A 1,200-year perspective of 21st century drought in southwestern North America Connie A. Woodhousea,b,1, David M. Mekob, Glen M. MacDonaldc, Dave W. Stahled, and Edward R. Cooke aSchool of Geography and Development, Harvill 409, bLaboratory of Tree-Ring Research, 105 West Stadium, University of Arizona, Tucson, AZ 85721 cDepartment of Geography, 1255 Bunche Hall, University of California, Los Angeles, Los Angeles, CA 90095 dDepartment of Geosciences, 113 Ozark Hall, University of Arkansas, Fayetteville, AR 72701 and eLamont-Doherty Earth Observatory of Columbia University, Palisades, NY 10964 Edited by B. L. Turner, Arizona State University, Tempe, AZ, and approved December 11, 2009 (received for review September 28, 2009) A key feature of anticipated 21st century droughts in Southwest North America is the concurrence of elevated temperatures and increased aridity. Instrumental records and paleoclimatic evidence for past prolonged drought in the Southwest that coincide with elevated temperatures can be assessed to provide insights on tem- perature-drought relations and to develop worst-case scenarios for the future. In particular, during the medieval period, ���AD 900��� 1300, the Northern Hemisphere experienced temperatures warmer than all but the most recent decades. Paleoclimatic and model data indicate increased temperatures in western North America of approximately 1 ��C over the long-term mean. This was a period of extensive and persistent aridity over western North America. Paleoclimatic evidence suggests drought in the mid-12th century far exceeded the severity, duration, and extent of subsequent droughts. The driest decade of this drought was anomalously warm, though not as warm as the late 20th and early 21st centu- ries. The convergence of prolonged warming and arid conditions suggests the mid-12th century may serve as a conservative ana- logue for severe droughts that might occur in the future. The severity, extent, and persistence of the 12th century drought that occurred under natural climate variability, have important implica- tions for water resource management. The causes of past and future drought will not be identical but warm droughts, inferred from paleoclimatic records, demonstrate the plausibility of exten- sive, severe droughts, provide a long-term perspective on the ongoing drought conditions in the Southwest, and suggest the need for regional sustainability planning for the future. climate change ��� water resources ��� paleoclimatology ��� medieval period Calready limate-change projections clearly indicate what observations suggest: Temperatures everywhere will be warmer in the future due to anthropogenic activities. General circulation models (GCMs) project continued warming, with annual tem- peratures 3���5 ��C above current levels by the end of the century (1). As previous articles in this Special Feature have discussed, warming temperatures, even without reductions in precipitation, will have far-reaching impacts on hydrologic sustainability in the Southwest. Twenty-first century droughts will occur under warmer temperatures with greater rates of evapotranspiration than occurred during the major droughts of the 20th century. Warming may also directly and indirectly increase the propensity for droughts in the Southwest (2���4). However, major 20th century droughts pale in comparison to droughts documented in paleo- climatic records over the past two millennia (5). Thus, warm droughts of the prehistoric past might provide evidence useful in understanding the current climatological changes, and for providing scenarios for worst-case droughts of the future and evidence of hydroclimatic responses in the Southwest to warmer climatic conditions. This paper examines recent temperature-drought relations and analyzes paleoclimatic data documenting droughts persisting for periods of a decade or more, develops evidence for drought linkages with elevated temperatures, and identifies ���worst-case��� scenarios for warm-climate drought to place the recent episode of drought in the Southwest in a long-term context. As the current early 21st century drought has occurred with elevated tempera- tures, warm-period paleo droughts may well be a preview of what can be expected for the future. The recent prolonged drought has already had significant impacts in the arid to semiarid Southwest. Currently, overallocated water resources are being further stressed by increased demands due to population growth, tribal settlements, changes in land use, recreation needs, and mandated requirements for instream flows for ecosystem func- tioning and endangered-species preservation (6���9). As a result, many water-supply systems have become increasingly vulnerable to drought impacts. The recent drought has underscored the cri- tical need for sustainable water-resource management and devel- opment (10). Such strategies should be informed by as long and complete a record of drought behavior and impacts as possible. Warm Droughts in the Southwest: Past Droughts as Analogues for the Future? The Role of Temperature. Elevated temperatures can have direct, local effects on drought as well as impacts on circulation features that promote large-scale droughts. Southwestern droughts are, typically, accompanied by above average temperatures because of factors such as subsidence, a lack of cloud cover, drying soils, and reduced evapotranspiration (e.g., 11���13). Major 20th century droughts, including the 1930s and 1950s, have occurred during periods of elevated temperatures, with persistence of high pres- sure leading to surface heating and drying in both winter and summer (11, 14, 15) (Fig. 1) and storm tracks displaced around the drought region (16). However, droughts do not always coin- cide with above average temperatures (17), as exemplified in the U.S. Southwest by the drought at the start of the 20th century (Fig. 1). Global or hemispheric warming may also strongly impact Southwest drought indirectly through influences on global sea surface temperatures (SSTs) and ocean/atmosphere dynamics. Increased radiative heating over the tropical Pacific has been shown to enhance the development of La Ni��a-like conditions that promote drought in the Southwest (4, 5, 18). It has been suggested that the influence of global warming on the western tropical Pacific and Indian Oceans may already be detectable, and along with cool SSTs in the eastern tropical Pacific, may have been a cause of drought conditions at the turn of the 21st century that affected regions including southwestern North America (19). One projected (and possibly already detected) result of global warming is an extension of the poleward arm of the Hadley cell that will cause an expansion of the area under the drying Author contributions: C.A.W., D.M., G.M.M., D.W.S., and E.C. analyzed data D.M., G.M.M., D.W.S., and E.C. contributed new reagents/analytic tools and C.A.W., D.M., G.M.M., D.W.S., and E.C. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1To whom correspondence should be addressed. E-mail: conniew1@email.arizona.edu. This article contains supporting information online at www.pnas.org/cgi/content/full/ 0911197107/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.0911197107 PNAS ��� December 14, 2010 ��� vol. 107 ��� no. 50 ��� 21283���21288 SUSTAINABILITY SCIENCE SPECIAL FEATURE

influence of subtropical high pressure (2, 20). Whereas some of these large-scale responses to warming may not have operated in the past others, such as SSTs anomalies in the tropical oceans, have been critical drivers. Past droughts best suited as analogues for the future are those accompanied by hemispherical tempera- ture changes favoring drought-inducing circulation and directly amplifying regional drought conditions and impacts. Warm Paleodrought. Paleoclimatic data for southwestern North America provide extensive documentation of past droughts (21, 22). Records collectively suggest a broader range of hydro- climatic variability than contained in instrumental records, parti- cularly with respect to drought extent, duration, and severity. Several notable droughts extended across much of western North America, including severe and sustained droughts in the late 16th century and the medieval period, between 900���1300 AD (23���25). In this period, episodes of extensive severe drought are documen- ted by a variety of proxy data, but most dramatically by evidence of trees rooted in lakes and river courses in the Sierra Nevada and northwestern Great Basin (26, 27). These droughts appear to have exceeded the duration and magnitude of any subsequent droughts in western North America (5, 25). Whereas the medieval period is now acknowledged as a time of increased aridity over western North America, it has more generally been known as a period of warmer temperatures, espe- cially over Europe (28, 29). An effort has been made to document the degree to which global and hemispheric temperatures were elevated at this time using a wide variety of proxy records, and with an emphasis on understanding the low-frequency compo- nent of variability (1, 30, 31). A recent analysis of a number of different proxy temperature records suggests that Northern Hemisphere decadal-scale averages over land may have been as much as approximately 0.2���0.4 ��C above the 1850���2006 mean from roughly 950���1150 AD (32). The medieval warming is, how- ever, markedly exceeded by late 20th and early 21st century warming, as temperatures now stand more than 0.8 ��C above the 1850���2006 mean (32). Ocean/atmosphere teleconnections provide a plausible causa- tive link between hemispheric-scale warm temperatures and drought in the Southwest during the medieval period. Associa- tions between SSTs and Southwestern drought during this period have been explored with paleoclimatic data and modeling (4, 33���35) and although the paleoclimatic data that document Pacific Ocean conditions during the medieval period are not in total agreement, most show temperatures in the eastern Pacific indicative of cool El Ni��o/Southern Oscillation, or La Ni��a-type conditions (22). More unequivocal evidence exists for a warm North Atlantic (36). Recent modeling efforts, assuming cool Pacific and warm Atlantic SSTS, have replicated the main fea- tures of medieval drought in North America documented in paleoclimatic data (36). It is worth noting that droughts of the 1950s and of recent years were both accompanied by cool Pacific and warm Atlantic SSTs (37). Although likely not matching the magnitude of the recent in- creases in global temperatures, the increased large-scale hemi- spheric warming in medieval times coincided with widespread and persistent aridity across the Southwest. On a regional scale, paleoclimatic data indicate that similar to the instrumental per- iod, warm and dry spells often concur in the Southwest, including during this period (13). Is it appropriate then to consider a medi- eval drought as a possible, although conservative (with respect to temperature), analogue for future warm droughts? The root causes of warming for the medieval period, increased solar irra- diance coupled with decreased volcanic activity (38, 39), and in recent decades, anthropogenic activities with some contribution from solar irradiance (1), are not identical. Although important differences must be acknowledged���for example, the causes and the amplitudes of the warming, and the probable impacts of land cover change on temperatures���the medieval droughts can provide some direct evidence of the Southwest hydroclimatic re- sponse to warming and a plausible, but conservative, worst-case scenario to be considered in sustainable water-resource planning. Medieval Drought and Temperatures in Southwestern North America The medieval period was characterized by widespread and re- gionally severe, sustained drought in western North America. Proxy data documenting drought indicate centuries-long periods of increased aridity across the central and western U.S. (Fig 2F) (25, 22 ). In the Colorado and Sacramento River basins, recon- structions show decadal periods of persistently below average flows during several intervals including much of the 9th, 12th, and 13th centuries (40���42) (Fig 2E). The 12th century episode, also reflected in precipitation and drought extent (13, 25, 43, 44), was particularly severe and persistent and was associated with a peak in solar irradiance and nadir in volcanic activity (4) (Fig. 2A). Most of these paleohydrological records primarily reflect winter and spring precipitation. Proxy records that document sum- mer precipitation are much less common and, of those that do exist, some suggest wetter summers during the medieval period (22, 45���47), whereas others indicate decadal variability of both drought and wetness (48). The temperature signal of the medieval period, though rela- tively strong in averages over the Northern Hemisphere (32) (Fig. 2B), is more complex at the regional scale (29). In contrast to paleohydrological records, there are fewer high-resolution paleotemperature records in the Southwest and evidence for anomalous medieval warmth in this region is less comprehensive (5). Tree-ring reconstructions of temperature for the Southwest suggest warmer temperatures for at least portions of the medieval period (13, 29, 49���51). These reconstructions usually represent growing-season temperatures and, because of limitations of the paleoclimatic indicators generally do not preserve centennial- scale variations (52, 53), at least on these regional scales. Along with evidence for multiyear periods of enhanced temperatures approaching 1.0 ��C during some intervals of the medieval period, records also indicate periods of normal to below average tem- peratures at other intervals (Figure 2D). Proxy records are con- sistent, however, in supporting periods of elevated warmth in the medieval period that coincide with periods of severe and wide- spread drought. At multidecadal and longer timescales, evidence from treeline, glacier, and chironomid studies suggests southwestern North 160 180 200 220 240 260 16.5 17 17.5 18 18.5 19 1900 1920 1940 1960 1980 2000 mm degrees C 120 140 160 180 200 2 2.5 3 3.5 4 4.5 1900 1920 1940 1960 1980 2000 mm degrees C Fig. 1. Total seasonal precipitation and mean seasonal temperature averaged over Colorado, Utah, New Mexico, and Arizona (17) five-year running means, 1900���2008. Precipitation in Blue Line (Horizontal Line is the average), temperature in Brown. Cool season (November���March), Top. Warm season (May���October), Bottom. Shading indicates periods of below aver- age precipitation. 21284 ��� www.pnas.org/cgi/doi/10.1073/pnas.0911197107 Woodhouse et al.