TEN

Late Classic Period Terrace Agriculture in the Lowland Maya Area

Modeling the Organization of Terrace Agricultural Activity

L. THEODORE NEFF

INTRODUCTION

The subject of this chapter is the organization of Late Classic period (ca. AD 550–800) ancient lowland Maya terrace agricultural activity. Agricultural terraces are embankments, typically constructed of stone but at times made of wood or augmented by living plants, placed perpendicular to hill slopes or drainages for the purpose of conserving or catching soil and catching or channeling runoff. Terraces are beneficial to agriculture because they create areas of wetter and deeper soils that are more conducive to plant growth (Beach et al. 2002:379; Donkin 1979:2; Kunen 2001:326; Treacy and Denevan 1994:95; Turner 1983). Terraces also allow greater cropping frequency than would otherwise be possible in hilly terrain and are thus typically associated with the process of agricultural intensification (Chase and Chase 1998; Dunning and Beach 1994; Healy et al. 1983; Johnston 2003; Kunen 2001; Murtha 2002; Neff 2008; Turner 1983). The ancient Maya civilization was located in what today is eastern and southeastern Mexico, Guatemala, Belize, and western Honduras (Figure 10.1). The Maya area is typically divided into highland and lowland zones with the highlands consisting of the mountainous areas of Mexico, Guatemala, El Salvador, and Honduras to the south and the rest of the terrain forming the lowlands to the north. Maya civilization began with the occurrence of early agricultural villages around 1500 BC and was decimated with the Spanish conquest in the early part of the sixteenth century. During the period running from ca. AD 550 to 800, known as the Classic period, Maya civilization was at its height in terms of population, sociopolitical complexity, and agricultural intensification.

Figure 10.1. Eastern Mesoamerica showing the study area location

Relic terraces are common features in the lowland Maya area (Dunning and Beach 1994; Kunen 2001; Turner 1974, 1983), as well as the rest of the Americas (Donkin 1979), and have been recognized as such since the first half of the twentieth century. Indeed, as early as 1936, as part of the Michigan-Carnegie Botanical Expedition to British Honduras (Belize), Cyrus Lundell observed numerous agricultural terraces on the Vaca Plateau in the far west-central part of the country. “On one hillside I counted no less than 51 terraces, and this was not an exceptional condition” (Lundell 1940:9). Lundell’s observations confirmed those of Ower (1927) and Thompson (1931), who had visited the same area a decade or so earlier, and all three concluded that the terraces were remains of the ancient lowland Maya civilization. Lundell (1940:11) made the following remarks about what these features indicated about the agricultural system employed by the ancient Maya:

That a shifting type of agriculture, such as the milpa system, would be employed in a terraced area is unbelievable. The building of stone retaining walls and the filling-in with soil call for an investment in labor which would not be expended for a form of agriculture where the land would be abandoned for eight to twelve years after one or two crops. Terracing indicates continued occupation of land and at least a form of semipermanent agriculture.

In the five-plus decades since the publication of Lundell’s remarks, ongoing settlement-pattern research has confirmed his observation regarding the Vaca Plateau area of Belize, as well as for numerous other parts of the lowland Maya area. Numerous agricultural terraces have been documented as dispersed among Late Classic period (ca. AD 550–800) structures (Chase and Chase 1998; Dunning and Beach 1994; Fedick 1994; Healy et al. 1983; Kunen 2001; Murtha 2002; Neff 2008; Turner 1974, 1979, 1983). Agricultural terraces make up roughly half or more of the surface-visible settlement traces in many areas (see Ashmore et al. 1994; Neff 2008; Neff et al. 1995) and constitute a substantial component of the built environment, along with residential and civic-ceremonial architecture.

Because of the large number of agricultural terraces associated with households in many parts of the lowland Maya area, agricultural activity associated with them must be an important component of household production. Beginning with the pioneering work of von Thünen in 1826 and continuing to the present, research on preindustrial, small-scale agrarian landscapes indicates that distance to fields is a primary variable regarding land use. Working with this premise, researchers in Mesoamerica have proposed models that characterize agrarian land use from an all-encompassing, top-down landscape perspective, as well as from a more focused, bottom-up household viewpoint. These perspectives tend to characterize agricultural areas adjacent to and interspersed among households, areas that often contain agricultural terraces (Chase and Chase 1998; Fedick 1994; Healy et al. 1983; Kunen 2001; Murtha 2002; Turner 1983; Wyatt 2005), as either zones of permanent or semi-permanent cultivation from the perspective of the larger landscape or as garden areas beyond the core area of the household. This chapter focuses specifically on these permanently cultivated areas. Little research, in particular archaeological excavation, has focused on the agricultural terraces that make up large portions of these zones in many parts of the lowland Maya area. In an effort to address this lack of research attention, a model positing a spatial continuum of terrace agricultural activities outside of the core area of the household is presented. Points along this continuum are defined as “adjacent agricultural space,” “transitional agricultural space,” and “outlying agricultural space.” The model is evaluated using terrace excavation data from areas near Dos Chombitos, a lowland Maya minor center located in far west–central Belize, Central America (see Figures 10.1, 10.2, and 10.3). The result is a successful preliminary test of the model and illustration of its usefulness. The overarching goal of this study is to lay the groundwork for further research on the organization of terraced agricultural activity.

BACKGROUND

Two extant models provide points of departure in formulating the model of terrace agricultural activities in relation to the household. Patricia McAnany’s (1995:69–78) biotic continuum model provides perspective from the larger landscape viewpoint. She defined points along a continuum running from the household out into the surrounding landscape. At one end, the continuum is anchored by pristine rain forest. This term is meant to designate forest areas that have not been affected by human activity. McAnany (1995:69) notes, however, that it is unclear what constituted “pristine” rain forest during the Maya Classic period (ca. AD 250–900) because macrobotanical and pollen evidence indicate a high rate of deforestation during this time. The next point on the continuum consists of lands that are part of a fixed-plot farming system with varying rates of fallow. McAnany (1995:69–70) describes this zone as an “ecotonal band along the continuum from forest to field.” Even though such lands may have been left fallow for a considerable period—long enough to look like true forest or pristine forest—their composition is such that they reflect the impacts of human use, such as the increased presence of economically valuable species. Fields that are currently in cultivation but are prone to intermittent fallow constitute the next point along the continuum. With regard to proximity to residences (household core areas), these areas are not infields (Turner and Sanders 1992:266–267) but are located at distances of an hour or more away without the aid of modern transportation (McAnany 1995:72). The other end of the continuum is composed of permanently cultivated fields, gardens, and orchards (McAnany 1995:74). These areas “envelop the house and are so thoroughly managed and continuously cropped that the term ‘field’ seems to be a misnomer” (McAnany 1995:77). Included in this category are features used to reclaim marginal lands, such as the use of agricultural terraces on steeper slopes.

Figure 10.2. Area surveyed by the Xunantunich Settlement Survey (XSS) in the upper Belize River valley area (the Dos Chombitos study area is marked)

Figure 10.3. Area surveyed by XSS near the minor center of Dos Chombitos (intensive mapping and excavation areas are marked)

Perspective from the household scale of analysis is provided by Thomas Killion’s (1992b:124) houselot model. According to this model the ideal configuration of a Mesoamerican houselot consists of “the structural core, a clear area of debris-free space surrounding the core, an intermediate area of fairly concentrated refuse enclosing the clear area, and a peripheral garden of mixed vegetation and debris” (Killion 1992b:124). The model specifies that clear spatial and material patterns are present around household structures, resulting from the maintenance of clear and trash-filled areas. Additionally, Killion made several important general points regarding the spatial and material properties of small-scale agricultural systems. First, Killion (1992a:4) noted that a lot of cultivated space is also heavily used residential space, and therefore it reflects material traces of both cultivation and habitation. Second, he stated that “as population reached progressively higher levels within circumscribed territories . . . more uniformly intensive systems of production would have replaced the infield-outfield structure” (Killion 1992a:6). Research in the lowland Maya area suggests that this process occurred during the Late Classic (ca. AD 550–800) period (Drennan 1988; Dunning and Beach 1994; Johnston 2003; Neff 2008; Sanders 1981; Turner 1983). Third, Killion (1992a:7) observed, following von Thünen (1966 [1842]), that a basic characteristic of an agricultural system is distance between residence and field. Farmers locate the most labor-intensive farming practices as close to residences as possible to reduce labor costs. Therefore, “the differential use of nearby or more remote locations requires different groups of people provisioned and organized for a different set of tasks” (Killion 1992a:7).

How can the biotic continuum and houselot models, coupled with Killion’s observations about the spatial and material properties of small-scale agricultural systems, aid in conceptualizing the organization of terrace agricultural activity in the Maya Lowlands? Beginning with the houselot model, Cynthia Robin (1997:4; 1999; 2006) conducted research on the specific layout of houselots in the vicinity of the Chan center located ca. four kilometers to the east-northeast of the Dos Chombitos area (see Figure 10.2). Her research provides information on the actual layouts of houselots near the Dos Chombitos study area and how they relate to the ideal constructs of the houselot model. Her research documented these distinct areas: (1) a mostly artifact-free area (except immediately adjacent to structures) in the form of a fifteen- to twenty-meter radius around structures (the cleared area around each house structure or the structural core); (2) an area of increased artifact density along the edge of this radius (the intermediate area of fairly concentrated refuse enclosing the clear area); and (3) a generally artifact-free area beyond this (the garden area). Robin also noted that low densities of artifacts were found adjacent to agricultural terraces and chultunob (subterranean storage pits). In the specific instances that Robin studied, agricultural terracing occurred beyond the fifteen- to twenty-meter-wide ring around house structures. Robin’s work demonstrates the usefulness of the houselot model for conceptualizing activity in the vicinity of household core areas. With respect to the houselot model and agricultural terracing, the following question may be asked, within houselot garden areas, which include terracing, is artifact patterning indicative of different kinds of agricultural activities? The answer is that it is not immediately apparent how the specifics of Killion’s houselot model—the density of artifact assemblages indicating areas of specific activity like trash disposal or maintaining cleared spaces—apply to terraced space in the vicinity of household core areas.

The definitions of two terms, “household core” and “residential agricultural space,” are important in modeling the organization of terrace agriculture activity. These terms make a distinction between space where exclusively agricultural activities likely occurred and where they likely did not. By making this distinction, analytically meaningful space is defined, from the perspective of agricultural terracing, at the intersection of Killion’s houselot and McAnany’s biotic continuum models. The household core consists of both the houselot structures and the fifteen to twenty meters of cleared space that surrounds them. Thus, the term combines two of Killion’s houselot model terms, the structural core and the surrounding area of cleared space, and designates a specific-sized area based on Cynthia Robin’s research. The terrain outside of the household core, together with the area that contains the majority of the agricultural terraces, is called residential agricultural space. Put another way, residential agricultural space is the “sea” that contains numerous household core “islands.” Residential agricultural space is a combination of Killion’s ring of refuse and peripheral garden area of mixed vegetation and refuse, as well as McAnany’s permanently cultivated fields, gardens, and orchards. I utilize the terms “residential” and “agricultural” to describe the space instead of “houselot” or “domestic” because I want to emphasize that primarily agricultural activities were taking place here within a larger spatial context composed of both agricultural and domestic space.

The concept of the household core embodies an important distinction with respect to the similar-sounding and previously discussed “structural core” of Killion’s houselot model. The latter term has a structural, in the built-environment sense, connotation. This follows from the standard definition of the ancient Maya household, which is a single structure or a group of several structures in which the structures composing the group are closer to one another than they are to adjacent isolated structures or groups (Ashmore 1981:47–49). However, it is assumed, and has been demonstrated archaeologically (Johnston 2002; Killion 1992b; Robin 1999), that the spatial footprint of the ancient Maya household is larger than its surface-visible structures. Killion’s houselot model defines the household as consisting of a number of different parts, one of which is the structural core, which consists of the structure or structures of the household. According to this conceptualization, the structural core, as well as the surrounding cleared area and ring of trash, are not areas where agriculture occurred or agricultural activities took place. Rather, agricultural activities occurred in the garden areas beyond the ring of trash.

The houselot model, however, while critically valuable from the heuristic standpoint, is conceptually problematic with regard to households that are associated with agricultural terraces. The conceptual problems have to do with the specific spatial components of the houselot in relation to the terracing that surrounds and intermingles with many ancient Maya households and also the underlying premise of von Thünen that agricultural activities are governed by the law of diminishing returns with distance (Butzer 1982:216). In the households considered for this study, terraces are present not only in the garden area where agriculture and agricultural activities took place but also in the ring-of-trash area where refuse disposal occurred and even occasionally in the cleared area where non-agricultural activities are assumed to have taken place. As per the houselot model, the basic assumption is that agricultural activities took place in the garden area but not in the ring-of-trash and cleared areas. The presence of agricultural terraces, on which agriculture and agricultural activities took place, in parts of the household where the houselot model generally infers they did not occur is problematic. A way through this conceptual logjam is to define areas of the household where agriculture and agricultural activities were likely to have taken place, including agricultural terracing (residential agricultural space), versus areas where they were unlikely to have occurred (household core).

THE MODEL

Figures 10.4 and 10.5 illustrate the delineation of household core space (not shaded) from surrounding areas (shaded) in two areas near the Dos Chombitos center that contain domestic and terrace architecture (see Figure 10.3 for the location of these areas). The surrounding space contains the vast majority of the agricultural terraces and is the area where agricultural activities took place. Importantly, this is the area where von Thünen’s premise, that agricultural activities are influenced by distance from the edge of the household core, is operative. This space needs some kind of a label, but what should it be? Killion’s houselot model conceptualizes the extra-household core space as areas of refuse and peripheral gardens of mixed vegetation, and McAnany’s biotic continuum model conceptualizes it as part of a larger area of permanent cultivation. Neither of these conceptualizations is appropriate for the task at hand, which is defining the types of activities that took place. They were not, with respect to Killion’s houselot model, phrased with terracing in mind or specifically phrased to delineate space where agricultural activities took place from space where they did not. Further, the scalar perspective of McAnany’s biotic continuum model is too coarse-grained. Despite these specific shortcomings, aspects of these models are quite important to the initial model-building process that is the subject of this chapter (see below).

The term “residential agricultural space” is chosen as the label for the extra-household core space. “Residential” is used as opposed to terms like “houselot” or “domestic” because it is unclear which portions of residential agricultural space is actually part of the houselot. In fact, just as the distinction between infield and outfield is neither appropriate nor useful in the lowland Maya area (Drennan 1988; Sanders 1981), the distinction between household agricultural space (areas outside of the household core) from non-household agricultural space is not useful either. The critical variable is distance from household core areas out into terraced residential agricultural space. “Agricultural” is preferred over terms such as “garden,” “orchard,” or “field” because these terms have spatial implications with respect to domestic areas or connote the kinds of plants grown. The model needs to be flexible enough to account for agricultural activities and specific modes of plant husbandry that do not fit into preconceived notions about household agricultural production.

The next step in the model-building process is to conceptualize residential agricultural space in such a way that von Thünen’s premise may be applied. Two points are important with respect to this conceptualization. First, residential agricultural space is conceived as representing a continuum, not an area divided into discrete blocks of space with rigid boundaries. McAnany chose the term “continuum” for her biotic continuum model because she wanted to imply a sense of vegetation succession. Although her model does have labels for certain parts of the overall biome, the emphasis is on the continuous nature of the biome. Residential agricultural space is conceptualized as a continuum in the same sense. Second, Killion describes his houselot model as an “ideal” construct. By ideal, Killion means that actual households will vary with respect to specific spatial patterning but they will generally conform to the tenants of the model. The following conceptualization of residential agricultural space is an ideal one in the same sense.

Figure 10.4. The Terrace Set #110 area showing the delineation of residential agricultural space

Figure 10.5. The Terrace Set #191 area showing the delineation of residential agricultural space

The continuum of residential agricultural space is anchored on one end by terraced areas directly adjacent to the household core. These areas are called “adjacent residential agricultural space.” Excavations placed along the edge of the household core of T/A1-183 within Terrace Set #191 sample this space (see Figure 10.5). Terraced areas located well beyond the structural core of the household anchor the other end of the continuum and are termed “outlying residential agricultural space.” This endpoint includes terracing located well beyond an isolated household core area and terracing equidistant from two or more household core areas. All of the excavations within Terrace Set #110 are within this space with respect to T/A1-152 and 153 (see Figure 10.4). Terraced areas that are transitional in terms of position between these poles are referred to as “transitional residential agricultural space.” Excavations placed within Terrace Set #192 and EDM 95 Terrs. are located within this space. The terms “adjacent,” “intermediate,” and “outlying” were chosen because they convey clear spatial differences while at the same time are not so specific as to imply a non-continuum-like and spuriously rigid delineation of terraced space.

The following are expectations regarding the material remains recovered and the activity inferred from them, depending on location within residential agricultural space. These expectations are derived from Killion’s (1992b) house-lot model and from research by Patricia McAnany (1992) at Pulltrouser Swamp, Belize. First, in terraced space adjacent to household structural core areas, greater densities of artifacts are expected, as well as a mix of artifacts representing residential and cultivation activities (Killion 1992a:4). Also, the proximal and medial fragments of broken agricultural tools would be brought back to the house for recycling (McAnany 1992:205–206) and thus would be expected in this context. Second, in space further away from household structural core areas—that is, outlying residential agricultural space—lesser densities of artifacts are expected, as well as artifacts representing specific agricultural activities (Killion 1992a:7). Specifically, the distal fragments of broken agricultural tools, which would not be brought back to domestic contexts, should be found in outlying residential agricultural areas (McAnany 1992:205–206). Third, material-culture characteristics from areas situated between these endpoints on the continuum—intermediate residential agricultural space—should be characteristic of their transitional position. Artifact densities should be higher than outlying residential agricultural space but not as high as adjacent residential agricultural space.

METHODS

The following methods were used during excavation and subsequent laboratory analysis to test the model. Important considerations were the placement of test excavations, definition of excavation provenience, assignment of contextual designations to excavation proveniences, and the application of a technique that results in the expression of the number of artifacts per provenience as a density value per a standard volume of space. Subsets of artifacts from spatially and stratigraphically informative contexts are distinguished by the application of these methods and allow an assessment of the model. The expression of artifact numbers as densities per a standard volume of space allows comparison across groups despite factors such as differences in the aggregate volume of proveniences from different spatial contexts.

A primary goal of the excavations was to obtain artifacts from different residential agricultural spatial contexts. Outlying residential agricultural space, transitional residential agricultural space, and adjacent residential agricultural space have been defined to characterize the continuum of residential agricultural space (see above). Excavations were undertaken in two areas adjacent to Dos Chombitos center (see Figure 10.3). West-northwest of Dos Chombitos two test excavations, Operations 261B and 261C, were placed within a set of cross-channel terraces (see Figure 10.4). These terraces are located in an arroyo approximately 100 meters downslope to the south from domestic architecture. Thus, these excavations produced the sample of artifacts from outlying residential agricultural space. East-southeast of Dos Chombitos five test excavations were placed at varying distances from household core areas. One excavation, Operations 277B and 286B (a continuous excavation), was placed near a household core area and yielded the sample of artifacts from adjacent residential agricultural space (see Figure 10.5). Four excavations—Operations 275B, 285B, 278B, and 287B—were placed in areas intermediate in terms of distance from either household core areas or the furthest outlying terraces in the group and therefore yield the artifact sample from transitional residential agricultural space.

A provenience is defined as a three-dimensional unit of space for the purpose of this study. During excavation, proveniences were defined by stratigraphy. In strata thicker than 0.20 meter, proveniences were changed every 0.20 meter. Excavation in the seven locations described above resulted in the definition of 124 proveniences. Each provenience was assigned a contextual designation. Table 10.1 presents the contextual designations used in this study and their numerical codes. Contextual designations were defined to partition variability in both natural and cultural deposition and formation processes (Schiffer 1987). An important goal in the process of assigning contextual designations was to isolate proveniences inferred to be the result of agricultural activity. Contexts 261, 262, and 263 are reasoned to be the result of agricultural activity. Of the 124 defined proveniences, though, only 38 were assigned contexts 261, 262, or 263. These 38 proveniences could be further grouped depending on their spatial origin, adjacent (n =8), transitional (n = 9), or outlying (n = 21) residential agricultural space.

While the number of proveniences from adjacent and transitional space is approximately the same—eight and nine, respectively—the number of proveniences from outlying space is significantly larger (n = 21). This disparity presents an obvious difficulty with respect to making comparisons among the three spatial groups. Artifacts from the outlying residential agricultural spatial context will skew the comparison because there are more artifacts in this group compared to the other two spatial contexts, and, importantly, they come from a larger aggregate total of volumetric space. Each provenience has a volume that is calculated by multiplying the length times the width times the depth of the provenience. One way to engage in a comparison among the three spatial groups, despite the disparities in aggregate volumes, is to represent the number of artifacts from a particular provenience or group of proveniences as a density figure using a standard unit of volume. The density of artifacts within a standard unit of space, calculated in the same way for all the three groups, is comparable even if the group volumetric space totals are different. The standard unit of space chosen for this study is 0.40 cubic meters, which is equal to 1 m × 2 m × 0.2 m.¹

Table 10.1. Contextual descriptors for terrace excavation proveniences used in the analysis

RESULTS AND DISCUSSION

The spatial relationship of ceramic artifacts, lithic tools, and associated debitage aided a model-building process for the reconstruction of terrace agricultural activity. The artifact analysis revealed several robust lithic and ceramic patterns and supported the expectations presented above.

Lithic tool diversity is the first pattern examined (Figure 10.6). A technological perspective was employed to examine the morphology, function, and production stages of the lithic tool assemblage (Neff 2008). Adjacent to the household, we had evidence of expedient tools, cores, polishing stones, miscellaneous digging tools, general-utility bifaces, and small tabular-shaped, broad-based distal tools (called trowels). The presence of polishing stones, presumably used during pottery manufacture (Rice 1987:138–139, 150; Hayden 1987:212), and the occurrence of other tool types represent the diversity of activities, agricultural and otherwise, taking place near the household structural core. This finding is in line with the expectation of greater diversity and density in artifact assemblages closer to the household structural core.

Figure 10.6. Lithic diversity by residential terrace agricultural context

The outlying context contained only expedient tools and general-utility bifaces. Using ethnographic information, archaeological evidence from Pulltrouser Swamp, and edge-wear analysis, McAnany (1992) posited that oval bifaces were used as weeding and tilling implements. One complete general-utility biface was found on the surface in the outlying agricultural context. Another distal general-utility biface fragment from the outlying context exhibited wear similar to sickle gloss, which was readily apparent to the unaided eye. The highly polished edge and dorsal face were smoothed and rounded, perhaps the result of the abrasion of the tool surface with phytoliths from grassy plants (Clark 1995:128). These artifact characteristics are in line with the expectation that tools found in the outlying context represent more specialized agricultural activities, including weeding and forest clearing. However, general-utility bifaces were probably used for a variety of purposes, ranging from forest clearing (which includes the cutting of grass and wood) to weeding, tilling, and wood carving (Clark 1995). The macrowear observations made for this study can suggest only one functional interpretation and it is likely that these tools were used for a variety of tasks.

The transitional context contained only expedient tools and small trowel-like tools. This result is in line with the expectation that transitional areas would exhibit assemblage type and density characteristics between those of adjacent and outlying contexts. “Trowels” and other digging tools represented a new tool type. These tools were made out of locally available slate—a prominent geological feature located in the nearby Macal River Valley (see Figure 10.2). Use-wear observed on these tools consisted of edge damage with extensive crushing and some rounding and polishing. Irregular microflaking was present and occurred on both faces of the tools. The polish had a dull texture. Aoyama (1995) performed an extensive experimental microwear analysis using a variety of lithic material types on a number of different working materials. His examination of tools at Late Classic Copán, Honduras, revealed that dull polish with a matte texture resulted from soil abrasion. The trowels and other digging tools are analogous to the modern hoe. They are amenable to a transverse haft like a modern hoe, but could have also been mounted using a simple socket haft at a right angle to the blade. Their similar shape and occurrence in agricultural contexts, in conjunction with use-wear patterning, suggests their use was similar to present-day hoes. The trowel-like tool may have been better suited for transitional and adjacent residential garden contexts because continual intensive gardening would have made other tools unnecessary.

Figure 10.7. Direct freehand-percussion core-flake distribution by residential terrace agricultural context

Figure 10.7 illustrates the distribution of direct freehand-percussion core flakes across the residential agricultural continuum. These artifacts are present in only adjacent and transitional contexts. This pattern is in line with the expectation that the initial stages of formal and informal tool production took place closer to the household structural core.

Figure 10.8. General utility biface resharpening flakes by residential terrace agricultural context

Figure 10.8 illustrates the distribution of general-utility biface reduction flakes by residential agricultural context. A 3 to 1 distribution ratio between adjacent and outlying contexts suggests farmers performed most of the tool refurbishing closer to the household structural core. This result is consistent with the argument that the proximal and medial fragments of broken agricultural tools were brought back to the house for recycling (McAnany 1992:205–206).

Figure 10.9 illustrates ceramic form diversity across the residential agricultural continuum. Closed and open ceramic forms occurred in all the residential agricultural contexts, although higher relative densities occurred in adjacent contexts. It is significant that jars with restricted openings had a distinctive presence in the outlying agricultural area. These jars were probably used as watering jars because this form would have minimized spillage and evaporation. Possibly, specialized watering activities were taking place in outlying residential agricultural contexts. Similarly, the minimal amount of open forms present in outlying contexts, in contrast to higher densities in adjacent space, supports the idea of a more specialized agricultural activity in the outlying region. In sum, ceramic form diversity patterns are compatible with the expectation that artifacts representing specific agricultural activities should be found in space further away from household structural core areas (Killion 1992a:7).

In the paragraphs above, I defined key terms, delineated points along the continuum of residential agricultural space, and identified expectations regarding artifact patterning based on previous research. I then tested the model using data from terrace excavations in the vicinity of the Dos Chombitos site. Patterns in lithic tool diversity and density, DFP (direct freehand percussion) core flake density, general-utility biface resharpening flake density, and ceramic form diversity data conform to the model’s expectations. This exercise resulted in a successful but preliminary first test of the model. To thoroughly evaluate the model more excavation and artifact analyses are required.

Figure 10.9. Ceramic form diversity by residential terrace agricultural context

SUMMARY AND CONCLUSIONS

Activity associated with agricultural terracing is a significant component of household production in many areas of the Maya Lowlands. An understanding of influential factors in, and the spatial organization of, terracing agricultural activity is an important companion to descriptions, explanations, and understandings of how and why terracing agricultural activity is significant to household production. This chapter proposed a model to explore and define factors influential in, and the spatial organization of, terracing agricultural activity. The model posits a spatial continuum of terrace agricultural activities outside the core area of the household. Points along this continuum are defined as “adjacent agricultural space,” “transitional agricultural space,” and “outlying agricultural space.” The model was evaluated using terrace excavation data and the result was a successful preliminary test.

The usefulness of the model is apparent in two ways. First, in archaeological investigations it is essential to understand the interrelationships of scale, pattern, and process and how they have profound effects on data interpretation. Second, it is necessary to define the appropriate scale of analysis for the research question to be addressed. Beginning with the latter point, within the domain of ancient lowland Maya household economy, I argue that investigation of agricultural terraces is not only an interesting question but also a critical one. The upper Belize River Valley study area for this chapter is just one of many areas in the Maya Lowlands where agricultural terrace features outnumber all other types of residential architecture. More research on agricultural terracing is critical to gain a better understanding of household economic behavior in many areas of the lowlands. The terrace-oriented residential agricultural continuum model and its initial test with excavation data represent an important avenue of household research. Indeed, using the same spatial model and artifactual data set, Linda Neff (2002) explored gender divisions of labor in lowland Maya terrace agriculture. Using multiple lines of evidence, she concluded that a gender ideology associating men with agricultural work occurring away from the household and women with multitasking activity, including agriculture, closer to the household was operative in the Dos Chombitos area during the Late Classic period (ca. AD 550–800). The terrace-oriented residential agricultural continuum model was important to Neff’s study because it provided a spatial structure that facilitated questions of who was doing what where and implications for household economy in an era of agricultural intensification.

The second point made above about the necessity of defining scale, pattern, and process in the interpretation of archaeological data is germane to the subject of this chapter. The terrace-oriented residential agricultural continuum model was formulated to organize the study of household production from an analytical perspective not specifically treated by extant models of the spatial order of prehispanic agriculture. The result is a more nuanced analytical perspective that has aided, and should continue to do so, in further exploration of terrace agricultural activity.

The spatial patterning inferences presented in this chapter, while thought provoking and initially successful, should be viewed as only the beginning of a program of research focused specifically on terrace agricultural activity. More terrace research in conjunction with household excavations is necessary to evaluate whether we are on the right track with our hypotheses.

NOTE

1. The following example demonstrates how using artifact density per a standard unit of volumetric space can allow comparisons among proveniences or groups of proveniences with significantly different aggregate volumes. Suppose provenience A yielded twenty artifacts from a volume of space equaling 0.8 cubic meters and provenience B yielded sixty artifacts from a volume of space equaling 7.5 cubic meters. The following calculations are performed to determine what the artifact density is per 0.40 cubic meters (the standard unit of volumetric space) for each hypothetical provenience. The twenty artifacts in provenience A are divided by the volume of the provenience (0.80 m³) to derive the provenience artifact density, which is equal to 25. The provenience density (25) is then multiplied by 0.40 cubic meters, resulting in 10, which is the artifact density of provenience A. The same calculations performed on hypothetical provenience B yield a density of 3.2 artifacts per 0.40 cubic meters of space. Having engaged in these calculations we know that provenience A has a higher artifact density than provenience B, despite the fact that it has fewer artifacts than the latter. Therefore, the figures presented in this chapter illustrate differences among outlying, transitional, and adjacent agricultural space via artifact densities per 0.40 cubic meter.

REFERENCES CITED