Long term effects of climate on human adaptation in the middle Gila River Valley, Arizona, America
- First Online:
The Hohokam, an irrigation-based society in the American South West, used the river valleys of the Salt and Gila Rivers between 500 and 1500 AD to grow their crops. Such irrigated crops are linking human agency, water sources and the general natural environment. In order to grow crops, water available through rain and river flows needs to be diverted to land where the plants are grown. With a focus on the Gila River, this paper uses the potential harvest of maize (a main Hohokam crop) as a proxy for evaluating the influence of natural water availability and climatic changes on irrigation options for maize. Available climate variables derived from tree-ring proxies are downscaled. These downscaled data are used as input for a crop growth model for the entire sequence of Hohokam occupation along the Gila River. The results of the crop model are used to discuss the potential influence of climatic variability on Hohokam irrigation and society. The results will show that climatic change alone cannot be used as an explanation for developments in Hohokam irrigation. Societal development resulting in growing population and extensive irrigation systems increasing pressure on water sources over time would have been a key factor to include to understand Hohokam society between 500 and 1500 AD.
KeywordsAncient canals Evolutionary perspective Climate Crop productivity
Water is and was vital for societies, and as such its manipulation has been a key for societies to emerge and prosper. The intimate linkages between human society, human agents and their water environments are expressed in ancient irrigation systems, like those of the Hohokam in the American South West discussed in this paper. In general, the evolution of ancient canals provides a distinctive insight into human adaptation under strategies in response to climate. An irrigation canal is an intentional modification of the environment, performed with awareness and attention of human beings; at the same time, irrigation systems create human activities and social relations. Ancient irrigation systems are highly suitable to study these dynamics, given their importance for societal evolution as main producers of food for the population. The variation in human population is generally assumed to be correlated with the variation in food productivity in early agrarian society, with crop yields variation linked to water availability.
In this study we focus on the process of human adaptation under climate variation in the context of Hohokam irrigation on the long term between 450 and 1450 AD. The aim of the overall project is to use this irrigation history to quantify the co-evolution of humans and nature. Human actions, climate and food production are considered as the pivot. We apply a crop growth model (CGM) to calculate the irrigation demands for the potential harvest of prehistoric maize and consider this as a proxy for evaluating how difficult it may have been to grow the crop when only considering climate variables. Human dynamics are estimated through the study of the development of canals and settlements. Climate variables are derived from the reconstruction and downscaling of tree-ring data.
The study on Hohokam food production is situated within the general debate on human niche construction (HNC), to understand human bio-social evolution (Odling-Smee et al. 2000, 2003; Laland et al. 2001; Ihara and Feldman 2004; Borenstein et al. 2006; Laland and Brown 2006; Odling-Smee 2006; Lipatov et al. 2011; Riede 2011). The basic concept of HNC is that humans not just adapt to environments but in part also construct their environments; this constructed environment on its turn influences human evolution (Lewontin et al. 1982 Laland et al. 2001). Studying long-term process with quantitative methods is a prominent feature of HNC. Some attempts have been made to discuss the evolutionary process of early irrigation canals with the theory of HNC. For example, Lansing and Fox (2011) suggest that niche construction should account for the “engineered landscape” of irrigation canals, next to biological phenomena.
Ertsen (2010) indicates that such an engineered irrigated landscape has structuring properties. Hydraulic behavior of irrigation systems resulting from human action partly enables and partly constrains new human actions. As such, we can consider the constructed canal environment as a niche with potential influence on human evolution. Humans modify the landscape and hydrological environment in the process of canals construction and the operation of these systems. The evolutionary process of canal changes in turn affects human development. Application of HNC to the evolutionary process of development of ancient canals, can help to quantify how humans respond to and adapt their direct environment in the context of climatic variation. This paper focuses on smaller-scale canal systems, localized within and closely related to naturally occurring conditions, and organized by local communities. Although there is no reason to assume that large scale imperial water systems did not construct human niches in the meaning as discussed, a focus on emergence of smaller systems at longer time scales allows further study of short term human actions within this longer time frame.
The prehistoric Hohokam canals in the middle Gila River valley of Arizona, America can be tracked back to two thousands years ago. A large amount of archeological surveys and research have been done since the 19th century (Bandelier 1892; Larson 1926; Midvale 1935, 1946, 1963, 1965, 1972; Haury 1937, 1976; Motsinger 1993; Ravesloot and Waters 2002; Woodson 2009, 2010). Hohokam canals are known along the Gila and Salt Rivers. Compared to several systems on the Salt River, the systems in the middle Gila River valley are quite small in size. Most scholars agree that an ‘irrigation community’ existed to conduct inter-village management of irrigation on the canal systems of the middle Gila area, which was separate from village administrations (Doyel 1979; Crown 1987; Hunt et al. 2005; Woodson 2010). Irrigation communities are defined as ‘autonomous, self-contained organization’ without intervention from an ‘external authority structure’ (Hunt 1988; Doyel 1991; Woodson 2010). With regards to the scale, emergence and organization, therefore, the canals systems in the middle Gila River valley are highly suitable to study HNC.
In this study on interactions between climate (change) and human dynamics, we concentrate on potential food productivity. A combination of maize, beans and cucurbita pepo (the ‘maize-bean-pumpkin complex’) has been considered the major domestic crop over prehispanic Native American agriculture (Castetter and Willis 1942; Hunt and Ingram unpublished). Beans provide nutrients for the growing needs of maize and squash; the maize stalks enable beans to climb up in turn; and the big leaves of squash at low ground keep weeds down and reduce the evaporation of soil water. Information about traditional maizes is relatively abundant, but little is known on beans and cucurbita pepo. Hence, the paper studies maize productivity under irrigation. The approach is to directly relate climatic conditions (temperature and rainfall) to possibilities for crop production over this period. A scenario on the longer term context of irrigation development in the middle Gila River valley is simulated. The temporal and spatial development of ancient canals and settlements under study are discussed, as those canals are modified, dynamically functioning at which crops are grown. The total water requirements are calculated on a basis of the potential arable. Thus we do not discuss the options to distribute the available water with the canal systems. The study emphasizes the water demands rather than the water availability for farmers within that larger area. The aim is to assess the water amount needed for irrigating all the command land and see whether changes in the water demand would have been a potential problem for the Hohokam. Crop water requirements are simulated with a CGM along the temporal scale, using regional climate data as input based on tree-ring reconstruction data, as explained below and in Ertsen et al. (2014).
Niche construction theory, adaptation, and ancient canals
Niche construction theory (NCT) originated as a branch of evolutionary biology in the 1980s (Lewontin et al. 1982; Lewontin 1983). NCT could be fundamentally recognized as an improvement of standard evolutionary theory (SET). NCT states that evolutionary changes in both organisms and environments depend on the states of both the organisms and the environments (Lewontin et al. 1982; Lewontin 1983), as ‘an evolutionary process whereby organisms modify their environment and thereby influence their own and other species’ evolution’ (Odling-Smee et al. 2003; cited in Kendal et al. 2011, p. 785). That is to say, ‘through niche construction organisms not only influence the nature of their world, but also in part determine the selection pressures to which they and their descendants are exposed, and they do so in a non-random manner’ (Day et al. 2003).
NCT explains two processes of the interaction between organisms and environments: (1) organisms are partly adapting and partly constructing their habitat, with the consequence that (2) ‘organism-induced’ changes in environment selectively shape organisms in turn. The interaction of above two processes moves in endless ‘loops’. From this point of view, organisms adapt to and change environments as a consequence of the reciprocal and dynamic relationship between natural selection and niche construction (Laland and Brown 2006). Niche construction modifies natural selection, and natural selection determines niche construction; the combination of them contributes to adaptation of organisms (Odling-Smee et al. 2003; Laland and Sterelny 2006). Therefore, capturing the roles of NCT as an endogenous mechanism is an important way of achieving the process of adaptation, and of exploring the interaction between organisms and environments.
At present, the perspective of niche construction has been particularly explored to interpret human behavior and society as a quantitative evolutionary method; all the current theories of HNC emphasize the co-evolution of human biological and cultural aspects (Laland et al. 2001; Laland and Brown 2006; Lansing and Fox 2011; Riede 2011; Iriki and Taoka 2012). However, in the course of human evolution, the effects of NCT not only incorporate gene and culture, but also reflect on the tools and techniques used by humans, or landscapes changed by humans. Irrigation canals essentially are the products of humans adapting themselves to nature. The origin of canals might have been trial and error. When farmers found that delivering water flows by canals might increase their harvest, they were likely to extend the network of canals. The expansion of canals consequently amplified crop yields, possibly together with an increase of sediments in canals and on fields. Increase of food subsequently enlarged the human population; but at the same time, accumulative sedimentation in canals required more laborers to remove the silts. From the standpoint of HNC, food could be considered as a positive response and sediments as a negative response when humans modified their environment by means of canals. In this situation, humans themselves created both negative and positive outputs of an irrigation system. The balance of positive and negative consequences would have affected the rate or direction of the canals’ development; in such a way, humans’ adaptation was altered evolutionally.
Irrigation and settlement
The altitudes of the study region range from around 300 m above sea level in the northwest to about 950 m in the southeast, with 335 m at the Gila River near Pima Butte, 457 m on the top of Gila Butte, and over 945 m in the Santan Mountains (Woodson 2010). The area is characterized by low-gradient desert plains and low-lying mountain upland (Huckleberry 1993). The topography of the alluvial plains along the middle Gila River enables the construction of gravity controlled irrigation canals. Agriculture might originate in 300 AD in the Gila River basin, although evidence of earlier agriculture elsewhere in the larger region has been found (Ertsen et al. 2014). It is generally accepted that the main cultivated crops were maize, bean, squash and cotton along the whole Gila River (Bohrer 1970; Haury 1976; Gasser and Kwiatkowski 1991). The limited precipitation is incapable of supporting the water needs for full growth of these crops under natural conditions. Thus, agriculture is assumed to be dominantly dependent on irrigation. The initial canals are likely to have occurred as early as the appearance of crops. Irrigation being a necessity for yielding adequate food for humans’ survival must have been a critical component in the behavioral responses to environmental variation in the Gila area.
The area is conventionally known as a part of the Hohokam, as defined by its culture. Hohokam is notable for its irrigated agriculture, which has received considerable attention from different fields (Haury 1976; Graybill 1989; Howard 1993; Hunt et al. 2005). The Hohokam society has been overwhelmingly considered as a prehistoric agricultural society, achieving a remarkable adaption to their arid environment in the North American Southwest (Haury 1976; Doyel 1979; Bayman 2001). Reconstructions of canals and settlements in the study area has been substantially investigated in previous research (Bandelier 1892; Southworth 1914, 1915; Larson 1926; Midvale 1935, 1946, 1963, 1965, 1972; Haury 1937, 1976; Gladwin 1948; Woodbury 1976; Schiffer 1982; Dean 1991; Motsinger 1993; Gregory and Douglas 1994; Neily et al. 1999, 2000; Ravesloot and Waters 2002; Woodson 2006, 2009, 2010; Loendorf et al. 2007; Ravesloot 2007; Burt 2007; Fertelmes and Loendorf 2010). In this study, the recent Hohokam chronology in the Gila area completed by Woodson (2010) is followed.
The first canal in the Gila area appears to have been built in Snaketown during the Early Pioneer period (450–650 AD); its habitation would have been at hamlet-level. No settlements in the Gila Butte, Santan and Granite Knob were found yet. In the Late Pioneer period (650–750 AD), the existing canal in Snaketown was doubled in length to the west, and the previous hamlet had extended to a village-level settlement. One camp (with two more possible) appeared next to the new canal of Snaketown. Southeast of Snaketown, the Gila Butte canal was established as well. Located at the tail end of the canal, a hamlet called ‘Gila Butte Site’ was likely to be built simultaneously to the large settlements. Woodson (2010) assumes that the residents of the Gila Butte Site were largely focused on their own canal system for subsistence purposes rather than the Snaketown Canal System. In addition, two camps emerged north of Gila Butte and a third camp was probably established at the southwestern foot of Gila Butte. Southeast of Gila Butte, the first canal in the Santan was constructed; two hamlet-size settlements were occupied at the north of the Santan canal (see more details in Woodson 2010).
During the Early Colonial period (750–850 AD), the entire canal system was enlarged three times and almost reached its maximum extent. In Snaketown, a new extension to the west was constructed based on the canal of the Late Pioneer period, and a south branch canal was built too. Three primary villages existed at this time. The two possible camps in the Late Pioneer period had expanded to a village still with two recognizable settlements. A new village came into being along the new south branch canal. There were four hamlets in the south of the North branch of the Snaketown canal. Six new camps and one possible camp were established as well. In Gila Butte, some small canals were built to the west, and the earlier hamlet grew into a village. The Santan canal system was extended to connect with Gila butte. At the north of Santan, another canal was constructed almost as long as the south one. It had three villages, and several (possible) camps. Granite Knob only contained one canal and one small village at the time. During the Late Colonial period (850–950 AD), the configuration of canal systems and settlements had changed very little compared with the Early Colonial period. The three villages of Snaketown slightly grew in size and a small village was added at Granite Knob.
In the Sedentary period (950–1150 AD), there was not much new construction on the canal systems and human settlement reached a peak. In Snaketown, three villages had increased in size, and a previously occupied hamlet near to Gila Butte had grown to village size. Eight new possible camps might have occurred besides the six existing camps. The upper part of the village at Santan expanded in intensity of use as well. The settlements in Granite Knob included a dual village as the one in the Snaketown and added a hamlet. During the Early Classic period (1150–1300 AD), the settlement sizes declined. A new canal was built between the Gila Butte Canal and the Snaketown Canal. In spite of one new small village appearing at the far western side of the Santan south branch canal, two previous villages had been abandoned. The habitants redistributed themselves into a growing number of small-size settlements. Instead of more intensive occupation in villages, eighteen hamlets (four hamlet clusters) and thirteen camps dispersed along the Snaketown canals. The settlements in Gila Butte shrank much in size. The situation in Santan was similar to that in the Snaketown. The three big villages were replaced with three small dispersed villages and an amount of hamlets. The camps reduced by nearly half in number. The villages in Granite Knob regressed to hamlet-level. By the Late Classic period (1300–1450 AD), more settlements were abandoned.
the tree-ring sites should be relatively close to the study area;
the tree-ring should be distinctively sensitive to climatic variables (temperature and precipitation);
the dataset should cover the main period one is interested in (which would be 450–1450 AD for the main Hohokam period);
and the data set should correlate with the study area.
The original tree sites and Hohokam irrigated area are not on the same altitude. All tree-ring sites are located above 2000 meters, but the Hohokam area is located around 450 meter, both above sea level. This also yields different climatic zones: tree-ring sites are associated with the CD2 zone, with low temperatures and relatively substantial rainfall, whereas the Hohokam main area is within the hot and dry climate of the CD6 zone. Correlation between temperature and rainfall were applied to the two zones. The resulting correlation coefficient r for temperature was 0.99, somewhat higher than the coefficient for precipitation of 0.83. This means that rainfall and temperature in study area were strongly correlated with the same in the tree site zone.
In order to use climatic data on a meaningful scale to simulate crop productivity and water use, we downscaled the annual data into month-scale ones using the change factor (CF) method (Chen et al. 2011), which adjusts observed time series by adding the difference (for temperature) or multiplying the ratio (for precipitation) between future and present climates. One of the advantages of the CF method is its straightforward application. For application, the equations were modified slightly by linking the change factors between reconstructed tree-ring data and upland observed data. As such, the CF method could take into account climatic differences between high and low altitudes, by linking climate variability at high altitudes to the same in the lowland, along the equations below. The simulated series of precipitation were validated by comparing month-distribution percentage with the observations for the same percentage. As discussed in Ertsen et al. (2014), the model captured the precipitation frequency at monthly scale well.
Previous studies on traditional maize yields
Castetter and Willis (1942)
King 1923, King and Leding 1926, King and Loomis 1932; cited in Hunt and Ingram, unpublished
Muenchrath et al (2002)
Mesa county 1838–2434 m
Hunt and Ingram unpublished 2
551–974 season 1; 971–1777 season 2
Based on Burns (1983)
Benson et al. (2013)
Based on Burns (1983)
Full irrigation impact of moisture
Castetter and Willis (1942) performed an ethnographic study on traditional maize cultivation at Pima and Papago. Their study covered planting calendars, crop characteristics, production, and consumption. From the standpoint of energetic consumption, Diehl (1997, 2005) estimated the yield of early maize under the hypothesis of optimal energetic return rates for foragers. Several experimental studies are available on productivity of traditional maizes (several publications of King, Leding, and Loomis (1920s–1930s), Burns (1983), and Muenchrath et al. (2002)). Finally, simulation models based on experimental data have been used in Hunt and Ingram (unpublished); Kohler (2012), and Benson et al. (2013) to estimate maize productivity. Kohler (2012) normalized maize yields from 1931 to 1960 and applied tree-ring based reconstructions of temperature to estimate the palmer drought severity index (PDSI) to produce prehistoric maize yields. Benson et al. (2013) considers that Kohler’s model attempts to involve ‘all processes impacting maize productivity’, which makes some of the model’s parameterizations problematic. In the paper, a simpler modeling approach to estimating maize yields is presented, by estimating maize production as a function of soil organic-nitrogen concentrations and water-year precipitation (which is also based on tree-ring reconstructed data).
With regard to perceiving the variance of prehistoric maize yields in the long term, along with tree-ring reconstructed data is a relatively precise and feasible option as mentioned in paragraph 4. The standard crop water requirement model AquaCrop of the Food and Agricultural Organization (FAO) was used in this study. We focus on the impacts of climate on maize productivity by computing maize production under the assumption of full water supply. Thus, we calculate the potential yield. What is actually harvested cannot be computed. Modeling ancient crop yields for ancient agriculture is particularly different from the simulation of modern crops. For ancient crops, it is unrealistic to obtain actual yields.
Reconstructing maize production over 1000 year period has its complexities. Again, the spatial difference between the study area (lowland) and the tree-ring sites (high altitude areas) as well as the divergence of temporal scales between the tree-ring chronology (annual) and the requirements of the model (monthly) are issues to deal with. In addition, there are several key factors determining maize productivity, such as seed variety, climate, soil environment, farmer’s behavior, insects etc. The study restricts its focus to the impact of climate variables on maize productivity. The objective of this restriction is to assess the possible role of climate in structuring human adaptation. Furthermore, no other single factor except precipitation and temperature can be reconstructed for a series of one thousand years in ancient periods. Due to these reasons, the models that rely on field observations (Kohler 2012; Benson et al. 2013) are not suitable for the study, despite those being able to estimate the variance of maize productivity at a thousand year scale. Instead, a CGM is used for calculating the irrigation demands for the potential harvest of prehistoric maize. The CGM can describe the growth process of the maize under given prehistorical environment, which incorporates four input variables: climate, soil, crop and management. With the irrigation demands of the maize in the study area, it can be estimated how difficult it may have been to grow the crop when only considering climate variables. The higher the water demand for maize, the more difficult it would have been to grow the crop. That will provide a reasonable baseline analog for how climate changes may have influenced maize growth.
Clarifying the distinction between potential yields, available yields, actual yields and consumptive yields is pivotal to understanding the modeling setup. There are several definitions available to distinguish between the four categories. Gaines and Gaines (2000) define potential yield as “the yield which is influenced by climatic and edaphic effects and is treated as a per unit variable”, and realized yield (actual yield) is “the food value obtained from the harvest”. De Wit (1958, cited in Rockstrom and Falkenmark 2000, p. 321) stresses the hydrological perspective: “for a given crop and hydro-climate, the potential yield (assuming no deficiencies at all) will be determined by climatic factors (basically incoming radiation and temperature)”. In this explanation, the climate is placed in a vital position for potential yields. The difference between potential, available and consumptive yields is explained by Schroeder (2001), who points out that “the amount of the potential yield that is harvested is referred to as available yield; … the proportion of the available yield that is consumed by humans is referred to as consumptive yield.” Not only did Schroeder explain the relationship among potential, available and consumptive yields, but also he identified available yields as actual yields.
Potential yields are assumed as a distribution of available maize under possible circumstances of climate and soil. In developing a baseline to reveal moisture effects on crop growth process, the irrigation requirements are modeled based on an acceptable range of potential yields. We have set this range starting with ideas on potential consumptive yields. In the work of Castetter and Willis (1942), consumptive yields were calculated assuming that a family with 5 people in average cultivated 0.25–2 acre (0.10–0.81 ha) of land, and the corn yield per unit ha was 12 bushels/acre (753 kg/ha). One family thus consumed 0.6–4.8 bushel corn per year. We reversed this method to relate potential yields and consumptive yields. The basal metabolic rate is about 1300–1500 kcal/day for an adult female and 1600–1800 kcal/day for an adult male (Purves and David 2004). For farmers with their high intensity of labor, the energetic costs of normal life are at least 2–3 times that of the basal metabolic rate (Diehl 1997).
Taking a 5-person family with an average 2400 kcal/day per person, the whole family would consume 4380,000 kcal per year. Castetter and Willis (1942) stated that only about 25 % of the calories came from maize within an Akimel O’odham diet. If taking this percentage for the middle Gila, there would be 1095,000 kcal of energy deriving from maize. According to Foundation of Anasazi Culture (Reed 2000), one kilogram of traditional corn could support 3476 kcal of energy. That would be 315 kg of maize needed for consumption. Given 0.1–0.81 ha of arable land for each family (half for maize, half for beans and pepo) and two cropping seasons per year (Hunt and Ingram 2007), yields should be within the range of 389–3150 kg/ha to support the basic consumption of a family. Taking the most conservative number, one hectare of farmland would have to produce 389 kg of actual maize to keep the balance between demand and supply.
This 389 kg is the actual yield. To determine potential yields from this baseline figure, all kinds of possible factors resulting in this actual yield from a given potential yield should be considered. When growing and processing maize, typical losses that can occur result from soil salinity, insects, and other factors like seed varieties, agricultural technology, and human actions (even stealing!). The paper assumes that each of these factors has a 40 % loss rate on average (based on Edwards et al. 1988; Oberle and Keeney 1990; Ismail et al. 1994; Kaufmann and Snell 1997; Gianessi and Janet 1999). Furthermore, a threshing rate of 20 % is added. All this leads to potential model yields of around 7.6 t/ha per season in the simulation. Referring to this amount of maize production, we have adjusted the Aquacrop model parameters to calculate more realistic irrigation demands for prehistoric periods. The precipitation and temperature input is the downscaled tree-ring data discussed above. The reference evaporation (ET0) data was calculated with the T-based Penman–Monteith equation.
Hohokam cropping intensity was likely to be twice per year with a growing cycle of 120 days per crop; the most likely cropping seasons were middle or early March to middle July and late July to middle November respectively (Hunt and Ingram 2007). At the beginning of each planting season, certain moisture and a minimum temperature are required for germination. The soil at the study site involves multiple types in the upper layer in terms of soil texture (Web soil survey, USDA). Although the soil variability does influence water demand for agriculture, the study emphasizes the impact of climate variability on crops. Therefore, the soil profile is assumed to be homogeneous as sandy-loam in the model. In future work, we will focus on the diversities of soil properties and the potential influence on accuracy and reliability of results. In addition, bunds and mulches are not applied in the model.
The figure shows that the net irrigation demands (unit in mm) have a good match with the so-called anomaly index (Salzer and Kipfmueller 2005). The anomaly index identifies the extreme climatic events at decadal scale, which is the most reliable scale from tree-ring data reconstruction. A positive value of the anomaly index is interpreted as wet conditions, and negative as dry conditions. In wetter periods, less water is required for maize in at least one of two seasons each year compared to other periods. During droughts, both seasons broadly demand larger quantities of water. In addition, there seem to be more droughts as well as floods during the Sedentary period compared to other periods on average. The results of first season irritation demands suggest that the probability of a demand above 600 mm during the Sedentary is higher than that during other periods, but the same is not apparent for a probability of below 500 mm. This may suggest that simulated results perform better in extreme dry events compared to extreme wet ones.
The results reveal that there are two leaps in water demands in the chronology of Hohokam society along the Gila River. The first one is between the Early Pioneer and Late Pioneer periods; a later one occurs between the Late Pioneer and Early Colonial periods. The reason of this leaping is obviously because of discrete values in command land areas and associated water volumes. These transitions—which would have been more gradual in reality—are very crucial periods in HNC. However, the actual evolutionary path in these transit periods has many possibilities. Did land expansion follow population growth or was it the other way around? Figure 6 shows that the population sustains a relatively stable increase before the Classic period and reaches its peak at the end of the Sedentary period. The rate of population growth is relatively high during the Early Colonial period compared to those in the Pioneer and Sedentary periods. Contrary to other periods, the Classic period shows a decline in population.
In the simulation, potential maize yields are demonstrated to be in a similar range per hectare. Over the span of one thousand years, maize water requirements did not become a limiting factor. Changing climate conditions were not determining maize potential productivity. Total water demands are directly proportional to potential maize yields for human consumption in each period and as such an indicator of stress for society to reach a certain production. Therefore, relating water requirements and population size indicates the relationship between potential food productivity and population change, or between potential supply of and demand for food. Figure 6 reflects an upward trend in both water demand and population during the Pioneer and Colonial periods. In the Sedentary period, populations keep increasing, but water demands maintain a stable range. By the Classic period, water demands continue to be stable, but population regresses. In this respect, the period between the Sedentary and Classic periods is a crucial process of HNC as well. Some changes seem to have broken the (delicate) balance between positive and negative influences of niche construction in this period. What these changes were exactly needs further study.
Discussion and conclusion
In this study on interactions between climate (change) and human dynamics, we have studied the relation between potential food productivity and population growth. We assessed the water amount needed for irrigating all the command area to see whether changes in the water demand would have been a potential problem for the Hohokam. We considered water needs for a reference maize harvest as a baseline for evaluating how difficult it may have been to grow the crop when only considering climate variables, providing a reasonable baseline analog for how climate changes may have influenced maize growth. The analysis has shown the occurrence of two leaps in relative water demands, one between Early Pioneer and Late Pioneer, and one between Late Pioneer and Early Colonial indicating that these were crucial transitions in HNC. The image of ‘leaping’ is artificial because of the use of discrete values in command land areas, and there are many possibilities to explain the exact shapes of these transit periods. Detailed study is needed to explain the transiting process further. In addition, the transition from Sedentary to Classic appears to be another significant process of HNC. The balance between positive and negative influences of niche construction is likely to be broken in this period.
The variation of irrigation requirements is matched with the normal fluctuation of climate over all four periods. As mentioned before, irrigation demands do not vary extremely over the modeling period; therefore, we assume that the instability of the canal systems originates from internal factors. The expansion of the canal system may increase its vulnerability and extreme climatic events—which can be found in each period—could have triggered the regression, even abandonment of the canal (Pande and Ertsen 2014).
The foregoing analysis illustrates the feasibility of tree-ring-based data in long-term modeling, taking into account traditional maize yields, using a CGM to calculate irrigation demands for the potential harvest of prehistoric maize. The concepts of potential, available, actual and consumptive yields needed to be taken into account. The irrigation requirements were modeled based on an acceptable range of potential yields, taking into account consumptive yields. The seasonal production of maize needed to be considered as well. Water demands in the first growing season appeared to be higher than that in the second because of the rainfall-evapotranspiration pattern. The simulated irrigation demands showed a relative good match with extreme climatic events. However, the model seemed to have better performance in extreme dry events than in extreme wet ones. One explanation may be that the tree-ring data are more sensitive to droughts than wet periods.
A next step would be to focus on the transformations between the phases as indicated. The actual construction and use of irrigation systems in response to changing conditions will be a main focus. Humans modify the landscape and hydrological environment in the process of canals construction. The evolutionary process of canal changes in turn affects human development. These canal systems do not work on rainfall, but on river flows. Thus, a next challenge is to build a 1000 year record of flows in the Gila River on a sufficiently low time step to be used as boundary conditions for flows in the canals. Once we have included stream flow data in the middle Gila River and how these flows were distributed between and within irrigation systems, we could determine ranges for available yields. This would possibly yield more detailed understanding of the variation of food availability and as such population densities in study area.
The study was supported by the Gila River Indian Community. We would like first to thank Dr. Kyle Woodson and Wesley Miles, who gave us great help to access to data collection. Profound thanks are due to Prof. Robert Hunt, who provided very important comments on this paper.
- Bandelier AF (1892) Final report of investigations among the Indians of the Southwestern United States, carried on mainly in the years 1880 to 1885. Cambridge University Press, CambridgeGoogle Scholar
- Burns B. T (1983) Anasazi storage behavior using crop yields reconstructed from tree rings. Anthropology. Tucson, University of Arizona. Ph. D DissertationGoogle Scholar
- Burt E. S (2007) Results of archaeological testing at three proposed home sites on allotted lands, Gila River indian community, pinal and maricopa counties, Arizona. CRMP Technical Report No. 2007-05Google Scholar
- Castetter EFB, Willis H (1942) Pima and papago Indian agriculture. New Mexico, University of New Mexico Press, AlbuquerqueGoogle Scholar
- Dean JS (1991) Thoughts on hohokam chronology. In: Gumerman GJ (ed) Exploring the hohokam: prehistoric desert peoples of the American Southwest., Amerind foundation new world studies Series No. 1. Amerind FoundationUniversity of New Mexico Press, Albuquerque, pp 61–149Google Scholar
- Doyel DE (1979) Prehistoric hohokam of the Arizona Desert. Am Sci 67(5):544–554Google Scholar
- Doyel D. E. (ed.) (1991). Hohokam cultural evolution in the phoenix basin. In: Exploring the hohokam: prehistoric desert peoples of the American Southwest. University of New Mexico Press, AlbuquerqueGoogle Scholar
- Fertelmes C. M, and C. Loendorf (2010) Archaeological testing at seven cultural properties for proposed agricultural development, Gila River Indian community, Pinal County, Arizona. CRMP Technical Report No. 2009-23Google Scholar
- Fish SK, Nabhan GR (1991) Deserts as context: the hohokam environment. In: Gumerman GJ (ed) Changing views of the hohokam. University of New Mexico Press, Albuquerque. pp 35–54Google Scholar
- Gasser R. E, and Kwiatkowski, S.M (eds.) (1991). Food for thought: recognizing patterns in hohokam subsistence. In: Exploring the hohokam: prehistoric desert peoples of the southwest. Albuquerque, University of New Mexico PressGoogle Scholar
- Gianessi LPC, Janet E (1999) Agricultural biotechnology: insect control benefits. National Center for Food and Agricultural Policy, WashingtonGoogle Scholar
- Gladwin H. S (1948). Excavations at snaketown IV: review and conclusions. Medallion Papers No.38. Gila Pueblo, GlobeGoogle Scholar
- Graybill D. A, (ed.) (1989) The reconstruction of prehistoric salt river stream-flow. In the 1982–1984 excavations at las Colinas: environment and subsistence. University of Arizona, TucsonGoogle Scholar
- Gregory D. A, and Douglas D. L (1994). An archaeological survey of the blackwater area, vol. 2: site descriptions and related data. Cultural Resources Report No. 86. Archaeological Consulting Services, Ltd., TempeGoogle Scholar
- Haury E. W (1937) The snaketown canal. In: H. S. Gladwin, E. W. Haury, E. B. Sayles, N. Gladwin (eds.). Excavations at snaketown: material culture, Medallion Papers No. 25. Gila Pueblo, Globe, pp 50–58Google Scholar
- Haury EW (1976) The Hohokam, desert farmers and craftsmen. University of Arizona Press, TucsonGoogle Scholar
- Howard JB (1993) A paleohydraulic approach to examining agricultural intensification in hohokam irrigation systems. In: Scarborough VL, Isaac BL (eds) Economic aspects of water management in the prehispanic new world. University of New Mexico Press, New York, pp 261–322Google Scholar
- Huckleberry G (1993). Surficial geology of the middle Gila River area, North-Central Pinal County, Arizona. Open-File Report 93-3. A. G. SurveyGoogle Scholar
- Hunt RC (1988) Size and the structure of authority in canal irrigation systems. J Anthropol Res 44:335–355Google Scholar
- Hunt R. C, and S. E. Ingram (2007) Double-cropping maize on the middle Gila River: intra-annual river regime and canal management in the classic period. Paper presented at the arizona archaeological council fall 2007 symposium, going with the flow: Current Research in Prehistoric Irrigation Technology, MesaGoogle Scholar
- Hunt R. C, and Scott E. I. (unpublished 1) Plant and animal production calendar for phoenix basin prehistoryGoogle Scholar
- Hunt R. C, and Scott E. I. (unpublished 2) Surface water and crop productivity in the southwest: the middle Gila River and Maize in ArizonaGoogle Scholar
- Kohler T. A, (ed.) (2012) Modeling agricultural productivity and farming effort. in emergence and collapse of early villages: models of central mesa verde archaeology. University of California Press, BerkeleyGoogle Scholar
- Larson A (1926) Map of prehistoric canal system, Gila valley. Map on file, Arizona State Museum, University of Arizona, TucsonGoogle Scholar
- Lewontin RC (1983) Evolution now: a century after Darwin-Smith, JM. N Y Rev B 30(10):21–27Google Scholar
- Loendorf C, Woodson K et al (2007) a preliminary report on the results of phase i data recovery investigations of cultural resources in santan reaches ST-IB, ST-IC, and ST-FP of the pima-maricopa irrigation project, Gila River Indian Community, Pinal County, Arizona. P-MIP Technical Report No. 2006-03Google Scholar
- Masse WB (1991) The Quest for Subsistence Sufficiency and Civilization in the Sonoran Desert. In: Crown PL, Judge WJ (eds) Chaco and hohokam: prehistoric regional systems in the American Southwest., School of american research advanced seminar seriesSchool of American Research Press, Santa Fe, pp 195–223Google Scholar
- Midvale F (1935) A general map of the central Pima Country on the Gila. Prepared for arizona geography course. Ms. on file, Cultural Resource Management Program, Gila River Indian Community, SacatonGoogle Scholar
- Midvale F. (1946) the prehistoric canals and ruins of the casa grande-florence area of the Gila River in Southern Arizona. map on file, Frank Midvale collection, Arizona collection, Hayden library, Arizona State University, TempeGoogle Scholar
- Midvale F (1963) The prehistoric irrigation of the casa grande ruins area of the Gila River in southern Arizona. Unpublished map on file (No. 225), Frank Midvale collection, Arizona collection, Hayden library, Arizona State University, TempeGoogle Scholar
- Midvale F (1965) Prehistoric irrigation of the Casa Grande Ruins Area. The Kiva 30(3):82–86Google Scholar
- Midvale F (1972) Frank Midvale papers, 1865–1972, bulk 1930–1972. In Arizona collection, Hayden library, Arizona State University, TempeGoogle Scholar
- Motsinger T. N (1993) Archaeological investigations at the Gila butte site: hohokam irrigation and economic systems along the Gila River, Arizona. Archaeological Report No. 93-84. SWCA, Inc., PhoenixGoogle Scholar
- Muenchrath DAKM, Sandor JA, Homburg JA (2002) Observational study of maize production systems of zuni farmers in semiarid New Mexico. J Ethnobiol 22(1):1–33Google Scholar
- Neily R. B, Brodbeck M, and Wigglesworth K. S (1999). An archaeological survey of the santan canal reaches and adjoining areas in the santan management area, Pima-Maricopa Irrigation Project (P-MIP), Gila River Indian Community, Arizona. P-MIP Report No. 2. Cultural Resource Management Program, Gila River Indian Community, SacatonGoogle Scholar
- Odling-Smee FJ, Laland KN et al (2003) Niche construction: the neglected process in evolution. Princeton University Press, New JeseyGoogle Scholar
- Petersen K. L (1987) Tree-ring transfer functions for estimating corn production. Dolores archaeological program: supporting studies: settlement and environment. K. L. O. Petersen, J. D. Denver, U.S. Department of the Interior, Bureau of ReclamationGoogle Scholar
- Purves WKS, David M (eds) (2004) Life: the science of biology. Freeman, New YorkGoogle Scholar
- Ravesloot JC (2007) Changing views of snaketown in a larger landscape. In: Fish Suzanne K, Fish Paul R (eds) The hohokam millennium. School for advanced research Press, Santa Fe, New Mexico, pp 91–98Google Scholar
- Reed PF (2000) Foundation of anasazi culture-the basketmaker-pueblo transition. University of Utah Press, Salt LakeGoogle Scholar
- Schiffer MB (1982) Hohokam chronology: an essay on history and method. In: McGuire RH, Schiffer MB (eds) Hohokam and patayan: prehistory of southwestern Arizona. Academic Press, New York, pp 299–344Google Scholar
- Southworth C. H (1914–1915) Gila River survey, Pinal county, Arizona. U.S. Department of the Interior, U.S. Indian Service Irrigation, Washington, DC. Map copies on file, Cultural Resources Management Program, Gila River Indian Community, SacatonGoogle Scholar
- Van West CR (1994) Modeling prehistoric climatic variability and agricultural production in southwestern Colorado. Washington State University Department of Anthropology, PullmanGoogle Scholar
- Woodbury RB (1976) West Fork. In: Haury EW (ed) The hohokam: desert farmers and craftsmen, excavations at Snaketown. The University of Arizona Press, Tucson, pp 141–142Google Scholar
- Woodson M. K (2006) National register of historic places eligibility assessment of GR-436 along reach ST-IB of the Pima-Maricopa Irrigation Project Main-Stem Canal, Santan Management Area, Gila River Indian Community, Pinal County, Arizona. P-MIP Technical Report No. 2006-06Google Scholar
- Woodson MK (2009) Map of prehistoric hohokam canal systems in the middle Gila valley, 3rd edn. Cultural Resource Management Program, Gila River Indian Community, SacatonGoogle Scholar
- Woodson M. K (2010) The social organization of hohokam irrigation in the middle Gila River valley, arizona. tempe, arizona state university. Ph. D DissertationGoogle Scholar
- Web Soil Survey http://websoilsurvey.nrcs.usda.gov/app/WebSoilSurvey.aspx
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.