The history of climate and society: a review of the influence of climate change on the human past

Recent decades have seen the rapid expansion of scholarship that identifies societal responses to past climatic fluctuations. This fast-changing scholarship, which was recently synthesized as the History of Climate and Society (HCS), is today undertaken primary by archaeologists, economists, geneticists, geographers, historians and paleoclimatologists. This review is the first to consider how scholars in all of these disciplines approach HCS studies. It begins by explaining how climatic changes and anomalies are reconstructed by paleoclimatologists and historical climatologists. It then provides a broad overview of major changes and anomalies over the 300,000-year history of Homo sapiens, explaining both the causes and environmental consequences of these fluctuations. Next, it introduces the sources, methods, and models employed by scholars in major HCS disciplines. It continues by describing the debates, themes, and findings of HCS scholarship in its major disciplines, and then outlines the potential of transdisciplinary, ‘consilient’ approaches to the field. It concludes by explaining how HCS studies can inform policy and activism that confronts anthropogenic global warming.


Introduction
The idea that changes in climate influenced human history can be traced back to antiquity (Glacken 1973). Few sought physical evidence for these changes until the seventeenth century, when some scholars began to interpret marine fossils and erratic boulders as relics of ancient, radically different climates, while others started to identify fluctuations in the output of the Sun and other stars (Tassoul andTassoul 2004, Tierney et al 2020a). Many European intellectuals argued that climate-a condition they defined as synonymous with latitude or local air qualitydetermined human physical, mental, and moral characteristics (Livingstone 1991, Fleming 1998. It was an assumption they used to justify colonial conquestand it implied that a gradual change in climate could alter the fortunes of civilizations (Zilberstein 2016, Warde 2018). In the nineteenth century, scientists in newly professionalized disciplines began to speculate that these changes could unfold on human timescales. Their motivations included a growing awareness of solar variability; new efforts to monitor and exploit colonized landscapes; and, in the Austro-Hungarian Empire, a unique quest to legitimize empire by uncovering how atmospheric circulation connected otherwise isolated populations. (Coen 2018, Heymann and Achermann 2018, , Morgan 2018. In the closing decades of the century, astronomer A. E. Douglass realized that variations in the width of growth rings in trees could serve as proxies for seasonal precipitation (1914). By then, some economists and geographers had started to look for statistical relationships between trends in grain prices, weather, and hypothesized causes for weather (such as sunspots) (Jevons 1875, Brückner 1895. With their controversial studies as inspiration, geographer Ellsworth Huntington used the assumption that an ideal climate existed for civilization to argue that century-scale changes in climate, identified by Douglass in tree rings, had enabled the rise and triggered the downfall of past societies (Huntington 1913a(Huntington , 1917a(Huntington , 1917b. Meanwhile, oceanographer Otto Pettersson concluded that climatic changes-determined, he thought, by the position of the Moon and Sun relative to Earth-had shaped the history of Norse settlements in fifteenthcentury Greenland and Iceland (Pettersson 1914, Utterström 1955. These claims eventually spurred the development in distinct regional research cultures of new, radically interdisciplinary fields of study that gradually grew in sophistication and were recently synthesized as the History of Climate and Society, or HCS (McCormick 2019, Storozum et al 2019, Degroot et al 2021).
HCS may be defined as the study of human responses to trends, anomalies, and regular variability in past climate. It is a fast-growing field that is today undertaken primarily by archaeologists, economists, geneticists, geographers, historians and paleoclimatologists (Degroot et al 2021). Here, we provide the first survey of the field as it is practiced in all of these disciplines, and as it considers the entire, approximately 300 000 year history of anatomically modern humans. Since HCS depends on accurately identifying past climatic changes, we begin by explaining how these changes are identified, or 'reconstructed' , by paleoclimatologists and historical climatologists. We provide a broad overview of what reconstructions reveal about how climate changed over the past 300 000 years. We then introduce the distinct evidence, methods, and models used by scholars in major HCS disciplines. We identify key findings and themes in each discipline, by selecting what we consider particularly influential or representative case studies. We describe the potential of transdisciplinary, 'consilient' approaches to HCS scholarship (Lang et al 2012), and conclude by explaining how HCS scholarship may inform efforts to cope with today's global warming.

Reconstructing climate change
For the period before the late nineteenth century, when colonial states established weather stations with reliable meteorological instruments across their growing territories, our knowledge of past climate comes primarily from proxy data in the archives of nature. Examples of proxy records include the annual rings of trees, the geochemistry of cave formations and coral skeletons, the composition and contents of polar and high-altitude glaciers and ice sheets, and the organic and inorganic components of lake and ocean sediments (figure 1) (Baumgartner et al 1989, Jones et al 2009, Bradley 2014, Kaufman et al 2020b.
Proxy records from these archives are the result of biological, chemical, and physical processes by which organisms or systems sense the variability of their environment and record it in a material archive whose characteristics we can observe and measure (Evans et al 2013). Paleoclimatologists typically reconstruct past climate from these data by inverting this chain of processes, either quantitatively or qualitatively. To that end, they use measurements of the proxy system and an understanding of how it reflects its past environment to estimate the conditions under which it formed. This inverse process of paleoclimate reconstruction is often imperfect, however, necessarily making inference from our modern observations of the proxy toward the past environment without complete knowledge or representation of how the archive was formed and how accurately or uniquely the sensor recorded the climate signal of interest (Hughes and Ammann 2009, Inkpen and Wilson 2009, Tingley et al 2012, Evans et al 2013.
Past climate reconstructions may be relative and qualitative-classifying periods as wetter or drier, for example, and warmer or colder-based on a mechanistic or physical understanding of the proxy. Alternatively, reconstructions can be absolute and quantitative, with the proxy data measurements transformed statistically into climate metrics like temperature, precipitation, soil moisture, salinity, or features of the large-scale ocean or atmosphere circulation (Hughes and Ammann 2009, Tingley et al 2012, Hu et al 2017. Where proxy records overlap directly with instrumental observations, statistical estimates of past climate can be directly calibrated in time and validated on a portion of withheld instrumental data. In those cases where proxy records are either too low resolution (each measurement reflecting decades or centuries) or where the chronology is too uncertain (the time of formation constrained with decades or longer), statistical transfer functions using a spacefor-time calibration can be used to estimate climate variables from the proxy measurements (Birks et al 2010, Juggins and Birks 2012). Chronology is critical, as it provides evidence for the sequence of events and therefore is a prerequisite (although not sufficient) for determining both physical or social causality.
Every paleoclimate reconstruction comes with uncertainties caused by the imperfect reflection of climate variability by the proxy system; the assumptions and methods used to estimate past climate from natural archives; and the limits of chronological precision. Each type of proxy record has distinct advantages and limitations. For instance, tree-ring records are well-replicated and provide exact annual chronology and so can be calibrated directly against the overlapping instrumental record; however, they are largely limited in their length to the last millennium or less and most of these records Figure 1. Spatial distribution of paleoclimate and paleoenvironmental data for major proxy data types archived at the NOAA World Data Center for Paleoclimatology (data served from www.ncei.noaa.gov/products/paleoclimatology on 1 March 2022). As discussed in detail in the text, each proxy type has advantages and disadvantages for reconstructing past climate with respect to human history. In particular, low temporal resolution and chronological uncertainties may limit a proxy's applicability when assessing the link between specific events in human history and climate extremes, even though they can provide information on climate change and variability on centennial or longer timescales relevant to longer-term cultural change and human evolution. are from Northern Hemisphere terrestrial regions (George 2014, Anchukaitis and Smerdon 2022). Marine sediment records provide much longer records of past climate, potentially many millions of years, but are usually lower resolution (centuries to millennia), time-uncertain (by hundreds of years), and biased toward coastal marine environments.
Climate changes have provoked responses in human material culture that provide additional proxies for climate reconstruction, in what historical climatologists call the 'archives of societies.' These archives include either observations of past weather, usually as recorded in surviving textual evidence, or evidence of activities that must have been profoundly influenced by weather, such as autumn harvests or the use of canals in winter. Sometimes they are both: logbooks written to aid navigation in the age of sail, for example, include both wind or sea ice measurements, and descriptions of marine activities that reacted to the velocity and direction of wind or the thickness and distribution of sea ice (García-Herrera et al 2005, García-Herrera et al 2018, Degroot 2020, Degroot and Ottens 2021. It can be easy to confuse one kind of source with another; artistic depictions of extreme weather could be weather observations, for example, but more commonly were created through actions only indirectly influenced by weather (Brönnimann et al 2018.
The archives of society should also be considered with appropriate skepticism. Textual evidence for past weather should never be taken at face value, since observers usually had many motivations for recording weather and often limited means of measuring weather. Such evidence can be discontinuous unless archived by institutions, and typically focuses on weather extremes rather than averages. Records of activities influenced by weather should similarly be used with caution, since those actions were inevitably affected by many other forces. Historical climatologists accordingly use index systems to quantify and standardize their qualitative information on simple ordinal scales that can sometimes be calibrated to modern instrumental measurements. Reconstructions created with index systems can reveal climatic trends with uniquely high resolution, for places, times, or seasons that may be difficult or impossible to cover using the archives of nature (Brázdil et al 2005, Blöschl et al 2020, Nash et al 2021.

Climatic changes and causes in Homo sapien history
On multi-millennial timescales, climate change over the 300 000 year history of anatomically modern humans was paced by the changes in the Earth's orbit, which control how incoming solar energy (insolation) is distributed latitudinally across the planet and over the seasonal cycle. These orbital cycles include eccentricity (how elliptical the Earth's orbit is, a 100 000 year cycle), obliquity (changes in the tilt of the planet, a 41 000 year cycle) and precession (the timing of perihelion in the seasonal cycle, a 23 000 year cycle). By changing the seasonal amount of insolation in the high latitudes, these cycles, combined with ice-albedo feedbacks and changes in greenhouse gasses (Ruddiman 2006) drove three rounds of expansion and melting of the high latitude ice sheets in the last 300 000 years (figure 2). Feedback is an essential concept in climate scholarship; it refers to the outcome of a change that either amplifies (a positive feedback) or mutes (a negative feedback) the initial change. A positive feedback magnifies changes in the Earth system such that a small initial change results in an outsized outcome.
During glacial periods, global temperature was perhaps 6 • C colder than preindustrial conditions (Snyder 2016) (figure 2). Large ice sheets covered northern North America and Eurasia, leading to a drop in global sea level of around 120 m (Waelbroeck et al 2002). This sea level change exposed the Bering Strait land bridge, connected the British Isles and southern Scandinavian islands via a dry North Sea region known as Doggerland, and revealed shallow shelves in the tropics. In the Indonesian region, the emergence of the Sunda and Sahul shelves shifted the patterns of tropical rainfall, resulting in extreme drying (Windler et al 2019). Overall, global climate was drier (Bartlein et al 2011), since a cold atmosphere holds less water. However, there were regional exceptions. The southwest region of North America was much wetter, for example, as the Laurentide ice sheet deflected winter storms farther south (Oster et al 2015). Likewise, not all places on Earth were colder during glacial periods. Beringia (present-day Alaska and northeastern Siberia) experienced minimal cooling (Bartlein et al 2011, Tierney et al 2020b and remained unglaciated (Glushkova 2001) because the ice sheet caused a change in atmospheric circulation that made the region warmer and drier (Löfverström and Liakka 2016).
On millennial timescales, the Northern Hemisphere experienced a number of rapid fluctuations in temperature during the last glacial period, known as Heinrich Events (1988) and Dansgaard-Oeschger events (Johnsen et al 1992). Within a matter of years, temperatures over Greenland changed by more than 10 • C during these events, and these changes endured for more than a millennium (Johnsen et al 2001). There is evidence that such events influenced climate not only in the high latitudes but also in the tropics, where they are associated with weaker Indian-Asian monsoon rains (Wang et al 2001).
During interglacial warm periods, global temperature was similar to, or perhaps slightly higher than, pre-industrial (that is, late nineteenth-century) conditions (Snyder 2016). Northern Hemisphere summer temperatures, however, were 3 • C-5 • C warmer (Otto-Bliesner et al 2021), leading to partial melting of the Greenland ice sheet and 6-9 m of sea level rise (relative to pre-industrial time; Kopp et al 2009). A defining characteristic of interglacial periods is an intensification and expansion of the African and Asian monsoon systems (Otto-Bliesner et al 2021). The Last Glacial Maximum (LGM) peaked at about 20 000 BP ('before present,' a radiocarbon dating convention meaning years before 1950). Thereafter, global temperatures rose approximately 7 • C, an increase paused only temporarily by the Younger Dryas event, the last millennial-scale climate perturbation (Alley 2000, Osman et al 2021) (figure 2). As with previous glacial cycles, this deglacial warming was triggered by orbital forcing, but the direct radiative drivers were the rise in greenhouse gasses (carbon dioxide and methane) and the loss of the Northern Hemisphere ice sheets. In the wake of the Younger Dryas, the Pleistocene geological epoch (2580 000-11 700 BP) distinguished by alternating glacial and interglacial periods, gave way to the Holocene epoch (11 700 years BP to pre-industrial), the current interglacial period. The Laurentide ice sheet, which once covered a large portion of North America, lasted until about 7500 years ago (Peltier et al 2015). Shortly before its demise, a rapid cooling of approximately 6 • C occurred over Greenland at 8200 BP (Thomas et al 2007). This '8.2-kiloyear event' is seen in Northern Hemisphere climate records, and also may be associated with a 1000 year long drying period in Northern Africa (Rohling and Pälike 2005, Morrill et al 2013, Tierney et al 2017b.
From about 7000 years ago to the twentieth century, global mean annual temperatures were comparatively stable (figure 2), even if temperatures evolved differently at local to regional scales and for specific seasons. Reconstructions of the Holocene global temperature changes showed a 'Holocene Thermal Maximum' at around 6-7000 BP (Marcott et al 2013, Marsicek et al 2018, Kaufman et al 2020a. However, the recent reconstruction of Osman et al (2021) does not show this feature, and there is an ongoing debate about whether the Holocene Thermal Maximum is an artifact of spatial bias (Osman et al 2021) or reflects a seasonal (summer) signal (Bova et al 2021, Wanner 2021. In either case, while global mean annual temperatures were fairly stable through the Holocene, this period saw large shifts in the hydrological cycle (Mayewski et al 2004). In the early Holocene (11 000-5000 BP) the African monsoon intensified once again, resulting in an expansion of grasslands, shrubs, and perennial lakes into the Sahara Desert (DeMenocal and Tierney 2012). Monsoonal rainfall also intensified in India; meanwhile, the South American monsoon was much weaker (Brierley et al 2020). In the Mediterranean and parts of Asia there is speleothem evidence for a period of intense drying during a '4.2-kiloyear event' (beginning in 4200 BP) that, despite its name, differed greatly from the 8.2-kiloyear event (Carolin et al 2019).
Enhanced spatial coverage and improved temporal resolution allow for detailed paleoclimate reconstructions of the late Holocene and the Common Era (CE, from the year 1 to the present) using tree-rings, corals, ice cores, speleothems, sediment records, and other proxies with sufficient temporal resolution (Frank et al 2010, PAGES2k 2013, Smerdon and Pollack 2016, Emile-Geay et al 2017, Esper et al 2018, PAGES 2k Consortium 2019, Anchukaitis and Smerdon 2022. Prior to the onset of modern anthropogenic warming, temperature reconstructions spanning the last 2000 years reflect a mix of forced and internal variability across a range of time scales, from interannual to millennial, including the seasonally and spatially variable influences of orbital forcing (Esper et al 2012, Lücke et al 2021. The most important Late Holocene forcing prior to the Industrial Revolution came from explosive volcanic eruptions, which can reduce incoming shortwave solar radiation by injecting sulfate aerosols into the stratosphere, cooling surface temperatures, altering the hydrological cycle, and affecting ocean and atmospheric circulation (Timmreck 2012, McGregor et al 2015, Esper et al 2018. Temperature reconstructions show periods of widespread decadal-scale cooling linked to explosive volcanism in the 6th, 15th, 17th, and 19th centuries in particular. Forcing from volcanic eruptions was the most important contributor to initiating and sustaining the 'Little Ice Age' (LIA), which paleoclimatologists date from the 14th through mid-19th centuries (Robock 1979, Crowley 2000, Zhong et al 2011, Esper et al 2018, Slawinska and Robock 2018, Anchukaitis et al 2019 as well as the 'Late Antique Little Ice Age' (LALIA), a period of uncertain spatial and temporal extent that began with two large eruptions in 536 and 540 CE (Gunn 2000, Larsen et al 2008, Churakova et al 2014, Sigl et al 2015, Büntgen et al 2016, Toohey et al 2016, Newfield 2018. Helama et al (2017a) suggested that the LALIA was embedded within a somewhat longer but ambiguously dated 'Dark Ages Cold Period' (DACP), spanning from approximately 400-765 CE.
Solar variability is now widely thought to play a minor role in climate fluctuations over the Common Era (Schurer et al 2014), yet the mechanisms for and significance of solar influences on climate remain uncertain across a range of spatial and temporal scales (Gray et al 2010, Lockwood 2012. Cold summer temperature anomalies did coincide with Grand Solar Minima in the 14th, 16th, and 17th centuries (Anchukaitis et al 2017). Even during centuryscale events, however, internal climate system variability (which arises naturally without changes in planetary energy balance from the complex interaction of oceans and atmosphere) and earth system feedbacks were very important contributors to the temporal evolution of cooling; the spatial patterns of temperature anomalies; and interannual-to-multidecadal climate variability at the local and regional scale (Goosse The Medieval Climate Anomaly (MCA; sensu lato 800-1300 CE), a period of relatively warm temperatures spanning the first and second millennia of the Common Era, likely reflects a combination of a weak increase in radiative forcing and internal climate system variability (Goosse et al 2005, Fernández-Donado et al 2013. In particular, spatial climate reconstructions of medieval temperatures suggest internal variability played a strong role in determining the magnitude and regional timing of temperature maxima (Goosse et al 2005, Diaz et al 2011. Indeed while it is therefore possible to roughly identify periods of the Common Era corresponding to large-and centennial-scale temperature anomalies, such as the MCA, LIA, and LALIA, at the local to regional scale, and from interannual to decadal periods, internal variability would have played a very important role in the temperature anomalies experienced by human societies (Degroot et al 2021).
Internal variability played an even greater role for precipitation and soil moisture anomalies, which also had impacts on agricultural productivity. Proxy reconstructions show spatial and temporal variability in past hydroclimate dominated by the influence of internal variability, including large-scale modes of ocean-atmosphere like the El Nino Southern Oscillation (ENSO) (Cook et al 2010a, 2010b, Hunt 2011, Ault et al 2013, Tierney et al 2013, Ummenhofer et al 2013, Coats et al 2015, Stevenson et al 2015, PAGES Hydro2k Consortium 2017. Ultimately, analyses that assume that uniform, widespread, or persistent warm and cold, or wet and dry conditions persisted during amorphous climate periods like the LIA are seldom justified when examining the spatial and temporal scales of interactions that exist between climate and human society (Degroot et al 2021).

Uncovering human responses: evidence
Evidence for human responses to these climatic changes survives in everything from the genes and languages of present-day populations, to the ruins, trash heaps, archived documents, and art of ancient societies. HCS scholarship requires many disciplines primarily because this evidence is so diverse and so abundant.
Past demographic events (such as population bottlenecks and migrations) and episodes of natural selection for genetic variants (in for example genes for disease resistance or high metabolism), left signatures in the genetic makeup of contemporary populations. Because climates transformed local environments and thereby influenced how human populations could inhabit them, the DNA of contemporary individuals can suggest compelling links between human and climatic histories. Large databases of genetic markers have been assembled for studies of demography and selection, with much of the data being publicly available (e.g. The 1000 Genomes Project Consortium 2015). However, the large amount of information that can be gleaned from genomes, including inferences about health conditions, has raised a number of ethical issues of consent (Caulfield and Murdoch 2017). Current practices accordingly require full involvement with communities that donate genetic material to make sure that all participants are properly informed and involved in research (Claw et al 2018).
Over the last two decades, the development of techniques to extract DNA from ancient remains (such as bones and hair) has revolutionized our ability to reconstruct human history. It is now possible to build time series that directly record the progressive accumulation of genetic signatures from past events (Orlando et al 2021). However, preservation of ancient DNA is greatly dependent on environmental conditions. The permafrost has yielded DNA from hundreds of thousands of years ago (van der Valk et al 2021), but recovering DNA from the tropics is very challenging even when dealing with much shorter time horizons (Orlando et al 2021). Recent developments in techniques that recover DNA from sediment have further opened the possibility of building detailed time series of site occupation (Crump 2021), although fragmented material recovered in this way is not amenable to complex population genetic analysis (Sigsgaard et al 2020).
The archaeological record affords a wide variety of sources for the study of past interactions between environments and human societies. Most fundamentally, archaeological data in the form of diachronically and spatially structured site distributions can be used to relate human settlement to broad ecological parameters, and to changes in those parameters. At the core of archaeology is material culture, much of which can be understood as 'extra-somatic means of adaptation' (Binford 1962). Changing forms of hunting equipment, farming or herding practices, mobility, and political arrangements can be read from the archaeological record directly (Evans 2002, Hussain and Riede 2020).
In addition, archaeological excavations commonly yield not only human-made artifacts but also many 'ecofacts,' or 'geofacts.' These include plant and animal remains, soils, and bio-or geoarchaeological substances that require laboratory analysis. These substances are by no means limited to DNA, and include lipids, isotopes, and other residues that are interpreted by specialists in, for example, environmental archaeology, geoarchaeology, archaeozoology, and palaeoethnobotany (Clift et al 2011, Sandweiss andKelley 2012, Murphy andFuller 2017). Increasingly, archaeological sources provide a bridge between evidence in history and genetics.
Climate historians, along with many geographers and economists, primarily rely on written accounts of the past. Accordingly, their studies have long focused on times and places that are well documented by surviving textual evidence. To be useful for climate historians, such evidence must either mention weather directly, or describe environmental and social changes clearly influenced by weather. Documents may suggest interactions between climatic and social changes on very different scales. Diaries and correspondence, for example, may reveal the lived experience of weather and associated ecological phenomena, such as plant and animal phenology, while accounts archived by institutions and governments can record the shifting fortunes of entire societies or economic sectors confronted with climate change (Huhtamaa and Helama 2017a, García-Herrera et al 2018, White et al 2018b, Van Bavel et al 2019, Degroot 2020, Pfister and Wanner 2021.
Some sources interrogated by climate historians register human and environmental changes across many scales in time and place. Visual art, for instance, may depict discrete, local events-skaters on a cold winter's day, for example-while using pigments and appealing to regional markets in ways that registered decadal economic trends, and while depicting cultural values, ideas, or artifacts that took centuries to develop (Behringer 2010, Ray 2019). Climate historians work to identify the influence of weather and climate across different resolutions, especially when combining source types, and to isolate the effects of weather from those of cultural, political and socioeconomic forces (Bell and Ogilvie 1978, Chiari 2019, Van Bavel et al 2019, Williamson 2020.
Although the textual record useful for historians can seem overwhelmingly vast, documentary evidence is in fact unavailable for much of the globe before the eighteenth century (Brönnimann and Wintzer 2019). In regions and periods where textual sources are sparse or nonexistent, historians are beginning to employ historical linguistic data to understand human responses to past climate change (De Luna 2016, Hannaford and Nash 2016, Degroot et al 2021. Historians may also use oral histories to identify environmental changes, often at relatively low resolution but occasionally across centuries or even millennia (Cruikshank 2001(Cruikshank , 2007. Still, the limitations of historical sources across large scales in time and space today encourage collaboration between historians and scholars in other disciplines, particularly archaeology and paleoclimatology.

Uncovering human responses: methods
Population genetics provides tools both to reconstruct past demography and thereby to quantify human responses to climate change, and to study specific genes that underpin selective responses to climate. The techniques used for these purposes depend on the nature of the data available. For demographic reconstructions, earlier studies mostly relied on mitochondrial DNA (mtDNA). The small size of the mitochondrial genome and its haploid nature make it an easy target for sequencing, both in contemporary and ancient samples. However, the power of demographic reconstructions based on mtDNA is limited, since mtDNA is a single, non-recombining marker. While it is easy to model its inheritance, the genealogy of a single marker is highly stochastic and can be poorly representative of underlying demographic history (Balloux 2010). Nuclear markers can provide much greater power, but the large size of the nuclear genome meant that, until a decade ago, it was prohibitive to build extensive datasets of complete genomes.
Single nucleotide polymorphism (SNP) chips, which genotype pre-defined positions of the genome known to be variable, have been used extensively to quantify human population structure. A challenge when using SNP chips is that the process of choosing the positions of interest (known as 'ascertainment') is biased towards variants more common in the discovery panel (the group of individuals used to choose the SNPs), and towards variants with intermediate frequencies (as rare markers are unlikely to be present in the small discovery panel) (Nielsen 2004). This ascertainment bias prevents the use of many demographic modeling techniques that rely on unbiased estimates of variant frequencies (Nielsen 2004, Albrechtsen et al 2010. Geneticists once used microsatellites-repetitive regions in the genome-as markers for population genetics analyses. However, modeling their evolution is challenging, and these markers have fallen out of fashion as they are not cost effective. Given the sharp drop in the cost of sequencing over the last decade, whole genome sequencing has become the standard approach, with projects building datasets of hundreds or even thousands of genomes (The 1000 Genomes Project Consortium et al 2015). Yet geneticists studying ancient material with scarce, damaged, or fragmented DNA may have difficulty choosing which genetic markers to focus on (Orlando et al 2021). Shotgun sequencing of the whole genome is possible, but expensive as much of the DNA found in ancient remains originated in microbesrather than humans-and is therefore not relevant to most HCS studies. Targeted capture uses the same logic as SNP chips and focuses on specific positions (Orlando et al 2021), but the ascertain bias that results from concentrating on these positions means that the data produced by this method can only be used for certain types of analysis.
Archaeologists have also developed an extensive toolbox for approaching human-environment relations. As accurate chronologies are essential in virtually any such attempt, dating methods and the downstream statistical treatment of dating information has received much attention. Complementing chronology, archaeologists have long used Geographic Information Systems (GIS) to aid their analysis of distribution data. The integration of digital methods into environmental archaeological analyses is increasingly seamless, and offers useful points of contact with neighboring disciplines such as geography (Siart et al 2017). A similar form of methodological fellowship aligns environmental and geoarchaeology with the biological and geological sciences respectively (Pollard 1999). Basic and advanced field and laboratory techniques developed in the former can often be transferred to the specific study of anthropogenic soils and biological samples deriving from archaeological contexts.
Archaeology's grounding in fieldwork allows for the study of human responses to climate change across a range of spatiotemporal scales. Occasionally, a single site will yield evocative evidence of abandonment or adaptation. At other times, such responses will be more clearly visible in the large-scale patterns emerging from inter-site comparisons. Like many scientific disciplines, archaeology is experiencing a rapid shift towards computational methods and working with large quantities of data. The often-fragmentary nature of archaeological evidence and the variable granularity of the archaeological record sometimes poses challenges. However, on both epistemological (Currie 2018) and practical (Caseldine and Turney 2010, Izdebski et al 2016, Boivin and Crowther 2021) grounds there are good reasons to believe that assessments of past human-environment relations based on archaeological materials are becoming increasingly robust.
In general, climate historians employ the same techniques of source analysis and criticism that are followed by most other historians. To use these techniques, historians study the historical contexts in which sources were produced; learn the methods by which sources communicated information (in the case of textual sources, often through training in languages or paleography); and then identify sources in repositories (for textual sources, in archives or libraries through consultation with archivists and librarians). In order to ascertain the legitimacy and authenticity of these sources as evidence useful for confirming or rejecting historical hypotheses, climate historians identify which individuals or institutions were responsible for them, and when, where, how, and with what purpose they were created (Howell and Prevenier 2001, Berkhofer 2008). The historical method can discourage climate historians from assuming that historical sources provide reliable and transparent accounts of weather, or activities influenced by weather, which in turn can discourage them from making spurious connections between human and climatic histories. Scholars in other disciplines who make use of historical primary sources-including geographers and economistsmay use historians' interpretations of these sources without realizing that those interpretations are not universally agreed upon by historians (Van Bavel et al 2019).
Most HCS scholars who use historical sources employ one of two broad methods for establishing that climatic changes caused or helped cause societal changes (figure 3). Historians largely follow what White and Pei have recently coined the cause-of-effect method by deriving inferences from close textual analysis and contextualization; comparing outcomes in similar cases of social or climatic change; and considering counterfactual scenarios. These longstanding elements of the historian's toolkit are used by climate historians to develop narratives that establish climatic changes or anomalies as both necessary and sufficient conditions in particular instances of social change (White and Pei 2020). Geographers and economists, by contrast, largely use an effect-of-cause method by quantifying historical evidence and finding correlations between climatic and social trends (De Vries 1980, White and. Significant correlations, interpreted at times using statistical methods for identifying causation, can ostensibly reveal the control of the climate variable over diverse social variables (Zhang et al 2010, Brunt 2015, Huhtamaa and Helama 2017b. For geographers and economists, this method establishes climate change as a cause of different forms of social change, without being necessary or sufficient in all instances of that social change (White and Pei 2020).
These different ways of establishing causation are, in principle, mutually compatible. Yet in practice geographers and economists have criticized the causeof-effect method as being insufficiently rigorous or 'scientific,' and too limited to specific case studies that lack predictive power. At the same time many historians have dismissed the effect-of-cause method as insufficiently grounded in textual analysis, and lacking interpretive power in contingent historical case studies (Degroot 2018a. Subdivisions also persist between scholars using each method. While some historians employing the causeof-effect method, for example, conclude that modest climate changes definitely transformed past societies, others focus on smaller scales of analysis, and emphasize uncertainties that follow from gaps in surviving evidence. Some use probabilistic terminology to identify connections between climatic and social processes unfolding across different scales in time and space (Degroot 2018c).
Recently, scholars have identified systematic shortcomings in how many HCS studies have used both cause-of-effect and effect-of-cause methods. HCS studies using either method have, for example, misinterpreted climate reconstructions; focused excessively on large spatial or temporal scales; identified simplistic dichotomies between vulnerable and resilient societies; devoted insufficient attention to uncertainty; and equated correlation too easily with causation. To overcome these problems, scholars have introduced a research framework consisting of binary questions that encourage greater integration of methods and sources from distinct disciplines (Degroot et al 2021).

Uncovering human responses: models
The models used by geneticists to reconstruct demography depend on the type of data available. Models that reconstruct gradual changes in population sizes through time, such as Bayesian skyline plots of mtDNA or sequentially Markovian coalescent (PSMC) methods for whole genomes, tend to focus on a single population that is assumed to be isolated-meaning it does not receive any migrantsor otherwise on a pair of connected populations (Ho and Shapiro 2011, Mather et al 2020). These approaches rely on the logic that lineages are more likely to merge when populations are small, and less likely when populations are large. According to this reasoning, the rate at which lineages merge at different times can be converted to population sizes (although migration will also affect this rate, and thus confound size reconstructions).
Models that consider the relationship among multiple populations (and consequently allow for migration) often require geneticists to define specific events, which can then be tested and quantified. For example, geneticists can identify the presence of population bottlenecks and correlate them with periods of climate change that reduced the productivity and in turn the habitability of local landscapes. Inferred demographic changes, however, have to be interpreted with care. Population genetics models reconstruct effective population sizes (the sizes of an idealized, unstructured and randomly mating population), and these can differ greatly from census population sizes, owing for example to geographic substructuring or marriage practices (e.g. Palstra and Fraser 2012). Climate Informed Spatial Genetic Models have attempted to explicitly reconstruct the possible influence of climate change on demography by matching its predicted impact to the observed patterns of genetic differentiation among contemporary and ancient genomes (e.g. Erikson et al 2012, Delser et al 2021).
Different aspects of genetic data can be used to define how compatible a model is with empirical evidence. Geneticists can use an explicit likelihood framework, where the best estimates of demographic parameters and their uncertainties can be directly derived, mathematically, from data. Alternatively, they can use simulations, by comparing simulated to observed genetic data using frameworks such as Approximate Bayesian Computation to formally obtain best estimates of demographic parameters and their uncertainties (Sunnaker et al 2013). Natural selection can be inferred by looking directly at the frequency of a variant of interest (either by comparing populations under varying degrees of selection, or different time points), as well as the signature that is left behind in sites adjacent to the one of interest (as selection tends to reduce variability around the selected site) (Scheinfeldt andTishkoff 2013, Vitti et al 2013).
Various metrics have been designed to capture and emphasize such signals. A complication in assessing the role of selection is that demographic events (such as bottlenecks) can generate signals that are similar to those arising from selection. However, while demographic signals are spread throughout the whole genome, selection only acts on a few variants. Selection tests take advantage of this latter property by using an outlier approach, where the signature at a locus of interest is compared to the background levels across the whole genome. This approach provides 'candidate' genes that are likely to be under selection (though it has a high rate of false positives; e.g. Koropoulis et al 2020), and these candidates can then be indirectly linked with climatic drivers (Rees et al 2020). A few models have also been developed to directly link the frequency of alleles to climatic drivers by looking for correlations between the frequency of a genetic variant and a climatic variable across contemporary populations (e.g. Hancock et al 2008). Recently, techniques have developed to use the time series provided by ancient DNA, which allow a direct observation of the changes resulting from selection (e.g. Schraiber et al 2016, Loog et al 2017. While these approaches have yet to be used to investigate climatic effects, they hold much promise for the future given the exponential increase in the number of human ancient genomes being recovered. Correlation in space and time-at different resolutions and with variable methodological rigoris also often at the heart of archaeological humanenvironment studies. Yet many archaeologists believe that, unless combined with some form of model for human behavior and culture change, any resulting interpretations remain difficult to substantiate (Contreras 2016), and open to charges of a priori deterministic biases inherent in chosen study designs (Arponen et al 2019).
Use of terms such as 'resilience,' 'adaptation,' and 'vulnerability' has become widespread in archaeology to characterize human responses to climate change. Many archaeologists have defined resilience as part of Formal Resilience Theory (RT), or simply RT. Central to RT is a method for conceptualizing connections between behavioral adaptations in human populations and external environmental change. This method is based on the Adaptive Cycle (AC) Model, in which social-ecological systems pass through four stages-which may be called growth, conservation, release, and reorganization-that together constitute a single cycle (figure 4). Cycles are nested within each other across different spatiotemporal scales, a feature called 'panarchy,' and connect through 'revolt' and 'remember' feedback loops. These loops represent the nonlinear nature of social responses to climatic changes (meaning their inputs need not be directly proportional to their outputs) (Leroy 2006, Cumming et al 2008. The four domains in each cycle have different degrees of 'connectedness' and 'potential,' which in turn gives them different degrees of resilience. Resilience in the model declines as the system gains complexity, and only increases during the reorganization/renewal period (Bradtmöller et al 2017).
For HCS scholars, there are advantages to using RT and the AC model. RT arguably provides a powerful tool for overcoming deterministic interpretations of climate change as a direct cause of social change, and it appears to permit straightforward comparisons between case studies of resilience (Hegmon et al 2008). Yet archaeologists have long differed over whether the AC model is better suited as a heuristic device or analytical tool, and they have not used the same definitions for components of the model. The model has also been difficult to operationalize when applied to imperfect archaeological datasets (Redman 2005, Bradtmöller et al 2017, Brewer and Riede 2018, Heitz et al 2021. Much like geography, the discipline of archaeology is divided between those who see cultural change as largely responsive to internal, political, and ideational forces, and those who would prioritize external forces, such as climatic and environmental change ('internalists' still dominate the discipline of history). Nowhere is this division more obvious than in discussions of past examples of societal 'collapse,' which many archaeologists discern in evidence of widespread site abandonment that seems to suggest demographic decline, elevated rates of conflict, and the disintegration of elite class superstructures. Some scholars argue on the basis of archaeological evidence that past societies collapsed in the face of cli- . While these basic attitudes do have subtle consequences for study design and practice, models seeking to combine exogenous and endogenous forces are clearly most promising (Butzer 2012, Butzer andEndfield 2012).
Some archaeologists and historians now stress the potential of archaeological or historical research to conduct 'natural experiments' that compare chosen study units (such as villages, cities, and cultures) before, during, and after a climatic perturbation (Diamond and Robinson 2010, Riede 2014, Manning et al 2017, Bauch 2020. In such experiments, societal responses to climate change are always co-determined by the socioeconomic and political conditions in place at the onset of a given perturbation. Both social and climatic changes govern access to resources, which can be identified in archaeological proxies of risk management that in turn approximate vulnerability and resilience (Halstead and O'Shea 1989, Wisner 2004, Riede et al 2017. Proxies for adaptation can then be sought in changes visible in archaeological remains that postdate climatic perturbations.
Influenced in part by archaeologists, historians have also adopted the concepts of adaptation, resilience, and vulnerability (Endfield 2014, Izdebski et al 2018, Xoplaki et al 2018, Degroot 2018c, Degroot et al 2021. The widespread use of this terminology in HCS scholarship has helped make theoretical models more applicable across disciplinary boundaries (Engler 2012). Yet while the most influential models in archaeology prioritize change over time, models in history, geography, and economics typically focus on the uneven magnitude of changes across different elements of past societies. In history, for example, the longstanding impact-order model (figure 5) typically conceptualizes climatic changes as having progressively less direct influence on biophysical, economic, social, and cultural transformations within past societies (Pfister 1984, 2007, Kramer 2015, Luterbacher and Pfister 2015, Camenisch and Rohr 2018. Historians originally developed versions of this model to describe societal responses to climatic anomalies in particular times and places, then used it to conceptualize the regional or even continental consequences of climatic trends. They have added complexity to the model by, for example, representing adaptive responses to climate changes, or adding cultural reactions to every impact level (Camenisch et al 2016a, Ljungqvist et al 2020. Geographers and economists developed similar models to identify how climate change provoked wars and demographic crises, for instane, by reducing agricultural output, although their models more consistently include feedbacks between social changes (Zhang et al 2010, Pei et al 2014, Burke et al 2015, Cook and Wolkovich 2016. Recently, some historians have pointed out that the impact order model depends on the assumption that climatic changes had more direct impacts on agriculture or pastoralism than other sectors of society: an assumption that increasingly does not seem to have been accurate in many historical contexts (Degroot 2018b, Degroot et al 2021. Indeed, because historians often stress contingency and spatiotemporal particularity in establishing causation, many have been skeptical of normative models-such as the AC model and to some extent the Impact Order model-that generalize across outcomes. Climate historians often benefit from abundant documentary evidence that can reveal human responses to weather in granular detail. As a result, they do not necessarily require models to argue that correlations between climatic and societal changes reveal causation. What some historians call for instead are process models that generalize mechanisms, and thereby provide novel ways to conceptualize how climatic changes spur large-scale social changes. Several recent studies accordingly consider how climate changes influenced the availability, quantity, and character of energy accessible to human populations, but such approaches are still in their infancy (Degroot 2018c).
An older and more influential process model used by historians identifies a 'fatal synergy' between sharp reductions in agricultural output that were triggered or worsened by climatic anomalies, and a host of social ills that further reduced agricultural productivity (figure 6). Repeated harvest failures, according to

Figure 6.
A representation of the 'fatal synergy' that, in influential studies by climate historians, worsened societal crises across large spatial scales that were ultimately triggered by late Holocene climate changes. Feedbacks connect the consequences of harvest failures to the further devastation of agricultural regions (Parker 2013). the model, ultimately reduced the availability of food and thereby caused widespread malnutrition, starvation, and famine. By altering the range and reproductive rate of disease vectors, and by weakening immune systems in malnourished human and animal bodies, climatic anomalies also spurred the spread of epidemic or epizootic disease. Migration from the famine-stricken countryside circulated epidemics or epizootics, and compounded overcrowding in cities that accordingly grew more vulnerable to epidemic outbreaks. Mortality among human laborers and animals essential to pre-modern agro-economies further reduced food production, and exacerbated widespread depopulation. Amid these disasters popular discontent with ruling authorities-who in some cultures were held directly responsible for weatherspurred sedition, rebellion, and even wars between polities looking to exploit one another's weaknesses. Pre-modern wars further reduced food availability through, for example, the conscription of agricultural laborers, the redirection of grain stocks to armies, and the plundering of the countryside (Parker 2013, Degroot 2018b. Implicit in the concept of the fatal synergy is once again the assumption that climatic anomalies and changes had a particularly direct impact on agriculture and pastoralism. Yet HCS scholars have used the concept to identify feedbacks that are typically hidden in the Impact Order Model, and help explain how relatively small variations in the climate of the Holocene may have had profoundly destructive, nonlinear consequences for some pre-modern societies (Parker 2013).

HCS findings: genetics
Paleogenetic evidence can reveal compelling correlations between demographic changes in human populations and climatic trends. By reconstructing local environmental conditions, HCS scholars can then infer causal links between climatic changes and the migration, expansion, retrenchment, or extirpation of human populations. Such work can suggestbut not confirm-relationships between climatic and human histories in the Pleistocene, which may be difficult to discern using other evidence.
Population genetics played a key role in providing support for a recent 'out-of-Africa' origin of anatomically modern humans (e.g. Cann et al 1987, Fu et al 2013, Rieux et al 2014, Malaspinas et al 2016. The timing of the primary out-of-Africa migration inferred from genetics (65-50 000 BP) suggests a link between this range expansion and climatic change. Earlier waves of migration coincided with generally warm and wet conditions that may have 'pulled'that is, enticed-hunter-gatherers away from their African territories. Yet the primary out-of-Africa migration coincided with exceptionally cold and dry conditions in the Horn of Africa (Liu et al 2015, Tierney et al 2017a). These conditions may have 'pushed' migrants to leave Africa. Since cold water takes up less volume than warm water, and vast quantities of water were trapped in ice sheets, sea levels were low and a southern migratory route out of Africa was therefore more traversable than it had been (Beyer et al 2021, Groucutt et al 2021. Genetic dating of the Pleistocene arrival of modern humans into the Americas (e.g. Raghavan et al 2015), together with sedimentary DNA, similarly suggests that the opening of a coastal corridor in the melting Cordilleran ice sheet may have aided migration into the Western Hemisphere after approximately 12 600 BP (Pedersen et al 2016). Nevertheless, recent archaeological and genetic evidence suggests far earlier migration into the Americas, when the corridor had not yet appeared (Moreno-Mayar et al 2018, Becerra-Valdivia and Higham 2020). Paleogenetic evidence has encouraged HCS scholars to assume causation from approximate correlations, but it can also reveal that population dynamics have long been shaped by much more than climatic conditions. Population genetics is in fact blind to population movements that resulted in no descendants (or at least, too few to be detectable), and large-scale population expansions often left genetic signatures that are stronger than, and therefore override, those generated by local changes in population (Miller et al 2018). Nevertheless, time series of ancient DNA have allowed geneticists to focus on local dynamics, and thereby to uncover that the climatic oscillations of the Pleistocene and early Holocene coincided with profound changes in human communities. Analysis of time series data of aDNA in western Eurasia, for example, shows a clear population bottleneck during the LGM, as well as a major population replacement during the warming period after 14 000 BP (Fu et al 2016, Posth et al 2016. Populations that persisted in colder climates during the LGM, such as the Western Hunter Gatherers and Caucasus Hunter Gatherers, are characterized by shorter runs of homozygosity (a sign of smaller population sizes) compared to populations from further south, such as the Early Farmers, which may have sustained larger populations owing to comparatively benign climatic conditions (Jones et al 2015). Using genetic evidence, regional climatic changes in the Holocene, such as the 4.2 kiloyear aridification of the Arabian Peninsula, have also been correlated with bottlenecks in local populations (Almarri et al 2021).
Such correlations between climatic change and human demography cannot easily reveal causation, let alone the motivations for human action. As a result, HCS scholars have attempted more direct tests of the effect of climate on demographics, by linking environmental variables to the level of gene flow (and thus divergence) among contemporary populations. Analyses of modern genomes in Africa (Petkova et al 2016) and Eurasia (Pagani et al 2016) have suggested that deserts and mountains were long the principal barriers limiting human movement, while coastal areas acted as corridors. Such studies use a static approach in which modern climate is correlated with genetic divergence, but the diachronic effect of climate change is ignored. Recent publications have attempted to explicitly account for the historical influence of climatic changes on demography, by integrating climate reconstructions of the Pleistocene and Holocene within Climate Informed Spatial Genetic Models. In these models, population sizes are either predicted by fluctuating climatic variables such as aridity and temperature, or by primary productivity. The models are able to reconstruct the level of divergence found in both modern (Eriksson et al 2012) and ancient genomes (Delser et al 2021), and confirm the important role of deserts and mountains in shaping gene flow over time. Moreover, geneticists have found that, in the Holocene, the level of genetic admixture between hunter-gatherers and farmers increased in areas with low Growing Degree Days, a measure of climate that predicts the accumulated heat available for crop growth. Higher mixingsuggesting more interaction between both groupsoccurred when and where climatic conditions contributed to less successful food production (Betti et al 2020).
Geneticists have also connected shifts in the climates experienced by Pleistocene hunter-gatherers, owing either to migration or orbitally forced climatic changes, to genetic adaptations that persist in human populations. Some of these adaptations are controversial. For example, genetic research has shown that, sometime between 6 and 10 000 BP, a genetic mutation affected the OCA2 gene in European populations and led it to reduce the production of melanin in irises (Eiberg et al 2008). The result was the emergence of blue eyes in human populations. The OCA2 mutation may have been an indirect adaptation to climatic warming, because the end of the Younger Dryas in approximately 11 700 BP allowed European populations to live at higher latitudes, where there would have been less sunlight in winter. Blue irises absorb more light than brown irises, and the blueeyed may have been resistant to Seasonal Affective Disorder caused by lack of winter light. Other geneticists have proposed sexual selection as an important driver for the evolution of blue irises, but this explanation does not necessarily preclude a climatic cause (Workman et al 2018). Versions of genes associated with lighter skin-primarily SLC24A5 and SLC45A2-may have emerged at approximately the same time in European populations that had settled across high European latitudes in the wake of the LGM. These genetic versions may have benefited European populations by facilitating vitamin D production when ultraviolet light is limited (Jablonski and Chaplin 2010). However, aDNA has recently revealed that SLC24A5 was associated with the spread of early farmers out of Anatolia, and some Mesolithic hunter-gatherers in Europe had moderately dark skin (e.g. Olalde et al 2014, Ju andMathieson 2021).
Explicit tests for association between SNP frequencies and climatic reconstructions indicate that climate-influenced selection is the likely explanation for the rise in the frequency of some genetic variants (Hancock et al 2008, 2011). The variants with the strongest signals tended to be associated with pigmentation and UV radiation, as well as infection and immunity (Hancock et al 2008, 2011). Furthermore, such studies have highlighted a number of SNPs (such as LEPR R109K and FABP2 A54T) that are associated with phenotypes favoring cold tolerance (Hancock et al 2008). In some instances, variants found to be associated with climate adaptations are known to increase the risk of autoimmune or metabolic diseases, which are likely to be pleiotropic effects of selection for an advantageous phenotype. A number of ailments currently seen in contemporary populations may therefore be the product of past selection to better withstand challenging climatic conditions. For example, geneticists have argued that a mutation in the TRPM8 cold receptor, which is common across higher latitudes, has helped populations regulate body temperatures in cold weather while increasing the prevalence of migraines (Key et al 2018). Genetic research has therefore revealed intriguing correlations between climatic changes, local environmental transformations, and genetic mutations, but HCS scholars who use genetic evidence can for the most part only infer causal connections.

HCS findings: archaeology
There is now little doubt among archaeologists that, in the broadest sense, early human history was shaped and constrained by climate change. Archaeological evidence supports genetic evidence that correlates changes in hominin populations with climatic shifts during the Pleistocene (Hublin and Roebroeks 2009, Dennell et al 2011, Pedersen et al 2021, Timbrell et al 2022, Timmermann et al 2022. Archaeologists emphasize, however, that the climatic oscillations of the Pleistocene helped encourage both biological and cultural adaptations within these populations that ultimately enabled environmental modifications by human communities at everincreasing scales (Roberts and Stewart 2018, Riede 2019, Murray et al 2021. The roots of farming lifestyles stretch back into the Pleistocene, yet the Neolithic Revolution-the widespread dissemination of agriculture and the technological and cultural innovations that accompanied it-unfolded only with the onset of the Holocene (Simmons 2011). Interglacial periods during the Pleistocene were in all probability at least as warm as the Holocene. Yet archaeologists have long argued that the globally warm and stable climate of the Holocene permitted the expansion of agriculture once the selection pressure of earlier cultivation yielded cereals productive enough to sustain human populations (Richerson et al 2001). Some archaeologists, however, argue that many huntergatherer communities first adopted agriculture only during transitions from warmer and wetter to cooler and drier conditions, which occurred in some occupied regions during the Younger Dryas of the late Pleistocene, and the 8.2 kiloyear event in the early Holocene (Simmons 2011). According to this theory, hunter-gatherer communities that expanded in comparatively benign climatic conditions could not easily accommodate the influence of climate change on the fauna and flora that had sustained them, and many were compelled either to migrate or domesticate plants and animals (Lieberman and Gordon 2018).
The classic case of the transition to agriculture by Late Natufian communities in the Levant provides ample evidence for both theories. For decades, influential archaeologists have argued that regional aridification during the Younger Dryas-which is now disputed by paleoclimatologists-reduced the size of forests and the availability of wild cereals, compelling some Natufian communities to cultivate grains that were increasingly difficult to obtain otherwise (other communities adapted by developing new hunting technologies, but they did not survive) ( Bar-yosef 1998, Stein et al 2010, Liu et al 2013. Excavations of, for example, rye seeds at the Natufian site of Abu Hureyra may suggest that this transition to agriculture coincided with the onset of the Younger Dryas in approximately 13 000 BP. Yet many archaeologists have found evidence that contradicts the idea of a Late Natufian subsistence crisis (Colledge and Conolly 2010). Some conclude that the Late Natufian transition to farming occurred only after the Younger Dryas had subsided. According to this view, growing populations spurred the development of new and more reliable means of obtaining food, and a wet, warm climate made agriculture more viable than it had been (Willcox et al 2008(Willcox et al , 2009). Archaeological evidence can suggest causation more clearly than genetic evidence alone, yet it is often fragmentary and open to contradictory interpretations. Debates therefore persist over whether communities were pushed or pulled towards agriculture by climate change, and indeed over the extent to which climate change mattered at all (Maher et al 2011).
Agriculture where it developed demanded and enabled new forms of storage and stationary infrastructure, which in turn helped some populations withstand climatic trends and anomalies. Agriculture also transformed labor relations and political structures. Most agricultural societies developed to be hierarchical and territorial, a process that while neither universal nor inevitable, eventually culminated in the emergence of states and empires (Graeber and Wengrow 2021). Archaeologists have found that climatic anomalies and trends accelerated some of these developments. Archaeologists and paleoscientists have for example tied the gradual emergence of Pharaonic Egypt to the desertification of northern Africa, which likely inspired some pastoral communities to migrate to the fertile Nile delta Timpson 2014, Williams 2021). Indeed some centralized political systems developed because they were able to better coordinate responses to climatic perturbations (Wu et al 2016). Yet the commitment to stationary resources in such systems also raised the cost of migration as an adaptive response to climateinduced hardship. Archaeologists have long argued that communities developed distinct and at times contradictory risk management strategies with varying degrees of success, which together contributed to lags between climatic changes and social responses in many societies (Laland and Brown 2006).
Nevertheless, proposed subdivisions in the Holocene can also be understood as periods of largescale transformation in human societies (Walker et al 2019a(Walker et al , 2019b, even if the manifestations of these events varied regionally (LeGrande et al 2006, Morrill et al 2013). Archaeologists have devoted considerable attention to both the 8.2 and 4.2 kiloyear event. By 8200 BP, lower latitudes were mostly occupied by farming communities, and larger settlement agglomerations, such as Catal Huyük in presentday Turkey, already had populations numbering in the thousands. At higher latitudes, hunter-gathererfisher communities had pushed into remote reaches of the Arctic. Archaeologists have shown that the regional manifestations of cooling and drying led to the widespread abandonment of settlements in the eastern Mediterranean, for example, and migration towards the west (Weninger et al 2006), likely by depressing agricultural productivity. At the same time, marked cooling in northern Europe and a major North Sea tsunami generated by a massive undersea landslide off the Norwegian coast together led to dramatic population declines (Manninen 2014 (Kuper and Kröpelin 2006). Regional differences in the environmental expressions of this climatic event, coupled with the different background conditions of affected societies, generated widely divergent outcomes for contemporary populations as well as lags and asynchronies between climatic and cultural changes (see Maher et al 2011). Some societies were resilient, while others likely underwent some degree of societal collapse.
Ancient Egypt ultimately depended on irrigation agriculture, as did powerful states across Southwest Asia. Other regional states relied on rain-fed farming and extensive pastoralism, while early states across the Indian subcontinent exploited regular monsoon rains. The 4.2 kiloyear event appears to have caused changes in the frequency, timing, and extent of rainfall in these regions (see Jones et al 2019). By interpreting evidence from sites such as Tell Leilan in Syria, archaeologists have argued that prolonged drought provoked the collapse of contemporary empires, including most famously the socalled Akkadian Empire (Weiss et al 1993, Cullen et al 2000, Weiss and Bradley 2001. It now seems that these climatic changes had some influence over the demographic and political development of empires, but also that their impacts varied from region to region and did not devastate every polity (Weiss 2017. There were winners and losers in what clearly constituted a complex political landscape (see Scanlon 1988). Archaeologists have found that contemporary polities were both inherently unstable and capable of mounting effective and diverse adaptive responses to climatic perturbations. During the 4.2 kiloyear event, lags caused by these culturally mediated responses may again have weakened chronological correlations between climatic and some social changes, or the climate changes themselves may not have been coeval.
Climatic changes may have hastened the decline of some ancient empires by undermining their agricultural foundations, yet it is often difficult for archaeologists to determine the relative contribution of climatic influences to political and economic trends (Sinha et al 2019). These complexities are especially clear in archaeological studies of regional climatic changes and social responses along the Desert and Steppe Silk Roads, which long connected economies and cultures across Eurasia . Scholars have linked regional drying trends from 3500 BP to a broad movement towards urbanization and irrigation along the classical Silk Roads, and then a shift towards wetter conditions to the dawn of a golden era in Silk Road trade (Hill 2019). Yet they have also established that regional responses to climatic trends and anomalies were diverse and often effective, depending more on the internal characteristics of communities and societies than the nature or magnitude of environmental change (Panyushkina et  Archaeologists have uncovered similarly complex evidence for societal upheavals during the LALIA. In Northern Europe and the Eastern Baltic region, textual sources are absent but archaeological evidence abounds. Archaeological evidence for societal responses to the LALIA remains open to conflicting interpretations (Moreland 2018), but most archaeologists argue that the sixth century was a troubled time in the region, with major breaks in settlement structure, land-use, demography, and cultic practices (Price et al 2015, Newfield 2018. Cooling led to years of extreme frost, the spread of cereal fungi, and ultimately poor harvests that culminated in a widespread agricultural crisis (Widgren 2012, Bondeson and Bondesson 2014, Helama et al 2018. Rising mortality may have been exploited strategically by Swedish elites, who acquired abandoned land and thereby increased their political power (Löwenborg 2012). Across Norway and the eastern Baltic, archaeological evidence suggests a sharp decline in cultural activity and population during the LALIA (Tvauri 2014, Solheim andIversen 2019).
In Denmark, soils were more fertile, growing days were more abundant, and economic connections were closer to the rest of Europe. Archaeological sources suggest that the impacts of the LALIA on Danish communities were accordingly more diverse. They reveal, for example, ritual responses in the form of starkly increased depositioning of valuables related to the contemporaneous Sun-cult. They indicate that while some settlements declined, others were founded, and new decorative designs and power relations emerged (Hoilund Nielsen 2005).
The LALIA may have had even more profound cultural consequences in Scandinavia. Recorded in writing only many centuries later, Nordic eschatology around the 'Fimbulwinter' as omen and prelude to the end of the world-or at least to social order-may have emerged during the LALIA, and could constitute an effort to grapple with the social responses to climatic change that archaeologists have uncovered (Axboe 1999, Gräslund 2007, Gräslund and Price 2012, Nordvig and Riede 2018. While less information is available from the Americas during the LALIA and therefore the regional climate manifestation and the causes of societal change remains poorly resolved, this period did correspond with a sixth-century Maya 'hiatus' (Curtis et al 1996, Dunning et al 1999, DeMenocal 2001, Nooren et al 2017, Dull et al 2019.
Centuries later, during the Medieval Climate Anomaly (MCA), Norse settlers occupied Iceland and parts of coastal Greenland and Canada. Archaeologists once concluded that North Atlantic warming allowed the Norse to establish settlements in Greenland partly by reducing the extent of Arctic sea ice, which made it easier to navigate near Greenland, and by allowing Norse settlers to import agro-pastoral practices from Iceland (Hambrecht 2015). Following this logic, Norse settlements disappeared with the onset of the LIA in the fourteenth and fifteenth centuries because sea ice severed their links with Europe, and subsistence strategies that worked at lower latitudes could no longer function in cooler weather (McGovern 1980(McGovern , 1981. Yet many archaeologists-including some responsible for originally characterizing the Norse as inflexible-now emphasize that Norse settlers developed traditional ecological knowledge attuned to diverse environments across the northern Atlantic (Dugmore et al 2009(Dugmore et al , 2013. In Greenland, Norse settlers organized communal labor across dispersed settlements and abandoned cod fishing for communal hunts of migratory harp and hooded seals, which filled a springtime provisioning gap. They learned to build driving lanes for caribou; improve farming through drainage, irrigation, fertilization, and fencing; and acquire live polar bears, narwhal teeth, and walrus tusks and hides for European markets. New explanations for the rise and decline of Norse settlements in Greenland are multicausal (Hambrecht 2015). Summer sea threatened Norse efforts to organize communal labor and travel to walrus hunting grounds that sustained the Norse economy. Rising sea levels eroded lowland pasture, and together with cooling temperatures exacerbated the growing dependency of small farmers on large magnate farms. The relatively small western (northern) settlement disappeared in the fourteenth century CE. The larger eastern (southern) settlement endured by intensifying the communal seal hunt, yet more frequent storms made this an increasingly perilous activity (Dugmore et al 2012). The exquisite adaptations that the Norse made to their Greenlandic environment made them more vulnerable when that environment changed (Dugmore et al 2013, Jackson et al 2018a).
Economic and demographic changes beyond the control of the Greenlandic Norse compounded those vulnerabilities. The migration of potentially hostile and technologically more sophisticated Thule Inuit communities into the outer fjords around the eastern settlement may also have complicated the Norse seal hunt. Moreover, the declining popularity of walrus ivory in Europe-and the growing dominance in Arctic commerce of the Hanseatic League, whose merchants traded fish rather than luxury goodsundermined the foundations of the Greenlandic economy (Barlow et al 1997, Dugmore et al 2012). The old story of Norse inflexibility is also not entirely without merit. There is no evidence, for example, that the Norse adopted Thule technologies-such as the toggling harpoon-and individualistic hunting practices that would have allowed them to hunt plentiful bearded and ringed seals in winter. Yet the fate of the Norse is now widely interpreted as a case study in the often nonlinear consequences of climate change for populations, which can overwhelm even initially successful adaptation (Dugmore et al 2009). The story of human settlement in LIA Greenland is in any case as much a story of Thule resilience as it is Norse collapse, but the Thule experience of climate change has attracted far less attention from archaeologists (Grønnow et al 2011).
In recent years, archaeologists have similarly reimagined other charismatic case studies of collapse in the Common Era, outside of the simple framework of the LIA or MCA, that once seemed like straightforward consequences of climatic change or variability. Many of these involve the purported collapse of hydraulic societies, such as the Tiwanaku in the twelfth-century Andes and Angkor in fifteenthcentury Cambodia ( To cope with the lack of surface water in the upland area of the peninsula, the Maya constructed an intricate system of dams, reservoirs, wells, and canals-as well as terraces and check dams to control soil erosion and leachingthat may have left them vulnerable to longer or more frequent fluctuations in precipitation (Douglas et al 2016). Paleoclimatologists first discerned evidence for a series of catastrophic ninth-and tenthcentury droughts in sediments extracted from Lake Chichancanab, in northern Yucatan. Some paleoclimatologists and some archaeologists argued that these droughts forced the abandonment of Classic Mayan settlements and thereby caused the long-term decline of Mayan culture (Hodell et al 1995, Curtis et al 1996. Other scholars, including archaeologists, pointed out that Maya polities in the driest areas were abandoned last, and that the societal, political, and demographic manifestations of collapse had other plausible social causes. Some critics argued that the Classical Maya did not collapse at all; rather, their culture and population merely became less visible in the archaeological record (McAnany and Yoffee 2010). The persistence of Mayan culture and identity subsequently led other scholars and activists to emphasize the resilience of many Maya communities to even the sixteenth-century Spanish Invasion and the late nineteenth-century Caste War (Braswell and Alexander 2014).
Further work in the region has uncovered more and higher resolution paleoclimate evidencederived most importantly from speleothems-for repeated and severe droughts across the Yucatan Peninsula in and just after the Terminal Classic Period (Kennett et al 2012, Douglas et al 2016, Medina-Elizalde et al 2016. Archaeologists increasingly incorporate these droughts within multicausal explanations for the transformation of Classical Maya society. In these explanations, synergistic pressures including widespread deforestation, owing in part to overpopulation; conflict within and between polities; and the rerouting of trade networks that once passed through the Central Maya Lowlands all rendered many Classical Maya polities increasingly vulnerable to drought (Lucero 2002, Dunning et al 2012, Turner and Sabloff 2012. Some Mayan adaptations to climatic variability seem to have mirrored those of the Norse to climatic cooling, and created similar vulnerabilities. Widespread adoption of waterconservative maize cultivation across the lowlands amid droughts in the third through the sixth centuries CE, for example, resembled the adaptive switch to seal hunting by the thirteenth-and fourteenthcentury Norse (Ebert et al 2017). Owing in part to the inflexibility of Norse and Mayan elites, both strategies seem to have been inadequate and indeed maladaptive in the face of combined climatic, socioeconomic, and political stressors in later decades and centuries (Douglas et al 2015, Ebert et al 2019. In any case, there is limited archaeological evidence for mass starvation and mortality in Classical Maya polities, suggesting that these stressors encouraged adaptive migration and resettlement (Turner and Sabloff 2012). Indeed, archaeologists have recently argued that stratified societies in Greater Amazonia that, in pre-Columbian centuries, relied on extensive infrastructure to control water-such as the Classical Maya-were less resilient in the face of prolonged drought than egalitarian societies with diverse and flexible subsistence strategies founded on polyculture agroforestry (De Souza et al 2019). Because such societies may be less visible in the archaeological (or historical) record-much like the Thule in Greenland-the emphasis by many HCS archaeologists on collapse may reveal a statist bias. Some archaeologists have argued that what may at first appear in the archaeological record as evidence of catastrophe can also be interpreted as an indication of dramatic but successful adaptation to climatic change or variability.
In sum, the archaeological record offers strong evidence that climatic changes provoked profound transitions in many societies during both the Pleistocene and Holocene. Yet there is also ample archaeological evidence for the persistence of communities and especially cultures amid changing climates. With its wide array of proxies relating to both past economies; trade and power networks; and ideologies and cosmologies, the archaeological record offers insights into the risk management strategies that societies adopted in the face of past climate change. Different forms of societies prioritized different response strategies, with migration the leading strategy for inherently mobile populations of pastoralists and foragers, and the least appealing, costliest option for sedentary agricultural societies. The latter could often draw on more or less effective means of buffering against sustained climate changes, which either delayed the catastrophic impact of such changes or allowed agricultural societies to remain resilient sufficiently long for a given stressor to abate. Most episodes of societal collapse did not lead to the total disappearance of societies, let alone cultures, yet definitions of resilience rooted in cultural persistence should be mindful not to ignore the political, socioeconomic, and especially the demographic costs associated with climate-driven collapses in the archaeological record.

HCS findings: history
Historical studies in HCS have long concentrated on periods and places that are well documented by surviving documents and art. This focus partly accounts for the disproportionate attention given by climate historians to Europe, and to some extent its colonies, during the LIA. The widespread adoption in early modern Europe of the printing press coincided with the strengthening of state bureaucracies to encourage the widespread proliferation of documentary evidence for weather and societal responses to weather. Plentiful and diverse documents allow climate historians to identify causal connections between largescale climate changes, local or regional environmental transformations, and human responses at a much higher resolution than is usually possible for scholars in other disciplines (Pfister and Wanner 2021).
By using archives of societies alongside archives of nature, climate historians have established that temperature and precipitation extremes associated with the coldest decades of the LIA decreased the accessibility and in many cases the availability of food and fuel for communities across Europe. The weather conditions that proved most damaging to contemporary agriculture varied across European regions (White et al 2018a). Historians have concluded that in general temperature mattered most for agriculture in Northern Europe, where frost could shorten growing seasons and thereby devastate the autumn harvest. Temperature and precipitation were both equally important for agriculture in Western and Central Europe, where cold in spring and summer, and heavy rains in summer and autumn, typically posed the most severe risks to harvests. In the Mediterranean, springtime droughts were especially damaging, as were bitterly cold winters (Ljungqvist et al 2020).
Historians have found that precipitation extremes lowered the quality and quantity of hay for domestic animals, which not only provided food in the way of dairy and meat but also the commodities and labor that sustained agrarian economies (D'Arrigo et al 2020). Severe winters killed domestic animals directly and delayed when they could consume fresh grass in the spring, which in turn reduced their milk supply (Baten 2001). Winters are less commonly considered by climate historians than other seasons, partly because they are hard to reconstruct using the archives of nature and, in some cases, those of societies. Yet across Europe frigid winters characteristic of the LIA disrupted plant and animal phenology; ruined stores of food and beer; and interfered with shipments of cereals, dairy products, fish, and freshwater (Cavert 2017, Degroot 2018b). Historical studies suggest that climate anomalies and trends also influenced the distribution and accessibility of wild animals used for food and fuel, such as fish and marine mammals (Hacquebord 1999, Ogilvie and Jónsdóttir 2000, Hoffmann 2005, Degroot 2022).
While historians have long accused HCS scholars of environmental determinism, climate historians increasingly argue that populations were far from helpless when weather associated with the LIA helped cause harvest failures. They have shown for instance that citizens within resilient communities, such as the coastal cities of the seventeenth-century Dutch Republic, benefitted from diverse diets and robust traditions of civic charity (Pfister 1978, Dijkman 2018, Degroot 2018c, Curtis and Dijkman 2019. Governments across Italy, Spain, and France adapted in the immediate aftermath of poor harvests by, for example, distributing cereals from and regulating access to granaries; providing grain subsidies (occasionally by taxing the wealthy) and limiting grain prices; and banning grain exports while arranging for emergency imports (Bauch 2019, Degroot et al 2021).
Historians have demonstrated that European governments and communities also pursued long-term strategies that either purposefully or unwittingly helped them adapt to the climatic trends of the LIA. Farmers in Finland deserted newly unproductive farms or adopted new crop varieties and cultivation strategies in the wake of particularly cold decades of the LIA, for example (Huhtamaa and Helama 2017a, 2017b). Across Europe, farmers diversified their crops by adopting rye, oats, or barley, which tolerate cold and wet conditions, or took up animal husbandry and even apiculture (Landsteiner 2001, Mrgic 2011, Tello et al 2017. Dutch farmers and guilds developed, refined, or exploited new technologies and transportation networks that allowed them to move agricultural commodities from domestic zones of production to centers of consumption in nearly any weather. Dutch merchants deployed newly efficient ships to dominate much larger trade networks that integrated grain markets and thereby buffered populations from local grain shortages (De Kraker 2017, Degroot 2018c. Nevertheless, climate historians have long devoted special attention to climatic anomalies during the LIA that overwhelmed communal capacities for resilience and adaptation. Historians contributed to early efforts to link climate change to the disappearance of Norse settlements in western Greenland, for example, then joined archaeologists and paleoscientists to reinterpret what initially seemed like a straightforward case of climate-driven collapse (Utterström 1955, Barlow et al 1997, Dugmore et al 2012, Jackson et al 2018a. Historians have since connected so many other disasters across mainland Europe to the climate of the LIA that several studies now identify periods when temperature and precipitation anomalies, associated in some cases with explosive volcanic eruptions, helped provoke continent-wide subsistence crises (Degroot 2018b).
Historians have, for example, linked European subsistence crises to low temperatures or precipitation extremes in the mid-thirteenth century, the early fourteenth century, the early fifteenth century, and the late sixteenth century (Jordan 1996, Campbell 2016, Camenisch et al 2016a, 2016b. Influential studies argue that the seventeenth, eighteenth, and nineteenth centuries were all bookended by periods of extreme weather that helped raise food prices and thereby unleashed fatal synergies across Europe, and occasionally in European colonies (Davis 2002, Pfister 2007, Behringer 2017, Parker 2018. Some historians argue that, in many of these periods, economic, political, and demographic pressures created sources of vulnerability that greatly worsened the synergistic social consequences of frigid, wet, or dry weather. Influential studies identify overpopulation, endemic warfare, and reductions in trade or purchasing power-brought about partly through deflation or inflation-as contributors to vulnerability. Some also emphasize the shortcomings of contemporary governments, which either would not or could not effectively organize to relieve high food prices (Davis 2002, Parker 2013. Climate historians have therefore consistently argued that, even in recent centuries, fluctuations in temperature and precipitation that were far smaller in magnitude than those already caused by global warming nevertheless plunged large and powerful polities into crisis. Other historians claim however that this conclusion is based more on crude correlation than causation; assigns undue importance to exogenous versus endogenous causes for subsistence crises; or reverses the correct chain of causation by for example conceptualizing climate change as a cause of war, rather than war as a cause of vulnerability to climate change (Warde 2015, Kreike 2021. Some climate historians argue that the proliferation of studies linking the LIA to harvest failures and societal crises reflects how studies are designed and the evidence they use, more than the relationships between past climate change and social outcomes that were most common or consequential (Degroot et al 2021).
Nevertheless, in recent decades historians have uncovered many similar connections between LIA climate anomalies and subsistence crises beyond Europe. Historians have argued for example that drought (Cook et al 2010a) and perhaps cooling across China and its northern borderlands repeatedly ruined harvests, reduced tax revenues, and inspired revolts against the Ming Dynasty, while encouraging raids by pastoralists that ultimately led to the seventeenth-century ascension of the Qing Dynasty (Brook 2010). They have shown that the rising dependence of the sixteenth-century Ottoman Empire on winter crops grown on semi-arid farmland left it vulnerable to severe drought at the end of the century, which combined with inflation and military requisitions to devastate the countryside. Migrating peasants spread epidemics of bubonic plague and anthrax, and many gathered in rebellious armies that eventually threatened the survival of the empire (White 2011). Historians have claimed that precipitation extremes ruined harvests and thereby undermined empires in sixteenth-century sub-Saharan Africa, inspiring waves of migration just as the European slave trade entrenched itself (Miller 1982, Webb 1995. They have also argued that cooling, drought, and pluvials ruined crops and thereby undermined contemporary colonization efforts in North America, while weakening Indigenous polities that might otherwise have offered stiffer resistance to European settlers (White 2017, Skopyk 2020.
The most compelling metanarrative advanced by climate historians may therefore be that the LIA provoked one or more 'global crises,' when cooling and precipitation extremes caused prolonged food shortages and political upheaval on continental, even hemispheric scales. Yet historians have started to move well beyond their longstanding focus on subsistence crises in agrarian empires. For example, new histories consider not only-or not at all-how climate anomalies helped provoke war, but also how they affected the course of wars already underway (Wickman 2018b, Degroot 2020. Since most wars have a victor and a loser, such histories reveal that the telltale weather of LIA cold periods in different regions could offer both advantages and disadvantages to contemporary populations (Degroot 2014, Wickman 2015. Climate historians increasingly consider how populations found ways to thrive during the LIA, often in regions-such as the open sea or across Indigenous North America-that have been largely ignored in previous histories (Endfield 2012, Parker 2013, Wickman 2018a, Degroot 2018c. Some historians join scholars in other disciplines to identify strategies that diverse populations followed to exploit-or at least cope with-climatic anomalies and trends (Adamson et al 2018, Degroot et al 2021. Historians have also emphasized that some populations and institutions within societies were resilient or adaptive in the face of climate change despite-or because of-the vulnerability of other populations (Soens 2018, Van Bavel et al 2018. New histories uncover the agency of people and animals confronted with climate change, and thereby reveal how environments seemingly rendered less productive by the LIA could still be exploited with unprecedented intensity by human communities (White 2014, Degroot 2022). An emerging metanarrative in climate history therefore explicitly contradicts the field's longstanding theme of global crisis by emphasizing diverse, contingent, and occasionally counterintuitive relationships between climatic anomalies and human responses.
While climate historians have long focused on the LIA, which is rich in both historical and paleoclimate data, they increasingly consider earlier periods of climate change. New histories, for example, consider harvest failures and other social consequences of droughts, pluvials, and temperature anomalies during the MCA (Kiss 2019, Kiss and Pribyl 2020, Pribyl 2020. Recently, historians have uncovered and refined case studies of ancient societies that coped with and occasionally succumbed to regional droughts and cold spells (Haldon 2016, Erdkamp et al 2021. The LALIA is emerging as a period of special interest for climate historians, though confusion over its likely magnitude, duration, and spatial extenta partial consequence of the limited textual record of the sixth century CE-has inspired competing 'maximalist' and 'minimalist' interpretations of contemporary social responses (Sessa 2019). Maximalist interpretations draw on fragmentary archaeological, paleoenvironmental, and especially historical evidence to identify large-scale connections between cooling, the spread of plague, mass mortality, and the collapse of societies, especially the Western Roman Empire (McCormick et al 2012, Büntgen et al 2016, Harper 2017. Minimalist accounts emphasize resilience by concentrating on local relationships between climatic anomalies and social responses that can be clearly identified using surviving sources (Newfield 2016, Haldon et al 2018a, Erdkamp 2021. This division is, in some respects, an exaggerated version of the tension between metanarratives developed by climate historians for the later LIA. It reflects differences between historians over the appropriate integration of historical methods and sources with those of other disciplines, as well as differences over the value of correlation in suggesting causation. It reveals another theme in climate history: the sheer difficulty of establishing widely agreed-upon connections between climate changes and human actions, which have historically unfolded across very different scales in time and space. Similar challenges confront scholars-including increasingly historianswho attempt to use far more abundant evidence than is available for pre-industrial periods to identify links between anthropogenic global warming and, for example, migration and conflict (Selby et al 2017).
One of the most important developments in climate history has been a growing 'turn' towards studies that consider the cultural dimensions of climate change (Williamson 2020). Historians and other scholars have long argued that decadal changes in the subjects and even the colors of European paintings reflected the climatic trends of the LIA, or at least the volcanic eruptions that caused its coldest years (Sager 2006, Behringer 2010, Degroot 2018c. They have also claimed that popular understandings of the causes for extreme weather, in central Europe in particular, contributed to the persecution of Jews and accused witches during some of the chilliest decades of the LIA, although other historians have disagreed with these claims (Behringer 1999, Oster 2004, Bell 2008, Leeson and Russ 2018. Recently, some of the most compelling historical work on the cultural dimensions of past climate change examines how and why severe weather was remembered, forgotten, and used to justify political or military projects (Oberholzner 2011, Sundberg 2015, Zilberstein 2016, Skopyk 2020.
Cultural histories of climate change can be problematic, however, when they assume broad connections between modest climatic changes and cultural transformations that had more plausible social, economic, and political causes (Blom 2019). The impact order model has helped historians confront such determinism, at the cost of encouraging the assumption that cultural responses to climate change were inevitably filtered through more direct agricultural and economic responses (Ljungqvist et al 2020). Yet recent histories find increasingly diverse and plausible connections between periods of climate change such as the LIA and local cultural expressions, including in architecture, technological development, religious practice, and scientific theory (Degroot 2018c, Barnett 2019, Ray 2019, Pluymers 2020, Fisher 2021, Li et al 2021. Climate historians have therefore shown that the human consequences of even modest climate changes can be exceptionally far-reaching, affecting nearly every aspect of lived experience, albeit often in subtle ways that can be difficult to distinguish from other historical influences as they unfolded gradually over time.

HCS findings: geography and economics
Most studies in HCS, regardless of their discipline, start when scholars identify one or more correlations between climatic and social change. Many scholars who follow the cause-of-effect method-including nearly all historians-use correlations as a starting point for establishing causation in case studies that are typically constrained in time or space. For them, causation can be identified through inferences based on qualitative as well as quantitative sources, with the caveat that it can never be established with certainty (White and Pei 2020).
Geographers and economists, however, have long pioneered the alternative effect-of-cause method, in which correlations between quantified sources are often not the beginning but the end of research projects (figure 3). If enough statistically significant correlations exist between two datasets, those correlations according to the effect-of-cause method can reveal causation more reliably and precisely than inferences based on qualitative evidence. The challenge for geographers and economists has been to identify evidence for historical changes in human populations that either is quantified or that can be quantified on spatiotemporal scales large enough for them to identify many possible correlations with climatic trends (Degroot 2018a).
This challenge has motivated geographers and some economists to focus on whether wars were historically caused by trends in precipitation and drought. Wars can have precise, quantitative definitions, meaning that a conflict must cross a threshold of combat-related deaths to be labeled a war. Using such definitions, geographers and economists have assumed that for regions with a rich record of textual evidence, the spatial extent, duration, and destructiveness of wars can be identified reliably enough to chart their annual regional frequency over decades, centuries, and in some cases even millennia. They then compare fluctuations in the frequency of war with trends in reconstructed temperature or precipitation. Chinese geographers pioneered these methods by identifying significant correlations between drought, cooling, and conflict across imperial China (Zhang et al 2006(Zhang et al , 2010. Economists and geographers then found similar correlations in Europe during the LIA, and in sub-Saharan Africa during the period of recent anthropogenic warming (Burke et al 2009, Lee et al 2019. Together, these studies suggest that cold, dry conditions were perilous for pre-modern states, but also that growth in warm, wet decades could exacerbate destabilizing resource shortages when climatic conditions changed (Fang et al 2019).
In recent years, geographers and economists have used statistical methods, such as wavelet analysis and Granger causality tests, to attempt to confirm the causation implied by significant correlations. New work has employed superposed epoch analysis to account for the conclusions of many cause-of-effect studies, which identify lags between climatic and social changes that were caused both by societal adaptations and by the systematic complexities of cascading and compound climate change impacts (Gao et al 2021). Some of these methods seem to establish that climate did in fact control the frequency of war and even the likelihood of dynastic collapse in distinct regions, environments, and socioeconomic or political contexts (Gao et al 2021, Zhang et al 2022. However, scholars have used statistical methods, such as Bayesian modeling, that compensate for some of the shortcomings and uncertainties in war frequency datasets to suggest that correlations detected in earlier studies of, for example, war in Europe did not actually exist (Peregrine 2020, Carleton et al 2021a, 2021b.
Nevertheless, many scholars now attempt to find correlations between climatic trends and social changes that are in many cases harder to identify, let alone quantify, than war. Studies focusing on imperial China, for example, have found significant correlations that purportedly reveal causation between climate change and agricultural output, migration, economic performance, technological innovation, and even cultural efflorescence (Jin 2002, Carleton and Hsiang 2016, De Dreu et al 2018, Lee and Yue 2020. An advantage of such work is that statistical relationships in the past can, in theory, be used to model future connections between climatic and social changes. Yet critics in disciplines other than geography and economics argue for example that scholars have used effect-of-cause methods naively, to quantify what is difficult or impossible to quantify on the basis of surviving evidence (a tendency known as the McNamara Fallacy). Critics also claim that many such studies misuse accessible datasets, compiled in most cases by historians, that were never meant to provide comprehensive statistical information (an observational bias known as the Streetlight Effect) (Adams et al 2018, Van Bavel et al 2019, Degroot et al 2021.
Still, studies by geographers and economists do provide a means by which case studies unearthed by historians or archaeologists can be evaluated systematically across large spatial and temporal scales. Their most important finding so far is that climate anomalies have been inherently destabilizing for societies already suffering from considerable political or socioeconomic stress. While this conclusion seems to echo those of many cause-of-effect studies in history and archaeology, its implications are broader, as it is not limited to specific case studies. At present, publications that purport to show similar relationships between climatic trends and social changes that do not directly involve conflict are not convincing, although they may suggest correlations that can be used as starting points for interdisciplinary scholarship.

Consilient research teams and coupled modeling
While scholars in different disciplines have undertaken HCS studies in distinct ways, the field's divisions between disciplines are not always obvious. HCS scholars may, for example, have degrees in several disciplines that are relevant for reconstructing climate changes and identifying their influence on human populations. Close bonds have long connected scholars in two or more of these disciplines. The first HCS studies were developed through the partnership of a geographer with a paleoclimatologist, and paleoclimatologists have long had fruitful collaborations with archaeologists.
Similar connections do not bridge scholars in all the disciplines most involved in HCS studies, however, and many publications in the field consequently misuse climate reconstructions or simplistically assume human responses to climate change. Scholars in disciplines that are often associated with the humanities-such as history-have been largely isolated from paleoscientists, owing partly to the priority given to single-authored books for promotion in these fields; the lack of common standards of evidence, approaches to uncertainty, or understandings of causality between disciplines; and the systematic subordination in collaborative climate scholarship of data that cannot be quantified and modeled (Degroot et al 2021). Differences in vocabulary, theory, publication and presentation venues as well as a mismatch between publication incentives create further barriers to collaboration. Recently, climate historians have called for more collaborative, 'consilient' approaches to HCS scholarship that involve the development of research projects by teams with equal numbers of participants in relevant disciplines (Newfield andLabuhn 2017, Haldon et al 2018b). Such partnerships can require computational methods for aligning distinct forms of data across radically different scales in time and space. Consilient approaches can slow the career progression of some participating scholars, especially those in academic departments that are not structured to support interdisciplinary research. Yet they can help scholars better account for the complexity of both environmental and social systems, and thereby to identify more convincing causal connections between climatic and societal histories (Degroot et al 2021).
Many HCS scholars believe that comparisons between climate and social change in the past cannot provide reliable, transdisciplinary lessons for enhancing present-day sustainability without a robust, mechanistic understanding of underlying socio-environmental systems. Mechanistic models allow scientists to formalize our abstract assumptions and intuitions so they can be tested, communicated, and updated as new information arises. A community-wide approach to computational modeling and model-data integration through consilient research teams has been central to the success of the climate sciences over recent decades (Edwards 2010). However, this approach has yet to diffuse to the human historical sciences, in large part because of the lack of standardized generative models of past societies (Romanowska et al 2021).
Complex 'integrated assessment models' , rooted in economic theory, are commonly used for such purposes in contemporary sustainability studies but have high data requirements and lack the flexibility and dynamism needed to model diverse societies over their long-term history (Rounsevell et al 2014, Donges et al 2017, Schill et al 2019. These macroeconomic models operate in relative isolation from their environmental counterparts, for example by using the outputs of previously-run climate models as external inputs rather than running the two in tandem. Figure 7. Conceptual diagram of a consumer-resource network, after Qubbaj et al (2015), with a network of cities harvesting and exchanging food, water, and energy from their rural surroundings. The behavior of these systems is driven not only by the dynamics of each component in isolation (e.g. population growth in cities, environmental degradation in agricultural regions) but by the various links among these components. Indirect links, such two distant agricultural regions harvested by cities connected by trade, are often as important for system behavior as the direct links visible in the historical and archaeological records.
Alternatively, 'fully coupled' models capture the rich two-way interactions among 'social' and 'natural' subcomponents, exhibiting emergent behaviors such as thresholds, time lags, and feedbacks that are absent from loosely coupled alternatives (Robinson et al 2018). A promising coupled modeling approach represents human societies as complex 'consumerresource networks,' in which interconnected urban systems continuously consume and exchange food, water, energy, and other resources harvested from their rural surroundings (figure 7) (Wilson 2008, Qubbaj et al 2015, Dermody et al 2018, Dolfing et al 2019, O'Dwyer 2020. Although originally rooted in theory from ecology and geography, this basic framework has proven ideal for HCS studies in multiple disciplines including demography (Lee and Tuljapurkar 2008, Puleston and Tuljapurkar 2008, Lee et al 2009, ecological economics (Anderies 2006, Motesharrei et al 2014, Kuil et al 2019, and cultural evolution (Turchin et al 2013, Weinberger et al 2017. Consumer-resource models highlight the crucial role of scale and interconnectedness for understanding complex social and environmental systems in the past. These systems are nonlinear; for example, the simulated social response to a drought or flood may have little to do with the magnitude of the event itself, but rather difficult-to-observe details of the society's demography, technology, and social institutions (Anderies 2006, Kuil et al 2019. Often, largescale social responses at one time and place reflect the lingering impacts of seemingly minor shocks in the distant past (Eppinga et al 2021) or faraway resource systems (Qubbaj et al 2015). Spatial and social networks like roads or exchange systems can either dampen or heighten environmental impacts depending on precise details of the network's structure (Dolfing et al 2019). In spite of this potential complexity, these modeling approaches are powerful tools to provide us a 'crude look at the whole' (Gell-Mann 1995)-a relatively simple approximation a system's subcomponents and their mutual interactions-that enable consilient research teams to produce more actionable, real-world insights than would a narrow focus on each discipline in isolation.

HCS for the future
Many HCS scholars study the past partly to uncover universally applicable relationships between climatic and social change. In identifying these relationships, many hope to inspire climate activism or contribute to the development of climate policy. Despite this common motivation, scholars in different disciplines have developed distinct methods of using the past to encourage or shape action in the present.
Economists, geographers, and paleoscientists, for example, may draw attention to troubling relationships between climatic and social trends that they have identified in their statistical work. They quantify these relationships and thereby motivate activism that aims, for example, to reduce the likelihood of resource wars in a hotter future. Archaeologists, geneticists, and historians, by contrast, are more likely to develop narratives that chronicle the influence of climate change on a past population, and thereby reveal the extent to which rapid change and warming could shape human affairs. Narratives that explain the collapse or (less commonly) the survival of societies confronted with climate change have attracted considerable public attention (e.g. Diamond 2011) and the concept of collapse is a powerful trope in popular media.
To influence policy, geographers and economists have pioneered econometric methods that use correlations between past climatic and social trends to forecast continued correlations, and therefore causal connections, between these trends in the future (Hsiang et al 2013). Scholars in many disciplines have modeled past relationships between climatic and social changes, occasionally in order to identify which social responses to future climate change will be most destructive or most likely to occur (Nelson et al 2016, Bauch 2020. Archaeologists and historians have also sought to identify common characteristics in populations that made them either resilient or vulnerable in the face of past climate changes, partly to suggest which qualities governments should emulate or avoid today ( By drawing on historical examples of civilizational collapse, they have also contributed to studies that suggest scholars and policymakers should more seriously consider the possibility that ecosystems and in turn present-day states will be unable to survive rapid anthropogenic warming (Reichstein et al 2021, Kemp et al 2022. Some have also identified relationships between past climatic and social changes that scholars in other fields have largely ignored, but may be important in the future (Degroot 2022).
Nevertheless, it can be very difficult for scholars to derive policy-relevant conclusions from HCS scholarship, and then to publish these conclusions in formats that are sensible to policymakers and advisors. This is especially true for scholars in humanistic disciplines-such as history-that typically emphasize contingency and historical particularity. Recently, scholars have drawn attention to the occasionally haphazard and often generalized way that HCS studies draw lessons from inferred longterm changes in ancient societies to inform shortterm policy development in today's radically different world (Jackson et al 2022, Tubi et al 2022. Some HCS scholars even argue that discontinuities between past and present social and climatic contexts are so great that HCS studies can provide few if any lessons for the future (Ray 2019). Yet even the recent past provides many examples of complex communities and societies that coped with profound local and regional climate changes, and ancient populations confronted global changes of far greater scale (D'Alpoim Guedes et al 2016). The past inevitably offers an imperfect guide to the future, yet for climate policy-as in other fields-many HCS scholars believe it provides lessons that should not be discounted (Jackson et al 2018b, Van Bavel et al 2020, Jackson et al 2022. HCS scholarship, however, is generally underrepresented in assessments of the future of climate change and in policy development. When the field is consulted to help policymakers forecast or prepare for the future, statistical studies by geographers and economists are often given preference over qualitative work by scholars in other disciplines (Holm and Winiwarter 2017). Consilient collaborations that involve social scientists and policy advisors are currently being assembled that may help HCS scholars overcome these challenges, and thereby offer unique perspectives on the human future in a warming world (Thomas et al 2019).

Data availability statement
No new data were created or analysed in this study.