Reclaiming freshwater sustainability in theCadillac DesertJohn L. Saboa,1, Tushar Sinhaa,2, Laura C. Bowlingb, Gerrit H. W. Schoupsc, Wesley W. Wallenderd,e, Michael E. Campanaf,Keith A. Cherkauerg, Pam L. Fullerh, William L. Grafi, Jan W. Hopmansd, John S. Kominoskij,3, Carissa Taylork,Stanley W. Trimblel, Robert H. Webbm, and Ellen E. WohlnaFaculty of Ecology, Evolution and Environmental Sciences, School of Life Sciences, Arizona State University, PO Box 874501, Tempe, Arizona,85287-4501; bDepartment of Agronomy, Purdue University, West Lafayette, IN 47907-2054; cDepartment of Water Resources Management,Delft University of Technology, 2600 GA, Delft, The Netherlands; dDepartment of Land, Air and Water Resources, University of California,Davis, CA 95616-8628; eDepartment of Biological and Agricultural Engineering, University of California, Davis, CA 95616-8628; fDepartmentof Geosciences, Oregon State University, Corvallis, OR 97331-5506; gDepartment of Agricultural and Biological Engineering, PurdueUniversity, West Lafayette, IN 47907-2093; hUS Geological Survey, Southeast Ecological Science Center, Gainesville, FL 32653; iDepartment ofGeography, University of South Carolina, Columbia, SC 29208; jDepartment of Forest Sciences, University of British Columbia, Vancouver, BC,Canada V6T 1Z4; kSchool of Sustainability, Arizona State University, Tempe, AZ 85287-5502; lDepartment of Geography, University ofCalifornia, Los Angeles, CA 90095; mUS Geological Survey, Tucson, AZ 85719; and nDepartment of Geosciences, Warner College of NaturalResources, Fort Collins, CO 80523-1482

Edited by Glen M. MacDonald, University of California, Los Angeles, CA, and accepted by the Editorial Board November 10, 2010 (received for review July6, 2010)

Increasing human appropriation of freshwater resources presents a tangible limit to the sustainability of cities, agriculture, and eco-systems in the western United States. Marc Reisner tackles this theme in his 1986 classic Cadillac Desert: The American West and ItsDisappearing Water. Reisner’s analysis paints a portrait of region-wide hydrologic dysfunction in the western United States, sug-gesting that the storage capacity of reservoirs will be impaired by sediment infilling, croplands will be rendered infertile by salt, andwater scarcity will pit growing desert cities against agribusiness in the face of dwindling water resources. Here we evaluate theseclaims using the best available data and scientific tools. Our analysis provides strong scientific support for many of Reisner’s claims,except the notion that reservoir storage is imminently threatened by sediment. More broadly, we estimate that the equivalent ofnearly 76% of streamflow in the Cadillac Desert region is currently appropriated by humans, and this figure could rise to nearly 86%under a doubling of the region’s population. Thus, Reisner’s incisive journalism led him to the same conclusions as those rendered bycopious data, modern scientific tools, and the application of a more genuine scientific method. We close with a prospectus forreclaiming freshwater sustainability in the Cadillac Desert, including a suite of recommendations for reducing region-wide humanappropriation of streamflow to a target level of 60%.

Manifest Destiny and thewestward expansion of Eu-ropean civilization in theUnited States during the 19th

century were predicated on an adequatefreshwater supply. The assumption ofadequate freshwater in the westernUnited States was justified by the pre-vailing view of hydroclimate, which in-cluded a theory that agriculture wouldstimulate rainfall, or “rain would followthe plow.” Early stewards of freshwaterresources—like John Wesley Powell—warned that the American West wasa desert, only a small fraction of whichcould be sustainably reclaimed (1).* No-tably, Powell remarked that irrigationwould be required in the arid region westof the 100th meridian, to make the par-cels provided by the Homesteading Actlivable (3). Indeed, irrigation was neces-sary to create a sustainable society in thewestern United States. Today dams, irri-gated agriculture, and large cities are thehallmark of western US landscapes.There are more than 75,000 dams in theUnited States, and the largest five reser-voirs by storage capacity lie west of the100th meridian. The storage capacity ofUS reservoirs increased steadily between1950 and 1980—from 246 to 987 km3

(4)—and the beginning of these “go-goyears” of dam building (5) coincides withthe US “baby boom” (roughly 1943–1964). Since that time, there has been anexodus from east to west: population ofthe 15 largest eastern US cities has de-clined by an average of 51% but increasedby 32% in western cities (6, 7). Similarly,although 74% of the cropland in the co-terminous United States lies in the east-ern United States, 68–75% of the revenuefrom vegetables, fruits, and nuts derivesfrom western farms (8). Water—notrain—has followed the plow, exceedingthe expectations of even the mostzealous proponents of Manifest Destiny150 y ago.

Reisner and the Cadillac DesertNumerous critiques of the sustainabilityof freshwater infrastructure in the westernUnited States have appeared (5, 9–12).Most poignant of these is Marc Reisner’sbook Cadillac Desert: The AmericanWest and Its Disappearing Water. Reisnersketches a portrait of the political follyof western water projects; his principalargument is that impaired function ofdams, reservoirs, and crop lands, coupledwith rapidly growing western cities,would eventually pit municipal water

users against farms and catalyze anapocalyptic collapse of western US

Author contributions: J.L.S., L.C.B., and E.E.W. designed re-search; J.L.S., T.S., L.C.B., G.H.W.S., W.W.W., K.A.C., W.L.G.,J.S.K., C.T., S.W.T., and E.E.W. performed research; J.L.S.,T.S., L.C.B., G.H.W.S., W.W.W., and P.L.F. analyzed data;J.L.S., T.S., L.C.B., G.H.W.S., W.W.W., M.E.C., P.L.F., W.L.G.,J.W.H., J.S.K., C.T., S.W.T., R.H.W., and E.E.W. wrote thepaper; and T.S. coded, calibrated, and implemented hydro-logic models.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. G.M.M. is a guesteditor invited by the Editorial Board.1To whom correspondence should be addressed. E-mail:john.l.sabo@asu.edu.

2Present address: Department of Civil, Construction, andEnvironmental Engineering, North Carolina State Univer-sity, Raleigh, NC 27695-7908.

3Present address: Odum School of Ecology, University ofGeorgia, Athens, GA 30602.

This article contains supporting information online atwww.pnas.org/lookup/suppl/doi:10.1073/pnas.1009734108/-/DCSupplemental.

*Powell writes: “A rough estimate may be made that [404,686 square kilometers] can be redeemed at the rate of[$2,470 per square kilometer] that is for US $1 billion [in1890]. In this work vast engineering enterprises must beundertaken. To take water from streams and pour themupon the lands, diverting dams must be constructed andcanals dug.” The area of irrigated croplands as of 2000 is173,858 square kilometers, as referenced in: de Buys (2).

www.pnas.org/cgi/doi/10.1073/pnas.1009734108 PNAS | December 14, 2010 | vol. 107 | no. 50 | 21263–21270

SPECIA

LFEATURE:PERSP

ECTIV

E

society.† In this article we explore some ofthe trends described by Reisner morethan 2 decades ago using a more up todate and scientific approach. Specifically,we compare hypothetical calamity in theWest with a control by means of directcomparison with watersheds east of the100th meridian. The 100th meridian hassome historical importance because it wasthe line implicated by Powell—and ad-vocated by Reisner—as a dividing linebetween climates capable of supportingrainfed agriculture and regions where ir-rigation was necessary for dependableharvest. For the remainder of this articlewe use the 100th meridian as the dividingline between east and west regions in thecoterminous United States. Thus, we ex-plore whether the problems Reisner en-visioned in Cadillac Desert exist and areunique to western watersheds. More im-portantly, we present a suite of metricsand indicators that summarize freshwatersustainability (or departures from sus-tainability) in the Cadillac Desert region.We first synthesize a comprehensive

geographic dataset that allows us toquantify and compare regional patterns offreshwater sustainability east and west ofthe 100th meridian. In doing this wecombine data from humid western basins(i.e., Columbia) with those in more aridwestern regions (i.e., Colorado) for muchof our analysis, noting important excep-tions where necessary. Inclusion of theColumbia River basin was necessary fortwo reasons. First, the Columbia basin isa prime example of the grand scale of waterprojects that characterize development ofthe western United States. Second, theColumbia River originates in part in theSnake River headwaters on the ColumbiaPlateau, a semiarid region that illustratesmany of the same impacts associatedwith large-scale water projects as outlinedin Cadillac Desert. We define freshwatersustainability as renewable surface water—hereafter “streamflow”—and its allocationto people, farms, and ecosystems. We ex-clude groundwater as a source of freshwaterin our analysis because it is not as immedi-ately renewable as surface water, and it isless relevant to our objective because Re-isner’s focus was on harnessing surface wa-ter. Below we quantify patterns of meanannual streamflow in the coterminousUnited States. We then quantify freshwater

sustainability in terms of (i) human waterstress, (ii) the efficacy and lifespan of res-ervoir storage, (iii) the impact of salt loadsin croplands on agricultural revenue, and(iv) biodiversity and invasion of native fishfaunas. After analyzing broad patterns ofsustainability and comparing sustainabilityindices east and west of the 100thmeridian,we narrow our focus to the arid lands westof this divide and estimate water stress toassess the future for sustainable urbangrowth in the region.

ResultsClimate and Surface Water Supply Set theStage. One of Powell’s key observationswas that rainfall was insufficient to provideadequate vadose zone water storage dur-ing the growing season for nonirrigatedagriculture in much of the western UnitedStates. The upshot of this observation wasthat streamflow would need to be har-nessed to provide irrigation and sustainagriculture. Estimated streamflow nor-malized by area (runoff) is low (<10 cm)for most of the west and much higher(≥40–100 cm) for much of the easternUnited States (Fig. 1A), with two notableexceptions. First, the Pacific Northwestand the northern mountains of Californiahave the highest runoff in the coterminousUnited States. Second, the longitude ofthe east–west transition between high andlow runoff in the Great Plains varies bynearly 10°—from the 95th meridian in thenorthern plains to the 105th meridian nearthe Gulf of Mexico. However, there areclear differences in the distribution ofrunoff, cities, and farms in eastern andwestern US watersheds. Below we defineUS watersheds using boundaries of the US

Geological Survey (USGS) four-digit hy-drologic unit code regions or hydrologicsubregions (13). Cities and farms are morelikely found in hydrologic subregions withabundant surface water (runoff >40 cm)in the East (nearly 94% of the populationand 65% of the cropland in the East is ina hydrologic subregion with streamflowexceeding 40 cm, compared with 55% ofthe population and 41% of the croplandsin the West). More relevant to the thesisof Cadillac Desert, 23% of the populationand 28% of the cropland in the Westfalls within a hydrologic subregion whererunoff is <10 cm [compared with 1% and13% of the population and cropland, re-spectively, found in a hydrologic subregionwith similarly low (10 cm) mean annualstreamflow east of the 100th meridian].

State of Current Infrastructure. The impactsof dams and reservoirs include increases inhydrologic storage and fragmentation ofriver networks. Relative storage capacitygives a measure of the number of yearsof average streamflow stored in the reser-voir system, and dam density providesa proxy for fragmentation of river networksby impoundments. Total storage capacityof reservoirs does not differ east and westof the 100th meridian (Fig. S1). Storagein more numerous but smaller reservoirsin the East is nearly equivalent to that inthe generally fewer, larger reservoirs in theWest. As a result, dam density is higherin the eastern United States (Figs. 1B and2A). However, more than 73% of water-sheds with relative storage capacity values>1 are located in the West, and another19% straddle the 100th meridian between−90° to −100° W. Overall, relative storage

Fig. 1. Patterns of hydroclimateand freshwater infrastructure inthe coterminous United States. (A)Mean annual streamflow (cm) esti-mated using the VIC macroscalehydrologic model (SI Appendix). (B)Average number of dams per 100km of river length (color coded, seelegend) in each USGS hydrologicsubregion in the coterminousUnited States and total storage ca-pacity per unit streamflow, or rel-ative storage capacity (text) foreach USGS hydrologic region.

†The apocalypses sketched by Reisner in Cadillac Desert are(i ) that western reservoirs will fill with sediment soon, andreduced storage capacity will present unprecedented wa-ter scarcity issues; (ii) crop lands will be increasingly re-tired due to salinity issues, to the extent that waterprojects will ultimately poison the farmlands that westernsocieties depend on for food; and (iii ) growing urbanpopulations will draw increasing water away from agri-cultural areas, further reducing the capacity for the Westto feed its people.

21264 | www.pnas.org/cgi/doi/10.1073/pnas.1009734108 Sabo et al.

is 3.3 times higher in the West (Figs. 1Band 2B). Dams fragment riverscapes morein the East, but large reservoirs alter hy-drologic dynamics more in the West—holding more water relative to streamflow.

Focus Area 1: Human Water Stress. Total an-nual water withdrawals are 1.7-fold higher inthe East, but the withdrawals for agricultureare 3.2-fold higher in theWest (Fig. 2C). Toassess the sustainability of surface waterwithdrawals in the United States, we esti-mated the water scarcity index (WSI) (14,15) for each hydrologic subregion. WSI isthe ratio of total withdrawals of freshwaterfor human use (W) (16) to renewable sup-ply (mean annual streamflow, MAF). Wedefined supply as the sum of local and un-used upstream annual average streamflowestimated by the variable infiltration ca-pacity (VIC) model (SI Appendix). Ourapplication of WSI provides a measure offreshwater sustainability defined as the ca-pacity for locally generated and unusedupstream streamflow to meet local de-mand. Subregions with WSI ≈0 appropriatelittle of their streamflow. Higher WSI in-dicates greater appropriation of local re-newable freshwater resources. WSI values>1 are possible where streamflow is lowand withdrawals include a substantialgroundwater component. Water stress iscommonly defined as WSI 0.4 (14), in-dicating 40% appropriation of renewablefresh water resources. This threshold is setat less than half of available streamflow tobuffer against high spatial and temporalvariability in streamflow and to set asidewater for ecosystems, navigation, and rec-reation. Water stress occurs in 58% of

subregions in the West, compared with10% in the East (Fig. 3A), and withdrawalsexceed local streamflow by 2-fold (WSI >2)in 10 western watersheds. Nine of the top10 WSI values are in the West (Fig. 3A;average ± SE WSI: 0.85 ± 0.1 West and0.22 ± 0.03 East). In a few eastern sub-regions WSI is high because withdrawalsfrom large freshwater lakes (e.g., the Lau-rentian Great Lakes) in neighboring sub-regions exceed local streamflow. Finally,consumptive use values were not estimatedin 2000, such that our estimates of WSIinclude consumptive and nonconsumptivewithdrawals of freshwater (SI Appendix).

Focus Area 2: Efficacy and Lifespan of Res-ervoir Storage. One of Reisner’s key criti-cisms of western reliance on reservoirstorage was inevitable sediment infillingand subsequent storage deficits for grow-ing cities and agriculture (17).‡ The ques-tion we ask here is not whether, but howfast will the nation’s reservoirs fill withsediment? Assuming observed infillingrates over the last century are represen-tative and constant, we estimate that 276of the reservoirs (22%) in the ReservoirSedimentation Survey Information System

(RESIS-II) are already completely filledwith sediment or have been dredged tomaintain function. However, only 1 ofthese reservoirs is >0.123 km3 (moderatelysized), and only 11 are even within an or-der of magnitude of this size (2 in theWest and 9 in the East >0.0123 km3; Fig.S2). Predicted minimum lifespans for theremaining (unfilled) reservoirs are lowestin the central United States and DesertSouthwest (Fig. S2); however, estimatedminimum lifespans are all ≥1.5–2 centu-ries. Thus, although Reisner was correctthat reservoirs fill with sediment, observedinfilling and complete loss of storage func-tion is by no means exclusively a westernphenomenon and will not likely occur formost large reservoirs in the foreseeablefuture. Given the long time horizon forcomplete infilling, we extrapolated esti-mated capacity losses from single struc-tures to entire hydrologic subregions toconstruct a metric of storage loss morecomparable to available water supply. Wenormalized this estimate of regional stor-age loss by MAF because this metric bet-ter quantifies the change in the region’sability to withstand prolonged drought orflooding (15). Relative capacity losses forthe 95 (of 204) subregions with adequatedata from RESIS-II range from 8 × 10−4

to >11 (units = mean annual streamflowequivalents) and are higher by a factor of≈11.7 in the West (Figs. 2D and 3B).Storage loss in a water supply reservoir

directly impacts the firm yield, or thelargest withdrawal rate that the reservoircan reliably provide. The relationship be-tween firm yield and active storage isgenerally nonlinear, with an initially

Fig. 2. Comparison of infrastructureand impacts of infrastructure east andwest of the 100th meridian. (A) Aver-age number of dams per 100 km ofriver length. (B) Relative storage ca-pacity (total reservoir storage/meanannual streamflow). (C) Sum of waterwithdrawals by category—municipal,industrial, agricultural, and total. (D)Average storage capacity losses in res-ervoirs as a result of sediment infillingexpressed relative to mean annualstreamflow (Left) and in absolute terms(km3, Right). (E) Estimated average re-ductions in firm yield (km3) for large(>1.23 km3) and small (<1.23 km3) res-ervoirs. Numbers above error bars (±1SE) are sample sizes in each category.(F) Estimated average revenue losses(millions USD) as a result of salt accu-mulation in croplands. (G) Average ra-tio of nonnative to native fishes (color)and number of nonnative species(text). (H) Average per capita virtualwater footprints (VWF) for all metro-politan statistical areas >100,000 insize. Footprints are negative if virtualwater is exported in crops from the watershed hosting the city, or positive if the city requires imports of virtual water (in crops) to feed the population. Errorbars are SEs using hydrologic subregions as the unit of replication (127 and 77 east and west of the 100th meridian, respectively), unless otherwise indicated.

‡In the second printing of Cadillac Desert, Reisner writes (p473), “As a result of [intensive machine based agricultureand loopholes allowing for agriculture on Class VI land]—andbecause itwas inevitable anyway—the dams are siltingup.” He then lists infilling statistics for 12 reservoirs in theUnited States, including Lake Mead, writing (p 474), “Inthirty five years, Lake Mead was filled with more acre feetof silt than 98% of the reservoirs in the United States arefilling with acre feet of water. The rate has slowed consid-erably since1963,because the silt is nowbuildingupbehindFlaming Gorge, Blue Mesa and Glen Canyon dams.”

Sabo et al. PNAS | December 14, 2010 | vol. 107 | no. 50 | 21265

shallow—but accelerating—rate of declinein firm yield as active storage capacitydecreases (Fig. S2). Firm yield for largereservoirs (>1.2 km3, storage capacity) hasalready diminished by ≈1.9% relative toyield at original capacity, and up to 6.25%for small reservoirs (1.2 km3 > storagecapacity > 0.12 km3). The absolute declinein firm yield since dam closure was notsignificantly different in the East and Westfor the reservoirs we analyzed (Table S1)except for small reservoirs, in which de-cline in absolute firm yield was marginallyhigher in the East (Fig. 2E). Although thedifferences in sediment related reductionsin firm yield across the country were notgenerally significant in eastern relative towestern reservoirs, estimated reductions inthe absolute volume of firm yield losses inthe West even for the small number ofstructures analyzed here are formidable.Estimated reductions in firm yield in thefive large reservoirs we analyzed west ofthe 100th meridian (firm yield volume≈0.584 km3 · y−1) are larger in sum thanmaximum annual conveyance by theLos Angeles Aqueduct (≈0.25 km3 · y−1)and Moffat Tunnel diversion to Denver(≈0.43 km3 · y−1), and equivalent to 60%and 32% of the annual conveyance ofthe Salt River Project (≈0.97 km3 · y−1)

and Central Arizona Phoenix Project(≈1.85 km3 · y−1), respectively.

Focus Area 3: Impact of Salt Loads in Crop-lands on Agricultural Revenue. Salinity isa worldwide threat to the sustainability ofirrigated agriculture (17). Both the accu-mulation of salt and the extent of salt-af-fected soils are more prevalent in the West(Fig. 3C and Fig. S3). Total estimatedrevenue losses experienced by the agricul-tural sector are ≈2.8 billion US dollars(USD) annually. Estimated revenue lossesare nearly an order of magnitude higher inthe West (2.55 billion USD · y−1, West vs.267 million USD · y−1, East). Crop yieldsand revenue have been disproportionatelyaffected in western watersheds, particularlyin regions with extensive areas of vegetablecrops and orchards (Fig. 3C). Revenuelosses are ≈60-fold higher per acre ofcropland in the West (Fig. 2F).

Focus Area 4: Biodiversity and Invasion ofNative Fish Faunas. The sustainability offresh water supplies can be measured interms of human water security and the ca-pacity of freshwater ecosystems to supportbiodiversity (18). These two sustainabilitygoals are not mutually exclusive—bio-diversity provides valuable ecosystem

services ranging from the food and eco-nomic benefits of inland fisheries (19) tothe maintenance of water quality (20) andregulation of gas exchange between fresh-water ecosystems and the atmosphere (21).Discharge magnitude and variation de-termine biodiversity in rivers across theglobe (e.g., refs. 22 and 23), and dams,water diversions, and human appropriationof streamflow homogenize this variation(24), thereby altering key components ofbiodiversity, including food chain length(25) and the number of nonnative species,especially fishes (26, 27).The proportion of all species that are

nonnative provides a proxy for the impactof freshwater infrastructure on nativebiodiversity because dams and reservoirsfacilitate invasion by nonnative fishes bycreating new habitat (e.g., still reservoirsrather than flowing water) and altering theflow and temperature regime in dam“tailwaters” (26, 28, 29). Further, this ratiois one of four drivers used in broad-scaleanalyses of threats to human water secu-rity and biodiversity (18, 30). This ratio ishigher in the West (Figs. 2G and 3D),and this is not a byproduct of higher nativespecies richness in the East, because theabsolute number of nonnative species isalso higher (Fig. S4). Thus, dominance of

Fig. 3. Assessment of current freshwater sustainability. (A) WSI for 204 coterminous hydrologic subregions. Here, WSI = W/MAF, where W is total withdrawalsbased on USGS estimates from 2000, and MAF is total mean annual streamflow, including locally generated streamflow and flow unused by upstream hy-drologic regions. (B) Estimated relative (storage loss/streamflow, color coded) and total losses (km3, text) of storage capacity in each USGS hydrologic regiondue to infilling by sediment. NA in Hydrologic Region 1 indicates no sediment surveys available for reservoirs over 1.23 × 10−2 km3 in this region (SI Appendix).(C) Agricultural revenue lost (in million USD) at HUC 4 scale due to soil salinization (color coded). (D) Ratio of nonnative to native fish species (color coded) andtotal number of observed nonnative species (text).

21266 | www.pnas.org/cgi/doi/10.1073/pnas.1009734108 Sabo et al.

western fish faunas by nonnative fishesresults from higher absolute numbers ofestablished nonnative fishes and low spe-cies richness of native fishes (Fig. 2G).Moreover, of the 25 most widespreadnonnative fishes west of the MississippiRiver drainage,§ 56% (14 of 25) are pis-civores native to lakes or rivers in hydro-logic regions east of the 100th meridianwith a less variable hydrologic regime, and6 of these 25 are capable of eating notonly native insectivores but also nonnativepiscivores on the basis of body and gapesize (Table S2). Eastern faunas not onlydominate the species roster in westernrivers, but they likely occupy one or moreunique trophic levels at the apex of foodwebs in heavily modified western rivers.This artificial increase in food chain lengthis due in part to a reduction in dischargevariability below dams (24, 25).

Civilization, If You Can Keep It. The centraltheme of Cadillac Desert is that the hydro-climate of the American West is not gen-erous enough to sustain cities and agri-culture, especially in the Southwest. Belowwe attempt to quantify this claim in twoways. First, we estimate agricultural waterfootprints of all large (>100,000 in size)US metropolitan statistical areas (MSAs)in the East and West. Our water footprintsexplicitly consider the net transfer of vir-tual water needed to feed the local pop-ulation. Second, we estimate the totalappropriation of surface water by humansfor a seven-region area constituting theDesert Southwest. We estimate humanappropriation of surface water for this“superregion” under all possible combi-nations of the following scenarios: usingcurrent withdrawals (as in Fig. 3A but ata larger resolution) or total water demandincluding the virtual water required forfood production; and using Census 2000population estimates, or assuming a dou-bling of the Census 2000 population.Disproportionately large water footprints ofcities in the US Southwest. Agricultural wa-ter footprints are the volume of waterneeded to meet the food demand for a cityor region (31). Here, we normalize thewater volume by runoff to find the equiv-alent land area to supply the water de-mand, analogous to a “carbon footprint”(32). Total water footprints based on wa-ter withdrawals are captured in Fig. 3A,where WSI indirectly represents the frac-tion of a subregion’s land area necessary to

generate the streamflow to sustain thosewithdrawals. Here we estimate net agri-cultural water footprints of the 332 largestUS MSAs [>100,000 in population as of2000 (33)]. Virtual water in food includeswater transpired during production [viaactual evapotranspiration (AET), i.e.,“green water” (14)], is higher in arid re-gions with higher prevailing rates ofevapotranspiration, and is ≈80% of allconsumptive water use worldwide (14).Net virtual water represents the differencebetween virtual water import and export,or alternatively, the difference betweenthe virtual water in locally grown food andthe virtual water locally consumed. Wedefine a virtual water footprint (VWF) ofa city as the land area necessary to capturethe streamflow required to satisfy thenet virtual water transfer (i.e., to grow theadditional crops needed to feed the pop-ulation of that city that are not grownlocally). Thus, our virtual water footprintsdiffer from WSI in two ways: (i) theyquantify the total land area equivalents ofstreamflow needed to feed cities via localagriculture, and (ii) they allow us toquantify net virtual trade in terms of im-port of virtual water to cities (positiveVWF) and export of virtual water fromwatersheds with extensive crop area (neg-ative VWF). Cities in the Desert South-west United States had disproportionatelyhigh net water import (large positiveVWF) (Fig. 4A). Urban areas with the topfive total positive VWF (indicating netimports of virtual water) were Los An-geles, Las Vegas, Phoenix, New York, andRiverside, in that order. The VWF ofLos Angeles is larger than the combinedVWF of the eight largest VWFs east of the100th meridian, including New York,Chicago, Dallas, Detroit, Philadelphia,Houston, Boston, and Washington, DC.Combined VWF for these eight metroareas is 206,585.5 km2, compared with214,922 km2 for the Los Angeles metro-politan area. Watersheds exporting themost virtual water in the form of foodproducts typically hosted smaller cities,many in the corn belt (Fig. 4B); subregionswith the highest net virtual water exportwere those hosting Wichita (KS), SiouxFalls (SD), Omaha (NE), Havasu City(AZ), and Colorado Springs (CO), in thatorder. The virtual water demand of thesesmaller population centers is dwarfed bywater used for agriculture in their water-sheds, which is exported as food products.Average VWFs were positive west ofthe 100th meridian (indicating net import)and negative east of the 100th meridian(indicating net export; Fig. 2H). Westerncities with net positive VWF had 7-foldlarger footprints than cities with net posi-tive VWF east of the 100th meridian(Fig. 2H). Western watersheds also export1.7-fold more virtual water than water-

sheds dominated by cropland east of the100th meridian (Fig. 2H), although thislatter difference is not significant (TableS1). In summary, western cities have muchlarger virtual water footprints, largelyowing to the more arid climate, andwestern crop lands export at least an equalmagnitude of virtual water as cities andcroplands east of the 100th meridian.Some but not all of the virtual water ex-ported from productive farmland in thewestern United States (e.g., Central andImperial valleys of California) offsets largefootprints of cities in the desert Southwest,because these farmlands produce tablevegetables, tree fruit, and nuts for much ofthe United States.Human appropriation of streamflow in the USSouthwest. Six major watersheds in the USSouthwest are connected by aqueductsand water transfers (Fig. 4B). Water fromthe upper Colorado basin is divertedacross the continental divide to the SouthPlatte, Arkansas, and Rio Grande riversfor municipal use by front-range cities aswell as agriculture. Snowmelt from theupper Colorado basin is collected lower inthe basin and diverted to Las Vegas, cen-tral and southern Arizona, and southernCalifornia. Snowmelt from the Sierra Ne-vada is diverted to San Francisco andReno and to cropland in the southernCentral Valley of California via the Cen-tral Valley Project (CVP). Finally, snow-melt from the northern Sierra Nevada andwater from Trinity River in California’sNorthern Coast Range is diverted southto the Central Valley (via the CVP),some portion of which reaches southernCalifornia via the State Water Projectof California. The resulting human-engineered watershed connects stream-flow generated in the mountains of Colo-rado and California to cities in at least sixstates, representing a combined urbanpopulation of more than 50 million, and toone third of all western croplands.Here we quantify the human appropri-

ation of streamflow in this superregion(Fig. 4B). The simplest index of humanappropriation is WSI, or withdrawalsnormalized to MAF across the super-region (SI Appendix). Humans currentlyappropriate the equivalent of 76% ofMAF in this superregion (WSI 0.76; Fig.4B). This number is equivalent to >90% ofstreamflow when we use the virtual waterdemand for agriculture instead of theactual withdrawals associated with currentagricultural practices to calculate WSI(Fig. 4B). This “virtual WSI” accounts forall water needed to grow crops to sustainthe entire population in the superregion,assuming food is grown within and no foodis transported out of the super region (i.e.,“regional food production”). Higher vir-tual WSI suggests that much higher ap-propriation of streamflow would be

§Here we focus on hydrologic regions 13–18, quantifyingprevalence of nonnative fishes in 8-digit hydrologic unitcode basins or accounting units. We chose a finer resolu-tion for this analysis to illustrate the comprehensive na-ture of fish invasion in western watersheds. At this finerlevel of resolution, we can record not only whether a par-ticular nonnative fish is present in a 4-digit subregion butalso how widespread it is within that subregion.

Sabo et al. PNAS | December 14, 2010 | vol. 107 | no. 50 | 21267

required for the superregion to persist onlocally grown food alone and that thereare likely important, but not as of yetquantified, ecological tradeoffs betweenwater footprints associated with regionally

produced agriculture and carbon foot-prints associated with food imports fromagricultural lands outside the superregion.Finally, population in the Cadillac

Desert superregion is projected to increase

significantly over the next 25–40 y (allprojections available from the US CensusBureau; ref. 34). For example, the pop-ulation of the state of California is pro-jected to grow by 50% by 2050, thepopulation of southern Nevada is pro-jected to grow by as much as 57% by 2030,and the Phoenix metropolitan area (heredefined as Maricopa County) and pop-ulation centers on the Front Range of theRocky Mountains of Colorado are pro-jected to double by 2050 and 2040, re-spectively. This suggests that populationdoubling is possible in many populationcenters in the superregion within a cen-tury. Hence, we estimated WSI and virtualWSI as above, but assuming twice thepopulation in the superregion. Usinga conservative (more recent) trend for therelationship between population and wa-ter use, we estimate that humans willwithdraw ≈86% of current MAF undera population doubling (WSI 0.86). With-drawals could be as high as 99.4% ofcurrent MAF according to the extrapo-lated virtual water demand required forregional food production (virtual WSI0.99; Fig. 4B).

DiscussionJohn Wesley Powell provided the earliestsketch of sustainable development in thewestern United States. Powell’s conclusionin 1876 was that water scarcity would placelimits on the growth of a new civilizationin the region (3). Marc Reisner pursuedthis conclusion in Cadillac Desert a centuryafter Powell’s explorations (5). Reisner’sdiagnosis was that the water demands ofagriculture and growing western citieswere at odds and precariously dependenton static conditions—optimistic estimatesof streamflow, unchanging reservoir stor-age capacity, and soils buffered againsthigh salt loads. In this article we use dataand methods unavailable in Reisner’stime to reevaluate this diagnosis. We findthat the characteristics and impacts ofdams and reservoirs differ considerablybetween the eastern and western UnitedStates, suggesting that the Cadillac Desertenvisioned by Marc Reisner has a strongscientific basis. Specifically, the US westof the 100th meridian is characterized by(i) low mean annual streamflow; (ii) largereservoirs spaced more distantly withinriver networks, but storing a more than4-fold higher proportion of mean annualstreamflow than in the East; (iii) 3-foldhigher surface water withdrawals as a pro-portion of streamflow; (iv) net virtualwater footprints at least seven times thearea of those of eastern cities; (v) largereservoirs with estimated minimum life-spans exceeding 1.5–2 centuries that havenevertheless already experienced losses infirm yield greater in volume than the an-nual conveyance of critical water delivery

Fig. 4. Water footprints for agriculture and human appropriation of streamflow by urban areas in theUS Desert Southwest. (A) Net virtual water footprints of metropolitan statistical areas >100,000 in size.Footprints represent the land area equivalent of streamflow generation required to grow the food tofeed the MSA population (following ref. 45); positive numbers (blue) indicate net export (i.e., the MSA’shydrologic subregion produces more food than is required by the MSA population), and negativenumbers (red) indicate net import (the MSA requires more food than is produced within the local hy-drologic subregion) for each MSA. See SI Appendix for more details. (B) Appropriation of streamflow bylarge urban areas (Census 2000 MSAs >100,000 in size) under a population doubling scenario of thesecities. Map shows the paucity of streamflow across five southwestern USGS hydrologic regions (Regions13–16 and 18) and the natural and engineered causeways for this streamflow. Pie charts show pro-portion of streamflow appropriated by large urban areas in the same five-hydrologic-region area underfour scenarios. Scenario A is the current water scarcity index (WSI = W/MAF) for the entire five-regionarea using USGS water use data from 2000 (W). Scenario B estimates the capacity for streamflow tosupport municipal and industrial withdrawals in Scenario A in addition to the virtual water needed forregional production of all food. The difference in human appropriation between Scenarios A and Bhighlights the degree to which the Desert Southwest imports streamflow (contained in food) from moredistant hydrologic regions. Scenarios C and D project human appropriation of streamflow under a re-gional doubling scenario of all MSAs >100,000 in size in 2000, assuming only changes in W associatedwith population increase (SI Appendix). In Scenario C, projections are based on water use data (in Sce-nario A), whereas in Scenario D, projections are based on municipal and industrial withdrawals andestimated virtual water for agriculture. All scenarios rely on current VIC streamflow estimates (MAF,based on average annual climate forcings from 1950 to 1995). Arrow width is proportional to themagnitude of water diversion associated with numbered major water projects in the Southwest: (1)Duchesne River Diversion, (2) Blue, Fraser, and Williams Fork River diversions, (3) Frying Pan and EagleRiver diversions, (4) San Juan River diversion, (5) Middle Rio Grande River diversions, (6) Salt River Project,(7) Central Arizona-Phoenix Project, (8) Colorado River flow exiting United States to Mexico (1.85 km3 ·y−1) as mandated by The Mexican Water Treaty of 1944, (9) All American Canal, (10) Colorado RiverAqueduct, (11) Boulder Canyon Project, Lake Mead, (12) Los Angeles Aqueduct, (13) State Water Project,California Aqueduct, (14) Central Valley Project, (15) Tuolumne River diversion, (16) Truckee River di-versions, (17) Central Valley Project: Trinity River diversion.

21268 | www.pnas.org/cgi/doi/10.1073/pnas.1009734108 Sabo et al.

systems (e.g., Los Angeles Aqueduct);(vi) >60-fold greater reductions in agri-cultural revenue due to inefficient irriga-tion practices and soil salinity; and (vii)faunas with nearly six times the ratio ofnonnative to native fishes than those in theEast. Our storyline, although hopefullymore measured, is in line with the oneReisner crafted in 1986.

Interaction Between Reclamation and ClimateChange. Our synthesis of water resourcesdata ignores important interactions withclimate change. Increased temperatures,higher water demand by crops, greaterrainfall variability, reduced snowpack andstreamflow, earlier snowmelt and peakstreamflow timing, and a doubling of majorurban populations are very likely scenariosin the next 100 y. Less certain but likelyscenarios include reduced average annualprecipitation in the southwest UnitedStates and climate/population-inducedwater withdrawal increases. Our analysisprovides some insight about interactionsbetween water storage systems, climatechange, and population growth scenarios.First, continued sediment accumulation

will result in lower active storage andfurther reductions in yield of water fromreservoirs. Reductions in firm yield due tosediment will be exacerbated by declines instreamflow, increases in variability, andchanges in the timing of peak streamflowassociated with climate change. Second,agricultural revenue losses due to salini-zation are likely to rise. Increasing tem-peratures would increase crop waterdemand and crop transpiration, leading togreater soil concentration of salts. Seasonalshifts and reductions in western watersupply will require greater reliance on sa-line/brackish or nonrenewable freshgroundwater as a source for irrigationwater. This double squeeze, from bothsupply and demand sides, is expected toincrease soil salinization in much of theWest. Third, invasion of rivers by nonnativefishes is ongoing. Native species in heavilyinvaded ecosystems will become in-creasingly threatened by nonnative speciesand flow regimes further altered byclimate change.In closing, we note that the capacity for

water to support cities, industry, agricul-ture, and ecosystems in the USWest is nearits limit under current management prac-tices. For an urban population doublethe Census 2000 size, we estimate thatwater withdrawals necessary to meet mu-nicipal, industrial, and agricultural demandwill exceed 86% of the current streamflowacross parts of seven hydrologic regionsin the southwest United States (Fig. 4B andSI Appendix). Our estimate of human ap-propriation is >99% of the streamflowgenerated by this region if we include thewater needed to produce all food to feed

a doubled population in the region (Fig.4B). These estimates are conservative fortwo reasons. First, our population dou-bling scenario does not include supply re-ductions due to climate change. Second,we assume conservative increases in waterwithdrawals as the urban populationgrows. Even these most-conservative esti-mates suggest that renewable freshwaterresources will not comfortably supporta population beyond two times the currentlevels in the western United States whilestill providing adequate flows to maintainvital ecosystems.To reclaim freshwater sustainability

in the Cadillac Desert, we suggest an ini-tially modest target of a 16% reduction inthe fraction of streamflow withdrawn, orWSI = 0.6 before the realization ofa projected population doubling across theentire geographic region (Fig. 4B). Thisimproved regional WSI represents a com-promise between reductions that wouldalleviate water stress altogether (WSI 0.4)and those that would significantly diminishalready insufficient freshwater resources inriver and delta ecosystems (WSI >0.8).Meeting this target will require a regionalwater conservation policy coordinatedacross seven US states addressing ata minimum: (i) continued improvementsin urban water use efficiency, (ii) im-plementation of desalinization by coastalcities, (iii) continued improvements inland-use practices that minimize erosionand sediment infilling of the region’s res-ervoirs, (iv) technological advances in-creasing water application efficiencyduring irrigation, (v) modified crop port-folios that include only salt tolerant andcash crops, (vi) effective reallocation ofsalvaged surface water to ecosystems asfarmlands are retired and cities shift todesalinization, and (vii) endorsement ofmarket-based rather than government-subsidized water pricing for all uses exceptthose that fulfill the most basic daily hu-man needs. Further, Reisner’s bookCadillac Desert and our analyses do notconsider the impact of water use ongroundwater reserves. A regional policy offreshwater sustainability should bridge thisgap and (viii) implement aquifer storageand recovery and artificial rechargeschemes for water storage and manage-ment, and (ix) endorse only judicious useof groundwater with minimal impact onsurface flows in pursuit of our suggestedtarget (WSI 0.6). This regional policy offreshwater sustainability will imposea cost, and this cost—as Reisner noted—will most likely include more expensivewater at the tap and on the farm.

Materials and MethodsMacroscale Hydrology. We used a macroscale hy-drologic model—the Variable Infiltration Capacity(VIC) model (35, 36)—to quantify patterns in mean

annual streamflow volume (km3) across the co-terminous United States at a resolution of 1/8 de-gree using observed meteorological forcings from1950 to 1999 (37). We used the VIC model to esti-mate streamflow (as opposed to available data)because the VIC model provides estimates of virginflow, whereas empirically measured streamflowincludes the effects of withdrawals and river regu-lation by dams. In contrast to previous continental-scale applications of the VIC model (37–39), thecurrent version was implemented with seasonallyfrozen soils, improving energy and water balanceestimates during thewinter season (40, 41). TheVICmodel was calibrated to monthly naturalized andobserved streamflow for 12 watersheds within sixmajor representative hydrologic regions across thecoterminous United States for a 10-y period andthen evaluated for the remaining (independent)observational period of 10–40 y, between 1950 and1999. Model bias was low and positive on average(9.2% ± 5.9% of naturalized or observed stream-flow) with reasonable variation across basins(Table S3).

Patterns of Infrastructure. Using data from theNational Inventory of Dams we summed the totalnumber and storage capacity of reservoirs in eachUSGS subregion. We then estimated the averagenumber of dams per 100 km of river length[according to USGS’s HYDROGL020 layer (US Na-tional Atlas Water Feature Areas)] (42) or damdensity. We also quantified the total storage ca-pacity relative to mean annual streamflow (rela-tive storage capacity) for each subregion usingstreamflow estimates from the VIC model (Fig. 1).We then made East vs. West comparisons in thissection and all others to follow at the watershed,or USGS four-digit hydrologic code (HUC 4) sub-region resolution. For all East vs. West compar-isons, we used geographic centroids of HUC 4subregions to determine their location relative tothe 100th meridian.

Sediment Infilling. We quantified infilling rates,reservoir storage capacity losses, and lifespansusing the RESIS-II database (43). This database in-cludes repeat bathymetric surveys for >1,200 res-ervoirs in the United States. We estimated singlestructure storage capacity losses from closure topresent (2010). Total capacity losses for hydrologicsubregions were then estimated by multiplyingthe subregion’s total reservoir capacity [from theNational Inventory of Dams (NID)] by the mini-mum observed proportion of capacity lost (fromRESIS-II). We expressed this capacity loss as a pro-portion of the subregion’s mean annual stream-flow (i.e., relative capacity loss).

Firm Yield Analysis. For 24 reservoirs in the RESIS-IIdatabase ranging in size from 0.04 to 35.5 km3, weestimated the change in firm yield via sequentpeak analysis based on our estimates of currentactive storage and observed monthly streamflowdata from nearby USGS stations.

Agricultural Revenue Losses to Salinity. We esti-mated revenue losses as a result of diminished cropyields in saline soils for all 204 hydrologic sub-regions in the coterminous United States. Weidentified salt-affected soils using the nationwideState Soil Geographic Database (STATSGO). Wethen estimated revenue losses according to na-tionwide crop type and soil salinity maps and dataon crop salt tolerances, crop yields, and prices.

Sabo et al. PNAS | December 14, 2010 | vol. 107 | no. 50 | 21269

Fish Invasion. We cataloged patterns of invasionby nonnative fishes using the USGS Non-indigenous Aquatic Species (NAS) database (44)and NatureServe’s Distribution of Native Fishes byWatershed database (45). For nonnative fishes inthe NAS dataset, we used only established, locallyestablished, and stocked nonnative species in theNAS dataset to avoid spurious single sightings ofnonindigenous species that might inflate our es-timates of invasion. We recompiled presence/ab-sence data at the resolution of hydrologicsubregions in the lower 48 states and estimatedα diversity for native, nonnative, and all (nativeand nonnative) fishes in each subregion (Fig. S4).

Water Footprints. To estimate a water footprint foreach city, we used the annual per capita waterrequirements based on a published averageUS diet(46). We also calculated per capita crop water usefor each subregion using estimates of AET fromcropped areas under natural rainfall together withknown quantities of irrigation water withdrawals.Per capita values were multiplied by the MSApopulation size from the 2000 census (US CensusBureau). The difference in virtual water demand

and supply was normalized by streamflow depthfrom the VIC model to estimate the land area re-quired to capture the net virtual water demand.

Extrapolation of Human Appropriation of Stream-flow Under Population Doubling. To extrapolatewater use and human appropriation of streamflowunder a population doubling scenario, we de-veloped a regression relationship between total USpopulation size and total annual water with-drawals. The slopeof this relationship at the timeofpublication of Cadillac Desert was ≈2, indicatinga doubling in water extraction with populationgrowth. Estimates of water withdrawals over thelast 25 y (1980–2005) indicate that withdrawalshave increased much less dramatically with pop-ulation (slope = 0.23 per unit population). We ex-trapolate WSI and virtual WSI under a populationdoubling scenario for the superregion assuminga constant relationship between water use andpopulation (slope = 0.23) over time frames consis-tent with population doubling (40–90 y). We rec-ognize the perils of linear extrapolation of currentwater rates and thus rely on themore conservative,

flatter relationship between population andwaterwithdrawals to estimate future WSI.

ACKNOWLEDGMENTS. We thank J. Baron,N. Baron, G. Basile, R. Glennon, J. Holway,T. Lant, C. Magirl, C. Perrings, S. Postel, R. Reeves,M. Scott, J. Reichman, and M. Wright for ideas,encouragement, data processing, and construc-tive comments on previous versions of themanuscript. J.L.S. thanks S. Bacon, J. Elser, J. Fink,and P. Gober for funding via Arizona State Uni-versity (ASU)’s College of Liberal Arts and Scien-ces, the School of Life Science Research andTraining Initiatives Office, the Global Institute ofSustainability, and the Decision Center fora Desert City, respectively. This project was sup-ported by National Science Foundation GrantEAR-0756817 (to J.L.S.) and ASU’s Office of theVice President for Research and Economic Af-fairs. This work was conducted as a part of the“Sustainability of Freshwater Resources in theUnited States” Working Group supported by theNational Center for Ecological Analysis and Syn-thesis, a center funded by National ScienceFoundation Grant EF-0553768, the University ofCalifornia, Santa Barbara, and the Stateof California.

1. Powell JW (1890) The irrigable lands of the arid region.Cent Mag 39:766–776.

2. de Buys W (2001) Seeing Things Whole: The EssentialJohn Wesley Powell (Island Press, Washington, DC).

3. Powell JW (1876) A Report on the Arid Regions of theUnited States, with a More Detailed Account of theLands of Utah (US Government Printing Office,Washington, DC).

4. Graf WL (1999) Dam nation: A geographic census ofAmerican dams and their large-scale hydrologicimpacts. Water Resour Res 35:1305–1311.

5. Reisner M (1986) Cadillac Desert: The American Westand its Disappearing Water (Penguin Books, New York).

6. US Census Bureau (2008) Population of the 100 largestcities and other urban places in the United States: 1790to 1999. Available at http://www.census.gov/popula-tion/www/documentation/twps0027/twps0027.html.Accessed November 30, 2010.

7. City Mayors (2010) The fastest growing US cities: Citiesranked 1 to 100. Available at http://www.citymayors.com/gratis/uscities_growth.html. Accessed November30, 2010.

8. US Department of Agriculture (2010) National agri-cultural statistics service. Available at http://www.nass.usda.gov/QuickStats/indexbysubject.jsp?Pass_-group=Economics. Accessed November 30, 2010.

9. Glennon RJ (2002) Water Follies: GroundwaterPumping and the Fate of America’s Fresh Water (IslandPress, Washington, DC).

10. Glennon RJ (2009) Unquenchable: America’s WaterCrisis and What To Do About It (Island Press,Washington, DC).

11. Pearce F (2007) When the Rivers Run Dry—The DefiningCrisis of the 21st Century (Beacon Press, Boston).

12. Worster D (1986) Rivers of Empire: Water, Aridity, andthe Growth of the American West (Pantheon Books,New York).

13. Seaber PR, Kapinos FP, Knapp GL (1987) HydrologicUnit Maps: U.S. Geological Survey Water-Supply Paper2294 (USGS, Reston, VA).

14. Oki T, Kanae S (2006) Global hydrological cycles andworld water resources. Science 313:1068–1072.

15. Gliek PH (1990) Vulnerabilities of water systems. In Cli-mate Change and U.S. Water Resources, ed Waggoner P(J. Wiley and Sons, New York), pp 223–240.

16. U.S. Geological Survey (2010) Estimated use of water inthe United States county-level data for 2005. Availableat http://water.usgs.gov/watuse/data/2005/. AccessedNovember 30, 2010.

17. Ghassemi F, Jakeman AJ, Nix HA (1995) Salinisation ofLand and Water Resources: Human Causes, Extent,

Management and Case Studies (Univ New South WalesPress, Sydney).

18. Vörösmarty CJ, et al. (2010) Global threats to humanwater security and river biodiversity. Nature 467:555–561.

19. Anonymous (2009) North Pacific Salmon FisheriesEconomic Measurement Estimates (Wild SalmonCenter, Portland, OR).

20. Carpenter SR, Kitchell JF (1988) Consumer control oflake productivity. Bioscience 38:764–769.

21. Schindler DE, Carpenter SR, Cole JJ, Kitchell JF, Pace ML(1997) Influence of food web structure on carbonexchange between lakes and the atmosphere. Science277:248–251.

22. Guegan JF, Lek S, Oberdorff T (1998) Energy availabilityand habitat heterogeneity predict global riverine fishdiversity. Nature 391:382–384.

23. Muneepeerakul R, et al. (2008) Neutralmetacommunitymodels predict fish diversity patterns in Mississippi-Missouri basin. Nature 453:220–222.

24. Poff NL, Olden JD, Merritt DM, Pepin DM (2007)Homogenization of regional river dynamics by damsand global biodiversity implications. Proc Natl Acad SciUSA 104:5732–5737.

25. Sabo JL, Finlay JC, Kennedy T, Post DM (2010) Therole of discharge variation in scaling of drainagearea and food chain length in rivers. Science 330:965–967.

26. Johnson PTJ, Olden JD, Vander Zanden MJ (2008) Daminvaders: Impoundments facilitate biological invasionsinto freshwaters. Front Ecol Environ 6:359–365.

27. Stohlgren TJ, et al. (2006) Species richness and patternsof invasion in plants, birds, and fishes in the UnitedStates. Biol Invasions 8:427–447.

28. Bestgen KR, Bundy JM (1998) Environmental factorsaffect daily increment deposition and otolith growth inyoung Colorado squawfish. Trans Am Fish Soc 127:105–117.

29. Seegrist DW, Gard R (1972) Effects of floods on troutin Sagehen Creek, California. Trans Am Fish Soc 101:478–482.

30. Halpern BS, et al. (2008) A global map of humanimpact on marine ecosystems. Science 319:948–952.

31. Hoekstra AY, Champaign AK (2008) Globalization ofWater: Sharing the Planet’s Freshwater Resources(Blackwell Publishing, Oxford).

32. Wackernagel M, Rees W (1996) Our EcologicalFootprint: Reducing Human Impact on Earth (NewSociety Publishers, Philadelphia).

33. US Census Bureau, Population Division (2010) Metro-politan and micropolitan statistical areas. Availableat http://www.census.gov/population/www/metroareas/metrodef.html. Accessed November 30, 2010.

34. US Census Bureau, Population Division (2010) US pop-ulation projections. Available at http://www.census.gov/population/www/projections/st-prod-proj-list.html.Accessed November 30, 2010.

35. Liang X, Lettenmaier DP, Wood EF (1996) One-dimensional statistical dynamic representation ofsubgrid spatial variability of precipitation in the two-layer variable infiltration capacity model. J GeophysRes 101(D16):21403–21422.

36. Liang X, Lettennmaier DP, Wood EF, Burges SJ (1994) Asimple hydrologically based model of land surfacewater and energy fluxes for general circulationmodels. J Geophys Res 99(D7):14415–14428.

37. Maurer EP, Wood AW, Adam JC, Lettenmaier DP,Nijssen B (2002) A long-term hydrologically baseddataset of land surface fluxes and states for theconterminous United States. J Clim 15:3237–3251.

38. Mitchell KE, et al. (2004) The multi-institution NorthAmerican Land Data Assimilation System (NLDAS):Utilizing multiple GCIP products and partners ina continental distributed hydrological modeling sys-tem. J Geophys Res 109:1–32.

39. Nijssen B, O’Donnell GM, Lettenmaier DP, Lohmann D,Wood EF (2001) Predicting the discharge of globalrivers. J Clim 14:3307–3323.

40. Cherkauer KA, Bowling LC, Lettenmaier DP (2003)Variable infiltration capacity (VIC) cold land processmodel updates. Global Planet Change 38:151–159.

41. Cherkauer KA, Lettenmaier DP (1999) Hydrologiceffects of frozen soils in the upper Mississippi Riverbasin. J Geophys Res 104:19599–19610.

42. United States National Atlas (2010) Streams and wa-terbodies of the United States. Available at http://nationalatlas.gov/mld/hydrogm.html. Accessed Novem-ber 30, 2010.

43. Water Information Coordination Program (2010) Thereservoir sedimentation database (RESSED). Availableat http://ida.water.usgs.gov/ressed/background/index.cfm. Accessed November 30, 2010.

44. United States Geological Survey (2010) NAS-non-indigenous aquatic species. Available at http://nas.er.usgs.gov/. Accessed November 30, 2010.

45. NatureServe (2010) Distribution of native US Fishesby Watershed. Available at http://www.natureserve.org/getData/dataSets/watershedHucs/index.jsp. AccessedNovember 30, 2010.

46. Renault D, Wallender WW (2000) Nutritional waterproductivity and diets: From “crop oer drop” towards“Nutrition per drop”. Agric Water Manage 45:275–296.

21270 | www.pnas.org/cgi/doi/10.1073/pnas.1009734108 Sabo et al.