CHAPTER 1: HUMAN VULNERABILITY TO CLIMTATIC DRY PERIODS IN THE PREHISTORIC U.S. SOUTHWEST Return to Table of Contents
A fundamental challenge in studies that investigate the relationship between climate and human behavior is explaining climate’s differential influence on human behavior over time and space. That is, why do particular climatic hazards, such as a dry periods, at some times and places appear to stimulate a particular behavioral response and at other times and places do not? In the arid and semi-arid prehistoric U.S. Southwest, this problem is especially evident in regional-scale studies when long-term paleoclimatic records of specific climatic hazards are compared to a record of expected behavioral responses. For example, dry periods that reduce resource productivity are expected to increase the risk of food shortfalls and stimulate a variety of responses to manage these risks. The movement of people out of areas of low productivity to areas of relatively higher productivity that often results in settlement abandonment is one potential response to these increases in the risk of shortfall. An investigation of population movement and settlement abandonment, however, reveals that some settlements are abandoned during dry periods, some are not, and many settlement abandonments show no relationship with climatic conditions. This uneven long-term relationship between climatic hazards and expected responses can be incorrectly interpreted as evidence of little to no influence of climate hazard on human behavior. This uneven relationship may be, instead, evidence of spatial and temporal variation in human vulnerability to the hazard. The specific conditions that contribute to or ameliorate this vulnerability are not, however, well understood. Deficiencies in understanding are caused by a lack of empirical scrutiny of existing conceptual models of vulnerability to climatic hazards and the reliance of these models on an unverified assumption of widespread resource marginality in dry climates.
The purpose of this study is to address this fundamental challenge of understanding climate’s differential impact on human behavior by advancing understanding of conditions that affect human vulnerability to dry periods in arid and semi-arid regions. I focus on dry periods because they are a common climatic hazard in these regions and are frequently argued to influence human behavior through increasing the risk of food shortfalls. My approach to this study is to evaluate prominent and often implicit conceptual models of vulnerability to dry periods used by archaeologists and other scholars investigating past human adaptations in dry climates. These models are used to explain and predict spatial and temporal variation in human vulnerability to dry periods. Each model emphasizes different demographic and environmental conditions assumed to influence this vulnerability. A contribution of this study is the identification of these largely implicit models of vulnerability to dry periods used in archaeological studies of the U.S. Southwest and elsewhere. In this study, I make these models explicit, subject them to empirical scrutiny to evaluate their veracity, and identify the spatial scales they might apply. This is the most spatially and temporally comprehensive examination of vulnerability to dry periods, and factors that contributed to this vulnerability, yet conducted in the U.S. Southwest.
I evaluate four conceptual models of vulnerability to dry periods. These models, discussed in detail in Chapter Three, are as follows.
1. Aridity Model. The aridity model emphasizes resource marginality and widespread vulnerability to dry periods across all demographic and environmental conditions because of low precipitation conditions inherent in dry regions. In this model, differences in vulnerability over time are expected to be directly related to differences in dry-period severity. Vulnerability to dry periods is expected to be highest when dry-period severity is greatest. Evaluating this model also identifies the spatial distribution of vulnerability to dry periods allowing the regional-scale resource marginality assumption to be evaluated throughout the study area.
2. Demand model. The demand model emphasizes the influence of settlement population levels and catchment population density on differences in vulnerability to dry periods. These demographic conditions affect resource demands, the rate of consumption of resources, and the extent of labor available to invest in strategies to manage the risk of shortfalls. People living in large settlements and/or densely populated watersheds are expected to be the most (and sometimes the least) vulnerable to dry periods.
3. Supply model. The supply model emphasizes the influence of settlement proximity to perennial rivers and long-term average precipitation levels of settlement locations to explain differences in vulnerability to dry periods. These environmental conditions affect the local-scale potential supply of resources and thus the extent of resources people may have access to during dry periods. Greater access to and availability of water is expected to decrease vulnerability to dry periods.
4. Demand and supply model. The demand and supply model emphasizes the combination of demographic (settlement population levels, catchment population density) and environmental conditions (proximity to perennial rivers, precipitation levels) to differences in vulnerability to dry periods. These conditions affect the balance between resource demands and supplies and the extent of resources people have access to during dry periods. Vulnerability to dry periods is expected to be greatest where and when demands were high and supplies low.
I evaluate these models using data on demography, residential abandonment, environment, and dry periods for central Arizona during the A.D. 1200 to 1450 period. Central Arizona during this period is an excellent context to investigate the vulnerability of subsistence agriculturalists to dry periods because of the diverse cultural, environmental, climatic, and social conditions. I use the Coalescent Communities Database (Wilcox et al. 2003), the most comprehensive source of settlement data for central Arizona, to identify demographic conditions and residential abandonments. I use modern environmental data to identify local-scale differences in potential resource productivity among settlements and watersheds. I use tree-ring precipitation and streamflow reconstructions to identify dry periods during the period of study. As vulnerability is not a directly observable phenomenon, I use residential abandonment and associated population movement as an indicator of potential vulnerability to dry periods. To evaluate the models, I assess the long-term relationship between dry-period severity and residential abandonment among settlements by differences in their demographic and environmental conditions as emphasized by the vulnerability models. If differences in the relationships are detected when conditions are varied (e.g., high vs. low settlement population levels), I attribute these differences to the influence of these conditions on vulnerability to dry periods. This long-term approach is in contrast to studies that rely on single space-time coincidences to suggest a relationship between climatic conditions and potential behavioral responses (M. Ingram 1981:19). A long-term approach is valid because the effectiveness and use of residential abandonment and movement as a response and strategy for managing dry-period risks of shortfall is assumed to be consistent over the 250-year study period.
Evaluating models of vulnerability to dry periods is important because the specific conditions that contribute to or ameliorate this vulnerability, now or in the past, are not well understood or are in dispute (e.g., Cutter et al. 2003; Knight and Jager 2009; Meyer et al. 1998:238-243; Ribot 1995). Furthermore, no standard framework exists for identifying the fundamental sources of differential vulnerability (Meyer et al. 1998:240). In the absence of such a framework, archaeologists in the U.S. Southwest rely on common sense notions and unverified models that are logical and intuitively appealing but lack systematic empirical scrutiny. For example, it is reasonable to expect that people living distant from perennial rivers or in areas of relatively low precipitation should be the most vulnerable to dry periods in arid climates because they have the least access to water. Likewise those living near perennial rivers or in areas receiving the highest precipitation levels should be least vulnerable to dry-period declines in resource productivity. Does a long-term settlement history demonstrate this relationship between access to water and vulnerability to dry periods? In the U.S. Southwest, after 100 years of archaeological study of climate-human behavior relationships, this question has not been systematically addressed. Similarly, increasing population levels can be argued to increase stress on local environments thereby increasing vulnerability to dry periods and the number of people at risk. More people, however, can reduce vulnerability because larger populations may have more resources (human and environmental) to cope with climatic challenges. Both climatic and archaeological datasets in the U.S. Southwest are now sufficiently detailed that we can empirically address these questions and evaluate models and assumptions that rely on unverified notions.
Singular among these unverified notions is the assumption of resource marginality that equates low annual precipitation conditions inherent in arid and semi-arid regions with widespread resource marginality and vulnerability to dry periods. This assumption treats vulnerability to dry periods as a "pre-existing condition" (Cutter 1996: 530-531) of living in an arid region. Resource marginality, by definition, occurs where and when resource productivity (wild and cultivated) is inherently low relative to human food needs and oscillates around a threshold above which there was enough food to eat and below which there was not. Where resources are marginal, changes in any condition that increases the demand or decreases the supply of resources can increase the risk of food shortfalls and motivate a behavioral response. Models of vulnerability to dry periods rely on a marginality assumption to make variation in demographic (demand), environmental (supply), and dry periods meaningful for human behavior. Where resources are not marginal, changes in these conditions cannot be linked to changes in the risk of food shortfalls, thus, human behaviors argued to be linked to changes in demographic or environmental conditions cannot be interpreted as responses to shortfall risks. As I discuss further in Chapter Two, there are a number of reasons to question the assumption of marginality in the U.S. Southwest. It is based primarily on perceptions of the challenges of living in a dry climate and supported by limited indirect evidence. Weaknesses in the marginality assumption imply weaknesses in models of vulnerability that rely on this assumption. In this study, by evaluating the long-term relationship between dry-period severity and residential abandonment at different spatial scales, I identify the spatial distribution of vulnerability to dry periods and by implication the extent of resource marginality across the study area.
Evaluating models of vulnerability to dry periods is also important because it allows us to appraise arguments that rely on these models. For example, the depopulation of the northern Southwest during the late 1200s is coincident with a well-known dry-period and the impact of this dry-period on the depopulation has been debated for almost a century (e.g., Ahlstrom et al. 1995; Benson et al. 2007; Douglass 1929; Jett 1964; Judge 1989, Kohler et al. 2008; Lipe 1995; Varian et al. 1996; Van West and Dean 2000). Among the many factors now considered influential in the depopulation are decreased mobility options due to increased settlement population levels and increases in catchment population density caused by settlement aggregation (Van West and Dean 2000; Varien et al. 1996). Decreased mobility could have increased vulnerability to dry periods by increasing reliance on limited local productive capacities and decreasing access to arable land. This explanation suggests that the effects of the late 1200s dry-period could have been greater than similar or more severe dry periods because of, at least in part, these changes in demographic conditions. I consider this a ‘demand model’ of vulnerability to dry periods because it uses changes in demographic conditions that affect the demand for resources to explain changes in vulnerability to dry periods over time or differences in vulnerability among settlements. While the model is plausible and well reasoned, it is also partially amenable to testing using a long-term paleoclimatic record of dry periods and a demographic history of mobility. That is, do we have any evidence that settlement population levels or catchment population density influenced mobility or vulnerability to dry periods? If so, the argument is strengthened; if not, we cannot lessen the possibility that the changing demographic conditions, the dry-period, and the depopulation were simply space-time coincidences.
This identification and empirical evaluation of prevailing conceptual models of vulnerability to dry periods will affect the use of and confidence we have in these models. It will affect the use of the models by identifying the spatial scales at which they might apply. For example, I examine differences in potential resource productivity and the effects of these differences on vulnerability at the scale of individual settlements (e.g., near or far from a perennial river) and at the scale of entire watersheds (e.g., averaged watershed precipitation levels). I also examine the influence of differences in the demand for resources at the settlement scale with settlement population levels and at the watershed scale with population densities (rooms per square kilometer). Our confidence in the models should be affected by the results of this study. Whether our confidence increases or decreases, results should stimulate other methods and places for evaluating these and other models. Finally, this study demonstrates the opportunity and necessity of empirically examining models and notions regarding vulnerability to dry periods and the assumptions upon which they rely.
Organization of this Study and Summary of Results
This study of the influence of demographic and environmental conditions on vulnerability to dry periods and the models of vulnerability that rely on these conditions is presented in ten chapters. I first define and describe the relationship between the key concepts of vulnerability, risk, dry periods, and resource marginality. I argue that there are reasons to question the resource marginality assumption and discuss the consequences the assumption has had on interpretations and analyses of the challenges and opportunities of living in a dry climate (Chapter Two). I also argue that residential abandonment is an appropriate indicator of potential vulnerability to dry periods. Next, I describe the four vulnerability models I evaluate and provide examples of their use in archaeological studies in the U.S. Southwest and elsewhere and in modern studies of vulnerability to natural hazards (Chapter Three). I then describe the cultural, demographic, and environmental context of the central Arizona study area during the 1200 to 1450 period. I argue that the diversity within each of these domains benefits this study by providing a range of social and environmental conditions to investigate vulnerability to dry periods (Chapter Four).
My approach to evaluating the vulnerability models is to identify the extent to which demographic and environmental conditions influenced the relationship between dry-period severity and residential abandonment in central Arizona from A.D. 1200 to 1450. This evaluation involves data on demography, environment, dry periods, and residential abandonment (Chapter Five). I identify demographic conditions with this database by using the number of identified rooms in each settlement to represent settlement population levels and the number of identified rooms in each watershed divided by each watershed's area to represent watershed population density. I identify the environmental conditions of each settlement and watershed with modern GIS datasets of the location of perennial rivers and average annual precipitation and streamflow levels. I identify dry periods during the period of study with tree-ring precipitation and streamflow reconstructions. I identify residential abandonment from reductions in the number of settlements and rooms using settlement data from the Coalescent Communities Database (Wilcox et al. 2003). If differences in the sensitivity and strength of the relationship (as identified by the slopes of regression lines and correlation coefficients) between dry-period severity and residential abandonment among settlements with different demographic and environmental conditions are detected, I attribute these differences to the influence of these conditions on vulnerability to dry periods. Results support or challenge the veracity of the models to the extent that the long-term relationship between dry-period severity and residential abandonment is an effective indicator of potential vulnerability to dry periods.
I find that, for most of central Arizona during the 1200 to 1450 period, results (Chapters Six through Nine) support conceptual models that emphasize the contribution of high watershed population density and watershed-scale population-resource imbalances to relatively high vulnerability to dry periods. Models that emphasize the contribution of: (1) settlement population levels, (2) settlement locations distant from perennial rivers, (3) settlement locations in areas of low average annual precipitation; and (4) settlement-scale population-resource imbalances to relatively high vulnerability to dry periods are, however, not supported by this study. Results also suggest that people living in watersheds with the greatest access to and availability of water were the most vulnerable to dry periods, or at least most likely to move when confronted with dry conditions. Thus, commonly held assumptions of differences in vulnerability due to settlement population levels and inherently water poor environmental conditions are not supported. The assumption of regional-scale resource marginality and widespread vulnerability to dry periods due to low average precipitation in the region is also not consistently supported throughout the study area. Critically examining the relationship between precipitation conditions and migration sparked by the potential risk of resource shortfall due to demographic and environmental conditions is an essential step toward advancing understanding of the variable influences of changing climate conditions on human behavior in dry climates. I discuss the contribution of this effort to archaeological research in the U.S. Southwest, climate and human behavior studies, and analytical methods in the final chapter (Chapter Ten).