Early Sedentism and Agriculture
in the Northern Tucson Basin

Suzanne K. Fish, Paul R. Fish, and John H. Madsen

By the eighth century A.D., when emergent community organization can first be discerned in the northern Tucson Basin, regional inhabitants had already been cultivators for at least 1500 years. Only with the recent proliferation of radiocarbon dates for Late Archaic cultigens within the Hohokam domain has it become clear that farming preceded the appearance of pottery by as much as a millennium; therefore there are conceptual difficulties in considering the earliest ceramic occupations as times of "initial" agriculture. Both Late Archaic and Pioneer period settlements are without doubt underrepresented in survey data due to the relative scarcity of diagnostic artifacts and prolonged exposure to postoccupational processes. Nevertheless, the similar distributions of sites within this broad combined time span is striking (Fig. 2.1). This chapter emphasizes implications of Late Archaic settlement, locational continuities into early ceramic times, and the establishment of basic patterns of agricultural settlement.


Agriculture is already a significant factor in the earliest period for which settlement patterns can be obtained in the northern Tucson Basin. There is a virtual absence of evidence for the cultural and demographic background from which these first farmers arose. A long line of local predecessors is suggested by a Clovis point fragment (Agenbroad 1967) and a handful of Middle Archaic projectile points. Middle Archaic points, dating between approximately 5000 B.C. and 1500 B.C., occur in both upper basin and riverine zones of the Marana study area. Intensive later reuse of key locations such as springs and the scarcity of earlier Holocene alluvial surfaces on the bajadas, some of which experienced mid-Holocene scouring (Schuster and Katzer 1984), inhibit survey detection of pre-agricultural populations.

Although the Sonoran Desert today is exceptional for its array of edible plants for hunters and gatherers, packrat midden studies indicate that current distributions of species became established in the low southern deserts of Arizona only about 4000 years ago (Van Devender and Spaulding 1979; Anderson and Van Devender 1991). Sites dating substantially prior to 1000 B.C. have not been studied in the Tucson Basin and evidence for subsistence is lacking regionally. However, elsewhere in the southern portion of the Southwest probable structures (Martin and Rinaldo 1950: 430; O'Laughlin 1980), storage pits (Windmiller 1973; O'Laughlin 1980), and large sites with diverse artifact assemblages and high densities of well-made ground stone (Whalen 1971; Bayham and others 1986; Sayles and Antevs 1941; Agenbroad 1970; P Fish 1967) suggest extended seasons of residence in some locations before the addition of cultigens.

The perception of Late Archaic occupations and often even the earliest ceramic ones as constituting a unitary "stage" in Southwestern cultural development, coupled with fragmentary regional data, tends to encourage the use of broad hunter-gatherer analogies in interpretation. For example, a generalized model of mobile band economy as reconstructed by Steward (1938) for the Great Basin has been widely applied across the regionally diverse environments of the Southwest. The possibility of a significant degree of sedentism in this early portion of the Southwestern archaeological record is seldom raised. For preceramic time, pronounced mobility in a seasonal round could be described as a widespread article of faith. However, analogy should be used even more judiciously for earlier than for later prehistoric times. Ethnographic observations of hunters and gatherers and less committed cultivators may not be representative of pre-agricultural settlement and subsistence in favored locales (Freeman 1968; S. Fish and R Fish 1991; Mueller-Wille and Dickson 1991). Historic hunters and gatherers in the Southwest have been geographically

Figure 2.1. Late Archaic through Pioneer period settlement in the Marana survey area.

marginal to the territories of successful agriculturalists or have been strongly influenced by the presence of domesticated animals and other postcontact elements.

Direct ethnographic analogs for early cultivator sites are lacking in the accumulating evidence from survey and excavation in the Tucson Basin. Hunter-gatherer subsistence patterns have not been recorded historically or ethnographically in those areas of the Sonoran Desert where optimal environments for initial cultivation occur. The Seri and Sand Papago, Sonoran Desert people to the south of former Hohokam territory who practiced little or no agriculture, lived in regions receiving half to one-third of the relatively generous 250 mm to 300 mm (10 to 12 inches) of annual rainfall in the Tucson area. The Tohono O'odham (Papago) Indians near Tucson, who in historic times moved between winter villages and summer farming settlements, did not also occupy stretches along the Santa Cruz River with sustained surface flow; Hispanics, Anglos, and missionized Indian populations dominated those locales. Furthermore, the high water requirements of numerous cattle curtailed the duration of Tohono O'odham settlement in the vicinity of moderate water sources that might otherwise have sustained human needs (S. Fish and Nabhan 1991). By the time of ethnographic record, Apache residence patterns reflected horses and other livestock, guns, and a postcontact economy heavily influenced by raiding and trade with non-Indians.

Due to a lack of comprehensive settlement patterns for the Hohokam tradition, other than outlines derived from large irrigation networks, it has not previously been possible to place the relatively few identified sites of Late Archaic and early ceramic date within local or regional frameworks of contemporary settlement. In the northern Tucson Basin, sites recorded by systematic survey are modest in number but nevertheless exhibit a cohesive subset of locational correlates within the broader range encompassed by later occupations. In turn, the broader array of situations and more complete patterns during subsequent periods aid in understanding the overall agricultural potential of basin settings. In this way, regional settlement patterns provide a supplement to broad analogies for the interpretation of environmental selectivity and other aspects of early site distributions.

Similarity in some settlement pattern attributes with those of later times, and the occurrence of cultigens at most excavated sites, provide a basis for considering extended residence a likely correlate of food production for many Late Archaic and early ceramic residents of the Tucson Basin. This position is counter to the weight of recent archaeological opinion concerning early cultivators in the Southwest (for example, Simmons 1986; Minnis 1985; Gilman 1987; Gumerman 1969; Rice 1980; Hard 1990; Nelson 1990) and undoubtedly is one of the more controversial conclusions arising from the present study. Issues regarding degrees of sedentism may not be conclusively resolved even with excavated data, which in the Marana study area are drawn from limited exposures at a few sites. Therefore, the proposed implications of Late Archaic and Pioneer period settlement patterns should be considered preliminary but are based on: (1) the unique availability of systematic distributional evidence from full-coverage survey; (2) attributes of regional environment in light of ethnographic analogy and later prehistoric land use; and (3) the model of agricultural transition and the potential for sedentism advanced in the following discussion (see also S. Fish, R Fish, and Madsen 1990a; S. Fish and R Fish 1991).


Variables Favoring Sedentism

In terms of multiseasonal abundance, diversity, and storability of plant foods, the Sonoran Desert has been described as one of the truly rich areas for the gatherer in North America (Nabhan 1985; Felger and Moser 1985; S. Fish and Nabhan 1991). Two seasonal peaks of rainfall in southern Arizona support a distinctive array of productive leguminous trees and succulents in addition to edible annuals. Linear valleys of the basin-and-range topography create elevational diversity within short trans-basin distances; most resources could have been acquired on a daily basis from settlements near permanent water at the river or mountain edge. In southern Arizona, constellations of conditions in favorable environments are consistent with both a substantial degree of preexisting sedentism and the early adoption of agriculture into compatible economic frameworks.

Of the critical combination of water, staples, and diverse resources in proximity that would permit extended residence in pre-agricultural times, perhaps the most limiting element in the Tucson Basin is a dependable water supply. Prolonged or permanent water in the Santa Cruz River occurs where igneous intrusions force underground flow to the surface. A second topographic location of permanent sources fed by upland watersheds is at the bases of the larger mountain masses bordering the basin.

The advantages of topographic and ecotonal diversity are well recognized (Flannery 1968; Gumerman and Johnson 1971). Diversity within short horizontal distances promotes the efficiency of logistical (Binford 1980, 1982) or more sedentary hunting and gathering strategies by increasing the range of resources within a convenient radius about a source of major dietary staples or longterm water. An optimal environmental constellation could be defined as: (1) a sustained water supply; (2) a dependable resource concentration, including an abundant staple; and (3) convenient access to environmental diversity. Indeed, such constellations should predict optimal locations for the earliest transitions to agricultural economies.

Both redundancy and diversity of resources contribute to the probability of a high degree of pre-agricultural sedentism in favorable locales of the Tucson Basin. Riparian species such as mesquite or saltbush occur widely but are densest near long-term water sources. Duplicate sets of diverse resources on opposing valley slopes or bajadas are accessible without residential mobility. Cacti such as saguaro, cholla, and prickly pear may be locally most dense on mid to upper bajadas, but distributions extend to the valley floor in many locations. The few additional species on relatively low mountain ranges such as the Tucsons and Tortolitas in the study area are generally within a day's round trip from a residential base near water. Only the occasional massive ranges of southern Arizona such as the Catalinas (Fig. 1.5) and Rincons east and south of the northern Tucson Basin study area offer unique (and, archaeobotanically, seldom documented) high-elevation resources. In fact, the mid to upper elevations of these mountains provide sparse evidence for prehistoric use during any period despite extensive survey (Arizona State Museum Site Survey File; Coronado National Forest Survey File).

In prime locations, no more than biseasonal movement and even year-round residence should be considered as possibilities for low-mobility hunters and gatherers and earliest farmers. Archaic occupations in these locations would have embodied a precondition of minimal conflict for the addition of cultivation to existing economic activities and schedules. Substantial duration of residence is a relevant variable both for situations of primary domestication and for those of secondary acceptance as in the Southwest. From this perspective, restricted residential mobility may be as much a prerequisite for a successful transition to agriculture as the result of such a transition.

Staple Resources

A few exceptional resources are usually pivotal in supporting the most sedentary lifestyles among groups who hunt and gather, such as salmon among tribes of the Northwest Coast or acorns among central California Indians. Dependable wild resources likely played a similar role as staples among those populations who were the earliest to adopt domesticates in any region. For example, in the Tehuacan Valley of Mexico, foxtail millet or Setaria was so consistently abundant in coprolites that it has been considered a potential early cultigen, although not cultivated in historic times (Smith 1967: 249; Callen 1967: 287). Sites with early corn in the Chaco Canyon area of New Mexico occur in dunes with dense stands of Indian rice grass. Caches of the seeds have also been recovered in contemporary rock shelters (Simmons 1982, 1986). These key resources recall the wild grasses that were heavily utilized just prior to the agricultural transition in the Old World, some of which gave rise to the first domesticates. Focused dependence and intensive tending of indigenous species are now demonstrated among Archaic populations of the eastern United States well prior to the advent of corn and other Mesoamerican domesticates (Smith 1989). If such practices constituted a widespread late Archaic pattern in North America, the potential for residential permanence in optimal locales of the Southwest would have been enhanced.

Mesquite beans exemplify a pivotal resource in southern Arizona. The pod yield from riparian mesquite groves is sufficiently prolific to serve as a staple for moderate populations. Except for the most severe droughts, freezes, or floods, and in spite of annual variability in individual trees, mesquite groves along larger drainages produce harvests year after year (Felger and Moser 1971, 1976; Aschmann 1959: 53; Bell and Castetter 1937: 18). In historic times, trees were pruned (Bean and Saubel 1972: 108) and yields are known to have been augmented by ditch irrigation (Bean and Lawton 1973: 27). The processed flour is highly nutritious and may be stored for up to two years. A single Cahuilla Indian worker might gather up to an estimated 80 kg of beans per day (Bean and Saubel 1972: 112). Average Seri production figures for flour are 40 kg per day with one man harvesting and two women pounding and winnowing (Felger and Moser 1971: 57). Raw or parched pods were stored for a year or longer in large basketry granaries by more sedentary historic groups (Castetter and Bell 1951; Bean and Saubel 1972; Bartlett 1854). Mesquite groves in some locales are thought to have permitted nonmigratory lifestyles, among the California desert Cahuilla (Bean and Saubel 1972: 108).

Mesquites are distributed widely in the Tucson Basin but the largest and most abundant trees grow along the wide floodplains of drainages in the upper basin as well as in riverine groves. Both upland and riverine zones share optimal environmental combinations of potable water, accessible resource diversity, and this abundant staple. Thus, subsistence advantages involving both gathered and cultivated resources would have permitted extended residence at Late Archaic sites in riverine and upland zones of the study area.


Visibility and Discovery Bias

Archaeological visibility presents a significant obstacle to the reconstruction of Tucson Basin settlement patterns of Late Archaic through Pioneer period age. Populations closely preceding the adoption of cultigens in the Late Archaic are virtually invisible due to the lack of chronologically sensitive diagnostic artifacts or features. It is not known whether projectile point styles associated with Late Archaic cultigens, were also manufactured by hunters and gatherers during the several prior centuries. Furthermore, when isolated Archaic points are encountered at a site that also contains ceramics, it is difficult to distinguish between the presence of a Late Archaic component and the common Hohokam practice of retrieving older points.

Diagnostics are also a problem in the earliest ceramic assemblages of the Hohokam area, which contain no pottery types with painted or plastic decoration. In the recent past, initial plain wares were not readily distinguished from later ones in the Tucson area, although distinctive vessel forms are now known to occur (Huckell and others 1987; Bernard-Shaw 1990a). It is significant that the age of the earliest Tucson ceramic site now known was recognized only through the results of chronometric dating (Huckell and others 1987). Even for those subsequent Pioneer period occupations with decorated pottery, diagnostic sherds constitute no more than two percent of the ceramic assemblage and sherds assignable to particular phases are fewer still (see Chapter 6; Czaplicki and Ravesloot 1989b; Bernard-Shaw 1989; Kelly 1978).

Continuous occupation at hydrologically prime locations compounds recognition problems. Small artifact samples from alluvial profiles seldom contain decorated ceramics or diagnostic projectile points. Late Archaic and early ceramic sites are not all at great depths, but even in shallow contexts the rarity of diagnostics makes it difficult to distinguish early features interspersed among later ones. To achieve distributional data even partially comparable with settlement patterns of later ceramic times, both large numbers of sites and large artifact collections are essential.

Northern Tucson Basin Survey Patterns

The quantity of Late Archaic through Pioneer period sites in Figure 2.1 must be considered an abbreviated representation of occupation for the long time span in question. Surface indications of such age persist in a wide range of geomorphological settings, although site burial is greater for these earlier time periods in areas of active late Holocene deposition such as alluvial fans and drainage floodplains. The rarity of diagnostics, along with site burial, undoubtedly accentuate the perception of dual settlement clusters with few outliers, one concentration along the river near the end of the Tucson Mountains and another along the skirts of the Tortolita Mountains, separated by substantial unoccupied and unused territory. Loci of short-term activities that generated few artifacts are least likely to yield rare diagnostics, leading to an overemphasis on residential sites in these early settlement patterns.

Maximum size of individual sites and overall densities reflect the potential for concentrated populations in the vicinity of the Santa Cruz River. Domestic water, riparian resources such as mesquite groves, and water diversion opportunities impart long-standing attractions. As in ensuing ceramic phases, the largest sites occur near the river along the Tucson Mountains where igneous intrusions raise underground flow. Late Archaic facilities in the form of shallow wells or small reservoirs were constructed on the Santa Cruz floodplain to enhance and prolong water sources (Bernard-Shaw 1988).

An additional agricultural orientation for the band of settlement paralleling the river corresponds to cultivation of lower basin alluvial fans (described in Chapters 4 and 5). Alluvially active portions of the fans support dense stands of weedy annuals for gatherers and desirable conditions for floodwater farming through diversion of ephemeral drainages following storms. Recent excavations at three localities on these fans in the northern basin have revealed occupations containing Late Archaic cultigens (Chapter 6; Roth 1988, 1989; Mabry 1990). Aggradation and lateral movement of drainages across fans tend to obscure surface remains. Therefore, quantitative indications of fan settlement are underrepresented in overall distributions. All known Late Archaic and early ceramic fan settlements are located within 2 km (1 mile) of high water tables along the Santa Cruz. Although seasonal occupation of these sites is a possibility, structural remains are comparable with counterparts in floodplain sites of the Tucson Basin, and such distances for domestic water transport from the river are well within the ranges of ethnographic analogs.

The second concentration of Late Archaic through Pioneer period settlement is along the Tortolita flanks, again evoking a linkage with permanent rather than seasonal water. This is a preferred situation throughout the prehistoric sequence because accessible flow from mountain watersheds persists in drainages that provide

Figure 2.2. Cumulative frequency comparison of Late Archaic tool assemblages from a site
on an alluvial fan near the Santa Cruz River (AZ AA:12:486) and a site on the upper bajada
(AZ AA: 12:284). Adapted from Roth 1989: 171-195. Using the Kolmogorov-Smirnov
two-sample statistic, there is a probability of less than .001 that these samples
were drawn from different populations.

domestic and agricultural water. Early sites tend to be located in the vicinity of easily diverted, small drainages as they emerge from the mountain front, and a few occur farther down the bajada along major water courses.

Although the largest upper bajada Archaic and early ceramic sites do not attain the size of the largest contemporary settlements along the river, maximum size exceeds 15 ha (35 acres). Artifact densities tend to be lower at higher elevation sites, but artifact assemblages, including frequencies of ground stone, are similar in some intensively occupied sites of both zones. This is illustrated by a comparison (Figs. 2.1, 2.2) of systematic artifact collections from an alluvial fan locality near the river (AZ AA:12:486) and from a site at the base of the Tortolitas (AZ AA:12:284). Early ceramic sites in both upper basin and river-oriented locations have produced figurines and exotic items such as shell and include instances of trash mounds, the localized accumulations that typify Hohokam disposal of residential refuse.

Duration of Residence

A late historic Tohono 0'odham or Papago analogy for riverine and upper basin settlement clusters would suggest annual movement between upland winter villages near permanent wells at the mountain edge and more temporary summer residences with seasonal reservoirs near low basin fields. However, this sort of biseasonal movement in the late nineteenth and early twentieth centuries described by Underhill (1939) and others was only one alternative among contemporary residence patterns (Fontana 1983a, 1983b). Additionally, the water needs of livestock seriously limited the duration of summer supplies at lower elevation settlements that were dependent on artificial water impoundments (S. Fish, P. Fish, and Madsen 1990b; S. Fish and Nabhan 1991).

Locales of early upper basin sites duplicate the same constellation of prolonged water availability and concentrated resources found at the river. Riparian vegetation is less areally extensive along upland drainages than along lusher stretches of the Santa Cruz floodplain, but mesquite and other species are abundant. Nonwoody plants that produce edible seeds, including saltbush, grasses, and various annuals, are similarly concentrated along both the river and drainages of the mountain edge. Mountain hunting and upland plant species such as wild grapes or yucca add supplementary resources only in the more massive Tortolita Mountains on the eastern basin edge. These minor resources could have been obtained through seasonal camps of brief duration. However, outside the mountain canyons, vegetation similar to that of the upper basin interior is predominant on all but the highest peaks bordering the northern basin. The subsistence advantage of seasonal movement of residence from either one of the settlement bands to the other would have been minor in view of the largely redundant wild resources and similar seasons of availability.

Environmental variables and archaeological remains suggest a potential for comparable degrees of persistence for both upper basin and riverine settlement clusters. Tohono O'odham ethnographic analogy would identify the higher sites with more extended occupations as winter settlements. As previously noted, upper basin sites reach substantial size but the largest are along the river. There are no excavated examples that can be used to compare details of settlement structure or subsistence. However, the diversity and proportions of artifact types for some upper basin settlements are similar to those of riverine counterparts.

Sites could have been advantageously located for both natural resource catchments and agricultural opportunity in a relatively uncrowded landscape. Lower population levels during the Late Archaic through Pioneer period apparently permitted habitation to be confined to optimal zones to a greater degree than in succeeding times. Undisturbed habitats for hunting and gathering also would have been more widespread and pressure on exploited resources lower.


The linkage of Late Archaic as well as early ceramic settlement patterns in the northern Tucson Basin with the requisites of farming is based on both direct evidence and on the emerging picture of Late Archaic subsistence throughout the Tucson Basin and surrounding areas of southern Arizona. Corn has been recovered from Late Archaic proveniences in excavations of limited scale on the Santa Cruz floodplain (Bernard-Shaw 1988), on two alluvial fans in the Marana study area (Chapter 6; Roth 1989), and on a third alluvial fan just to the south (Mabry 1990). By early ceramic times, botanical analyses in both these zones clearly demonstrate reliance on the range of later cultigens, with only minor exceptions that may reflect limited sample size (Chapter 6; Bernard Shaw 1989).

An appreciable number of excavated Late Archaic sites with reliable complements of radiometric dates, subsistence remains, and artifact assemblages has now been recorded for the Tucson Basin and surrounding areas (Fig. 2.3, Table 2.1). However, these cases cannot be uncritically accepted as representative of overall spatial patterns. Compared to the systematic coverage in the northern basin, this sample registers a more pronounced influence of discovery bias. Recorded site locations largely coincide with intense modern development that prompts investigation and with floodplains of the Santa Cruz and major tributaries where channel cuts have created exposures. Excavations of any scope are few in the upper basin zones of permanent water and along mountain flanks that were used continuously for farming. Similar locational bias is evident among investigated ceramic sites through the Pioneer period.

Late Archaic sites for which subsistence analyses have been performed are dated as early as the beginning centuries of the first millennium B.C. and have yielded cultigens with remarkable consistency (Fig. 2.3, Table 2.1). When evidence is sufficient to first evaluate Late Archaic subsistence, domesticates are widespread. Almost every instance of analysis at a range of Tucson area site types and locations has revealed the presence of corn. Recovery at even small and ephemeral sites indicates the role of domesticates as important dietary elements. Furthermore, measures for cultigens at the earliest Late Archaic sites (Huckell 1987, 1988, 1990) are within the range of early ceramic settlements, and both compare favorably with later Hohokam riverine occupations (Miksicek 1988). Subsistence reconstruction suggested by these findings seems inconsistent with a predominance of seasonally mobile lifestyles in the Tucson Basin. Moreover, recent studies of diverse materials such as coprolites, charred plant remains, and human bone (Matson and Chisholm 1991; Minnis 1989; Wills and Huckell, in press) suggest that a number of Late Archaic and earliest ceramic populations across the Southwest shared a dietary reliance on corn.

Figure 2.3. Locations of Late Archaic sites in Table 2.1.

Table 2.1. Late Archaic Sites in the Tucson Area with Evidence of Corn
* Direct dates of corn remains.

The foregoing interpretation of Late Archaic and early ceramic remains in the northern Tucson Basin has been shaped primarily by the nature of settlement patterns rather than by comprehensive excavation at one or a few key sites. Perceptions based on study area survey have been subsequently strengthened by the accumulating densities of recorded sites throughout the Tucson area; by the presence of substantial structures, storage facilities, and burials in excavated residential contexts; and by the abundance and consistency of associated cultigens. Earlier versions of settlement history that postulated a peopling of the Tucson Basin by Preclassic immigrants from the Phoenix core and posited relatively late expansion away from riverine zones (Grebinger 1971; Haury 1976; Doyel 1977) have not been supported. Although the origin and incorporation of later organizational, ceremonial, and stylistic modes is not resolved by these findings, it is evident that a basic and persistent pattern of agricultural subsistence and settlement in the Tucson Basin was established well prior to the appearance of these distinctively Hohokam cultural forms. The locational and agricultural bases of the Marana Community are clearly foreshadowed in the settlement distributions of Late Archaic and early ceramic farmers.

Next: Chapter 3 | Previous: Chapter 1

UAP Online Books | UAP Home Page

The University of Arizona Press, 2/10/01 5:03PM