8. Violence, Conflict, and Societal Disruption

We can’t build a peaceful world on empty stomachs and human misery.

—Norman Borlaug, father of the Green Revolution

The universality of terms such as “hotheaded,” “hot-tempered,” “hot under the collar,” and “hot-blooded” demonstrate the widespread perception that there is a link between temperature and violence. Cultural icons such as Rod Steiger in Sidney Poitier’s film In the Heat of the Night and the Tennessee Williams play Cat on a Hot Tin Roof provide additional support for the validity of the link that seems almost intuitive between heat and violence. The data that examine linkages between heat and violence reinforce that concept. Will the increase in temperatures predicted to occur as a result of climate change lead to more violence? Or will recent trends that show reductions in violence prevail? For example, the FBI reports that there was a 14.5 percent reduction in violent crimes between 2004 and 2013, during which time a total of 368 murders, non-negligent manslaughters, rapes, robberies, and aggravated assaults were reported for each one hundred thousand people in the country.

For the title of his book The Better Angels of Our Nature: Why Violence Has Declined, Steven Pinker drew from the last words of Abraham Lincoln’s first inaugural speech. The title frames his thesis that there is less violence now than in the past. Lincoln used these words to create an optimistic vision of the future for a nation facing civil war after the secession of the southern states; Pinker uses the same words to help convince us that changes in culture and society have made violence less prevalent.

There is still a great deal of violence among individuals and societies, a fact that Pinker would not deny. It is a rare local television news broadcast that fails to present lurid details of a recent murder. Fear of kidnapping keeps parents from permitting their children to walk to school. Every bank seems to have been robbed. Crimes of all sorts seem to abound. Terrorists recruit suicide bombers, including young women and girls, to strike fear into their target populations. Governments repress, vilify, and may even attempt to exterminate specific populations that differ in characteristics that seem to have little meaning. Although Pinker therefore was not without his critics, he researched and compiled an enormous amount of data to make and reinforce his argument.

Numerous factors lead individuals and societies toward violent actions. A body of evidence that is largely out of sight but growing in acceptance suggests that violence, weather, and climate are closely related. In perhaps the best known of these, investigators examined examples of conflict in which it was possible to infer a causal link between weather or climate and conflict.1 The authors found that increases in temperature systematically increased the risk of violence. The question is whether evolving cultures and societies of the future will be adequate to promote the trend that Pinker describes and that is reported by the FBI, or whether changes in climate will overwhelm and reverse this trend.

Individual Acts of Violence

Baseball is a sport characterized by an almost endless list of statistics, including the use of the beanball. A beanball is a pitch thrown with the intent to strike a batter on the bean (where bean is a slang term for head). The relationship between the beanball and temperature was evaluated in 1991 in a report based on an analysis of 826 randomly selected Major League Baseball games.2 At that time, the authors found a positive and significant relationship between temperature and the number of batters hit per game. As is often the case, these authors produced a much more comprehensive and detailed follow-up report in 2011, in which they analyzed data from an astounding 57,293 Major League Baseball games to determine whether there was any relationship between the temperature at game time and the probability that a batter would be beaned by the opposing pitcher after a batter on his team was hit by a pitch (deliberately or not).3 When this happens, baseball tradition and custom have it that “eye for an eye” form of justice must prevail. In this follow-up study, the researchers found that there was indeed a significant interaction between the temperature and what may have been a retaliatory beanball. When the temperature was 57°F and a player on the pitcher’s team was beaned in the first inning, the probability that the pitcher would hit a member of the opposing team was approximately 0.22. Under identical circumstances but at a temperature of 95°F, the probability rose by around 23 percent. The risk of getting hit by a pitch also rose steadily when one, two, or three batters had been hit by the opposing pitcher.

This is not the only example of a link between temperature, sports, and violence. Football (American football, not soccer) is a sport that is certifiably more violent than baseball. In an analysis of weather reports, data extracted from over 750 police agencies in the National Incident-Based Reporting System and information gathered from twelve years of National Football League Sunday games, an upset (or an unexpected loss by the home team) was associated with a 10 percent increase in the number of police reports of male-on-female “intimate partner” violence.4 Heat appeared to be a contributing factor. They defined an unexpected loss as a loss by the home team when it was expected to win by four or more points, according to the pregame point-spread prediction. Wins and expected losses were not associated with additional reports of violence. Additional analyses of the data showed that intimate partner violence was 8 percent higher when the temperature exceeded 80°F.

To illustrate the complexity of the issue, researchers also found that holidays exerted large additional effects on intimate partner violence. Although alcohol and drug abuse are widely accepted as contributing factors in interpersonal violence, there was no conclusive evidence that this played a role in post-football-upset violence. Statisticians frequently create mathematical models of the phenomena that interest them; this group was no exception, and mathematical formulas filled with Greek letters are found throughout their report.

Who hasn’t been trapped in traffic? The impact of heat on aggressive behavior by motorists was studied by a group of investigators in Phoenix, Arizona, where temperatures frequently exceed 110°F in the spring, summer, and fall. In this fertile ground for heat-related research, a team deliberately blocked traffic for up to twelve seconds and observed the behavior of the delayed motorists.5 These intrepid researchers found a linear increase in horn honking as the temperature increased. This effect was accentuated when just those cars with the window rolled down were studied. They hypothesized that these drivers lacked air conditioning and were presumed to be hotter than their fellow motorists who had rolled-up windows. Luckily, that was the extent of the violence they reported. There are places where one might literally risk his or her life by conducting this experiment—or worse, precipitate an episode of road rage directed at the motorists up ahead.

A final and classical example of the link between violence and temperature at an individual level comes from data collected during firearms training exercises for police officers in the Netherlands.6 This study involved thirty-eight police officers with an average of five years of on-the-job experience. Each officer carried a laser pistol in his or her holster. During the exercises, participating officers were confronted by life-sized television images that showed a suspect holding a crowbar in a threatening position. The video was constructed so that the officer could take evasive action to avoid an attack if he or she chose to do so. The officer’s responses were videotaped and scored after the exercise had ended. The temperature on the training site, the independent variable in experiment, was under the control of the investigators. It was set randomly to either 21°C or 27°C. Trainees completed questionnaires that described their reactions to the videos after the exercise. Four outcomes were significantly more probable at the higher temperature: attribution of a negative affect to the subject, the impression that the subject was aggressive, the impression that the subject was threatening, and the tendency to shoot. The officers kept their weapons in their holsters more often in the cooler than in the hotter condition, 15 percent versus 41 percent. Although the officers “shot” the suspect more often in the warmer condition, 62 percent versus 45 percent of the time, this difference did not reach statistical significance.

The investigators made the point that, according to Dutch law, shooting the suspect was not an appropriate action under the circumstances, since the scenario was constructed in a manner that allowed the officer to avoid harm. The report did not describe any ethnic differences among the trainees or the subjects portrayed in the videos, but these are subjects for research with a different but important focus. Based on the number of high-profile shootings by US police officers, it seems quite possible that in a real-life situation, ethnic stereotyping might well magnify any heat-related effect and increase the probability of a perceived threat.

Investigations of the relationships between temperature on one hand and aggressive behavior and violence on the other are often based on so-called field studies—that is, an analysis of actual behavior under real-life conditions. This places constraints on the ability of investigators to control other relevant variables, to quantify other relevant outcomes, or both. Studies performed under laboratory conditions seek to minimize these confounding problems. This is one of the great reasons to conduct laboratory versus field studies.

One study conducted in a laboratory environment involved video games and exercise.7 The game was similar to Pac-Man, a game in which an operator-controlled token “eats” an object. The level of frustration while playing the game was altered by having some participants use the joystick in the normal position (low frustration) while others played with an inverted joystick (high frustration). Temperatures were set at one of three levels at constant humidity, creating a two-by-three matrix (two frustration levels by three temperatures). In a second, virtually identical experiment, participants stepped up and down on a stool for a minute as a part of the study. The participants completed questionnaires, and physiological indicators (blood pressure and heart rate) were measured at three different times (before exercise and at two points after completion of the exercise). The results of both of these experiments produced increases in what the investigators termed hostile affect, hostile cognition, and physiological arousal in hot temperatures compared to temperatures that were cooler. They concluded that heat increases the likelihood of bias in a hostile direction during social events that are not typically viewed as hostile or nonhostile.

Craig Anderson, director of the Center for the Study of Violence at Iowa State University, has been one of the most influential and prolific behavioral scientists investigating the relationships between heat and violent behavior. In a 2000 review, his group presented the results of an analysis of aggressive behavior in seven different locations in North America.8 The group members divided their data into segments that were each one month long. The annual low point for violence occurred in January, when just over 6.5 percent of all assaults took place. The high point came in August, when almost 10 percent of the assaults they analyzed took place. They performed complex analyses of violent crimes in the United States that occurred in 1980. In addition to the crimes themselves and temperature data, they created a factor they called southernness. This factor was in part a reflection of what they referred to as a culture of honor but was based on more easily quantified elements, such as migration patterns from the South to the North, US Census Bureau data, and the fraction of the 1968 presidential vote that went to George Wallace, an outspoken racist. They also included a measure of social and economic status (SES). Among the three variables, temperature, southernness, and SES, temperature was the one that had the most significant relationship to crime. This was followed by low SES. Southernness did not have a significant relationship to crime, and the researchers concluded that it should be dismissed from consideration. Restated, the significant relationship between temperature and violent crime was stronger than low SES, which was also significant.

They also evaluated data on rape and domestic violence from Minneapolis, Minnesota. The relative rate for both of these forms of interpersonal violence rose smoothly from their lowest points when the temperature was just below 0°F in January to their highest values when the temperature was 95°F in August.

In a 2001 review of the heat hypothesis, which seeks to answer the question “Does excessive heat increase violence?,” Anderson makes the following statement: “My colleagues and I believe that most heat-induced increases in aggression, including the most violent behaviors, result from distortion of the social interaction process in a hostile direction.”9 They proposed a four-stage model of behavior to explain aggression and how it is affected by heat.10 The first stage consists of personal variables, such as aggressive personality, and situational variables, such as uncomfortable heat. These elements feed into the second stage of their model, which they refer to as the present internal state. This includes cognitive elements, affective states (such as hostility), and arousal, as measured by heart rate and other physiological variables. These three elements trigger the third stage, which Anderson refers to as appraisal processes, or automatic and controlled actions. These appear to be similar to what Kahneman refers to in his book Thinking Fast and Slow as system 1, which is fast, emotional, and intuitive, and system 2, which is more deliberative, slower, and more logical.11 The final stage of the Anderson model is the outcome. This is where behaviors may trigger a response, such as a violent act. In this comprehensive review and presentation of original work, the group marshals evidence from multiple sources to support the hypothesis that heat breeds violence.

In their work, Anderson and colleagues address climate change directly.12 They refer to studies that showed a significant relationship between the hotness of the year and the murder and/or assault rate. When they quantified these relationships and subjected them to a statistical analysis, they found that there were 4.58 additional murder/assault crimes per one hundred thousand people in the United States for an increase in the temperature of one degree Fahrenheit. Using that relationship between temperature and violent crimes and a population of 270 million, a temperature increase of two degrees Fahrenheit led the researchers to predict that there will be twenty-four thousand additional murders or assaults per year in that hotter future. An eight degree increase would produce about one hundred thousand additional crimes.

Anderson injects a note of caution by reminding us that this prediction assumes that social and other systems, such as those that supply food and water, adequate policing, a fair and just government, and other elements that we take for granted in modern society, will remain intact and that other elements of a future climate, such as drought, severe storms, or so on, that place stress on governments and social systems are not present.

Intergroup Violence

A great deal of the research on violence in societies has been made possible by a database maintained by two Swedish groups: the Peace Research Institute Oslo (PRIO) and the Uppsala Conflict Data Program (UCDP). These data are broken down into several categories: events not intended to damage property or cause injury are classified as nonviolent events, and violent events implies that the event in question was designed to cause injury or property damage. In addition, a threshold of twenty-five or more deaths is commonly applied to define a significant event. Significant violent events include riots—such as antigovernment riots—or repression of populations by a government. Violent events are subdivided into government-targeted and non-government-targeted events, depending on whether a central government was the objective of the violence. Some food riots fall into the government-targeted category and others into the non-government-targeted category, as discussed in chapter 5. Finally, the two Swedish groups add all violent events together to yield a total for violent events. The UCDP maintains a website that provides interactive maps that allow an interested observer to view the distribution of violence on a country-by-country basis. A quick look at these maps shows that in the 1975–2013 time interval, virtually every nation was involved in violence that was included in the UCDP data set. These data have been used in conjunction with weather and climate data to examine the relationships between weather, temperature, and violence.

Virtually every climate change study predicts that in the future there will be large changes in precipitation in many parts of the world, after which enormous consequences could follow. Most stable societies are adapted to the amount of rain that falls in typical years. Increased amounts of rain can cause flooding that delays planting or harvests, damages crops in the fields, affects livestock, or causes damage to property that is often associated with morbidity or mortality. Droughts also have the potential to disrupt societies in ways that affect the behavior of individuals and groups by causing loss of income, food shortages, price increases, and so on. This relationship has been referred to in the literature as environmental security.13 Investigators have identified at least five mechanisms whereby unexpected departures from the average amount of rainfall may give rise to social conflict:

• Consumers of water may be affected directly. These include individuals or industries that unexpectedly must compete for a scarce resource or face threats posed by floods.

• Price disputes between agricultural interests and consumers may follow droughts or floods. Either droughts or floods are likely to increase the cost of food.

• When areas become uninhabitable, residents must move, usually into a city. Competition for jobs, housing, and services, such as those provided by law enforcement and utilities, may become sources of tension that can trigger conflict. Some refugees are likely to cross national boundaries.

• Confronted by internal problems related to the supply of commodities, governments may intervene in markets in one or more ways that create tension. Interventions are often related to exploitation of markets to increase profits or political advantages. This is more likely to occur in countries run by corrupt officials or by a single strongman.

• Macroeconomic effects in a nation or region may take a human or financial toll.

Sadly, Africa has proven to be a laboratory for investigating violence and its relationship to a variety of stressors, including climate. Using the UCDP and PRIO data, investigators have conducted a detailed analysis of over six thousand social conflicts in Africa that took place over a twenty-year period.14 These investigators combined these data with rainfall data for each country from the Global Precipitation Climatology Project. Their analysis showed that there was a significant positive association between all forms of political conflict with deviations from the average annual rainfall. It did not matter whether there was too much or too little rain. However, civil wars and other violent conflicts were more common during the wet extremes than during the dry extremes.

Because rainfall in Africa ranges wildly between Egypt, where 3.1 cm of water falls in a typical year, to Sierra Leone, where the annual expectation is 233.3 cm, the investigators normalized rainfall for each country. This process results in a similar standard value for each country, as well as similar values for deviations from this average. They use the term rainfall deviation to describe these normalized data. This procedure makes it possible to include many nations with widely divergent average rainfall values in a single analysis. After the normalization was complete, they combined data from multiple countries and expressed the percentage by which conflicts changed as a function of departures from this normalized amount of rain. The results of this quite complex analysis are shown in figure 8.1. The strongest association was found for the total number of events, as defined previously, and rainfall deviations. For example, an increase of just over 33 percent (one standard deviation) in the normalized rainfall was associated with a 6.1 percent increase in social conflicts, whereas the decrease of a similar amount was associated with a 9.2 percent increase. More extreme deviations in rainfall were, by definition, less common, but were associated with larger increases in the number of conflicts. When rainfall increased by just over 47 percent and almost 50 percent, conflicts increased by 38.1 percent and 103.5 percent, respectively. Decreases of similar amounts were also associated with increases of 30.7 percent and 88.2 percent, respectively. In other words, they found a U-shaped curve: A departure from expected amounts of rainfall, whether it was an increase or decrease, was associated with an increase in the probability of violence.

Some of these fluctuations in rainfall and attendant violence are related to predictable fluctuations and prolonged warming of the sea surface along the western coast of South America, known as El Niño, and the atmospheric pressure component associated with the temperature changes, known as the Southern Oscillation.15 Collectively, these two phenomena are referred to as ENSO (see also chapter 4 for ENSO effects on malaria). Variations in ENSO are associated with climate fluctuations in many tropical regions worldwide and along the northwestern coast of North America even though they are at some great distance from the Pacific Ocean where ENSO arises. A relationship between distant phenomena is referred to as a teleconnection, a term used by atmospheric scientists to describe features of the climate that are related to each other over large distances, often thousands of miles.

Researchers have correlated fluctuations in ENSO with the probability of a civil conflict in these regions that are teleconnected to ENSO.16 They found that the probability for a civil conflict doubles during intervals when the influence of El Niño is strong compared to intervals when it is weak (referred to as La Niña). They conclude that from a statistical perspective, changes in the ENSO pattern were linked to 21 percent of conflicts that developed between 1950 and 2004. Although additional studies should be performed that test the hypothesis that ENSO and conflict are linked, this study at least raises the possibility that the researchers’ conclusion is accurate. The potential importance of seemingly disparate linkages was addressed in a discussion of vulnerable populations in a landmark paper published in 2015 in The Lancet, in which the authors emphasized the “interconnected nature of climate systems, ecosystems, and global society.”17

Increases or decreases in rainfall commonly lead to reductions in the harvest of agricultural commodities and increases in the cost of food. As a corollary to the study of violent and nonviolent social conflict, another report focused on food riots and food prices in Africa and the Middle East.18 The results of this study are shown in figure 8.2, which shows clustering of food riots around peaks in the Food Price Index as reported by the Food and Agriculture Organization.

The World Bank Group monitors global agricultural production and prices, reporting these results periodically in its newsletter, Food Price Watch. The May 2014 issue addresses the nexus between food supplies and violence by addressing food riots. They note that others have used the term food riot imprecisely and with no real consensus as to its definition. Is loss of life necessary? How long must a disturbance last? What about peaceful demonstrations? The World Bank begins with a proposed set of defining criteria: “A violent, collective unrest leading to a loss of control, bodily harm, or damage to property, essentially motivated by a lack of food availability, accessibility or affordability, as reported by the international and local media, and which may include other underlying causes of discontent.”19 The organization would exclude most of the 2011 Arab Spring uprisings, because food was not a major aspect of these events, but it would include prerevolution events in Algeria and Tunisia, because inflated food prices were primary precipitating factors. The World Bank subdivides food riots into two types: Type 1 riots are the most frequent and are directed at governmental authorities; Type 2 riots are devoid of political aspects, are not directed at governments, and typically target food repositories such as warehouses, shops, and trucks. The authors of the Food Price Watch report note that food prices are frequently not the exclusive cause of violent unrest. Predisposing conditions also are common precursors to riots. These conditions include poverty, weak or inadequate governments, and an absence of planning for disaster. There does not appear to be a simple answer or solution to what may become an increasingly common and difficult problem as population growth, rising temperatures and increases or decreases in rainfall alter the socioeconomic landscape of the future.

Societal Collapse

The studies summarized in the first part of this chapter addressed the short-term effects of weather and violence, such as horn honking during hot spells in Phoenix, controlled environments during police weapons training, and laboratory environments in which the temperature was manipulated during the execution of tasks. Then, I considered events that last for a longer period, such as the relationship between El Niño and the Southern Oscillation, and violence in Africa and across the globe, food riots, and other forms of civil unrest. When considering the total collapse of a civilization, which may take a substantial period—decades or even centuries—considerations of weather (short term) turn to considerations of climate (long term).

In his book Collapse: How Societies Choose to Fail or Succeed, Pulitzer Prize–winning author Jared Diamond develops a thesis to explain the end of past civilizations.20 He postulates five elements that operate to varying degrees:

• Environmental damage caused by inhabitants. This may take the form of stripping an island of its trees, or agricultural practices that cause a loss of topsoil that leads to crop failures.

• Hostility of neighbors, often preceded by weakening of the affected society. Here, Diamond cites the fall of Rome to barbarian invaders. Roman society was weakened and eventually fell after a series of unsuccessful invasions by hostile neighbors.

• Related to the hostile neighbor scenario, a reduction in support from a neighbor.

• Inadequate responses to societal problems. Here, Diamond cites the deforestation of Greenland by the Norse settlers and a similar loss of trees on Easter Island as examples of how environmental damage combined with inadequate societal responses led to collapse. By contrast, the Japanese and other island civilizations developed programs to buttress their societies, making them stronger.

• Climate change, the subject of this book.

These are undoubtedly not independent variables. For example, a prolonged drought may weaken a society, which makes inadequate responses to problems, which eventually leads to a takeover by another. Climate change may combine with self-inflicted environmental damage, which weakens a society, which makes poor decisions concerning resource allocation. These decisions, in turn, lead to conditions under which continued success of the society is impossible. Based on what we know about the relationship between stress and violence, it seems likely that interpersonal, intrasocietal, and intersocietal violence increase as a society collapses.

This appears to be the rationale exercised by Solomon Hsiang and his colleagues in their quantitative analysis of climate and conflict that spanned twelve thousand years, beginning ten thousand years BCE and continuing to the present.21 They used a broad definition of conflict that ranged from examples of short-term violence between individuals, such as the police-training episode reviewed earlier, to political instability and civil war. Their comprehensive study drew on data collected by multiple disciplines in the social sciences, including psychology, criminology, economics, archeology, geography, history, and political science. They evaluated sixty studies of forty-five sets of data about conflicts published in twenty-six journals by over 190 researchers. These scholars had well-defined inclusion and exclusion criteria in their study. Only those publications and data sets that made it possible to make quantitative assessments were used. The researchers ultimately concluded that climate had a major influence on conflict between individuals and societies as a whole. When they focused on studies published after 1950, they found that “climate’s influence on modern conflict [was] substantial and highly significant.” A temperature increase of about 34 percent from the mean of the time (one standard deviation) was associated with a 4 percent increase in interpersonal violence and a 14 percent increase in intergroup conflict.

To illustrate the importance of the work done by multidisciplinary researchers in the examination of the fate of early American civilizations, I turn to expeditions sponsored by the National Geographic Society, as reported in 1929 in their journal, National Geographic.22 Andrew Douglass, the leader of the National Geographic expeditions, began his career as an astronomer who specialized in sunspots. Based on earlier reports, he thought that his astronomical expertise would allow him to link the study of growth rings to variations in sunlight caused by fluctuations in the number of sunspots. This hypothesis was ultimately shown to be untrue, but his observations of tree rings were of seminal importance.

Douglass is now revered as the father of a scientific discipline known as dendrochronology. His efforts have made it possible to use tree ring data to establish accurate links between rainfall and seminal events in various civilizations, such as those that once flourished in the Southwestern United States, the Angkor area of Southeast Asia in what is now Cambodia, and elsewhere. He recognized that trees are not “mechanical robots,”23 but are living things, and a tree’s “food supply and adventures through life all enter into its diary.” He wrote, “Trees are nature’s rain gauges.”24 He describes finding a log near Show Low, Arizona, with the field designation HH-39 that, in his words, proved to be the Rosetta stone for tree-ring research. (Note: The name Show Low is believed to be the result of a lengthy card game in which one player bet his 100,000-acre ranch on the turn of a card. His opponent showed the lowest card and won.) The HH-39 log unlocked the secrets of tree rings and rainfall. Using tree ring spacing that can be accurately linked to rainfall, it can be shown that widely spaced rings indicate large amounts of rain and narrowly spaced rings indicate drought. This turns out to be a much more precise dating method than carbon-14 (¹⁴C) dating and other commonly used techniques. Using the science of dendrochronology, it is possible to create timelines that link salient historical information about a society with the exact year of occurrence. Douglass’s examples linked specific events in the Southwest to the time of the Sixth Crusade, the reign of Charles Martel, and other events that took place in other parts of the world. Later in this chapter, I will discuss the use of tree ring data to chronicle the decline of the Angkor civilization.

Diamond also looks at the American Southwest. He tells the story of the Anasazi, who came to live in what is now called the Chaco Canyon, located in the northwestern portion of New Mexico.25 This society had its heyday between 600 and 1200 CE. From a series of seemingly disparate observations that are based on studies of packrat feces, isotopes of strontium in logs, and other trace evidence of the society, scientists found that the Chaco Canyon site was the focal point of a society that spread outwards. As the community grew, local resources were exhausted. More trees were cut than could be replaced by normal growth. In this time of plenty, the region became unable to produce enough food locally. The inhabitants therefore had to import food from adjacent areas where water and the soil were able to support the growth of corn, their principal crop. Trees were cut in distant forests and used as building material. The beginning of the end came with the onset of a severe drought that led to the collapse of agriculture. Starvation conditions ensued, and people began to eat mice rather than deer, which had vanished from the area. Within-group strife and warfare, including evidence of cannibalism, marked the final days of this once-prosperous society.

Other American civilizations have been studied using a variety of other archeological tools. With a combination of validated strategies, anthropologists and geologists have studied human populations in what is now known as the Bighorn Basin in Wyoming.26 These scientists combined ¹⁴C radiocarbon dating, an analysis of fossilized pollen samples (used to form another critical database for the study of ancient civilizations), and the analysis of stable oxygen isotopes in corals (similar to the oxygen isotopic studies discussed in chapter 2) to study the remnants of the societies in that area. They found evidence for five episodes in which there were major increases or decreases in the population of the area over a time span of thirteen thousand years. Using artifacts left behind by these inhabitants and temperature and rainfall data, they were able to link the rise and fall of this population to climate. They reported that 45 percent of the variation in the number of people living in the area could be explained by reconstructed concordant temperature and moisture records. They found that on average there was a three-hundred-year interval between a climatological change and a change in the size of the population. Other studies show that this is the predicted length of time between a major change in the climate and an effect on the population of the area. Whatever the explanation, these investigators report clear evidence that the population adjusted to environmental conditions on a continuing basis.

On the other side of the world, in Angkor—the capitol city of the Khmer Empire in what is now Cambodia—another group of scientists used climatological data to time the rise and fall of this fourteenth- and fifteenth-century civilization.27 Their study relied heavily on tree ring data. From 7.5 centuries of evidence, they were able to establish the dates for a series of droughts and monsoons. They found that droughts lasted decades and were interspersed with periods of much higher than usual rainfall during intense monsoons. Droughts damaged agricultural production and had adverse effects on intricate systems of canals and components of the Angkorian water-management infrastructure. Floods from intense monsoons also damaged the water-handling systems by inundating them with sediment. The scientists linked these environmental disasters to changes in the surface temperature of the Pacific, concluding that a warm ocean and El Niño events led to drought on the time scale that coincided with the downfall of the Angkorian civilization.

The Trajectory toward the Future

The four climate-modeling scenarios used by the IPCC predict different degrees of warming by the end of the century.28 The amount of warming, of course, depends on the efforts to mitigate climate change—that is, to reduce greenhouse gas emissions, particularly in nations that emit huge amounts of these gases. The representative concentration pathways used to describe the climate of the future (see chapter 2) include considerations of how land is used, whether or how much populations grow, the characteristics of energy use and production, and social and economic factors in addition to greenhouse gas emissions.29 Based on the current responses (or nonresponses) to the threats posed by climate change, it seems virtually certain that significant worldwide warming will occur and that there will be substantial changes in rainfall across the globe. Some regions will experience large increases in precipitation, whereas others will be gripped by megadroughts. Neither of these precipitation alternatives holds the promise of a better world.

Most researchers who have investigated the relationships between climate and weather on one hand and violence, conflict, and social disruption on the other have looked to the past. It is difficult enough to develop, validate, and apply the varied tools of history, anthropology, economics, geology, sociology, psychology, physiology, and others to the study of what has happened; it is even more difficult to predict what will happen in the future.

Hsiang and his colleagues are exceptions to this rule.30 Operating from the premise that past behavior is the best predictor of future behavior, they have risked a look into the future. Recall that they found that temperature increases were associated with increases in interpersonal violence and intergroup conflict. Using an entirely plausible model of the future climate, they found that huge portions of the inhabited world could expect mean temperatures to increase significantly. Some of these regions are clustered about the tropics and subtropics, where the El Niño and Southern Oscillation climate–conflict linkages are the strongest.31 These are also regions where childhood malnutrition is the highest and tropical diseases are rampant. In other words, these are the parts of the world where the five elements that Diamond believes lead to societal collapse are the strongest. These elements may become the factors that determine the fates of these societies.