Addressing the challenges of anthropogenic warming is an urgent responsibility of the scientific community. Archaeologists know that the past is an archive of human decision-making that includes responses and adaptations to climate challenges. Investigating what worked, what did not, and why can lead to helpful insights as humanity confronts the challenge of adapting to our warming world. The importance of archaeological studies in this domain has been argued for more than two decades (e.g., Boivin and Crowther Reference Boivin and Crowther2021; Kohler and Rockman Reference Kohler and Rockman2020; Rick and Sandweiss Reference Rick and Sandweiss2020; van der Leeuw and Redman Reference van der Leeuw and Redman2002; and many others). What has not been clear is how archaeologists and others who document and interpret the past can generate these insights. How to investigate potential climatic influences on human behavior in the past is rarely presented or taught as a distinct mode of inquiry. As a result, published case studies are the primary but indirect method for learning how to conduct this work. That there is little published critique to guide future efforts or consensus on the basis to critique it (other than correlation is not causation) is a barrier to advancement in this domain. Climate and human behavior studies in archaeology are an established but underdeveloped domain of study and practice, mostly bereft of published guidance and methodological clarity.
This article aims to address this gap by providing a practical introduction to and tool kit for some of the conceptual models, methods, data, and challenges of climate and human behavior studies. The audience for this article includes students seeking foundational guidance on how to do this work and archaeologists, anthropologists, historians, environmental social scientists, geographers, and others who are looking for new ideas, methods, and resources to inform their work. By expanding competence in this domain, we can enhance documentation and interpretations of the past, enabling policy-relevant insights to emerge that will contribute to preparing for and responding to our warming world.
Climate and human behavior studies characterize and interpret the climatic context of behavior and potential climatic influences on this behavior at scales ranging from households to regions. Human behavior here includes recorded (orally, archaeologically, historically) evidence of human actions that comprise histories and their trajectories. Episodic actions that are responses to short-term weather events are not the focus because they are unlikely to result in sustained changes in what people do or be detectable in the archaeological record. Climate and human behavior studies exist within the long history of anthropological and archaeological studies of human–environment interactions subsumed within environmental and ecological anthropology. Research questions, models, data, and expectations typically arise from frameworks developed through related perspectives referred to as cultural ecology (Steward Reference Steward1972), human ecology (Butzer Reference Butzer1982), environmental archaeology (e.g., Dincauze Reference Dincauze2000), and historical ecology (Crumley Reference Crumley and Carole1994).
Climate and human behavior studies are part of a rich legacy of careful anthropological thinking, documenting, and interpreting that, in its most applicable form for modern problem solving, keeps the community as the primary unit of analysis and recognizes human agency. These studies need not be a throwback to environmental determinism: explanations of culture and behavior that rely primarily on environmental factors and ignore multicausality. Instead, climate and human behavior studies can illuminate the importance of human choices and options in creating historical trajectories within various environments and climates.
This article is organized into seven steps, the key components of a research design to develop a climate and human behavior study. The five associated guides (available through tDAR [Ingram Reference Ingram2025a]) are an extensive tool kit containing directions, keystrokes, examples, and references to expand the presentation here. I focus on what is of practical value for those who want to do this work and for those looking for new ideas and resources. The models, methods, and data presented apply to many regions globally and are drawn from my training and the long history of climate and human behavior studies in the North American Southwest (SW). The steps of a research design and the associated sections include guidance on how to:
1. Identify a research problem and question. These questions can be stimulated by a desire to understand climate–human interactions and trajectories in the past, present, or future.
2. Identify climate and human behavorial data. Instrumental (meteorological station) and tree-ring–based studies are the focus here. These data are abundant and publicly accessible.
3. Select a conceptual model to inform the research design, analysis, and interpretation of results. Models help our work progress with rigor, transparency, and replicability and discourage deterministic assertions.
4. Investigate how climate affects resource (plant and animal) productivity. The impact of climate on people is usually through its impacts on resources that people rely on.
5. Consider the social and environmental context. Changes in context often affect human vulnerability and resilience to climate extremes, such as dry, wet, warm, and cool periods.
6. Select methods and analyze potential relationships. Some analytical challenges and suggested procedures are identified.
7. Present the results to and beyond archaeologists. The benefits of using terms that are understood beyond archaeology and sharing climate stories from the past are asserted.
The research design process is iterative, and the steps need not be followed in the order presented here. My first climate–human behavior study began with seeing an Excel spreadsheet containing a thousand years of tree-ring–derived, retrodicted annual streamflow data.
Step 1: Identify a Research Problem and Question
Problems and questions in climate and human behavior studies can originate in the past or present. If we begin with a concern about current anthropogenic warming and related droughts, then we can identify past periods and places where warming or drought was identified and can investigate the range of human responses. If we are stuck on an archaeological problem that originates in the past, then it is likely there are approaches beyond archaeology that might be helpful. When I began reading reports written by the Intergovernmental Panel on Climate Change (IPCC), I learned new ways of thinking about climatic influences on behavior and a new conceptual vocabulary such as vulnerability, adaptive capacity, mitigation, and climate extremes (see Kohler and Rockman [Reference Kohler and Rockman2020] for a “primer” for archaeologists on the IPCC). International organizations have conducted extensive research and developed material on most contemporary problems. Their reports are a tremendous resource for thinking about many problems in the past.
Archaeologists can make important contributions to the present and future by investigating, with archaeological data, the assumptions on which modern policies rely (e.g., Cooper and Sheets Reference Cooper and Sheets2012; Ingram and Patrick Reference Ingram and Patrick2021; Nelson et al. Reference Nelson, Ingram, Dugmore, Streeter, Peeples, McGovern and Hegmon2016). A primary argument for archaeological contributions to present concerns is their ability to document long-term cycles of stability and change in human and environmental systems. Such a long-term view can reveal thresholds, emergent behavior, and controlling variables within systems and histories. Policymakers and practitioners working toward solutions in the present are also informed by learning from past successes and failures, but their accumulated policy-relevant information will be limited to shorter time spans.
Some ongoing and future questions for climate and human behavior studies to consider include the following:
• Indigenous peoples whose traditional and modern territories are often in environmentally marginal areas, are especially vulnerable to the impacts of climate change due to socioeconomic and political marginalization (Redsteer et al. Reference Redsteer, Bemis, Chief, Gautam, Rose Middleton, Tsosie, Garfin and Jardine2013). How do we work together with Indigenous people to address challenges associated with climate change?
• What social, cultural, and environmental conditions led to human resilience and vulnerability to past climate extremes?
• What adaptation strategies and plant resources relied on in the past can be helpful today (Jackson et al. Reference Jackson, Dugmore and Riede2017; Minnis Reference Minnis2014; Pailes et al. Reference Pailes, Norman, Baisan, Meko, Gauthier, Villanueva-Diaz, Dean, Martínez, Kessler, Towner and Lal2024)?
• What does climate injustice look like in the past, and what were its consequences?
• What interdisciplinary issues are arising in the study of climate and human behavior, and how can studies of the past contribute to and benefit from these studies (e.g., Sommer Reference Sommer2015)?
• How can archaeologists and other historically focused social scientists engage in transdisciplinary and global change research (Rick and Sandweiss Reference Rick and Sandweiss2020) with scholars in different fields who already have the attention of policymakers (Smith Reference Smith2021), including those at the IPCC (Kohler and Rockman Reference Kohler and Rockman2020)?
• There are millions of smallholder farmers worldwide who continue to practice subsistence agriculture (Lowder et al. Reference Lowder, Skoet and Raney2016) using methods similar to those used in the past. Databases and archives of human behavior with strong potential for synthetic insights include CyberSW (Mills et al. Reference Mills, Ram, Clark, Ortman and Peeples2020), the Digital Index of North American Archaeology (e.g., Ritchison and Anderson Reference Ritchison, Anderson, Robert and Aaron2022), Human Relations Area Files Archaeology, the Digital Archaeological Record, Crow Canyon Archaeological Center, Chaco Research Archive, and the site databases of each state historic preservation office. What can large-scale archaeological databases of settlement locations, dates of occupation, and other material culture, when combined with spatially explicit paleoclimate records, tell us about human decision-making in the context of climate extremes? These data offer opportunities for generalizable insights.
Step 2: Identify the Available Climate and Human Behavorial Data
There are abundant publicly accessible climate data, but their volume and diversity are often overwhelming (e.g., National Centers for Environmental Information 2024). Instrumental (meteorological station) and tree-ring data are available for most of North America and are valuable sources of information on precipitation, temperature, streamflow, and other factors that affect resource productivity. However, tree-ring data available for download from the International Tree-Ring Data Bank can be difficult to interpret. A guide to data sources, instructions, and interpretations of both tree-ring and instrumental data is provided in Ingram (Reference Ingram2025b:Guide 1). It is often helpful to find collaborators and reviewers with an intimate knowledge of the data, but specific expertise in climate science may be unnecessary if the strengths and weaknesses of the data are identified and considered in a research design.
Many types of human behavioral data are appropriate for a climate and human behavior study as long as a plausible chain of causality is established (and argued) between some aspect of climate and a materially identifiable behavioral response. Multiple lines of behavioral evidence will always be stronger than single lines of evidence. For example, for investigations of drought impacts in a study area, increased food-storage efforts (Dean Reference Dean, David and Jeffrey2006), nutritional deficiencies evident in skeletal remains (Martin et al. Reference Martin and George1994), and documentary and ethnographic evidence describing drought impacts would provide strong evidence of those impacts on behavior. The spatial scale of a study can range from a single site to thousands of sites in a region. The temporal scale will be a function of the climate and behavioral data available and the research question. For contributions to modern climate change efforts, generalizable results that engage large-scale archaeological databases will likely be the most persuasive.
Step 3: Select a Conceptual Model
Climate and human behavior studies require more than curve-matching a change in climatic conditions with a coincident change in human behavior (see Contreras Reference Contreras and A. Contreras2016; McGovern Reference McGovern1991; and many others). As with all interpretive efforts, equifinality challenges climate and human behavior studies. There must be methodological rigor to identify causal relationships. As Kohler and Rockman (Reference Kohler and Rockman2020:640) succinctly state about archaeological contributions to the IPCC, “We cannot be taken seriously by other scientists if they do not judge our causal arguments to be credible.” The first assumption of a climate and human behavior study should be that the climatic condition of interest did not affect behavior (the null hypothesis). Prolonged climatic extremes are frequent in climate systems worldwide; thus, temporal correlations between an extreme and a behavioral change are inevitable. How do we determine when a climate extreme stimulated a behavioral response (causation) and when it did not (correlation)?
A disclosed conceptual model that links some aspect of climate to plausible behavioral responses is a necessary first step to building credibility and avoiding spurious conclusions (de Vries Reference De Vries, Kates, Ausubel and Berberian1985; Ingram et al. Reference Ingram, Farmer, Wigley, Wigley, Ingram and Farmer1981). A conceptual model hypothesizes or assumes connections, interactions, and influences among a limited set of variables. It identifies potential mechanisms for the proposed chain of causality between independent (climatic) and dependent (behavioral) variables and provides the basis for inferences about expected relationships. Models clarify and limit the range of proximate (nearest in time or space) and ultimate causes, so one can investigate potentially influential variables in a specific study. They are not intended or designed to reveal the range of options that might explain a behavior or history.
Research questions, problems, and theoretical frameworks drive the selection of models and variables. The utility of a model can be judged by the questions it prompts a user and reader to ask and the transparency it provides for replication and evaluation. Variable boxes can represent model relationships, and connecting arrows can show a plausible chain of causality (Figure 1; see Kates [Reference Kates, Kates, Ausubel and Berberian1985] for some climate and society models). Conceptual models represented in this way are simplifications that are not meant to capture the full complexity of a phenomenon but rather to make it analytically tractable.
A conceptual model of a plausible causal relationship between dry periods (droughts) and a human behavioral response. This risk model relies on an assumption of resource marginality.
Conceptual models are necessary because the impact of climate on human behavior is usually indirect and mediated by other factors. Climate directly affects the growth of plants and animals (discussed in Step 4) and other resources people rely on. Resource security and insecurity can stimulate behavioral responses. Purely deterministic models—for example, drought causes conflict—often fail to reveal the proposed chain of causality, making them difficult to evaluate. Droughts did not directly cause people to fight, but perhaps competition for diminishing arable land initiated the conflict; see Hsiang and colleagues (Reference Hsiang, Burke and Miguel2013) for a meta-analysis of 60 quantitative studies of the effect of climatic events on conflict and plausible causal mechanisms. Models that identify potential mechanisms of causation and research that tests proposed relationships are an antidote to what McGovern (Reference McGovern1991:78) describes as the problem of “concluding statements about culture change that boil down to ‘it got cold/hot/dry/wet and they died/floresced/migrated/intensified.’”
Assumptions
Climate and human behavior models often rely on an assumption of resource marginality. Marginality exists where and when resource productivity fluctuates around a threshold above which there was enough culturally defined food to meet needs and below which there was not. This marginality creates the risk of resource shortfalls where and when resource productivity is low relative to actual or perceived human food needs. 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 may motivate a behavioral response. Resource marginality is not just an environmental characteristic of drylands because marginality is a function of human perceptions, expectations, and changes in resource supply and demand. It is also a function of social systems and the technologies available and how they are deployed. Evidence supporting a marginality assumption can include historical and ethnographic accounts of climate-related crop failures, relatively low productivity of resources defined as food, and evidence of human nutritional deficiencies.
Model Examples
The models in this section are often used in archaeological and anthropological studies of potential climatic impacts on behavior. Each model is briefly described to help new entrants into this domain identify potential models that are compatible with their interests and data. For those already doing this work, the examples may stimulate new ideas for their research. To clarify how each model can be operationalized, some independent and dependent variables and published examples of studies that are informed by the models are provided in Ingram (Reference Ingram2025c:Guide 2). The models and associated references are also an abundant source of research questions and, in general, create a research framework. No “best” model exists, and some studies engage multiple models.
Risk Models. A climate-related risk model links some aspect of climate to the risk of a negative consequence that may stimulate a behavioral response. For example, a drought-related risk of starvation can motivate various behavioral responses such as migration as people act to reduce the risk (see Figure 1). Risks may be actual or perceived. Risk has been defined and used in many ways and disciplines (e.g., Cashdan Reference Cashdan1990; Cochran et al. Reference Cochran, Miller, Wholey, Gougeon, Gaillard, Jane Murray and Parker2023). The IPCC defines risk as the “potential for adverse consequences for human or ecological systems, recognizing the diversity of values and objectives associated with such systems” (Reisinger et al. Reference Reisinger, Howden, Vera, Mathias, Margot, Sylvia and Katharine2020:5).
Adaptation and Mitigation Models. Adaptation and mitigation models identify, describe, and explain various behaviors as cultural responses to resource-related scarcity, such as those related to drought. These behaviors are often understood as responses to risk or uncertainty (Halstead and O’Shea Reference Halstead and O’Shea1989) and are usually persistent, culturally informed buffering strategies, rather than episodic responses to specific climate events. Among these strategies are resource (plant or animal) diversification or specialization, food storage, social networks and exchange, water management, and mobility.
Resilience Models. A climate-related resilience model identifies social and environmental conditions that affect the capacity of a socioecological system to absorb disturbances, such as those from climate extremes, and still retain its structure and function; that is, to be recognizable after a disturbance (Folke Reference Folke2006; Gunderson and Holling Reference Gunderson and Holling2002). The resilience of a system varies over time in an adaptive cycle, with periods of greater and lesser vulnerability to disturbance. Thus, resilience models can be used to explain differences in human responses to similar climate events. The analytical unit in resilience studies is a socioecological system wherein people and nature are understood as inseparable (Berkes and Folke Reference Bawden and Martin Reycraft2000). Folke notes, “The resilience approach emphasizes non-linear dynamics, thresholds, uncertainty and surprise, how periods of gradual change interplay with periods of rapid change and how such dynamics interact across temporal and spatial scales” (Reference Folke2006:253).
Vulnerability Models. A climate-related vulnerability model identifies social and environmental conditions that affect the extent of harm that people and communities experience from some aspect of climate. These conditions can explain climate’s differential impact on people and societies over time and space (Thomas et al. Reference Thomas, Hardy, Lazrus, Mendez, Orlove, Isabel Rivera‐Collazo, Rockman, Benjamin and Winthrop2019). The study of human vulnerability to climate change is a significant research emphasis of the IPCC and multiple disciplines (e.g., Thomas et al. Reference Thomas, Hardy, Lazrus, Mendez, Orlove, Isabel Rivera‐Collazo, Rockman, Benjamin and Winthrop2019; Turner et al. Reference Turner, Kasperson, Matson, McCarthy, Corell, Christensen and Eckley2003). Meyer and colleagues state, “Vulnerability of populations and activities is the most widely used umbrella concept for those factors that mediate between geophysical events and human losses” (Reference Meyer, Butzer, Downing, Turner, Wenzel, Wescoat, Rayner and Malone1998:239). There are many specific definitions of vulnerability (for a summary, see Cutter Reference Cutter1996:531–532): three definitions are the “potential for loss” (Cutter Reference Cutter1996:529), the “capacity to be wounded” (Kates Reference Kates, Kates, Ausubel and Berberian1985:17), and the “potential for negative outcomes or consequences” (Meyer et al. Reference Meyer, Butzer, Downing, Turner, Wenzel, Wescoat, Rayner and Malone1998:239). Vulnerability models are easier to archaeologically operationalize than resilience models because changes in human vulnerability are often materialized and inferred through changes in existing practices or conditions, whereas periods of resilience can appear as periods of stability due to a lack of change.
Push–Pull models. “Push–pull” models, used in many areas of study, are frequently used to investigate potential climatic impacts on people when the behavior of interest is population movement (migration). In this model, population movements are most likely to occur when there are push factors at the population origin (e.g., drought, conflict, population increases), pull factors at the population destination (e.g., social, climatic, and environmental advantages), and when the migration costs between the two are perceived to be acceptable (Anthony Reference Anthony1990; Burke et al. Reference Burke, Miguel, Satyanath, Dykema and Lobell2009; Hsiang et al. Reference Hsiang, Burke and Miguel2013). The push–pull model for archaeological research is well articulated and demonstrated by Cameron (Reference Cameron1995), Clark (Reference Clark2001), Cook and Comstock (Reference Cook and Comstock2022), and many others.
Supply and Demand Models. Many studies emphasize climate-related changes in the supply and demand for resources as an explanation for behavioral responses. Some supply-model studies invoke past global-scale warming and cooling periods—for example, the Medieval Warm Period, Little Ice Age, or Younger Dryas—or changes in atmospheric circulation patterns, such as the El Niño–Southern Oscillation, that cause large-scale changes in resource productivity as explanatory factors for behavioral responses. Others might identify variation in resource productivity due to a spatial variation in climatic or environmental conditions within a region. Demand models emphasize the contribution of demographic conditions to human responses to climate events without considering the supply of resources. Combined demand and supply models are essentially carrying-capacity models. However, the carrying capacity of any given area is not fixed, so these models are often challenging to develop and evaluate. In all versions of these models, research that demonstrates rather than infers the impacts of changes in supply and demand on human responses to climate challenges is stronger. Excellent examples of such demonstrations are paleoproductivity models–computational models (agent-based and dynamical) that integrate multiple variables that affect, for example, maize growth and usually integrate behavior data, such as settlement locations and population parameters (e.g., Bocinsky and Varien Reference Bocinsky and Mark2017; Schwindt et al. Reference Schwindt, Bocinsky, Ortman, Glowacki, Varien and Kohler2016).
Combined Behavior, Demography, and Paleoenvironment Models. Jeffrey Dean and colleagues (Reference Dean, Euler, Gumerman, Plog, Hevly and Karlstrom1985) propose understanding human responses to environmental fluctuations as conditioned by interactions among key low- and high-frequency environmental, demographic, and cultural processes. Although the combined model was developed for the southern Colorado Plateau in the US Southwest, it has wide applicability elsewhere. Low-frequency processes have periodicities longer than a single human generation (>25 years); for example, cycles of erosion and deposition, fluctuations of water tables, long-term population change, and cultural trends. High-frequency processes have shorter periodicities, usually within a single human generation (<25 years) and include annual climate variability, droughts, short-term population fluctuations, and risk-buffering strategies such as increasing food storage. These model variables form relationships that define an adaptive system at any point in time. Thus, similar climatic and environmental perturbations may result in different human responses depending on demographic and cultural conditions. The model does not support single-cause, deterministic explanations of cultural and historical change.
Sociocultural Models. Conceptual models need not be limited to purely functional relationships between climate variables, resource productivity, and human responses, such as migration away from resource shortfall risks. Ethnographies and ethnologies are abundant sources of models to explain climatic influences on behavior. Humans often make meaning based on their culturally mediated perceptions, rather than on an objective assessment of conditions (e.g., McIntosh et al. Reference McIntosh, Tainter and McIntosh2000). Climate events that create food insecurity or shortfalls can engage a range of social responses based on their perceived causes. Such causes can lead researchers to belief systems that link climate events and agricultural productivity to spiritual sources and rewarded and punished behaviors (Ember and Skoggard Reference Ember and Skoggard2023; Skoggard et al. Reference Skoggard, Ember, Pitek, Conrad Jackson and Carolus2020).
Disaster Models and the Anthropology of Climate Change. Anthropological studies of disasters (e.g., Bawden and Reycraft Reference Bawden and Martin Reycraft2000; Faas and Barrios Reference Faas and Roberto2015; Oliver-Smith and Hoffman Reference Anthony and Susanna2019) and the anthropology of climate change (e.g., Baer and Singer Reference Baer and Singer2015; Crate and Nuttal Reference Crate and Nuttall2016) provide detailed information on how climate affects people at specific places and times. These place-based studies generate nuanced descriptions and interpretations that can stimulate model development for interpreting the past and suggest strategies for the future. Cross-cultural studies can identify and communicate the range of coping strategies that people used in response to different climatic hazards to enhance future climate change adaptation (Pierro et al. Reference Pierro, Ember, Pitek and Skoggard2022).
In review, the importance of a disclosed conceptual model for a climate and human behavior study cannot be overstated. A model presents why and how some aspect of climate may or did affect human decision-making and the resulting behavior. The logic of a research design should build on the relationships specified in a model.
Step 4: Investigate How Climate Affects Resource Productivity
The research design of a climate and human behavior study must consider how climate affects resources that people rely on because climate-related resource productivity is the critical link between changes in climatic conditions and changes in behavior. Essential resources for people include plants (wild and cultivated) and animals necessary to meet food needs and social obligations. Water is a primary limiting factor on plant growth in drylands and elsewhere (Fischer and Turner Reference Fischer and Neil1978; Muenchrath and Salvador Reference Muenchrath, Salvador and Toll1995). Animals that rely on plant foods are also affected by changes in climate that influence plant growth (Osborn Reference Osborn1993; Wendt et al. Reference Wendt, McWethy, Widga and Shuman2022). For example, as archaeologists of the Great Plains know, it is important to investigate the impacts of climate on the distribution of bison and the plants they rely on because those impacts also affected, in some cases, the distribution and associated behaviors of people.
Understanding how aspects of climate can affect resource productivity identifies the opportunities and risks that people must work with to produce or acquire food. Without recognizing some of the details and nuances of climate–plant and animal–human relationships, we risk making simplistic and spurious assertions and interpretations of causality. For example, if the growth of plants perceived as food is determined mainly by winter precipitation and associated soil moisture, then a focus on proxies of summer or total annual precipitation may be irrelevant. Or, if a key resource is not substantially limited by climatic conditions, then there is no basis for investigating potential climatic impacts on behavior.
Where there is a strong relationship between aspects of climate and resource productivity, those aspects may be used as a proxy for variations in resource productivity. This link relies on the concept of limiting factors: “The growth and functioning of an organism is [sic] dependent upon the amount of the essential environmental factor presented to it in minimal quantity during the most critical season of the year, or during the most critical year or years of a climatic cycle” (Taylor Reference Taylor1934:378). Plant growth requires water, mineral nutrients, light, and air; excesses or deficiencies in any of these factors may stress a plant and affect its growth and productivity (Muenchrath and Salvador Reference Muenchrath, Salvador and Toll1995:309–310). Limiting factors of a plant vary based on its specific location. At higher elevations, temperatures might restrict the growth of some plants, whereas at lower elevations, precipitation may be the primary limiting factor. Here I focus on plants and later agriculture, but climate affects many other nonfood resources that people rely on, such as shelter and mobility.
Researchers can apply the method of identifying climatic impacts on specific plant resources presented in Guide 3 (Ingram Reference Ingram2025d) to many plants that people relied on, such as maize, rice, cotton, wheat, amaranth, millet, beans, squash, melons, and agave. Maize (Zea mays L.) is the example developed because it is sensitive to climatic conditions, grown and relied on worldwide, and extensively studied within and beyond archaeology. For information on the limiting climatic and environmental factors of other plants, studies from early twentieth-century US Agricultural Experiment Stations are good sources of information (e.g., McClatchie Reference McClatchie1904). Experiment Stations were tasked with providing basic and useful information to farmers because many did not have accurate information about climate or farming in specific places (Libecap and Hansen Reference Libecap and Kocabiyik Hansen2002). The Food and Agriculture Organization of the United Nations (2024) is an excellent source for the water, temperature, and other requirements of major plants grown worldwide. Note that modern hybrids may have requirements that differ from traditional varieties grown in the past.
Climate-related resource productivity is strongly affected by plant cultivation and agricultural practices (e.g., Doolittle Reference Doolittle2002; Ingram and Hunt Reference Ingram, Ingram and Hunt2015). Some primary variables that affect this productivity include the selection of soils and their characteristics and interactions with water, field locations that through human locational choices seek to manage climatic and environmental variables for the benefit of plants, and water management strategies that move and aggregate water at or away from planting locations. These related components interact with climate and culture throughout a solar year and are sometimes ethnographically documented in a food production calendar. One benefit of learning about climatic variables that affect food production in a study area is that it clarifies why precipitation, streamflow, and temperature values (seasonal, annual, or total) must be used with caution when they are deployed as proxies for resource productivity. Humans have many ways to influence productivity. For more information about maize-related agricultural variables, food production calendars, and related human strategies, see Ingram (Reference Ingram2025d:Guide 3).
Most of us interested in climatic impacts on behavior are not specialists in agronomy and plant biology, and the details provided here are intended to improve and support rather than overwhelm or discourage future efforts. Some studies will not require retrodicted annual values of past climate variables or detailed knowledge of the water requirements of specific plants. For example, first-principle associations between water and plant growth support the hypotheses, expectations, and models of most dryland climate and human behavior studies. This is because evapotranspiration—water loss due to evaporation from the soil, water bodies, and other land surfaces, as well as transpiration from plant leaves—averages 100% in arid and semiarid climates (Hanson Reference Hanson, Paulson, Chase, Roberts and David1991). The result is a near-linear relationship between net plant production and growing season rainfall that has a similar slope in semiarid lands worldwide (Scholes [Reference Scholes2020] and references therein). Thus, increases in precipitation and streamflow levels, up to some threshold, improve plant and animal productivity. Conversely, decreases in precipitation and streamflow cause declines in resource productivity. Therefore, climate proxies that identify relative changes in precipitation and streamflow may be sufficient to track changes in resource productivity in some environments.
Step 5: Consider the Social and Environmental Context of Potential Climatic Influences on Human Behavior
An effective research design for a climate and human behavior study considers the social and environmental context of potential human responses to some aspect of climate. This step has several aims: it generates research questions, calibrates expectations, identifies the range of material indicators of behavior, and offers the long-term variation in climate and behavior necessary to identify a pattern and associated causality. It is also necessary because similar climate extremes often do not result in similar behavioral responses over time (e.g., Ingram Reference Ingram, Ingram and Hunt2015; Kintigh and Ingram Reference Ingram, Sulas and Pikirayi2018). The result is an uneven relationship between climate and behavior, possibly explained by changes in context. In other words, an argument for the influence of a specific drought on behavioral change must explain why all droughts with similar characteristics did not result in similar impacts on behavior.
The social context of climatic challenges can increase people’s vulnerability to climate extremes by affecting the range of available behavioral responses to a climate extreme. For example, if conflict increases in a region, the size of the area from which people obtain resources may decrease if people respond to an increasingly dangerous landscape by limiting their mobility and associated range of resource acquisition. Trade relationships may also change if alliances are reconfigured in a hostile environment. Similarly, social challenges in one area can affect vulnerabilities and insecurities in other areas (e.g., Ingram and Patrick Reference Ingram and Patrick2021). Hill and colleagues (Reference Hill, Clark, Doelle and Lyons2004), for example, argue that depopulation of the northern SW in the late thirteenth century AD created social and environmental changes in destination areas in the southern Southwest that resulted in a century-scale demographic decline. People facing such challenging circumstances were likely more vulnerable to climate-related resource declines if traditional adaptation and buffering strategies were weakened. Thus, an investigation into climatic impacts on behavior during this tumultuous period of decline that did not consider the changing social context would risk producing spurious results.
Considering the environmental context often reveals that environmental variables are linked and always changing. People are often confronted with simultaneous changes in multiple environmental variables, all of which may create challenges to resource productivity and food security. In a classic study of the thirteenth-century depopulation of the northern SW, Gumerman (Reference Gumerman1988) documented multiple environmental and climatic challenges to resource productivity created by changes in the deposition and erosion of floodplain sediments, lowering of the water table, declining precipitation and temperature trends, and demographic and cultural changes. Until these variables were documented, the prevailing view was that a single severe drought was responsible for the depopulation.
Step 6: Select Methods and Analyze Potential Relationships
A primary methodological effort in climate and human behavior studies is to identify a pattern between some aspect of climate and some aspect of behavior (or histories) in an area of interest and to investigate the possibility of a causal relationship. Despite the analytical necessity to limit the number of climate and behavioral variables investigated in a study, most researchers accept that climate is rarely the single cause of complex social phenomena and histories. There is no expected set of methods or tools to identify causal relationships; however, plausible conceptual models linking climate and behavior are necessary (see Step 3). Methodological challenges are generally understood, and some processes for research in this domain have been proposed (e.g., Degroot et al. Reference Degroot, Anchukaitis, Bauch, Burnham, Carnegy, Cui, de Luna, Guzowski, Hambrecht and Huhtamaa2021; Kintigh and Ingram Reference Ingram, Sulas and Pikirayi2018; Kohler and Rockman Reference Kohler and Rockman2020). Some analytical considerations for a research design are presented next.
Temporal and Spatial Mismatches
Researchers must develop a method to address the mismatch between the temporal resolution of climate data (annual if based on tree-ring data; centuries to millennia for some climate proxies) and behavioral data that vary based on the available chronometric procedures. Comparing climate data and behavioral changes with curving (smoothing) lines in a scatterplot can be a weak approach, especially when the mechanisms of influence are not specified. It is also a weak approach if periods when the curves do not match are unexplained. In study areas with high climatic variability and relatively abundant behavioral changes, curve matches are inevitable (for a statistical method of addressing this problem, see Kintigh and Ingram Reference Ingram, Sulas and Pikirayi2018). Another approach is to characterize climate conditions in the same temporal scale as the behavioral data, such as in 50-year intervals (for an example, see Ingram Reference Ingram, Lucas and Eric2016). Researchers investigating possible spatial and temporal correlations should expect lagged human responses because of the variety of buffering strategies that people employ.
Spatial mismatches can occur when a global-scale climate event, such as the Medieval Warm Period, Little Ice Age, Younger Dryas, or El Niño–Southern Oscillation, is proposed to have affected human behavior that occurred at a much smaller spatial scale. Social and environmental landscapes are not homogeneous, so a global-scale climate event likely resulted in different impacts, if any, across space. Attempts must be made to account for variation in the strength or impacts of the climate event at the scale of a study area.
Systematic Correlation Analyses
A climate sensitivity assessment is one approach to investigate past climate–human behavior relationships. Sensitivity studies “attempt to identify climate-sensitive groups, activities and areas, linking them to the varied levels of climate extremes” (Kates Reference Kates, Kates, Ausubel and Berberian1985:30). They are part of the worldwide effort to improve understanding of climate change impacts, adaptation, and vulnerability (McCarthy et al. Reference McCarthy, Canziani, Leary, Dokken and White2001). Sensitivity assessment can identify the long-term relationship between climate extremes and behavioral responses at various spatial scales and under different conditions that might affect sensitivity, such as differences in the supply and demand for resources. For example, a systematic investigation of the long-term relationship between drought and population movement can help reveal causality if it exists (Ingram Reference Ingram, Lucas and Eric2016, Reference Ingram, Sulas and Pikirayi2018). If farmers decide to migrate during every prolonged drought over 200 years, then causality is likely. In contrast, if some droughts are associated with out-migration and some are not, then the task is to determine why these differential impacts occurred. Documented differences in conditions that affect people’s vulnerability and resilience to a climate extreme may offer an explanation (Step 5). Differences in the characteristics (duration, magnitude, etc.) of specific climate extremes may also explain differences in human responses, including no response.
Identifying Climate Extremes and Their Characteristics
Climate extremes are defined by the Intergovernmental Panel on Climate Change (Reference Field, Barros, Stocker, Dahe, Dokken, Ebi and Mastrandrea2012:5) as the “occurrence of a value of weather or climate variable above (or below) a threshold value near the upper (or lower) ends of the range of observed values of the variable.” Various thresholds and methodologies are acceptable to use to identify extreme periods if the methods are described, consistent with analytical objectives, and replicable. Quantitative methods require that start and end dates be identified, and the selected threshold strongly affects research results. Guide 4 (Ingram Reference Ingram2025e) explains how to identify climate extremes and their characteristics in a climate dataset, and Guide 5 (Ingram Reference Ingram2025f) shows how these characteristics can be identified using an Excel spreadsheet.
Climate extremes can be described and differentiated based on their characteristics, such as duration, magnitude, intensity, and the extent of an extreme’s “surprise” or rarity (Table 1). Characterizing extreme periods provides a “quantitative solution to the need for identifying the ‘strongest,’ ‘greatest,’ or ‘most remarkable’ periods, which would otherwise be selected differently by different observers” (Biondi et al. Reference Franco, Kozubowski and Panorska2002:24). Differences in the characteristics of extreme events will have different effects on social and ecological systems. Consider the magnitude difference between a 10-year drought with only slightly below-average annual rainfall compared to a 10-year drought with rainfall values one standard deviation below the mean. The extent of surprise (Streets and Glantz Reference Streets and Glantz2000) or rarity (Nelson et al. Reference Nelson, Ingram, Dugmore, Streeter, Peeples, McGovern and Hegmon2016) can also be identified in a climate data series. Adaptive and buffering strategies most likely develop over time based on actual climate challenges curated in social memories. Climate extremes with characteristics without precedent in multigenerational memories would likely be beyond the repertoire of strategies that people develop in response to lived experiences (Minc Reference Minc1986).
Example of Identifying the Characteristics of Climate Extremes.
Notes: See Guide 4 (Ingram Reference Ingram2025e) for how climate extremes and their characteristics are defined and calculated. Guide 5 (Ingram Reference Ingram2025f) shows how these characteristics can be identified using a Microsoft Excel spreadsheet.
1 Lower “Rank” numbers (i.e., 1, 2) are more severe (drier) than higher “Rank” numbers (i.e., 9, 13). Some ranks are missing in the table because extreme periods before 900 AD are included in the ranking, but those periods are not included in the table for brevity.
2 The “Surprise” value is the number of years since an extreme period of equal or greater duration, magnitude, or intensity.
3 The climate data are from Dean and Robinson (Reference Dean and Robinson1978). The Chaco tree-ring reconstruction retrodicts precipitation from AD 661 to 1988. Here, only dry periods from AD 900 to 1154 are shown.
Periods of high and low temporal and spatial variability of a climate variable are also important to consider in climate and human behavior studies because these periods affect farmers’ planting and related decision-making and the risk of food shortfalls (see risk models in Step 3). Differences in the spatial variability and variation in climatic conditions across a region create differences in potential resource productivity, thereby affecting options for trade, exchange, and mobility.
Computational Models
Because of the introductory focus of this article and the broad audience intended, I have not emphasized quantitatively sophisticated mathematical models and simulation efforts. Yet, these studies contribute innovative new methods and valuable insights into climatic influences on human behavior and histories (e.g., Bocinsky and Varien Reference Bocinsky and Mark2017; Schwindt et al. Reference Schwindt, Bocinsky, Ortman, Glowacki, Varien and Kohler2016; Strawhacker et al. Reference Strawhacker, Snitker, Peeples, Kinzig, Kintigh, Bocinsky, Butterfield, Freeman, Oas and Nelson2020). They can also be difficult to understand, evaluate, and replicate because of the plethora of variables, procedures, and assumptions built into the models (for a notable exception, see Anderson et al. Reference Anderson, Stahle and Cleaveland1995). This work is usually produced by interdisciplinary teams with multiyear grant funding and is often limited to archaeologists practicing within specific academic contexts. Interested new entrants into the study of climate and human behavior should not be discouraged by computational models because such approaches are not necessary for generating insights.
Natural Experiments, Comparative Archaeology, and Cross-Cultural Research
Some broad methodological approaches for consideration include those developed for natural experiments (Diamond and Robinson Reference Diamond and Robinson2010), comparative archaeology (Smith Reference Smith2012), and cross-cultural research (Ember and Ember Reference Ember and Ember2009). Natural experiments rely on the occurrence of similar initial conditions, such as those created by national borders (e.g., El Paso, Texas, and Ciudad Juarez, Chihuahua) and islands (Haiti and the Dominican Republic) to investigate why different historical or behavioral outcomes occurred in response to similar initial conditions. For example, why were the human responses and impacts to the same climate extreme different among people in a similarly affected area? A long-term study of a single population can also be compared to itself when it is exposed to different climate conditions at different points in time (Hsiang et al. Reference Hsiang, Burke and Miguel2013). Comparative methods can be used to isolate the conditions that affect different outcomes, such as human adaptations to climate change (e.g., Orlove Reference Orlove2005). Cross-cultural research methods rely on user-selected samples from the Human Relations Area Files (eHRAF) and can be used to statistically evaluate the influence of independent cultural and environmental variables on potentially dependent aspects of behavior. Depending on the sampling method selected, results may be generalizable.
Documenting the Climatic Context
An alternative to investigating causality is documenting the climatic context for human decision-making and historical events (e.g., Anderson Reference Anderson2001). Detailed investigations into the possibility of causality can wait until you have additional evidence. For example, Stahle and Dean (Reference Stahle, Dean, Hughes, Swetnam and Henry2011) identify many tree-ring reconstructed climatic extremes and potential impacts on humans in North America and elsewhere over the past 1,000 or so years. These correlations, without demonstrated causality supported by evidence, should inspire many future studies (see also Step 1). They state, “Testing these hypotheses [of causation] with improved climate reconstructions, better archaeological data, and modelling experiments to explore the range of potential social responses have to be the central goals of archaeology and high-resolution paleoclimatology”(Stahle and Dean Reference Stahle, Dean, Hughes, Swetnam and Henry2011:323).
The climatic context of past human action is best understood using a combination of paleoclimatic and modern instrumental climate data. Paleoclimatic data provide varying spatial and temporal resolution of past climatic conditions. Modern instrumental data from weather stations provide daily to annual records of multiple climate variables and relatively fine spatial resolution. Maps using modern instrumental data that show the variation in a climate variable across a landscape are necessary to understand landscape to settlement-level conditions that would have offered people different opportunities and challenges (Figure 2; see also Ingram Reference Ingram2025b:Guide 1). Climate conditions in the past can be understood with modern climate data if the human behavior under study occurred where and when atmospheric and physiographic controls on climate remain the same as modern conditions. For example, the “Southwestern climate type has been stable at least since the end of the Altithermal [ca. 7000 to 4500 BP], when atmospheric circulation patterns and other factors that regulate modern climate were established” (Dean Reference Dean, Varien, Kohler and Wright2010:328).
An example of a PRISM Climate Group, Oregon State University (2008) spatial climate dataset of modeled precipitation (based on weather station data) and a settlement distribution (red dots) from the cyberSW database (Mills et al. Reference Mills, Ram, Clark, Ortman and Peeples2020) displayed by ArcMap10 to identify the climatic context of each settlement. The climate dataset is a “climate normal,” a 30-year average of a climate variable commonly used to identify typical or baseline conditions for an area. See Ingram (Reference Ingram2025b:Guide 1), for more information about PRISM and other climate data.
To conclude this section on methods, it is worth remembering that studies of past climate and human behavior interactions are subject to the same standards as any form of scientific knowledge creation. Interpretations are judged on their ability to account for the data. Those explanations that account for the greatest number of observations are most accepted, and multiple lines of supporting evidence are more potent than a single line of evidence.
Step 7: Present the Results to and beyond Archaeologists
Everyone who investigates and interprets the past can contribute to preparing for our warming world. In addition to reports and publications, we can share the climatic context of past human actions and histories through our public education efforts, from site tours and presentations to everyday conversations with others. Climate stories connect and remind all of us that the story of humanity includes our ever-present companion—climate. Preparing for a warming world involves, in part, a shared awareness of this relationship to climate in the past, present, and future. For public presentations and conversations, I recommend the National Park Service’s “Every Place Has a Climate Story” approach (Rockman and Maase Reference Rockman, Maase, Dawson, Nimura, Lopez-Romero and Daire2017). Climate stories must always reflect the best available evidence. Researchers can organize stories based on successful or unsuccessful adaptation, mitigation, resilience, vulnerability, or other themes.
Communicating the results of our climate–human behavior studies beyond archaeology is also essential for insights to emerge. Cochran and colleagues (Reference Cochran, Miller, Wholey, Gougeon, Gaillard, Jane Murray and Parker2023) argue that archaeologists often fail to define or contextualize terms for stakeholders and others. In the context of climate change and heritage at risk, the stakes are too high to communicate ineffectively. To communicate beyond archaeology, it is best to use the vernacular of the audience we are trying to reach. Many archaeological terms and concepts have equivalents in international, multidisciplinary, scholarly, and practitioner discourse. Archaeological and anthropological problems of risk and resource shortfalls are understood as problems of “food scarcity” and “food insecurity” by non-archaeologists, settlement abandonment or population movement is “displacement” or sometimes “forced” or “climate-related migration,” and floods and droughts are “climate extremes” or “disturbances” to a socioecological system that affect the “resilience” and “vulnerability” of a “socioecological system.”
Conclusion
This article has aimed to stimulate and inform climate and human behavior work to generate insights to help humanity adapt to a warming world. Investigating potential past climatic influences on human behavior should not be the exclusive domain of those with specialist training in climate science. However, understanding and using conceptual models and appropriate methods and data are necessary, and collaborations with climate proxy experts and interdisciplinary studies are always helpful. I encourage readers to explore the extensive Guides (Ingram Reference Ingram2025a) that accompany this article to find and interpret instrumental and tree-ring climate data, identify climate extremes and their characteristics in these data, see a detailed example that demonstrates how and why understanding climate–plant–human interactions may be necessary, and find options for visually characterizing the climatic context of past human behavior. Expanding our knowledge of how and why climate has and has not affected human history is essential for maintaining hope and progress toward meeting the challenges of anthropogenic warming.
Acknowledgments
The scholarship of Professor Jeffrey S. Dean of the University of Arizona’s Laboratory of Tree-Ring Research singularly inspired my interests and informed my understanding of the study of climate and human behavior. I thank the anonymous reviewers for their insights and the resources they have shared, as well as the students who read and commented on previous versions of this article.
Author Contributions
Scott E. Ingram, PhD (Conceptualization: Lead; Data curation: Lead; Formal analysis: Lead; Funding acquisition: Lead; Investigation: Lead; Methodology: Lead; Project administration: Lead; Resources: Lead; Software: Lead; Supervision: Lead; Validation: Lead; Visualization: Lead; Writing – original draft: Lead; Writing – review & editing: Lead).
Funding Statement
The research presented in this article was supported, in part, by a Jackson Fellowship provided by the Helen Jackson and William S. Jackson Family Endowment, from the Hulbert Center for Southwest Studies of Colorado College, Colorado Springs, Colorado.
Data Availability Statement
No original data were used. Supplements to this article are available through a collection at tDAR, the Digital Archaeological Record: https://core.tdar.org/collection/72003.
Competing Interests
The author declares none.
Open access
