lable at ScienceDirect

Quaternary Science Reviews 30 (2011) 855e866

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Reconciling arroyo cycle and paleoflood approaches to late Holocene alluvialrecords in dryland streams

Jonathan E. Harvey*, Joel L. PedersonUtah State University, Department of Geology, 4505 Old Main Hill, Logan, UT 84321, USA

a r t i c l e i n f o

Article history:Received 13 April 2010Received in revised form22 December 2010Accepted 30 December 2010Available online 12 February 2011

Keywords:ArroyosPaleofloodsAlluvial recordsDrylands

* Corresponding author. Present address: UCSB D1006 Webb Hall, Santa Barbara, CA 93106, USA. Tel.: þ893 2314.

E-mail addresses: harvey@eri.ucsb.edu (J.E. Har(J.L. Pederson).

0277-3791/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.quascirev.2010.12.025

a b s t r a c t

A century of research into the environmental records of alluvial-valley fills in drylands has led to newtheories about landscape response to climate change and cultural evolution over the Holocene. Ina largely separate line of inquiry, paleoflood hydrologists have scoured bedrock canyons for slackwaterdeposits, extending the flood record and exploring their climatic significance. Both approaches rely onthe analysis and dating of late Holocene alluvium, sometimes along the same drainages, yet they differ inthe geomorphic setting in which they should be applied. Studies of arroyo cut-and-fill cycles are focusedon broad alluvial valleys, whereas paleoflood hydrology has been focused mainly in narrow bedrockcanyons.

With a focus on the southwestern U.S., we review and compare the fundamentals of these twoapproaches and their paradigmatic disconnect, and then discuss potential linkages that could lead toa more complete understanding of how dryland streams adjust to external stimuli. Recent regionalcompilations provide insight into the broader relation between the two record types over entire phys-iographic provinces. Meanwhile, new tools such as OSL dating, short-lived isotopes, and improvedhydraulic modeling are paving the way for refinement and reconciliation of the two approaches withinindividual drainages. The relation between past arroyo cut-and-fill cycles and paleofloods must bethoroughly explored if we are to fully understand how drainages will respond to a changing climate overthe coming decades to centuries.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Holocene alluvial deposits stored along dryland streams providerecords of environmental change in their watersheds. Carefulinterpretation of these alluvial archives can reveal past landscaperesponses to changes in climate, hydrology, and cultural activity(e.g. Bryan, 1925). Such records are especially pertinent becausethey often record changes similar in scale to those predicted for theupcoming century (e.g. Hughes and Diaz, 2008).

Most studies of alluvial deposits in drylands fall under one oftwo paradigms. The first stems from a long history of study ofalluvial cut-and-fill cycles and the “arroyo problem” as reviewedbelow. The second, younger approach is based on the analysis ofslackwater sediments to reconstruct paleoflood frequency and/or

epartment of Earth Science,1 805 893 7242; fax: þ1 805

vey), joel.pederson@usu.edu

All rights reserved.

magnitude (e.g. Patton et al., 1979; Baker, 2008). Although eachapproach is recognized by its practitioners as applicable to Holo-cene alluvial records in particular settings, confusion remains abouttheir appropriate application and relation to one another along andacross drainages. In fact, these contrasting approaches may lead toquite different interpretations of the same alluvial deposits, yetthey have the potential to be complementary rather than contra-dictory. Reconciliation of these distinct yet fundamentally relatedapproaches is essential for a complete and systematic under-standing of sensitive dryland streams in a changing climate.

Geomorphic setting is the key factor distinguishing these twoscientific approaches and record types (Fig.1). In the example of thesouthwestern U.S., stream valleys range from broad, unconstrictedreaches a kilometer or more wide to bedrock slot canyons onlymeters wide. Often, such variation in valley geometry can be seenalong individual drainages. Large volumes of valley-fill alluviumcan be stored along the broader reaches, where bedrock exerts littleto no control on channel form. Early workers in the southwesternU.S. recognized that these valley-fills record episodes of aggrada-tion and degradation on the scale of centuries ormillennia, and thatthese have key relations to paleoclimate and archeology (e.g.

Fig. 1. Contrasting geomorphic settings of alluvial records in drylands: (A) the arroyo cut-and-fill approach, typically applied in wider valleys with alluvial channel boundaries, and(B) the paleoflood hydrology approach, typically applied in constricted bedrock canyons with immobile channel boundaries.

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866856

Huntington, 1914; Gregory, 1917; Bryan, 1922; Hack, 1942; Antevs,1952). In stark contrast to broad, alluvial reaches, very little sedi-ment can be stored in narrow bedrock canyons. The few placeswhere sediment can be accommodated (alcoves, confluences,expansions, etc.) are utilized by paleoflood hydrologists to recon-struct individual flood events and Holocene-scale flood chronolo-gies (e.g. Baker, 1977; Kochel and Baker, 1982; Ely, 1992). Thus, twoend-member paradigms now exist in the literature: reconstructionof arroyo cut-and-fill histories from deposits in broad alluvialvalleys, and reconstructions of paleoflood frequency and magni-tude from deposits stored in constricted bedrock canyons.

Despite the number of contributions demonstrating eitherparadigm in the study of dryland alluvial records, the twoapproaches remain poorly reconciled. For example, paleofloodhydrologists often assume that channel boundaries are stablethrough constricted bedrock canyons over millennial timescalesdespite the pervasiveness of arroyo cutting and filling cycles in thealluvial reaches of the same streams. Further, our experience is thatdrainages vary continuously in valley geometry and often containlaterally-transitional reaches with a more ambiguous affinity.Workers in such cases must make the difficult determination ofwhether the deposits therein record only episodes of large floods ina stable channel or instead a more complex suite of aggradationalevents relating to changes in sediment load and/or hydrology.Finally, the temporal relation of deposition across these geomor-phic settings is poorly understood; leaving a major gap in ourunderstanding of how different drainage components respond tochanging environmental factors (Hereford, 2002).

Here we review the body of work representing both end-member approaches to dryland alluvial records. We explore thenature of the disconnect between these two paradigms andpossible linkages that could lead to a greater model of drylandfluvial response to external forcing. Finally, we advocate additionalstudies based on focused stratigraphy, sedimentology, geochro-nology, and modeling to clarify the relations between approachesand record types within individual drainages. Though we focus onthe much-studied southwestern U.S., the issues we discuss arerelevant to drylands worldwide.

2. Holocene arroyo dynamics

2.1. Historic arroyo cutting

Geologists and archeologists have studied the alluvial records ofsemiarid streams for over a century (Dodge, 1902). Much of this

work was motivated by awidespread episode of channel incision inthe American Southwest from w1880e1920 AD. Over these fewdecades, drainages incised up to 30 m into their alluvium, leavingformer floodplains behind as terraces. Because many farms andsettlements had been built on these surfaces in the precedingdecades, erosion accompanying arroyo cutting resulted insubstantial property damage, agricultural losses, and abandonmentof entire communities (Bryan, 1929; Peterson, 1950; Cooke andReeves, 1976).

2.2. Mechanisms of arroyo formation

The causes of historic arroyo cutting have been the subject oflongstanding debate in the literature. A number of reviews discussthe variety and evolution of hypothesized mechanisms over thepast century (e.g. Cooke and Reeves, 1976; Graf, 1983; Miller andKochel, 1999); and though we include a brief review below, wedefer to these works for a more exhaustive treatment of the issue.

2.2.1. AnthropogenicEarly workers suggested that land use changes during

Anglo-American settlement of the Southwest triggered arroyoincision (e.g. Dodge, 1902; Huntington, 1914; Bailey, 1935). Specifi-cally, the introductionof livestock grazing accompanying settlementhaddisturbed vegetation and compacted sedimentonhillslopes andvalley floors, hypothetically leading to increased storm runoff, moreerosive floods, and channel entrenchment (Bailey, 1935; Antevs,1952). The construction of roads, irrigation ditches, and otherfloodplain modifications have also been cited as potential anthro-pogenic influences on arroyo formation (Cooke and Reeves, 1976;Webb, 1985). However, infilled paleoarroyos exposed in the wallsof modern cutbanks indicate that there have been several cycles ofalluviation and erosion prior to the influence of Anglo settlers(Bryan, 1925; Peterson, 1950). Though attribution of incision toanthropogenic forcings persisted in the literature formany decades,it is now recognized that human activity alone cannot explain allHolocene arroyo cut-and-fill cycles (Cooke and Reeves, 1976).

2.2.2. ClimaticDecades of stratigraphic and geochronologic study in the

southwestern U.S. have revealed broad regional correlations in thetiming of stream aggradation and degradation in the late Holocene(Haynes, 1968; Knox,1983; Karlstrom,1988; Hereford, 2002). Thesecorrelations have motivated research into the hydroclimaticchanges that could explain the regionally synchronous stream

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866 857

behavior during these cycles. At least four episodes representingtwo cycles over the past millennium have been correlated acrossseveral catchments in Arizona, New Mexico, and Utah: (1) a periodof aggradation before w1200 AD coincident with the rise ofPuebloan cultures; (2) prehistoric arroyo cutting aroundw1200 ADthat coincides roughly with the Puebloan abandonment; (3) anepisode of aggradation from w1300e1880 AD and (4) historicarroyo cutting from w1880e1920 AD (Hack, 1942; Haynes, 1968;Hall, 1977; Euler et al., 1979; Graf, 1987; Hereford, 2002;Huckleberry and Duff, 2008) (Fig. 2A). Incipient refilling of thehistoric arroyos began around 1940 AD in many southwesternstreams (Leopold, 1976; Hall, 1977; Hereford, 1984, 1986; Graf et al.,1991). Though several episodes of arroyo cutting and filling prior to1200 AD have been recognized (e.g. Waters and Haynes, 2001;Mann and Meltzer, 2007; Harvey et al., in press), records of theseolder events are more rare and generally lack the temporal reso-lution necessary for inter-drainage comparison.

Mechanisms linking decadal- to centennial-scale climatechanges to arroyo cut-and-fill cycles have long been debated. Someearly workers speculated that arroyo formation was simply asso-ciated with episodes of enhanced precipitation and a resulting

Fig. 2. Climatic and stratigraphic reconstructions for the southwestern U.S.: (A) Late HolFrequency of dated paleoflood slackwater deposits in the southwestern U.S. (redrawn fromdrought as derived from tree-rings (black line, smoothed over 60 yr). Two-tailed 95% bootst(redrawn from Cook et al., 2004); (D) Columns: Frequency of El Niño events per 100-yr binfrom Stahle et al., 1998); and (E) Cumulative probability density functions for radiocarbonHarden et al., 2010).

increase in runoff (Dutton, 1882; Huntington, 1914; Gregory andMoore, 1931). Bryan (1925) and Hack (1942) suggested as analternative that it was a change to more arid conditions thatincreased surface runoff by decreasing vegetation cover on hill-slopes. In more recent decades, the increasing density of meteo-rological and streamflow data has fed a number of studies thatattempt to link decadal-scale changes in precipitation frequencyand intensity to floodplain growth in modern arroyos. Hereford(1984, 1986), Balling and Wells (1990), and Graf et al. (1991) eachused such records to show that the arroyo refilling of the mid-twentieth century generally took place during a time of droughtand reduced streamflow. This is in contrast to the historic arroyocutting of the preceding few decades, which was characterized bymore intense precipitation interspersed with drought (Herefordand Webb, 1992).

In addition to a changing hydrology, some have pointed tochanges in sediment supply from hillslopes and upper fluvialsystems as a primary driver of channel change. Graf (1987) studiedthe patterns of sediment storage and evacuation for streams ofvarying drainage area in the Colorado Plateau. He concluded thatthe storage and transport of sediment are most sensitive to climatic

ocene arroyo cycles in the Colorado Plateau (generalized from Hereford, 2002); (B)Ely, 1997 e note abbreviated height of last column); (C) Percent of western U.S. area inrap confidence intervals shown as dashed lines, mean percent area shown as gray line(modified from Moy et al., 2002). Line: Reconstructed Southern Oscillation Index (data-dated deposits in alluvial (dashed) and bedrock (solid) river reaches (redrawn from

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866858

and land use changes in local, lower-order streams (drainage areasof 101e103 km2), whereas regional, higher-order streams(103e104 km2) store and evacuate sediment in response to thechanging supply from these smaller local tributaries. This suggeststhat sediment storage propagates down-system from local streamstoward regional streams over time, a finding supported by morerecent work in the Great Plains by Mandel (2008). This patternshould be discernable chronostratigraphically as an inverse corre-lation in sediment age between upper alluvial valleys and down-stream reaches.

Hereford (2002) suggested that the cooler, wetter climate of theLittle Ice Age increased freeze-thaw action on hillslopes, therebyincreasing sediment yield and promoting regional channel aggra-dation. A different link between hillslopes and valley floor alluvi-ation is described by McAuliffe et al. (2006), who argue that majorhillslope-stripping events have occurred during transitions fromdrought to episodes of increased precipitation. Further, Graf (1987),Graf et al. (1991), and Pederson (2000) concluded that sedimentstorage and mobilization within upper fluvial systems is a keyprocess effecting downstream changes in sediment load. Thus,there is broad agreement that late Holocene cut-and-fill cycleslikely involve changes in both hydrology and upstream sedimentdelivery.

2.2.3. AutogenicOther workers have emphasized the role of autogenic adjust-

ments in driving cycles of aggradation and erosion. They documentfield examples and experimental studies wherein pulses of incisionfollowed by widening and alluviation migrate up-network inresponse to the formation of local slope anomalies (Schumm andHadley, 1957; Schumm and Parker, 1973; Patton and Schumm,1981). These anomalies represent local base level changes that donot necessarily occur simultaneously across regional catchments.For example, local oversteepening may result from deposition ofa tributary debris fan or deposition resulting from downstreamlosses in sediment transport capacity due to bed infiltration. Theresulting oversteepened reaches are subject to higher shearstresses and bed incision during large floods, sending upstreamtransient waves of incision followed by widening and renewedalluviation. The signature of such internal complex response isa sequence of relatively small-scale, time-transgressive (youngingupstream) terraces (Schumm and Parker, 1973; Daniels, 2008).Waters (1985) concluded that such “intrabasin geomorphicparameters” were stronger influences than external climaticdrivers on the timing of alluviation and erosion for a network ofstreams in southeastern Arizona. In a different example of internalresponse, Patton and Boison (1986) showed that recent landslide-generated debris overwhelms the fluvial system of Harris Wash,a tributary of the Escalante River in southern Utah, causing thespatial and temporal patterns of alluviation and incision to differfrom that seen in other regional streams. Despite these scatteredcases, complex response is unlikely to outweigh other allogenicforcings such as climate and land use over late Holocene timescalesin the southwest U.S. (Daniels, 2008).

2.2.4. Working hypothesisThough theremay be no simple “smoking gun” to explain all late

Holocene arroyo cutting and filling cycles (Cooke and Reeves, 1976),there has been a recent convergence of hypotheses in the literature.Many authors now agree that a shift toward an episode of increasedflood frequency was an important factor in prehistoric and historicarroyo cutting (Webb, 1985; Webb et al., 1991; Hereford, 2002;Mann and Meltzer, 2007; Huckleberry and Duff, 2008). Otherspoint out that the effect of such a shift would be enhanced if thewatershed were already in a sensitive state due to extended

drought and/or land-use changes (Cooke and Reeves, 1976; Webbet al., 1991; Miller and Kochel, 1999; Tucker et al., 2006). There isless agreement on what specific climate shifts could have alteredhydrology in such away as to cause synchronous incision of arroyosover such a large region. For amore detailed treatment of that issue,we point the reader to Webb (1985).

As reviewed by Hereford (2002), many streams in the ColoradoPlateau region saw peak streamflow around the time of historicarroyo cutting. These observations are based on tree-ring derivedstreamflow reconstructions (Hereford et al., 1996a), anecdotalevidence, and in some cases, discharge reconstructions based onflood-stage indicators. While this supports the working hypothesis,performing a similar comparison of paleoclimatological and pale-ohydrological proxies with prehistoric arroyo cutting is moredifficult for several reasons. First is the rough temporal resolution ofclimatic reconstructions covering the late Holocene (Fig. 2). Tree-ring reconstructions are helpful in that they generally capturevariations in cool-season precipitation at an annual resolution.However, their utility is limited by their insensitivity to precipita-tion intensity, which is arguably more important for floodproduction. Further, large, destructive floods can occur duringextended droughts without affecting tree-ring growth, just asmoist intervals do not necessarily require anomalously large floods(Pederson, 2000; Huckleberry and Duff, 2008).

The types of storms that produce extreme floods in thesouthwestern U.S. include localized, diurnal thunderstorms asso-ciated with the summer monsoon; dissipating eastern Pacifictropical cyclones; and slow-moving winter and spring frontalsystems (Ely, 1997). A general rule is that drainages are mostresponsive to storms of a spatial scale similar to the size of thedrainage basin (Leopold, 1942). Therefore, monsoonal storms aremore likely to dominate flood hydrology in smaller headwaterdrainages, while larger frontal systems and tropical cyclones aremore likely to trigger large floods in higher-order trunk drainages(Baker, 1977; Graf, 1983; Webb, 1985).

Many workers have cited a centennial- to millennial-scale cyclein the frequency of El Niño-Southern Oscillation (ENSO) events asa potential driver of changes in the frequency and intensity of largerfall and winter cyclonic storm systems in the southwestern U.S.(Hereford andWebb,1992; Ely,1997; House and Hirschboeck,1997;Waters and Haynes, 2001; Hereford, 2002). In addition to causingfloods enhanced by saturated soils, these storms are also respon-sible for much of the winter snowpack that accumulates in higher-elevation drainages (Cayan et al., 1999). Further, Webb andBetancourt (1992) argue that El Niño years are related to morefrequent landfall of tropical cyclones from the eastern Pacific,which have been known to produce floods-of-record in drainagesin the southwestern U.S. (Webb, 1985). While historic arroyocutting appears to be coincident with an episode of increased ElNiño activity (Fig. 2B and Hereford, 2002), it is difficult to recon-struct the record of earlier El Niño events at a resolution highenough to confirmwhether or not prehistoric arroyo cutting can berelated to anomalous ENSO behavior (Fig. 2C).

Recognizing the extreme precipitation intensities that occurduring late summer convective thunderstorms, Mann and Meltzer(2007) tie their reconstructed histories of aggradation and incisionin New Mexico to changes in the strength of the North Americanmonsoon. In contrast to prehistoric variations in cool-seasonprecipitation, which can be reconstructed with tree-rings; short-duration, high-intensity monsoonal storms are very difficult toreconstruct.

Limitations in length and resolution of existing paleoclimaticreconstructions preclude confident determination of the drivers ofnon-stationarity of large floods. Regardless, one could test thehypothesis that arroyo cut-and-fill cycles are driven by variations in

Table 1Examples of drainages featuring both arroyo cycle and paleoflood hydrologyresearch in the southwestern U.S.

Drainage Alluvial cycles Paleoflood

Kanab Creek, UT/AZ Smith (1990);Summa (2009)

Webb et al. (1991),Ely (1992)

Paria River, UT/AZ Hereford (1986);Hereford (2002)

Webb et al. (2002)

Escalante River, UT Webb (1985); Pattonand Boison (1986)

Webb (1985)

Buckskin Wash, UT/AZ Hereford (2002);Harvey et al. (in press)

Ely (1992); Harvey et al.(in press)

Little Colorado River, AZ Hereford (1984) Ely (1997)San Juan River NM/UT Hall (1977) Orchard (2001)Virgin River, UT Hereford et al. (1996a) Ely (1992); Enzel et al.

(1994)Colorado River, AZ Hereford et al. (1996b) O’Connor et al. (1994)Verde River, AZ Johnson et al. (1997) Ely and Baker (1985);

House et al. (2002)Gila River, AZ Waters and

Ravesloot (2000)Ely (1992)

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866 859

the frequency and magnitude of flood-producing storms bycomparing dated cut-and-fill cycles with historic and prehistoricflood records from the same drainages (Webb, 1985; Webb et al.,1991; Ely et al., 1996; Hereford, 1999).

3. Paleoflood hydrology

3.1. Background

The field of paleoflood hydrology is based on using depositionaland erosional evidence to reconstruct floods that occurred beforethe instrumented record (Patton et al., 1979; Kochel and Baker,1982). Like for arroyo cycles, there are existing reviews of paleo-flood hydrology (Costa, 1987; Baker et al., 2002; Benito andThorndycraft, 2005; Baker, 2008). Early paleoflood studies werefocused on characterizing catastrophic flood events such as glacialoutburst floods (Dana, 1882; Bretz, 1923). Modern applications ofthese methods began in earnest in the late 1970s, and includeextending the flood record to refine flood recurrence intervals (e.g.Baker, 1977; Kochel and Baker, 1982; O’Connor et al., 1994) andpaleoflood reconstructions related to climatic trends (e.g. Ely, 1997;Harden et al., 2010). Depositional evidence for paleofloods includesslackwater sediments, perched woody debris, and silt lines plas-tered on bedrock walls. Erosional evidence includes scarring ontrees within the flood zone, abrasion of bedrock walls, and trun-cation of landforms such as debris fans.

3.2. Application in bedrock canyons

In the Southwest, paleoflood studies rely heavily on slackwatersediments preserved in bedrock canyons. These sand and siltdeposits rapidly fall out of suspension in areas of flow separationsuch as alcoves, backwaters upstream of severe constrictions, andinterior channel bends. Repeated deposition in these zones byprogressively larger floods produces series of distinct flooddeposits, constituting a paleoflood package (Baker et al., 1983;Kochel and Baker, 1988; Benito, 2003; Benito and Thorndycraft,2005). The stratigraphy of these packages can then be describedand dated using some combination of radiocarbon, optically stim-ulated luminescence, short-lived isotopes, associationwith culturalmaterials, and dendrochronology. Workers are often interested indetermining the discharge of specific paleofloods, for which theyrely on careful identification and surveying of paleostage indicatorssuch as silt lines and slackwater deposit surfaces. Water-surfaceprofiles reconstructed from these stage indicators are then inputinto a hydraulic modeling package such as HEC-RAS to estimatepaleoflood discharges (e.g. Webb and Jarrett, 2002). Approachesutilizing the above methods have produced paleoflood records fordozens of rivers worldwide; each constructed with the goal ofbetter assessing the magnitudes of floods a stream is capable ofproducing as well as their non-stationarity in recent millennia.

A critical factor in the calculation of paleodischarges is knowl-edge of channel geometry at the time of deposition (Patton et al.,1979). Most workers make the assumption that the modernchannel geometry is a reasonable approximation (Baker et al., 1983;Webb et al., 1991). This assumption would certainly be invalid inthose stream reaches undergoing arroyo cutting and filling cycles,where grade changes on the order of 10 s of meters in decades havebeen recorded. Only a few researchers have attempted to correct forchanges in stream grade and/or channel geometry since depositionusing geologic evidence (e.g.Webb et al., 2002). Indeed,minimizingthis uncertainty is one reason paleoflood hydrologists have focusedtheir efforts on constricted bedrock canyons with relatively stablelateral channel boundaries. However, the thickness of the alluvialveneer is rarely known, and some have noted the possibility of

episodic channel aggradation and sediment storage in bedrockcanyons during the late Holocene (e.g. Tinkler and Wohl, 1998;Webb et al., 2002; Harvey et al., in press).

3.3. Non-stationarity of dated flood deposits

The southwestern U.S. benefits from an abundance of intactslackwater deposits in bedrock canyons, resulting in a large body ofwork on paleoflood frequency and magnitude. Regional compila-tions of this work suggest large floods tend to cluster into particularepisodes during the Holocene (Ely, 1997; Harden et al., 2010;Fig. 2B, E). These workers have interpreted this record to suggestthat episodes of increased flood frequency generally occur duringcooler, wetter periods such as the Little Ice Age (w1500e1900 AD).In contrast, overall drier and warmer episodes like the precedingMedieval Warm Period (w1100e1400 AD) are poorly representedin the flood record. Ely (1997) suggested that these centennial-scaleclimatic variations are related to shifts in the frequency of El Niñoevents, which have been statistically linked to cooler, wetterconditions in the U.S. Southwest (Cayan et al., 1999). However,dependable millennial-scale El Niño reconstructions remainelusive, and this proposed correlation is not yet tested. In fact, localpaleoflood records for several rivers in the Colorado Plateau argueinstead that the LIA was not characterized by frequent, large floods(Webb, 1985; Webb et al., 1991; Webb and Jarrett, 2002).

Indeed, the source of non-stationarity in regional floodfrequency records is not simple or clear. In addition to any primaryclimate drivers, preservation bias and resolution issues obscure therecord. Regarding preservation, there is a significant skew towardyounger ages that is difficult to de-trend. Harden et al.’s (2010)analysis includes streams ranging from the large trunk drainageof the Colorado River that heads in the Rocky Mountains to lower-order streams of the Colorado Plateau, to streams that flowprimarily through the Sonoran desert in southern Arizona, tostreams in southwest Texas. Such a blending of drainage locations,sizes, and hypsometries can be problematic for two reasons: (1) Theeffectiveness of any given storm at producing a flood is dependenton drainage area. Smaller drainages may experience their flood-of-record during localized monsoonal thunderstorms, while the samestorm may have little or no geomorphic effect on a trunk drainagedownstream (Graf, 1983; Knox, 1983). The inverse is also possiblefor broader regional storms such as stalled winter frontal systems(House and Hirschboeck, 1997). (2) Climate changes that couldcontrol the frequency of large floods may be manifest differently

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866860

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866 861

across physiographic provinces. For example, streams draining theuplands of the Colorado Plateau would likely respond differently toan increase in winter snowpack than streams in the low deserts ofsouthern Arizona. Though these complications make them difficultto interpret, such regional reconstructions suggest that floodfrequency and magnitude have seen dramatic variability in theHolocene.

4. Exploring the disconnect between paradigms

4.1. Introduction

Though the study of arroyo cut-and-fill histories and theconstruction of paleoflood chronologies generally have beenseparate lines of inquiry, they are inherently related. Both rely uponthe analysis and dating of Holocene alluvial deposits. Additionally,since many dryland streams flow for only short periods followingstorm events, the study of dryland alluvial deposits is in most casesthe study of flood deposits (Graf, 1983; Tooth, 2000). Herein lies animportant distinction e paleoflood hydrology focuses on large,unusual floods, whereas arroyo stratigraphies may be composed ofdeposits from any range of flow magnitudes. Finally, the twoapproaches have been applied in neighboring or even the samereaches of a single drainage without much regard to the other. Thefollowing discussion reviews some of these cases, as well asprevious attempts to link the two record types from the scale ofa single watershed to the entire southwestern U.S.

4.2. Settings between end-members

There are many drainages that contain both end-member valleygeometries shown in Fig. 1. In some cases the transition betweenthem is abrupt and obvious, while in others it is more gradual anddiffuse. These more ambiguous, transitional reaches presenta challenge, as it may not be clear whether perched alluvialdeposits therein were emplaced by large floods or are remnantsfrom an episode of aggradation. At what point in such a transitionzone do the cycles of aggradation and degradation that characterizealluvial reaches give way to a stable canyon reach that preservesonly larger floods? It is possible that hydrologic modeling couldhelp define the geometric transition between these two end-member reach types.

The Colorado River is a classic example of a drainage thatfeatures varying valley geometries along its length, as well asdeposits that have been interpreted as both paleoflood archives andfill terraces. O’Connor et al. (1994) interpreted a sequence ofdeposits at Axehandle Alcove in Marble Canyon as a sequence ofpaleoflood deposits interbedded with locally-derived sediment andcultural material. They use the surface of the deposits as paleostageindicators to calculate discharges, assuming that the channelgeometry in 1990 AD reflects the geometry at the time ofemplacement. They acknowledge that this assumption is theirlargest source of uncertainty, and suggest that their reconstructeddischarges should be interpreted as minima due to probablechannel scour during floods. They do not, however address thepossibility of longer-term grade changes over the >4000 yr-longrecord.

Representing the cut-and-fill approach, Hereford et al. (1996b)studied the debris fans and alluvial terraces that line theColorado River at Nankoweap Rapids, w80 km downstream from

Fig. 3. Potential linkages between upstream-alluvial reaches and downstream bedrock ca bedrock canyon downstream are: (A) deposition of slackwater deposits with no change in gincision; and (C) downcutting in kind with upstream reach, leaving behind former streambreaches before (solid) and after (dotted) arroyo cutting.

the site studied by O’Connor et al. (1994). They interpreted alluvialdeposits 5e17 m above the modern river as a series of three insetaggradational terraces deposited between w800 BC and 1880 AD.They associate deposition of the alluvial terraces with changes inthe grade of the river related to increased activity on nearby debrisfans. Interestingly, the dated alluvial sequences appear to correlateto the aggradational sequences described elsewhere in the Colo-rado Plateau (Hereford, 2002). In a more recent study aimed ataddressing the confusion surrounding alluvial deposits along theColorado River, Tainer (2010) showed that Holocene alluvialdeposits in (constricted) Glen Canyon include strong evidence forboth changes in river grade as well as more recent paleoflooddeposits. This result suggests that despite the narrowness of thecanyon at the study site, bed aggradation and degradation hastaken place in recent millennia.

Clearly, the interpretations of the alluvial record in GrandCanyon by Hereford et al. (1996b) and Tainer (2010) are remarkablydifferent than that of O’Connor et al. (1994). Considering theoverlapping ages and landscape positions of the studied deposits; itis possible that the paleoflood sequence studied by O’Connor et al.(1994) was a vertical cross-section through one or more of thealluvial terraces described by Hereford et al. (1996b) and Tainer(2010). Alternatively, the suite of inset fill terraces described byHereford et al. (1996b) could be flood packages formed while theriver underwent overall lowering of grade. A third solution is thatboth approaches are valid, and terrace formation is associated withlocalized aggradation around tributary debris fans. In any case, theColorado River is an example where it is not always clear whetherthe Holocene alluvium should be interpreted as a stack of paleo-flood deposits or as the result of a more complicated sequence ofchanges in channel grade. We suggest that such ambiguous reachesmay be as common as those that are clearly recognizable asbelonging to either end-member stream geometry.

4.3. Unreconciled results from both approaches in single drainage

There are several examples where both approaches have beendemonstrated on different geomorphic reaches of the samestreams. In fact, a series of these cases in the southern ColoradoPlateau was the subject of an AMQUA field trip in 1996 (Ely et al.,1996). Field trip participants visited several locations where thealluvial deposits of both alluvial and constricted reaches had beenstudied through separate lines of inquiry. For example, the PariaRiver in southern Utah has been the subject of both upstreamarroyo stratigraphy (Hereford, 1986, 2002) and downstream pale-oflood research (Ely, 1992; Webb et al., 2002). Additional examplesexist throughout the Southwest (Table 1), yet little work has beendone to reconcile these results and their distinct interpretationsregarding channel processes and controls. In most cases, suchreconciliation is precluded by inadequate temporal resolution, yetworkers sometimes do not acknowledge the results and implica-tions of the alternative approach.

Aworthwhile exercise for the cases mentioned in Table 1 wouldbe to compile and compare the chronostratigraphic results fromboth approaches on a stream-by-stream basis, highlight the datagaps that currently inhibit direct comparison, and attempt to fillthose gaps using all available techniques. The working hypothesisdescribed above in Section 2.2.4 and originally described byHereford (1999) predicts that, in a given drainage, the ages ofarroyo-fill deposits should be inversely correlated with paleoflood

anyons. During arroyo cutting along the alluvial reach, three potential responses inrade (working hypothesis e see Section 2.2.4); (B) aggradation in response to upstreamed deposits as a “false” paleoflood package. Profiles show stream grade through both

Fig. 4. Summary of alluvial geomorphic events in both the upstream-alluvial and downstream-constricted reaches of Buckskin Wash, Utah (U.S.) as revealed through OSL,radiocarbon, dendrochronology, and short-lived isotopes (Harvey et al., in press). In this case study, constricted-reach deposits are inversely correlated with upstream valley-fills.

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866862

slackwater deposits (Fig. 3). As an example, in the alluvial reachesof the Paria River basin, Hereford (1986, 2002) identified twomajorarroyo-cutting events centered around 1300 AD and 1880 AD, eachbounded by longer-lasting aggradational episodes. Downstream, inthe narrows of Paria Canyon, Webb et al. (2002) identified 11slackwater flood deposits. The highest deposit surface was deter-mined to be historic (pre-1955 AD), and was linked to one of thelargest known historical floods (either in 1862, 1884, or 1909 AD).The next highest deposits date to around w1420 AD and 1280 AD,respectively. Thus, these highest three flood deposits appear to bebroadly correlatedwith upstream incision events, though the link isnot strong. Webb et al. (2002) also noted that several large historicfloods were not represented in the alluvial record of the bedrockcanyon at all. As with most of the other cases from Table 1, moredetailed, higher-resolution age control must be produced beforereconciliation of the records in the two end-member settings of theParia River is possible.

4.4. Regional-scale comparisons

Direct comparison of regional paleoflood compilations with therecord of arroyo dynamics is difficult, in part due to the contrastingtemporal resolution between them and uncertainties associatedwith radiocarbon-based age control (Ely et al., 1996; Hereford,

2002; Huckleberry and Duff, 2008) (Fig. 4). Arroyo cut-and-fillrecords are more numerous and supported by a close associationwith archaeological studies, radiocarbon ages, and dendrochro-nology. The paleoflood record, on the other hand, is limited toa smaller dataset. Additionally, the regional flood chronology con-structed by Ely (1997) bins large flood events into 200-year inter-vals (Fig. 2B). We know, however, that arroyo headcuts can passthrough a drainage in a matter of a few years to decades (Gregoryand Moore, 1931; Webb, 1985). If the hydroclimatic conditionsassociated with arroyo cutting occur over only years to decades,they could be lost in 200-yr bins.

A newer method of synthesizing Holocene alluvial records overentire regions at higher resolutions has been developed in recentyears (e.g. Macklin and Lewin, 2003; Lewin et al., 2005;Thorndycraft and Benito, 2006; Benito et al., 2008; Harden et al.,2010). Focused mainly on rivers throughout Europe, these studiesinvolve the construction of large databases of dated alluvialdeposits across regions. They first classify the dated units based ontheir geomorphic and stratigraphic contexts. For example,Thorndycraft and Benito (2006) break their dated deposits into thefollowing settings: alluvial overbank, flood basin, alluvial channel,fluvio-torrential, and slackwater flood deposits, before convertingthem into cumulative probability distribution functions (CPDFs)and comparing them across geomorphic contexts and over time.

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866 863

Studies of this type have the potential to reveal the broaderadjustments of fluvial systems to anthropogenic and climate forc-ings over the mid-late Holocene as a function of geomorphicsetting.

Harden et al. (2010) builds upon Ely’s (1997) record by applyingthis methodology to the abundant dated deposits in the south-western US. They compiled a database of 724 radiocarbon datesfrom 37 locations, making the important step of distinguishingbetween “bedrock” and “alluvial” geomorphic settings. From thisdatabase, they constructed CPDFs for alluvial deposits found ineither setting. The temporal pattern of these data illustrates a roughinverse correlation between deposits in the two reach types, atleast toward the latter part of the record (Fig. 2E). Such a patternwould appear to support the hypothesis that episodes of increasedflood frequency are associated with regional arroyo-cutting events.Harden et al. (2010) suggest that alluvial reaches may not capturethe largest floods due to channel scour and enlargement duringsuch events. Alternatively, the more stable channel boundaries inthe bedrock reaches allow for more complete preservation of largerfloods (Baker, 1977). They conclude that episodes of frequent, largefloods as preserved in bedrock reaches correlate with cooler, wetterperiods; whereas flood deposits are more likely to be preserved inalluvial reaches during cool, dry periods.

For the purposes of reconciling the temporal relations betweenthe two main types of dryland alluvial records, the Harden et al.(2010) study is subject to the limitations described earlier insection 3.3. Those issues could be addressed by comparing onlystreams of a similar drainage area and physiography. Anotherelement of uncertainty associatedwith this method is the challengeof distinguishing between climatically significant flood depositsand those that record minor, less significant events. Since paleo-flood magnitudes are calculated using flood-stage indicators,factors that influence the apparent stage of a paleoflood such assyn- or post-depositional incision or aggradation can havea dramatic effect on resulting interpretations.

The preservation potential of a fluvial deposit is intricately tiedto the particular process and geometry in the reach of interest(Lewin and Macklin, 2003). As an example, in the case of a channelthat spends hundreds of years migrating laterally across an alluvialvalley at a steady grade, most flow events will not be preserved. Inbedrock canyons, deposits are likely to be preserved only if a floodevent is large enough to passively overtop existing deposits ina sheltered zone, yet is small enough not to scour out the entirepackage. Effectively, however, a greater range of flow events islikely to be preserved in alluvial reaches, due to overall increasedaccommodation space along the drainage.

Despite existing limitations and uncertainties, regional paleo-flood compilations can provide a helpful context in which to viewlate Holocene arroyo cycles, especially once they begin to includepaleoflood deposits whose ages are constrained by other datingtechniques such as OSL, dendrochonology, and short-lived isotopes.

4.5. Single-drainage comparisons

Few workers have addressed the chronostratigraphic relationbetween record types in a single drainage. For his doctoral disser-tation, Webb (1985) attempted to directly reconcile alluvial recordsfrom both broad and constricted valley geometries within theEscalante River drainage of south-central Utah. The Escalante hasa pattern common to the Colorado Plateau, consisting of anupstream reach with an arroyo cut into a broad valley fill anda downstream reach that is significantly constricted by bedrockwalls. Webb studied deposits in the upper alluvial reach to recon-struct its arroyo cut-and-fill history and in the constricted reach todevelop a paleoflood chronology, constraining ages of deposits

with radiocarbon dating, the instrumented record, and historicphotographs. He identified three episodes of arroyo formation:w900 AD,1600 AD, andw1909 AD. The most recent event could betraced to a large flood observed by settlers that caused a knickpointto migrate several kilometers upstream. He notes the roughcorrelation between clusters of large floods between1200e900 yr BP and 600e400 yr BP with the development of thefirst and second paleoarroyos, respectively, although this link islimited by uncertainty in radiocarbon analysis.

Working on Harris Wash, a tributary in the Escalante Riverbasin, Patton and Boison (1986) similarly identified sandy flooddeposits that they temporally correlated with incision of valley-fillsupstream. They noted that the flood deposits they linked toupstream incision were generally best-preserved in areas of flowseparation near bedrock walls, yet, the sedimentological charac-teristics that distinguish these deposits from the alluvial-valley fillswhich they overlie were not described in detail. They suggest thatfurther identification and dating of these incision-related floodpackages could help refine the complex alluvial history of theregion.

Another interesting case is that of Kanab Creek, one of thedeepest and most dramatic arroyos in the Colorado Plateau. Byexamining anecdotal evidence and flood-scarred trees, Webb et al.(1991) document that the modern arroyo was cut by a series ofexceptionally large floods beginning the late 19th century. Despitethe magnitude of channel change caused by floods in the upperalluvial reach, re-photography of several historical photos suggeststhat there was very little channel scour or fill in the downstreambedrock canyon. Though Smith (1990) and Summa (2009) haveinvestigated earlier arroyo cycles in Kanab Creek, contemporaryflood deposits have not been identified in the canyon downstream.

Following the lead of Webb (1985) and Webb et al. (1991),Harvey et al. (in press) performed a similar study along BuckskinWash, a tributary to the Paria River in the Colorado Plateau thatfeatures an alluvial reach draining into a slot canyon. They identifyfour episodes of arroyo formation since w3 ka, with the last twooccurring at w1300 AD and w1880 AD (Fig. 4). The majority of theslackwater deposits in the slot canyon downstreamwere depositedbetween w1850 and 1950 AD, suggesting that the bedrock canyonmay record the same floods that caused arroyo cutting upstream.However, noting the complete absence of deposits fromw1000 ADto w1850 AD in the slot canyon, Harvey et al. argue that preser-vation of flood deposits in the slot canyon was temporarilyenhanced by extreme sediment loading resulting from the incisionof upstream valley-fills. This phenomenon would serve to exag-gerate the frequency andmagnitude of floods during arroyo cuttingand underestimate those that occurred during upstream aggrada-tion. This has important implications for paleoflood studies, as thesite in Buckskin Wash has been referred to as a classic locality forslackwater deposits in the southwest U.S. (Ely et al., 1996; Knox,2000).

5. Reconciliation

The studies summarized above are examples of how alluvialrecords have been used to reconstruct the late Holocene behavior ofstreams in the southwestern U.S. They also set the context for thequestions that remain in order to reconcile the differentapproaches. Are constricted-reach channels truly stable overHolocene time while neighboring alluvial reaches experience bedelevation changes of tens of meters? How do alluvial records (andassumed stream behaviors) geometrically transition from onerecord type to the other along a single drainage? What are theeffects in both reach types of the largest floods vs. smaller, morefrequent events? Can we define the hydraulic and geometric

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866864

transition or threshold between record types and clarify theirdistinctions?

We suggest that resolving these issues will require a series offocused chronostratigraphic studies like Webb’s (1985) and Harveyet al.’s (in press), wherein records from both end-member reachtypes are studied and compared in detail within a single drainage.Attentionmust also be paid to the transitional reaches that separateend-members. In order to achieve the precise, high-resolution agecontrol necessary to compare the temporal pattern of deposition,a suite of geochronologic methods must be employed. Newer toolslike optically-stimulated luminescence and short-lived isotopes canaugment AMS radiocarbon, dendrochronological, and archeologicalmethods; especially where datable organic or cultural material isabsent.

Equally important, more sedimentological detail is needed toachieve a process-based understanding of deposition across thetwo major reach types. There is a notable lack of such data in theliterature. For example, one can use the sedimentology of flooddeposits in bedrock canyons to distinguish between the scenariosshown in Fig. 3. Typical slackwater deposits consist of silt- andsand-size particles that fall rapidly out of suspension in areas of lowflow velocity. If coarser material and bedforms that resemble thehigher-velocity channel bed are preserved high in outcrop, it islikely that significant changes in bed elevation have occurred.Further, analyses of flood deposit composition and grain sizedistributions could be helpful in correlating flood packages overmany kilometers, allowing exploration of the control of valleygeometry on flood deposit preservation.

Thanks to advances in hydraulic modeling techniques, explo-ration of the role that valley geometry plays in controlling long-term stream behavior in constricted vs. alluvial reaches is becomingincreasingly possible. New generations of two-dimensionalmodeling packages may be better able to capture the complexboundary conditions and across-channel variations in velocity orshear stress than existing one-dimensional models (Miller andCluer, 1998). Such modeling may also allow exploration of thetransition zone between end-member geomorphic settings. If wecan quantify the thresholds in valley geometry that determine thestability of channel geometry and control the preservation poten-tial of events of various magnitudes, it could help guide those thatwork in such settings toward the correct interpretation of thealluvial deposits stored therein.

6. Conclusions

There is a notable disconnect between the approaches taken bythose studying arroyo cut-and-fill cycles in broad alluvial valleysand those studying paleoflood records in bedrock canyons. Whileeach approach has been demonstrated dozens of times in end-member settings, little effort has been made to relate the tworecord types in a given drainage. Similarly, little attention has beenpaid to those reaches that lie morphometrically between the twoend-member geomorphic settings. However, the two seeminglydisparate datasets are fundamentally related in that they rely onthe analysis of late Holocene alluvial deposits, and are oftenundertaken in different parts of the same streams. Further, recon-ciliation of the two approaches can serve as an important test of theworking hypothesis in the arroyo cut-and-fill literature: thatarroyos are cut during episodes of unusually frequent, large floodstied to external climatic phenomena.

As these datasets continue to grow and new tools are employed,it is becoming possible to combine them into a greater model offluvial response to environmental change in drylands. Regionalsyntheses must be complemented by more focused, watershed-scale studies. Every geochronological tool available must be

embraced in order to achieve sufficient temporal resolution tocompare records both along and across drainages. The value of suchstudies would also benefit from greater sedimentological analysisof alluvial deposits in either end-member setting as well as thetransitional zone between. Finally, improved numerical modelingcapabilities may permit quantitative exploration of the control thatreach geometry has on the preservation of alluvial deposits therein.The collective results of these additional studies will help usapproach a more complete understanding of how dryland fluvialsystems have adjusted to past environmental changes and howthey might behave in the future.

Role of the funding source

This work was partially funded by a graduate student researchgrant from the Geological Society of America, the Arthur D. Howardaward from the Quaternary Geology and Geomorphology divisionof GSA, and student support from the Utah State UniversityDepartment of Geology.

Acknowledgments

Thanks to Richard Hereford, Varyl Thorndycraft, and JamesKnox, whose thoughtful reviews improved this manuscriptconsiderably.

References

Antevs, Ernst, 1952. Arroyo-cutting and filling. Journal of Geology 60, 375e385.Bailey, R.W., 1935. Epicycles of erosion in the valleys of the Colorado Plateau

province. Journal of Geology 60, 375e385.Baker, V.R., 1977. Stream-channel Response to Floods, with Examples from Central

Texas. In: Geological Society of America Bulletin, vol. 88, pp. 1057e1071.Baker, V.R., 2008. Paleoflood hydrology: origin, progress, prospects. Geomorphology

101, 1e13.Baker, V.R., Kochel, R.C., Patton, P.C., Pickup, G., 1983. Palaeohydrologic Analysis of

Holocene Flood Slack-water Sediments. In: Special Publications of the Inter-national Association of Sediment, vol. 6, pp. 229e239.

Baker, V.R., Webb, R.H., House, P.K., 2002. The scientific and societal value ofpaleoflood hydrology. In: House, P.K., Webb, R.H., Baker, V.R., Levish, D.R. (Eds.),Ancient Floods Modern Hazards e Principles and Applications of PaleofloodHydrology. American Geophysical Union, Washington, DC, pp. 1e19.

Balling Jr., R.C., Wells, S.G., 1990. Historic rainfall patterns and arroyo activity withinthe Zuni River drainage basin, New Mexico. Annals of the Association ofAmerican Geographers 80, 603e617.

Benito, G., 2003. Sedimentology of high-stage flood deposits of the Tagus River,Central Spain. Sedimentary Geology 157, 107e132.

Benito, G., Thorndycraft, V.R., 2005. Paleoflood hydrology and its role in appliedhydrological sciences. Journal of Hydrology 313, 3e15.

Benito, G., Thorndycraft, V.R., Rico, M., Sanchez-Moya, Y., Sopena, A., 2008. Palae-oflood and floodplain records from Spain: evidence for long-term climatevariability and environmental changes. Geomorphology 101, 68e77.

Bretz, J.H., 1923. The channeled scabland of the Columbia Plateau. Journal ofGeology 31, 617e649.

Bryan, K., 1922. Erosion and Sedimentation in the Papago Country, Arizona. In: U.S.Geological Survey Water Supply Paper, vol. 499.

Bryan, K., 1925. Date of channel trenching (arroyo-cutting) in the arid Southwest.Science 62, 338e344.

Bryan, K., 1929. Floodwater farming. Geographical Review 19, 444e456.Cayan, D.R., Redmond, K.T., Riddle, L.G., 1999. ENSO and hydrologic extremes in the

western United States. Journal of Climate 12, 2881e2893.Cook, E.R., Woodhouse, C.A., Eakin, M.C., Meko, D.M., Stahle, D.W., 2004. Long-term

aridity changes in the western United States. Science 306, 1015e1018.Cooke, R.U., Reeves, R.W., 1976. Arroyos and Environmental Change. Clarendon

Press, Oxford, p. 213.Costa, J.E., 1987. A history of paleoflood hydrology in the United states, 1800e1970.

In: Landa, E.R., Ince, S. (Eds.), History of Hydrology. History of GeophysicsNumber 3. American Geophysical Union, Washington, DC, pp. 49e53.

Dana, J.D., 1882. The flood of the Connecticut River valley from the melting of theQuaternary glacier. American Journal of Science 123, 179e202.

Daniels, J.M., 2008. Distinguishing allogenic from autogenic causes of bed elevationchange in late Quaternary alluvial stratigraphic records. Geomorphology 101,159e171.

Dodge, R.E., 1902. Arroyo formation. Science 15, 746.Dutton, C.E., 1882. Tertiary History of the Grand Canyon District. In: U.S. Geological

Survey Monograph, vol. 2, p. 264.

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866 865

Ely, L.L., 1992. Large Floods in the southwestern United States in relation to late-Holocene climatic variations. Ph.D. Thesis, University of Arizona, Tucson, AZ, p.326.

Ely, L.L., 1997. Response of extreme floods in the south-western United States toclimatic variations in the late Holocene. Geomorphology 19, 175e201.

Ely, L.L., Baker, V.R., 1985. Reconstructing paleoflood hydrology with slackwaterdeposits: Verde River, Arizona. Physical Geography 6, 103e126.

Ely, L.L., Hereford, R., House, P.K., Baker, V.R., 1996. Effects of Paleohydrology andClimate Change on Bedrock and Alluvial Channels, Southern Colorado Plateau.Field trip guidebook, American Quaternary Association, 14th Biennial Meeting,Northern Arizona University, Flagstaff, Arizona, p. 31.

Enzel, Y., Ely, L.L., Martinez-Goytre, R., Vivian, R.G.,1994. Paleofloods and a dam-failureflood on the Virgin River, Utah and Arizona. Journal of Hydrology 153, 291e315.

Euler, R.C., Gumerman, G.J., Karlstrom, T.N.V., Dean, J.S., Hevly, R.H., 1979. TheColorado Plateaus: cultural dynamics and Paleoenvironment. Science 205,1089e1101.

Graf, J.B., Webb, R.H., Hereford, R., 1991. Relation of Sediment Load and Flood-plainFormation to Climatic Variability, Paria River Drainage Basin, Utah and Arizona.In: Geological Society of America Bulletin, vol. 103, pp. 1405e1415.

Graf, W.L., 1983. The arroyo problem e Paleohydrology and paleohydraulics in theshort term. In: Gregory, K.G. (Ed.), Background to Paleohydrology. John Wileyand Sons, New York, pp. 279e302.

Graf, W.L., 1987. Late Holocene Sediment Storage in Canyons of the ColoradoPlateau. In: Geological Society of America Bulletin, vol. 99, pp. 261e271.

Gregory, H.E., 1917. Geology of the Navajo Country. In: U.S. Geological SurveyProfessional Paper, vol. 93.

Gregory, H.E., Moore, R.C., 1931. The Kaiparowits Region, a Geographic and GeologicReconnaissance of Parts of Utah and Arizona. In: U.S. Geological SurveyProfessional Paper, vol. 164, p. 161.

Hack, J.T., 1942. The changing physical environment of the Hopi Indians of Arizona.Peabody Museum Papers 25 (no. 1), 85. 12 plates.

Hall, S.A., 1977. Late Quaternary Sedimentation and Paleoecologic History of ChacoCanyon, New Mexico. In: Geological Society of America Bulletin, vol. 88, pp.1593e1618.

Harden, T.M., Macklin, M.G., Baker, V.R., 2010. Holocene flood histories in south-western U.S.A. Earth Surface Processes and Landforms 35.

Harvey, J.E., Pederson, J.L., and Rittenour, T.M. Exploring relations between arroyocycles and canyon paleoflood records in Buckskin Wash, Utah: reconcilingscientific paradigms. Geological Society of America Bulletin. in press.

Haynes Jr., C.V., 1968. Geochronology of late-Quaternary alluvium. In:Morrison, R.B., Wright Jr., H.E. (Eds.), Means of Correlation of QuaternarySuccessions. University of Utah Press, Salt Lake City, pp. 591e631.

Hereford, R., 1984. Climate and Ephemeral-stream Processes: Twentieth-centuryGeomorphology and Alluvial Stratigraphy of the Little Colorado River, Arizona.In: Geological Society of America Bulletin, vol. 95, pp. 654e658.

Hereford, R., 1986. Modern alluvial history of the Paria River drainage basin.Quaternary Research 25, 293e311.

Hereford, R., 1999. A geological model linking episodic valley-fill aggradation, hill-slope processes, and paleoflood chronologies. In: House, P.K. (Ed.), The SecondInternational Paleoflood Conference, September 26e29, Prescott, AZ, p. 18.

Hereford, R., 2002. Valley-fill Alluviation During the Little Ice Age (Ca A.D.1400e1880), Paria River Basin and Southern Colorado Plateau, United States. In:Geological Society of America Bulletin, vol. 114, pp. 1550e1563.

Hereford, R., Webb, R.H., 1992. Historic variation of warm-season rainfall, southernColorado Plateau, southwest United States. Climatic Change 22, 239e256.

Hereford, R., Jacoby, G.C., McCord, V.A.S., 1996a. Late Holocene Alluvial Geomor-phology of the Virgin River in the Zion National Park Area, Southwest Utah. In:Geological Society of America Special Paper, vol. 310, p. 41.

Hereford, R., Thompson, K.S., Burke, K.J., Fairley, H.C., 1996b. Tributary Debris Fansand the Late Holocene Alluvial Chronology of the Colorado River, Eastern GrandCanyon, Arizona. In: Geological Society of America Bulletin, vol. 108, pp. 3e19.

House, P.K., Hirschboeck, K.K., 1997. Hydroclimatological and paleohydrologicalcontext of extreme winter flooding in Arizona, 1993. In: Larson, R.A.,Slosson, J.E. (Eds.), Storm-induced Geologic Hazards: Case Histories from the1992e1993 Winter in Southern California and Arizona. Geological Society ofAmerica Reviews in Engineering Geology, vol. XI, pp. 1e24.

House, P.K., Pearthree, P.A., Klawon, J.E., 2002. Historical flood and paleofloodchronology of the lower Verde River, Arizona: stratigraphical evidence andrelated uncertainties. In: House, P.K., Webb, R.H., Baker, V.R., Levish, D.R. (Eds.),Ancient Floods, Modern Hazards: Principles and Applications of PaleofloodHydrology. Water Science and Application, vol. 5. American Geophysical Union,Washington, DC, pp. 267e293.

Huckleberry, G., Duff, A.I., 2008. Alluvial cycles, climate, and Puebloan settlementshifts near Zuni Salt Lake, New Mexico, USA. Geoarcheology 23, 107e130.

Hughes, M.K., Diaz, H.F., 2008. Climate variability and change in the drylands ofwestern North America. Global and Planetary Change 64, 111e118.

Huntington, E., 1914. The Climatic Factor as Illustrated in Arid America. CarnegieInstitution of Washington Pub. 192, p. 341.

Johnson, W.C., Arborgast, A.G., Homburg, J.A., 1997. Alluvial Geomorphology andGeoarchaeology of the Lower Verde River. In: Vanishing River: Landscapes andLives of the LowerVerde Valley. The Lower Verde Archaeology Project, vol. 2, pp.17e32.

Karlstrom, T.N.V., 1988. Alluvial chronology and hydrologic change of Black Mesaand nearby regions. In: Gumerman, G.J. (Ed.), The Anasazi in a ChangingEnvironment. Cambridge University Press, New York, pp. 45e91.

Knox, J.C., 1983. Responses of river systems to Holocene climates. In: Wright Jr., H.E.(Ed.), Late Quaternary Environments of the United States. The Holocene, vol. 2.University of Minnesota Press, Minneapolis, pp. 26e41.

Knox, J.C., 2000. Sensitivity of modern and Holocene floods to climate change.Quaternary Science Reviews 19, 439e457.

Kochel, R.C., Baker, V.R., 1982. Paleoflood hydrology. Science 215, 353e362.Kochel, R.C., Baker, V.R., 1988. Palaeoflood analysis using slackwater deposits. In:

Baker, V.R., Kochel, R.C., Patton, P.C. (Eds.), Flood Geomorphology. Wiley, NewYork, pp. 357e376.

Leopold, L.B., 1942. Areal extent of intense rainfalls, New Mexico and Arizona.American Geophysical Union Transactions 23, 558e563.

Leopold, L.B., 1976. Reversal of erosion cycle and climatic change. QuaternaryResearch 6, 557e562.

Lewin, J., Macklin, M.G., 2003. Preservation potential for late Quaternary alluvium.Journal of Quaternary Science 18, 107e120.

Lewin, J., Macklin, M.G., Johnstone, E., 2005. Interpreting alluvial archives: sedi-mentological factors in the British Holocene fluvial record. Quaternary ScienceReviews 24, 1875e1889.

Macklin, M.G., Lewin, J., 2003. River sediments, great floods, and centennial-scaleHolocene climate change. Journal of Quaternary Science 18, 101e105.

Mandel, R.D., 2008. Buried paleoindian-age landscapes in stream valleys of thecentral plains, USA. Geomorphology 101, 342e361.

Mann, D.H., Meltzer, D.J., 2007. Millennial-scale Dynamics of Valley Fills over thePast 12,000 14 C Yr in Northeastern New Mexico, USA. In: Geological Society ofAmerica Bulletin, vol. 119, pp. 1433e1448.

McAuliffe, J.R., Scuderi, L.A., McFadden, L.D., 2006. Tree-ring record of hillslopeerosion and valley floor dynamics: landscape responses to climate variationduring the last 400 yr in the Colorado Plateau, northeastern Arizona. Global andPlanetary Change 50, 184e201.

Miller, A.J., Cluer, B.L.,1998.Modeling considerations for simulation offlow inbedrockchannels. In: Tinkler, K.J.,Wohl, E.E. (Eds.), Rivers Over Rocke Fluvial Processes inBedrock Channels. American Geophysical Union, Washington, DC, p. 323.

Miller, J.R., Kochel, R.C., 1999. Review of Holocene hillslope, piedmont, and arroyosystem evolution in southwest United States: implications to climate-inducedlandscape modifications in southern California. Technical Reports. In:Rose, M.R., Wigland, P.E. (Eds.), Proceedings of the Southern California ClimateSymposium: Trends and Extremes of the Last 2000 Years. Natural HistoryMuseum of Los Angeles County, Los Angeles, CA, pp. 139e192.

Moy, C.M., Seltzer, G.O., Rodbell, D.T., Anderson, D.M., 2002. Variability of El Niño/southern oscillation activity at millennial timescales during the Holoceneepoch. Nature 420, 162e165.

O’Connor, J.E., Ely, L.L., Wohl, E.E., Stevens, L.E., Melis, T.S., Kale, V.S., Baker, V.R.,1994. A 4500-year record of large floods on the Colorado River in the grandcanyon, Arizona. Journal of Geology 102, 1e9.

Orchard, K.L., 2001. Paleoflood Hydrology of the San Juan River, Southeastern Utah,USA. M.S. Thesis, University of Arizona, Tucson, AZ, p. 138.

Patton, P.C., Schumm, S.A., 1981. Ephemeral-stream processes: implications forstudies of Quaternary valley fills. Quaternary Research 15, 24e43.

Patton, P.C., Boison, P.J., 1986. Processes and Rates of Formation of Holocene AlluvialTerraces in Harris Wash, Escalante River Basin, South-central Utah. In:Geological Society of America Bulletin, vol. 97, pp. 369e378.

Patton, P.C., Baker, V.R., Kochel, R.C., 1979. Slackwater deposits: a geomorphictechnique for the interpretation of fluvial paleohydrology. In: Rhodes, D.D.,Williams, G.P. (Eds.), Adjustments of the Fluvial System. Kendall-Hunt, Dubuque,IA, pp. 225e252.

Pederson, J.L., 2000. Holocene Paleoloakes of Lake Canyon, Colorado Plateau:Paleoclimate and Landscape Response from Sedimentology and Allostratig-raphy. In: Geological Society of America Bulletin, vol. 112 147e158.

Peterson, H.V., 1950. The problem of gullying in western valleys. In: Trask, P.P. (Ed.),Applied Sedimentation. Wiley, New York, pp. 407e434.

Schumm, S.A., Hadley, R.F., 1957. Arroyos and the semiarid cycle of erosion. Amer-ican Journal of Science 255, 161e172.

Schumm, S.A., Parker, R.S., 1973. Implications of complex response of drainagesystems for Quaternary alluvial stratigraphy. Nature 243, 99e100.

Smith, S.S., 1990. Relationship of Large Floods and Rapid Entrenchment of KanabCreek, Southern Utah. M.S. Thesis, University of Arizona, Tucson, AZ, p. 82.

Stahle, D.W., D’Arrigo, R.D., Krusic, P.J., Cleaveland, M.K., Cook, E.R., Allan, J.,Cole, J.E., Dunbar, R.B., Therrell, D.A., Gay, D.A., Moore, M.D., Stokes, M.A.,Burns, B.T., Lillanueva-Diaz, J., Thompson, L.G., 1998. Experimental den-droclimatic reconstruction of the Southern Oscillation. Bulletin of the AmericanMeteorological Society 79, 2137e2152.

Summa, M.C., 2009. Geologic Mapping, Alluvial Stratigraphy, and Optically-stimu-lated Luminescence Dating of the Kanab Creek Area, Southern UT. M.S. Thesis,Utah State University, Logan, UT. p. 183.

Tainer, E.M., 2010. High-resolution Holocene Alluvial Stratigraphy at ArchaeologicalSites in Eastern Grand Canyon, Arizona. M.S. Thesis, Utah State University,Logan, UT. p. 165.

Thorndycraft, V.R., Benito, G., 2006. The Holocene fluvial chronology of Spain:evidence from a newly compiled radiocarbon database. Quaternary ScienceReviews 25, 223e234.

Tinkler, K.J., Wohl, E.E., 1998. A primer on bedrock channels. In: Tinkler, K.J.,Wohl, E.E. (Eds.), Rivers Over Rock e Fluvial Processes in Bedrock Channels.American Geophysical Union, Washington, DC, p. 323.

Tooth, S., 2000. Process, form, and change in dryland rivers: a review of recentresearch. Earth-Science Reviews 51, 61e107.

J.E. Harvey, J.L. Pederson / Quaternary Science Reviews 30 (2011) 855e866866

Tucker, G.E., Arnold, L., Bras, R.L., Flores, H., Istanbulluoglu, E., Solyon, P., 2006.Headwater Channel Dynamics in Semiarid Rangelands, Colorado High Plains,USA. In: Geological Society of America Bulletin, vol. 118, pp. 959e974.

Waters, M.R., 1985. Late Quaternary alluvial stratigraphy of whitewater draw,Arizona: implications for regional correlation of fluvial deposits in the Amer-ican southwest. Geology 13, 705e708.

Waters, M.R., Ravesloot, J.C., 2000. Late Quaternary geology of the Middle Gila River,Gila River Indian reservation, Arizona. Quaternary Research 54, 49e57.

Waters, M.R., Haynes Jr., C.V., 2001. Late Quaternary arroyo formation and climatechange in the American Southwest. Geology 29, 399e402.

Webb, R.H., 1985. Late Holocene Flooding on the Escalante River, South-centralUtah, Ph.D. Thesis, University of Arizona, Tucson. p. 204.

Webb, R.H., Betancourt, J.L., 1992. Climatic Variability and Flood Frequency of theSanta Cruz River. In: U.S. Geological Survey Water-Supply Paper, vol. 2379.Pima County, Arizona, p. 69.

Webb, R.H., Jarrett, R.D., 2002. One-Dimensional estimation techniques fordischarges of paleofloods and historical floods. In: House, P.K., Webb, R.H.,Baker, V.R., Levish, D.R. (Eds.), Ancient Floods and Modern Hazards: Prin-ciples and Applications of Paleoflood Hydrology. Water Science andApplication Series, vol. 5. American Geophysical Union, Washington, DC,pp. 111e125.

Webb, R.H., Smith, S.S., McCord, V.A.S., 1991. Historic channel changes in KanabCreek, southern Utah and northern Arizona. Grand Canyon Natural HistoryAssociation Monograph 9, 91.

Webb, R.H., Blainey, J.B., Hyndman, D.W., 2002. Paleoflood hydrology of theParia River, southern Utah and northern Arizona, USA. In: House, P.K.,Webb, R.H., Baker, V.R., Levish, D.R. (Eds.), Ancient Floods and ModernHazards: Principles and Applications of Paleoflood Hydrology. Water Scienceand Application Series, vol. 5. American Geophysical Union, Washington, DC,pp. 295e310.