Case Studies 2 Convergence and Divergence as Problems of Explanation in Land Use Histories
Two Mexican Examples
Aleksander Borejsza and Arthur A. Joyce
Abstract
Similar results that arise from different processes and causes (convergence or equifinality) and different results that arise from similar processes and causes (divergence) are two common problems of explanation in any historical science. Two cases rooted in geoarchaeological fieldwork in highland Mexico exemplify such problems. A widespread entrenchment of the stream network of the Nochixtlan Valley occurred at ca. A.D. 1000. Potential proximate causes include changes in runoff, sediment delivery from slopes, adjustments of stream gradient, or channel form. Each hints in turn at a plethora of converging ultimate causes in realms such as climate, demography, agriculture, or warfare, but choosing between them is impossible without widening research to include issues usually considered to be beyond the scope of environmental archaeology. Tlaxcala experienced cycles of demographic explosion and collapse between 500 B.C. and A.D. 100, and again between A.D. 1200 and 1600. Each cycle was accompanied by agricultural intensification and disintensification, but the form they took were very different. In the first cycle, population growth resulted in severe land degradation, in the second in widespread land improvement. The divergence seems to be due to differences in cultural context and accumulated historical experience. In order to satisfactorily resolve problems of explanatory convergence and divergence, environmental archaeologists must constantly shift attention between process and history, and actively contribute to endeavors such as chronology building, on-site excavation, and the long-term curation of collections.
Introduction
This chapter sets out to demonstrate, by example, how easy it is to make up multiple and conflicting stories that purport to explain the alluvial record in terms of human land use. Our ulterior aim is to show how similar epistemological misgivings can be across the “Cartesian divide of nature and culture,” in this case the divide that separates disciplines as disparate as fluvial geomorphology and cultural anthropology. We use fluvial geomorphology to understand how different physical processes shape the alluvial stratigraphic record and cultural anthropology to understand how and why farmers decide between alternative forms of land use. We cross the divide between them, almost imperceptibly, when we assess the physical effects of agricultural activities such as tillage, grazing, or the building of terraces.
In our experience, many fellow archaeologists suffer from an inferiority complex that leads them to believe that the natural sciences rest on an infinitely more solid theoretical foundation, and are therefore more resolute when relating cause and effect. Stanley Schumm’s To Interpret the Earth: Ten Ways to Be Wrong (1991) is adept at dispelling this belief and inspired the title of our contribution. Schumm explores some of the common pitfalls of relating cause and effect in the earth sciences, especially when the effect is a fragmentary stratigraphic record and the cause something that existed in the past, but need not exist any longer in the present, a conundrum familiar to most archaeologists. Two problems Schumm specifically writes about are convergence and divergence. We illustrate each with an example drawn from our research in highland Mexico. The convergence example is negative, in that we cannot yet confidently choose between alternative explanations. The divergence example seems more positive, in that we are able to discount some explanations advanced by our predecessors. We end by pondering what makes the difference between the two examples, and why we have progressed more in one study area than the other.
Convergence in the Nochixtlan Valley: The Causes of Stream Incision at A.D. 1000
Convergence or equifinality refers to a situation where different causes produce similar effects (Schumm 1991:58). The effect we are concerned with is one that Schumm himself studied throughout his career, that of stream incision (e.g., Schumm 1957; 1999). Incised streams are common throughout the highlands of formerly Spanish America (Figure 2.1), and there is a corresponding wealth of vernacular terms—arroyos, barrancas, barrancos, cañadas, quebradas—that identify them in place names. They flow confined between almost vertical walls, and adjust to changing external inputs by filling up with sediment and raising the elevation of their channel (i.e., aggrading their narrow floodplain) or by incising (degrading, downcutting, trenching). If they remain stable, i.e., retain the same elevation, their floodplains develop soils. Because the channel experiences minor lateral shifts with each incision, the walls or cutbanks often expose the stratigraphic record of previous filling cycles (Borejsza et al. 2014a:fig. 1).
The cutting is usually much more rapid than the filling and can have dramatic effects on human land use. It drops water tables, makes irrigation impracticable, undermines bridges and dams, creates obstacles to movement, and may increase the magnitude and frequency of devastating floods. The causes of stream incision are thus of not only academic, but also practical concern. For example, a spate of stream incisions that began in the southwestern United States roughly at the time of Anglo-American takeover and adversely affected the settlers’ agricultural enterprises gave birth to the notion of an “arroyo problem” and a vast literature debating its exact timing and causes (Bryan 1925; Cooke and Reeves 1976; Elliott et al. 1999; Waters and Haynes 2001).
Figure 2.1 Incised streams in highland Mexico: (a) Yuzanu in the Nochixtlan Valley. The arrow points to the retaining wall of an agricultural terrace which crossed the channel at 150 B.C. (b) Barranca Tenexac in north-eastern Tlaxcala. This reach incised at the onset of the Holocene, and then again at ca. A.D. 100. Note the proximity of slope and channel.
In the stratigraphic record, one recognizes an incision by an erosional unconformity that laterally separates alluvial fills of differing age. By definition, the incision leaves behind a fluvial terrace, i.e., a surface that is no longer flooded. The terraces, however, need not be paired, nor form obvious steps. Instead, younger fills are often “inset” or “nested” within older ones. The age of an incision can be bracketed by dating the age of deposition of fills on either side of the unconformity, ideally something from just underneath the surface of the terrace, and something from near the base of the ensuing fill. This is the procedure we tried to follow in our research in the Nochixtlan Valley in the Mixteca Alta region of the state of Oaxaca (Mueller et al. 2012).
An incision spread throughout the stream network that drains this 500 km2 valley about a thousand years ago. Radiocarbon dates on the organic matter of buried palaeosols, artifact inclusions, and the position of a few small archaeological sites with respect to modern and former stream channels suggest that the incision started at the mouth of the valley close to A.D. 1000 and reached the smallest headwater tributaries some two centuries later. We have documented other incisions throughout the late Quaternary, but this one seems to have been particularly pervasive and rapid, especially considering the relatively large size and complexity of the fluvial system in question.
The three or four centuries centered on A.D. 1000 loom large in the imagination of scholars concerned with both environmental change and the culture history of Mesoamerica (e.g., Diehl and Berlo 1989; Sodi 1990; Demarest et al. 2004; Manzanilla 2005). They seem packed with dramatic events that revolve around ecological disaster and political turmoil. The most severe droughts of the Holocene are placed within this timeframe, both in central Mexico and the Yucatan (Metcalfe et al. 2000; Stahle et al. 2011). They have been held responsible for the “Great” Maya collapse of the Terminal Classic (see Kennett and Beach 2014), as well as for the contraction of the agricultural frontier on the northern fringe of Mesoamerica, and the resulting southward migration of people viewed by the old urban civilizations of central Mexico as barbarians (Braniff 1989; Beekman and Christensen 2003:148). In both the Mixteca Alta and the neighboring Valley of Oaxaca, this is a time of major shifts in settlement patterns and sociopolitical organization (Spores 1972; Kowalewski et al. 1989; 2009; Winter 1989; 1994; Blomster 2008), synthesized in terms like “balkanization” (Flannery and Marcus 1983) or “collapse and reemergence” (Joyce 2010). Do they bear any relation—as either cause or consequence—to our stream incision?
Geoarchaeological research suggests that the dramatic cultural transformations centered on A.D. 1000 may indeed bear some relation to stream incision in Nochixtlan. Figures 2.2 and 2.3 map out some of the plausiblecausal relationships triggering incision. Because flowing water is the physical force that cuts a new channel, the proximate causes of incision (entries [1,2,3] in Figure 2.2) are hydrologic. They are essentially a function of the balance between water discharge and sediment load. Reduced sediment loads [1] or increased average or peak discharges [2] could have tilted the balance in the direction of stream incision (Bull 1991:fig. 1.4; Knighton 1998:formulae 6.9–6.18; Schumm 1999:table 2.2D). Yet another possibility is the lowering of the base level at some point farther downstream. Formed by the confluence of the two major branches of the Nochixtlan Valley, the Río Verde descends to the Pacific Ocean through narrow canyons cut into bedrock. The breaching of a major obstruction somewhere along the way [4] at A.D. 1000, perhaps triggered by an earthquake [5], could have suddenly dropped the base level.
Figure 2.2 Convergent causes of stream incision in the Nochixtlan Valley.
Figure 2.3 Convergent causes of different land use choices in the Nochixtlan Valley.
If we focus on the more mundane factor of increased discharge [2], it could be caused in turn by more rainfall [32]; faster runoff from slopes [16]; increased connectivity (Fryirs 2013) of the stream network [7]; a steepened longitudinal stream gradient [8] (Schumm 1977:77–137; Bull 1979; 1997; Patton and Schumm 1981); or the abandonment of irrigation networks, with the water that was formerly diverted to fields now re-integrated in the overall stream discharge [13]. If we follow instead the lead of reduced sediment load [1], it may take us to reduced sediment delivery from slopes [14], often linked to their reduced geomorphic connectivity [20], perhaps brought about by the construction of agricultural terraces on slopes [28].
A different kind of agricultural terrace, the lama-bordos (Spores 1969; Pérez Rodríguez 2006; Mueller et al. 2012; Leigh et al. 2013; Pérez Rodríguez and Anderson 2013) may also have played a critical part. This ancient agricultural technology, known to date in the Mixteca as far back as 1500 B.C., is a form of cross-channel terracing (for agricultural terrace typology and related terminology see Whitmore and Turner 2001:133–64; Frederick and Krahtopoulou 2000). Farmers block a barranca with an obstacle of brush or stone rubble, which slows the movement of water and traps nutrient-rich fine sediment and plant litter. As new courses are added to the terrace riser, the tread grows in both thickness and surface area. The scale of these works in the Mixteca is truly monumental, with risers several meters tall, and flights of treads spanning several kilometers. Thousands of breached stone risers of different ages jut out from barranca walls (Figure 2.1(a)). We would expect the growth of lama-bordo systems [11] to result in reduced channel connectivity [6], and reduced sediment load downstream [1]. Their abandonment and collapse [12] would, conversely, result in increased connectivity [7], as the barranca re-established its natural continuous channel. This reasoning, however, is complicated by a technological variant we observed in some lama-bordos. Instead of blocking the barranca completely, farmers may shape the risers and the cultivation surfaces in such a way that excess water is shunted to one side of the field [10]. If repeated over a whole flight of terraces, this procedure promotes the appearance of a narrow but high-energy channel that hugs one side of the valley. The growth of such lama-bordos may instead lead to high channel connectivity and discharge [7→2].
It is clear from this discussion and from the tree-like shape of Figure 2.2 that, as we move from proximate to more distant causes, we can imagine a dizzying array of equifinal processes that converge on the same effect of stream incision. What is even more disconcerting is that there are pairs of exact opposites in the diagram: reduced and increased channel or slope connectivity [6/7, 20/22]; construction and abandonment of hill slope or cross-channel agricultural terraces [28/31, 11/12]; a wetter and a drier climate [32/33]. What we find interesting and somewhat reassuring is that there is no neat separation of natural factors near the base of the tree or (agri)cultural factors in the higher branches, nor vice versa. It is thus neither a simplistic scheme of people controlling nature, nor a case of environmental determinism. The progression from what we have marked as levels I to IV can be thought instead, in geomorphic terms, as moving from sediment sink (or outlet) to source. We shall see further on what it means in terms of archaeological practice.
So far, we have not even left the realm of the technological and rather mechanistic aspects of agriculture to explore those ultimate causes that may lie in the realm of human decision-making. If alluvial geoarchaeology is to shed some light on why people changed the way they used the land, we will need to graft, on each of the terminal branches of the tree in level II or IV, some segments cut from Figure 2.3. Several of the level II and IV entries of Figure 2.2 are therefore repeated in the two side rectangles of Figure 2.3. The left-hand rectangle [11,25,27,28] groups together those changes in land use that one would expect to have taken place, in isolation or in tandem, when the valley as a whole underwent a cycle of agricultural intensification. Conversely, those in the right-hand rectangle [12,13,26,29,30,31] can be thought of as effects of disintensification. Most of the new entries in this diagram [numbers higher than 33] are concerned with the possible causes of either, and therefore converge on one of ...