a series of planta tions that range in size from 20 to 100 ha. As well as dig
ging trenches, heavy machinery has leveled themajority of archaeological mounds in this zone. Behind the alluvial plain, a series of low hills forms the
fourth environmental zone. These in 14 km of the Pacific Ocean
391
low hills extend towith
between the Cahuacan
and
rivers (fig. i). In contrast, around theMazatan Naranjo in the northwestern area of the land between the two zone, no large swamps, there is topography to contain the over the centuries Coatan River which has meandered more than the other three major rivers in the area much
(Gutierrez in press). The low hills are well drained and sites in mostly planted with corn today. Archaeological these low hills have been less impacted by modern devel opment than those on the coastal plain or piedmont, and most Prehispanic mounds remain intact.Ground visibility is variable and land is divided intomuch smaller plots than on the coastal plain. The piedmont is the fifth environmental zone and be gins when the low hills risemore than 100 masl. This area receives as much as four times the rainfall as the coastal
plain, and today contains themajority of the population. Much of this land is covered by modern settlements, such as the city of Tapachula (fig. i), and the extensive highway connects that Central America to theUnited States. system Urban development has destroyed a large number of ar chaeological sites in this zone. The piedmont contained some of themost productive agricultural land prior to the intricate system of irrigation canals that now weaves itsway
through the coastal plain. The piedmont is defined on its northeastern side by the Sierra Madre mountains that be gin at 300 masl and rise to heights over 4000 masl.
Previous
Soconusco Surveys Five systematic surface surveys focused on documenting the estuary, swamp, and coastal plain zones have been un
dertaken in the Soconusco
re (fig. i): in theAcapetahua the western rivers of the half gion (Voorhies 1989), along of the Soconusco (Voorhies and Kennett 1995), in the Mazatan around the site of Paso de laAmada (Clark region 1994), Blanca
the Naranjo
(Love 2002a), the site of El Mesak
River
region around the site of La and the Jesus River region around (Pye 1995). Prior to this, Drucker (1955) carried out regional recon
(1948) and Lorenzo naissance on theMexican
side of themodern border, as did
Shook (1948, 1965) andCoe andFlannery(1967) on the
Guatemalan
side. In 1978 and 1979, Barbara Voorhies (1989) surveyed 44 sq km in the Acapetahua region nw of the Cantilena swamp. Eleven siteswere documented in 11 sq km in the estuary and 22 siteswere found in the 33 sq km surveyed on the coastal plain and edge of the piedmont below the modern Pan-American (Voorhies 1989: 107). Highway The survey universe was stratified into nine regions and 500 m transectswere walked crews by survey spaced 50 m
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
392
apart. Ten-person crews covered transects and sites were on aerial plotted photographs. While 33 siteswere found as part of systematic survey,Voorhies (1989: 112) reports that another 63 siteswere documented non-systematically. At all 96 sites, ceramics were collected judgmentally and were used to determine the phases of occupation at each site; Archaic through Late Postclassic sites were docu mented. Voorhies (1989: 112) notes that due to the lack of surface ploughing, often dense vegetation, and long oc same areas, earlier sites are likelyunder-rep cupation of the resented.
and Kennett (1995) conducted a two-month rivers between the towns of Pijijiapan and the of survey was de Mazatan during the winter of 1991. This survey to take advantage of rivers cutting their banks to ex signed pose sites buried by the extensive alluvium that can build up in the region. From the 333 linear kilometers of river Voorhies
were documented from the Early For surveyed, 22 sites mative through Late Postclassic periods. on work carried out by Ceja (1985) at Paso de BuUding laAmada, John Clark and Michael Blake undertook a sys tematic survey in theMazatan region (see Clark 1994: 51 sites in a 6 km In Blake documented 1985, 96-99). and covered ap stretch of estuary nw of Puerto Madero km. resurveyed Recendy, JohnHodgson proximately 5 sq the area covered by Blake using a GPS and extended the area of estuary nw toward the Cantilena swamp. During twomonths in 1992, Clark (1994) undertook a 100% cov
erage, systematic survey of 50 sq km of coastal plain in the crew walked fields at 25 m intervals Mazatan region. The and artifact scatters. Each and identified all mounds mound was surface-collected and collections were taken every 40 to 50 m when no mounds were present (Clark 1994: 98). In all, 204 siteswere located in the 50 sq km survey area and almost all time periods were represented. sur In 1983 and 1985, Michael Love (2002a: 21-49)
River area between the veyed 200 sq km in theNaranjo Suchiate River and theGuamuchal swamp to put the Con chas phase component of La Blanca into regional context. m apart and archaeolog Survey crew members walked 75 ical data were supplemented by landowners' knowledge (Love 2002a: 23). All ceramic rim sherds, decorated body sherds, and figurines were collected from each mound and when artifact scatterswere encountered without mounds, these artifactswere collected from 10 x 10 m or 20 x 20 m
areas. Sites were
defined based
on
the presence
of
mounds.
conducted a survey in the Mary Pye (1995: 116-195) area of El Mesak, se of the land the in River Jesus drainage In 1991 and two between the large swamps (figs, i, 2). four months of were documented during 1992, 48 sites
survey carried out along the river and the coast where the river empties into the ocean 322-342). (Pye 1995: Mounds were recorded along with surface artifact scatters, in cow pastures where trampled generally documented earthwas cut by erosion, providing a glimpse of subsurface remains (Pye 1995: 129). These five survey projects each identified sites and de fined their limits according to the presence of mounds
and/or surface scatters of artifacts. The spacing between survey crew members varied between 25 m and 75 m and survey crew size varied from two to 10 people. Sites were
located based on visually correlating the field with those visible on aerial tion maps were then generated with on base maps with major geographic The
Cuauhtemoc
distinctive features in
photos. Site distribu site locations plotted features indicated.
Region
has been published from the previous Cuauhtemoc (fig. 2). Philip region of the Soconusco Drucker (1948) passed through this area and conducted some of the first reconnaissance in the area. From 1963 to No
work
1974, Carlos Navarrete conducted reconnaissance in this part of the Soconusco and the Early Formative sites of Ca are shown on New World Archaeologi pulin and Dorado cal Foundation publications (e.g., Lowe 1975: fig. 1). In de Instituto Nacional addition, employees of Mexico's e Historia National (INAH) Registry of Antropologia out a carried and Sites revisited sites reported byNavarrete and survey in the estuary region between Puerto Madero the Suchiate River (Clark 1994: 95). No excavations were carried out at any sites in the region, and no sites encoun tered during reconnaissance were published. The site of Cuauhtemoc had long been known by An in tonio Arazate (of theMuseo Regional del Soconusco
and was visited by John Clark and Michael Tapachula) In 1996, Blake in the 1980s (personal communication). the sitewas trenched to begin cultivating bananas. This dis turbance revealed artifact rich layers and numerous burials. The site's centralmound, which was 5 m high, was leveled at this time. That year,Alejandro Tovalin of theCentro Re two weeks excavating four gional INAH, Chiapas, spent at the site. John Clark also visited the newly dis test pits turbed site at that time,made surface collections, and drew a number of sketch maps of artifact distributions. Clark documented materials dating to many Early Formative numerous surface remains from the phases and observed Based on this in earlyMiddle Formative Conchas phase. formation, and at the suggestion of Clark, I selected the area for research site and surrounding Cuauhtemoc (Rosenswig 2005). was Cuauhtemoc
systematically
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
surface-collected
in
Journal of'FieldArchaeology/Vol.33, 2008
n I
0
y
5km // f
/^ec^ite
f J f
^GUATEMALA
MEXICO ^\ ^
/
Cuauht6moc
/
HBHiKn^kX-
Figure 2. The Cuauhtemoc
\
\
J^P^^^~_~,~j^_~_~_~>qT^l
/
J
I
LbBlanca Aj
survey region.
2001 to test surveymethodology and a 220 m section of one of the deep drainage canals that cut through the site was profiled. In 2002 and 2003, excavations documented how the site was built on a raised sand levee that would have kept inhabitants above the level of seasonal flooding. This desirable location would have originally attracted peo ple to settle at the site during the Barra phase (1900-1700 cal. b.c.). The Cuauhtemoc site was occupied until the cal. b.c.) and nev the end of Conchas phase (1000-850 er reoccupied. In 2002, systematic surface surveywas con ducted for two months
in the area around Cuauhtemoc.
Methods Initial reconnaissance
site in around the Cuauhtemoc covered three environmental zones: estuary swamp, coastal plain, and the low hills below the piedmont. Sub sequent reconnaissance has documented many more sites
2001
393
in each of the five environmental zones outlined above. Sites from many time periods were located in each zone, but on the coastal plain, irrigation and drainage trenches provided a unique opportunity to document subsurface re
mains across the entire area (Orton 2000: 90). As most of the areas surveyed in 2002 had been trenched, it ismuch easier to assume a correspondence between surface and subsurface remains than is the case with most surface sur veys (Downum
and Brown
1998; Dunnell
and Dancey
1983: 269-270; Simmons1998; Tolstoy andFish 1975).
The well-known highland Mexico surveys began with extensive survey coverage (Blanton et al. 1982; Kowalews ki et al. 1989; Sanders, Parsons, and Santiey 1979). This extensive strategywas followed by a number of lowland
Gulf Coast
surveys (Santiey and Arnold 1996: 226; Sant iey,Arnold, and Barrett 1997; Symonds, Cyphers, and Lu nagomez 2002). The main advantage of an extensive ap
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
394
Figure e-w.
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
3. Photograph Perpendicular
site facing ne, and the 3 m deep drainage m are indicated arrows. spaced every 30 by
of the Cuauhtemoc 1m deep ditches
proach is that large areas are quickly covered at relatively low cost (see Kowalewski 1990). The disadvantage of an extensive approach to regional survey, however, is that the extents of temporal segments of a are determined given site ceramic in the identification field through phase by the sur
vey director (Blanton et al. 2005: 3). The results produced an approach are not replicable. The reliability of using such results can be biased by surveyor familiaritywith ceramic types,weather conditions on different days or in different
seasons, different land-use practices or ground visibility, and a host of other factors (e.g., Cowgill 1990; Terrenato and Ammerman 1996; Orton 2000: 103; Hawkins, Stew art, and Banning 2003). A more intensive approach was adopted by the SFP sur vey.While the SFP survey proceeded more slowly than if an extensive strategy had been employed, it produced on data both the temporal assignments of site replicable components and the artifact density fall-off at site limits. An intensive collection strategy helped ensure thatmateri als representing the fine-grained Soconusco chronology would be detected; extensive surveymethods are less like ly to recover diagnostic ceramics from all time periods
(Cowgill 1986; Redman 1987).
canal which
runs
80% of the area covered by the SFP in 2002 was employed for banana production at the time itwas sur veyed or had been trenched for banana production and re cently planted with mango or papaya trees.The production of each of these fruit crops entails keeping underbrush to a minimum using chemical defoliants and machetes. While this is the ideal, when plantation owners experience cash flow problems, the number of people employed to remove Over
vegetation decreases. The result is thatwhile the number of artifacts brought to the surface from cutting trenches may have been similar from one plantation to the next, the amount of ground cover varied considerably.
These fruitplantations ranged in size from 20 to 100 ha. Such large and clearly defined areas facilitated the acquisi tion of permissions to survey from landowners and pro vided a means to ensure complete survey coverage. Two survey crews operated simultaneously and usually covered a 40 to 50 ha plantation in one day. Ten people thus docu mented half a square kilometer each day, on average. Reg ularly spaced drainage canals had been dug on most of the coastal plain every 100 m to a depth of 3 to 4 m to reach a substratum that consists of thick gray clay. These deep can be many kilometers in drainage canals length and emp
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
Journal ofField Archaeology/Vol.33, 2008
ty either into adjacent rivers or into the swamp and estuary systems described above. Perpendicular to these canals, 1 m to carry excess m deep ditches had been dug every 30 rainwater away from the bananas' roots. Figure 3 shows a 3m deep drainage canal that runs through the center of the
site and empties into the swamp 2 km away. Note the regularly spaced ditches (indicated by arrows) that drain rainwater into the canals. Cuauhtemoc
The SFP survey documented the number and distribu tions of hectares of the surface thatwere covered by arti facts attributable to 20 ceramic phases (see table i). Re one numerous to of surface is docu survey ways gional ment past demography (see Schacht 1981; Kolb 1985; Paine 1997 for reviews of other methods), but it is the on
can on a lymethod that systematically record data region al scale (Fish and Kowalewski 1990; Sullivan 1998; Ban 10-12). The correspondence of site area to ning 2002:
population size has long been assumed by archaeologists (Childe 1951: 34; Schwartz 1956). There are studies that show a general correlation between settlement area and
population (e.g., De Roche 1983) while others argue that the number of houses or their floor areas provide a better
estimate of population 1981; Schreiber and (Hassan Kintigh 1996; Curet 1998: 371). It ishard to fathom how house floor area would be documented in the tropical al luvial environment of the Soconusco without excavation. As this is not realistic on a regional scale, site surface area remains the most
viable
(if imperfect) option for docu menting Prehispanic population dynamics. Yet it is not possible to interpret a surface scatter of cultural remains as or not in the absence of mounded architec being a house ture. Even when mounds are present, they could have been created as the result of non-residential activities thatwould require excavation to determine. One of themost serious problems with surface survey projects studying long-term settlement changes is that deeply buried sites are systemat
icallyunder-represented (Tolstoy and Fish 1975; Johnston 2002). The survey results presented here partially mitigate this problem because plantation trenching systematically brings deeply buried materials to the surface. In the tropical environment of the Soconusco, it is of ten difficult to determine if artifact scatters located near each other should be considered a single site. It is also dif ficult to know what proportion of sites remain buried in the alluvial environment. The method presented here was intended only to estimate site size based on surface re
mains. Although sites are concentrations of past cultural re that are documented in the present, we assume that there is a relationship between sites and communities of people in the past. This assumption is an area of archaeo logical inquiry still in its infancy,which I cannot test here.
mains
395
Systematic Walking SFP survey crews consisted of five people?a survey di rector with two workmen on each side; crew members walked in a line and spaced out at 15 m intervals in order tomaintain visual contact with each other through the veg etation. Survey crews used access roads and the 3 to 4 m deep canals to define the limits of sub-sections of planta tions within which to walk. The use of the drainage canals as survey boundaries avoided having to scramble in and
out of these sometimes formidable obstacles (see fig. 3). Further, covering the area between evenly spaced, 100 m wide sections of the plantation ensured that all land was covered. The survey crew members thus walked back and forth between drainage canals. This meant that theywalked m parallel to the smaller, 1 deep drainage ditches. As these ditches were the sources of much subsurface disturbance,
this ensured thatmost cultural remains were encountered. With a survey intensity of 15 m between survey partici pants and drainage ditches spaced every 30 m, coverage was total for sites larger than 0.1 ha (Orton 2000: 71). In other words, therewas 100% detectability (sensu Shennan 1997: 390-393) of sites larger than 0.1 ha, sufficient to en counter the remains of all households (Stark 1991: 42).
SurfaceCollecting and GPS Coordinates In the trenched banana plantations, sites are relatively easy to recognize by the presence of artifactson the surface. Equally important, the absence of cultural materials in these trenched areas makes the lack of sites obvious. Dur ing the initial 2001 at the Cuauhtemoc was used as a "dog units. The collection
season, surface collections were made site at 90 m intervals; a rope 4 m long leash" to define 50.24 sq m collection strategy for the 2002 season was sim ilar except that the interval between collection units was re duced to 30 m and the area collected was reduced to 28.26 sq m by using a rope 3 m long. This modification meant that surface collection areas (in sqm) were halved and col lection unit spacing was increased threefold to record each site at a finer resolution. We determined that halving the areas collected in 2002 allowed for the same chronological resolution as the 2001 results due to the quantity of tem porally diagnostic pot sherds recovered. A sitewas identifiedwhen any survey crew member en countered a significant number of (usually) pot sherds. team members Survey stopped walking so that the survey director could confirm that a sufficient artifact was density present and, in cases where itwas, assign the site a number. When this occurred, the other three surveymembers con on the survey director who verged spread them out ap m 30 each from other. The proximately precise location of
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
396
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
Table 1.Number of hectares of occupation for the 20 Soconusco ceramic phases identified in the Soconusco Formative Project survey (JohnClark, personal communication 2008).
Period
Calibrated years B.C./A.D.
Ceramic phase
1500
Late Postclassic_
Cuauhtemoc Survey ha ha ha
Late
04
Postclassic
04
Total
0
1350 04.1
Remanso
Postclassic
1000 Late 700
Peistal
Slsic
22-8
22"8 0
Metapa
600
Loros
500
Kato
400 Jaritas Late Formative/ 0 o n 0 Q
Classic Early
200
|stapa
v^ ^
100 Hat0
B.C./A.D.
100
Guillen
350 Frontera 550
Middle Formative
_
Escal6n
750
Duende
850
0 0.4 0.4
0
0
0
00 0
Conchas
17.9
7.7
Jocotal
9.5
16.57.0
Cuadros
0.2
4.0 3.8
Cherla_0_^8_3.8 0.5 Ocos
4.3 3.8
25.6
1000 l?teEa+r|y Formative
1200 _1300 1400 1500 early Early Formative
Locona
2 50
Barra
0.1
17
1
1700 3.9 3.8
~ 1900
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
Journal ofField Archaeology/Vol.33, 2008
each collection unit was established when the survey direc tor sunk a 0.5 m high wooden stake (that each survey member carried) into the ground. Locations were selected to avoid banana treeswhich were relatively small in the ear as ly spring most had just been planted. Next, each stake's location was recorded with a handheld Global Positioning was a assigned. System (GPS) and collection unit number were Garmin Etrek Venture GPS units employed and pro vided coordinates that had a 3 to 5 m horizontal error rat not obscure ing inmost plantations where vegetation did were also record our view of the sky.The limits plantation ed and were then checked against georeferenced aerial pho tographs. A rope 3 m long attached to each stakewas used to de fine the extent of each collection unit. All cultural materi als fromwithin these circular 28.26 sqm surface units were
collected. Machetes were used to cut vegetation and in crease visibilitywhen cover required it, thus ensur ground were that all surface remains encountered. Once arti ing factswere collected and stowed in their backpacks, survey members were assigned to collect artifacts from new col
lection units. Survey directors recorded GPS readings for each unit, and drew sketch maps of each site, which in cluded schematic representations of all collection units rel ative to permanent features such as drainage canals, roads, and buildings. Directors also filled out site forms that recorded basic information: plantation names and GPS co
ordinates, crops planted, degree of surface visibility, etc. Surface collection units extended beyond the obvious limit of dense artifact scatters to quantitatively document artifact fall-off at the site edges (see below). This method is much more economical than documenting all "non-site" areas (Dunnell 1992; Dunnell and Dancey 1983), yet still provides a quantitative basis to determine the distribution of artifacts across the landscape (see Kanter 2008: 46). For the purpose of this survey project, a sitewas thus defined simply "...as a contiguous spread of artefacts" (Orton 2000: 126). The goal of the SFP surveywas to determine the number of sites, the size of sites, and the distribution of sites across the landscape. The trenched banana planta tions provide one of those situations where".. .the edges of a site seem quite obvious because the archaeologist can eas ily distinguish the bounded area containing material cul ture from surrounding space with no artifacts at all" (Kan ter 2008: 44). A density of at least two artifacts per sq m was used as the threshold to define site edges in the heavi areas disturbed In where land-use results in ly plantations. fewer disturbances, a lower threshold may be necessary to provide comparable assessments of site size.
No knowledge of ceramic typology was necessary in the field as temporal assignments were made after the sherds
397
were washed. This reduced the time required to document each site and inter-observer error that could occur due to factors such as survey directors of differing experience, days with differentweather conditions, and other confounding (Hawkins, Stewart, and Banning 2003). This a requires minimum of five collection units to doc ument a 0.1 ha site because after an initial unit is collected, factors
method
fourmore
surface units must be collected to document
the
site's edges. coordinates of all collection units were down GPS loaded into a laptop computer. This was done everyweek or as the GPS unit's storage limit was ap 200-point were All collected artifacts washed in the field proached.
laboratory and each class of artifact was counted and weighed. During the two month 2002 season, 860 units were
sq km area around the Cuauhtemoc site. I later analyzed each collec tion unit and recorded the presence of temporally diag nostic sherds. collected and 80 sites defined
in a 28
Generating SpatialData The temporal assessment of each collection unit was en tered and linked to its respective GPS coordinates using ESRI's ArcView 3.2 software. Using the "Assign Proximi
ty" function in the Spatial Analyst extension, polygons were generated that plot artifact densities, and polygons with more than 56 artifacts (i.e., > 2 artifacts per sq m in a 28 sq m collection unit) were used to define site limits. The presence/absence of sherds from each ceramic phase was recorded as a "1" for presence or "0" for absence for
each collection unit. For each period, a map of the survey zone was generated. These results allow for changing site size to be plotted within the survey zone based on GPS co
ordinates. The temporal designations of sherds from col lection units were used to determine site occupation by phase.
The Cuauhtemoc site provides an example of how this works at the site level. Figure 4 shows the site represented as a series of dots that correspond to the GPS coordinates at the center of each collection unit. Each dot was linked to a table of sherd density values and used to create a polygon at limited themidpoint by the next dot with ArcView's As sign Proximity function. This function assigns each cell to the closest dot and treats entire same as the polygons the data linked to them. Therefore, the content of each collec tion unit is projected into the entire polygon. Collection units with less than two artifacts per sq m were eliminated
from further analysis and represented as white in Figure 4. This provides a consistent manner with which to establish the site's limits as themidpoints between collection units. While polygons with less than two artifacts per sq m were
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
398
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexko/Rosenswig
~ i
0
1
y
80m
_:_i Golection context
^^^^^m1111 . Sherds/sqm
: \zz\ 0-2
20.1-40 40.1-60
I
I*
^
il1>
11'm1
.
^^^^Jf^^^^^^^^^^^y
^HHHH
^^^^^^^^^^^Bffilhk. ^^^^^^^^^^^^Hfl^^^^^
Drainage canal
Figure 4. Densities pictured in Figure
of surface sherds from Cuauhtemoc
collected
in 2001.
3.
eliminated from further analysis, this does not mean that there were no temporally diagnostic sherds recovered. In the trenched, bulldozed, and plowed plantation environ ment, many more artifacts are visible on the surface than is case. normally the Determining site limits in thisway pro a vides conservative estimate of site size and attempts to minimize
the effectof artifact dispersal through plow 7ing
(Curet1998: 364).
Figure 5 represents the temporal associations of the sherds recovered in each collection unit plotted by each phase of the site's occupation. These represent the hectares of occupation through time in a quantitatively consistent manner (table i). Based on two seasons of excavation at (Rosenswig 2005), we know that the poly gon missing in themiddle of the site during the Conchas cal. b.c.) is the result of bulldozer ac phase (1000-850
Cuauhtemoc
tivity scraping away this final phase of the site's occupation. In addition, the three collection units in the eastern part of the site contain high numbers of Jocotal and Conchas
phase material but do not meet the 56 artifact cut-offpoint and so were eliminated from consideration in defining the site limits.However, from knowledge gained through ex
canal is that
Drainage
cavation, we know that thiswas an artificially raised area of the site. Artifact density from this construction fill is con
siderably lower than from habitation and midden deposits and so fell below the cut-off point. At the site level, these issues can be explored (e.g., Rosenswig in press b) but it is unrealistic to address such site formation issues at a re to our survey method, scale. According the gional sitemeasured 7.7 ha during itsmaximum ex Cuauhtemoc tent in the Conchas phase. This is less than the 10 ha that was determined during excavation (Rosenswig 2005). For the sake of regional consistency, such factors were not in corporated into the estimates of hectares of occupation for Cuauhtemoc. As stated above, thismeans that the results underestimate the number of hectares over which debris
was
spread.
Results Data from the 28 sq km survey zone are reported as the total number of hectares of occupation for 20 ceramic
phases (table i). The numbers of hectares of occupation are a presented in graph with three peaks: in the Locona phase, the Conchas and Jocotal phases, as well as the Late
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
Journal of^FieldArchaeology/Vol.33,2008
._Barra
Locona_ _066s
.';
Cuadroe
399
?
:"':.
l?|%ff^^
I__^_I
.
_Conchas_
Figure
5. Areas
of occupation
of Cuauhtemoc
I_]
I_
.
_J
N
by ceramic phase.
Classic period (fig. 6). Cultural remains from all periods were collected but the temporal units used to present the are uneven data due to my knowledge of the ce survey ramics and the goals of the SFP (Rosenswig 2005, 2006, 2007). The first 10 ceramic phases thatmake up the Early and Middle Formative periods were each identified; how ever, the ceramic phases in the Late Formative and Early Classic periods were lumped together. Late Classic and were identified, as Early Postclassic Plumbate ceramics were Late Postclassic red-on-white wares and distinctive "Aztec-style" avian effigy tripod vessel supports. As all sherds from each collection unit are curated at the New World Archaeological Foundation in San Cristobal de las Casas, Chiapas, the assemblage can be reanalyzed in the fu ture to focus more on other specifically periods and other research
Jocotal
;
issues.
In the tropical environment of the Soconusco, it is of ten difficult to determine ifdiscrete artifact scatters located near each other should represent a com single Prehispanic to know what proportion of such munity. It is also difficult
a commiinity may remain buried in the alluvial environ ment. By limitingmy discussion of survey results to the to tal number of hectares, I avoid the issue of defining site limits and converting sites found in the present to com inhabited in the past. I assume, all things being that older remains aremore likely to be buried than equal, aremore recent ones; if the number of hectares of occupa tion increases progressively over time this could be a prod uct of a systematic recovery bias; however, when debris from earlier periods covers more area (as is evident in fig.
munities
6), the patterns can be tentatively interpreted as changes in past demographic patterns. An additional aspect of analytic ambiguity is introduced
into the interpretation of survey results if the number of hectares of surface remains is converted to an estimate of
the number of people who lived in the region. While pop are necessary to figures explore issues such as if population surpassed carrying capacity of available land, it is unnecessary to simply document population changes over time in relative terms. Any universally applied con
ulation
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
400
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
irry_;_ Formative
Figure 6. Number
of hectares of occupation
s
MMda Early
by ceramic phase
and period
version figure used to transform hectares of artifact scatters to numbers of not alter relative patterns. Sur people will terms in of hectares of occupation vey findings by are summarized in phase/period Figures 6 through 10. In the remainder of this section, results are discussed in terms of local culture history (Blake et al. 1995; Love 2007) and changing environmental conditions (see Neff et al. 2006).
Barra,
Locona,
and Ocds Phases
(1900-1400
cal. b.c.)
is the largest site in the area during the Cuauhtemoc Barra through Ocos phases (fig. 7). The greatest number of hectares of occupation from this 500-year period be to the Locona longs phase which represents the first (and sur of three smallest) population peaks documented in the zone are These consistent the sit results with vey (fig. 6). zone are uation in theMazatan where rank societies docu at this time (Clark and Blake 1994). Clark (1994, 2004) documents a Locona-phase population increase and a two-tier settlement at the hierarchy. The excavations Cuauhtemoc site and survey from the surrounding area in dicate that at this time therewas a center of early complex ityat the southwestern end of the Soconusco. Cuauhtemoc
mented
measured
5.12 ha during the Locona phase and all other sites in the survey zone combined measured 2 ha, also pro a two-tiered settlement hierarchy. On the ducing side of the border, 21 siteswere documented Guatemalan for the early Early Formative period (Love 2002a: (Love 2002a: fig. 9). including 34 mounds
fig. 10)
.Late
Classic
Postetassic
in the survey region.
On either side of our study area population levels are lower. To the nw, Voorhies (1989: 116) documented 12 sites from the (combined) Early andMiddle Formative pe riods, but as noted previously, she expects sites to be un
due to lack of surface disturbance in the survey zone. Southeast of the Guamuchal/
der-documented
Acapetahua Manchon swamp, a few Early Formative mounds were documented in the Jesus River zone along the edge of the mangrove (Pye 1995: fig. 7.1). This dispersed settlement is all along theGuatemalan coast dur documented pattern the Formative ing Early period (Arroyo 2003). The land between the two large Soconusco swamp systems thus ap island of demo pears to correspond to a pronounced and graphic growth political development during the first half of the Early Formative period
(Rosenswig
in press a).
Cherla, Cuadros, and JocotalPhases (1400-1000
CAL.B.C.)
There was a decrease
in the number of hectares of occu
pation during the Cherla and Cuadros phases followed by a marked increase during the Jocotal phase (fig. 8). The site has the majority of occupation hectares Cuauhtemoc documented during the Cherla and Cuadros phases, but only 7 of the 16.5 ha of occupation during the Jocotal phase (table i). This initial drop and Jocotal spike paral in theMazatan lels the pattern documented survey zone
(Clark1994). Clark (1997; Clark andPye2000) proposed
that the center of political activity shifted out of his survey
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
^ Journal ofField Archaeology/Vol.33,2008
/^^^^^
401
BarraLocona
I
^
\
J
i?'
Cuauhtemoc
*'
7
v.
??
) *
J
[
km
0_l_2
/ Ocos
*
1
Figure
7. Survey findings from the early Early Formative
Barra, Locona,
zone to the Coatan River during the Cuadros and Jocotal Suarez 2002; Cheetham 2006). The drop in phases (Perez occupation during the Cuadros phase in theMazatan zone
thus be attributed to the survey block becoming po litically marginal. The dramatic increase in occupation hectares during the Jocotal phase may be due to the re survey zone or to a different population of theMazatan
may
type of settlement practice resulting in people moving more frequently and covering more area (see Curet 1998: and may ex 363). The same may be true at Cuauhtemoc plain the extent of ceramic sherds from the Jocotal phase a lower along with density of remains, possibly suggesting shorter occupation spans (Rosenswig in press b). In theNaranjo zone, 21 siteswere documented for the
(
and Ocos
phases.
entire lateEarly Formative period (Love 2002a: fig. 11) in (Love 2002a: fig. 9). With the same cluding 26 mounds number of sites and eight fewermounds, population may have also decreased at this time in theNaranjo zone. On
the other side of the Guamuchal swamp, no Cherla or Cuadros phase sites are reported in the Jesus River zone but a few Jocotal phase sites are documented in addition to El Mesak with its significant occupation during this final phase of the Early Formative
Conchas
Phase
(1000-850
(Pye and Demarest
1991).
cal. b.c.)
Hectares
of occupation increased further in the zone the Conchas phase (fig. 9). This during is the second peak in the numbers of occupation hectares
Cuauhtemoc
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
402
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
Cherla
/?i
/
/
Cuadros
Cuauhtemoc / (
"
/
I
Cuauhtemoc
*
Ell
/
L,
I-'
.
)
'?-/N km
0_|_2
/
H /-'
4
Jocotal / ^_
*
^_ ^^^^ #
^
7
Cuauhtemoc w 7
?1
/
I
Figure
8. Survey findings from the late Early Formative
1-1
Cherla, Cuadros,
(fig. 6) in the survey zone when the greatest area (25.6 ha) was occupied. The Conchas phase is the timewhen the La Blanca site, on the west shore of the Naranjo River, ex tended over at least 100 ha (Love 2002a). In theNaranjo
zone, 129 mounds were encountered dating to the early Middle Formative Conchas phase (42 of which are at La Blanca) (Love 2002a: fig.9) and 60 siteswere encountered from this phase (Love 2002a: fig. 12). The La Blanca poli at the time and the tywas the largest in the Soconusco as a to have been site Cuauhtemoc appears integrated to center La Blanca's rise third-tier regional promi during nence (Rosenswig 2007).
and Naranjo The results of the Cuauhtemoc surveys contrast with the pattern in both theMazatan and the Je
+
/ ^^^W
/
and Jocotal phases.
sus River survey zones where no Conchas phase sites have been found (Blake and Clark 1999; Pye and Demarest 1991). Blake and Clark (1999) interpret this as represent was drawn to the political a ing timewhen the population see Love 2002a). development of the La Blanca polity (and Blake and his colleagues (1995: 180) note that 56 of the 70 known Conchas phase sites are located in theNaranjo River drainage. The SFP survey has documented another zone two dozen Conchas phase sites in the Cuauhtemoc and provides quantitative data to compare with Clark's (1994) survey results. The beginning of the Conchas phase corresponds to in creased stability and productivity of the local environment (Neff et al. 2006:
397). Current
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
evidence also indicates
Journal ofField Archaeology/Vol.33, 2008
Conchas -^^^
0_|_2
/ ^^^^
Figure 9. Survey findings from theMiddle
that thiswas
Formative
Conchas,
Duende,
a time when
firstmade
graphic reorganization of the region, agricultural depen dence, and a three-tier settlement hierarchy (Rosenswig
2007).
Escaldn, and Frontera cal. b.c.)
Duende, (850-350 a
Phases
or Escalon single Duende phase settlement was encountered in the 28 sq km Cuauhtemoc survey zone, so this small region appears to have been abandoned after the Conchas phase population peak (figs. 6, 9; table i). No Not
Duende and Escalon
km
Frontera
a threshold was passed and the a to commitment population agri et culture (Blake al. 1992; Rosenswig 2006). Therefore, this adaptive transition accompanied a significant demo
Soconusco
403
Escalon,
and Frontera phases.
Formative period remains were encountered in the JesusRiver zone (Pye 1995: 151-153) or theNaranjo zone but this is likely due to these phases being poorly de fined at the time the surveyswere conducted and remains from these phases were likely lumped into the Late For mative Crucero phase (Michael Love, personal communi cation 2008). In theMazatan zone, the site ofHuanacastal was established as a on the coastal regional center plain during this time (John Clark, personal communication 2004) as was El Ujuxte in Guatemala ne of theNaranjo lateMiddle
and Jesus River zones (Love 2002b). It appears that dur ing this time a number of larger polities were first estab lished in the region with uninhabited zones between them. It was also during this time that Izapa emerged as the
This content downloaded from 169.226.144.89 on Wed, 21 Jan 2015 15:35:21 PM All use subject to JSTOR Terms and Conditions
404
Prehispanic Settlement in theCuauhtemoc Region, Soconusco,Chiapas,Mexico/Rosenswig
Late Formativeand Late Classic
i\
/
~-,_/
r
*
.\
J. * L
?"-/
1 /
* \ A
r' ??,
j
r1 r
/ J n
"
/
4 ^?* #?