ecology, management, and conservation implications of north...

Ecology, management, and conservation implications of NorthAmerican beaver (Castor canadensis) in dryland streams

POLLY P. GIBSON,* and JULIAN D. OLDENSchool of Aquatic and Fishery Sciences, University of Washington, Seattle, WA, USA

ABSTRACT

1. After near-extirpation in the early 20th century, beaver populations are increasing throughout many parts ofNorth America. Simultaneously, there is an emerging interest in employing beaver activity for stream restorationin arid and semi-arid environments (collectively, ‘drylands’), where streams and adjacent riparian ecosystems areexpected to face heightened challenges from climate change and human population growth.

2. Despite growing interest in reintroduction programmes, surprisingly little is known about the ecology ofbeaver in dryland streams, and science to guide management decisions is often fragmented and incomplete.

3. This paper reviews the literature addressing the ecological effects and management of beaver activity indrylands of North America, highlighting conservation implications, distinctions between temperate and drylandstreams, and knowledge gaps.

4. Well-documented effects of beaver activity in drylands include changes to channel morphology and groundwaterprocesses, creation of perennial wetland habitat, and substantial impacts to riparian vegetation. However, manyhypothesized effects derived from temperate streams lack empirical evidence from dryland streams.

5. Topics urgently in need of further study include the distribution and local density of beaver dams;consequences of beaver dams for hydrology and water budgets; and effects of beaver activity on the spread ofaquatic and riparian non-native species.

6. In summary, this review suggests that beaver activity can create substantial benefits and costs for conservation.Where active beaver introductions or removals are proposed, managers and conservation organizations are urged toimplement monitoring programmes and consider the full range of possible ecological effects and trade-offs.Copyright # 2014 John Wiley & Sons, Ltd.

Received 02 July 2013; Revised 04 November 2013; Accepted 17 November 2013

KEY WORDS: riparian; habitat management; restoration; reintroduction; mammals; vegetation; alien species; impoundment;erosion

INTRODUCTION

Once driven to near extinction, over the pastcentury populations of North American beaver

(Castor canadensis) and Eurasian beaver (Castorfiber) have rebounded to inhabit much of theirformer geographic range. Hunting restrictions

*Correspondence to: Polly Gibson, School of Aquatic and Fishery Sciences, University of Washington, 1122 NE Boat St., Seattle, WA 98105, USA.E-mail: [email protected]

Copyright # 2014 John Wiley & Sons, Ltd.

AQUATIC CONSERVATION: MARINE AND FRESHWATER ECOSYSTEMS

Aquatic Conserv: Mar. Freshw. Ecosyst. 24: 391–409 (2014)

Published online 13 January 2014 in Wiley Online Library(wileyonlinelibrary.com). DOI: 10.1002/aqc.2432

combined with deliberate reintroductionscontributed to this conservation success (Nolet andRosell, 1998; Baker and Hill, 2003); however,increases in beaver range and population size alsoraise new questions for ecosystem management.Beaver are widely recognized as ecosystem engineerswith major influence on landscape form and thestructure of aquatic ecosystems. Beaver dams createlentic habitat, promote landscape heterogeneity, andalter stream hydrology, sediment dynamics, andnutrient cycling (Naiman et al., 1986), withnumerous implications for flora and fauna (Rosellet al., 2005). There is an emerging interest inemploying or mimicking this remarkable engineeringpower in stream management efforts (DeVries et al.,2012; Pollock et al., 2012), but beaver activity canalso conflict with other ecological management goalsor create significant economic damage (Nolet andRosell, 1998). Discussions over whether and whereto promote beaver populations (Longcore et al.,2007; Carrillo et al., 2009) are now at the forefrontof the conservation debate, leading to new questionsabout the historical distribution, abundance, andecological effects of beaver.

Although the ecological effects of beaver activityhave been thoroughly studied (Naiman et al., 1986;Rosell et al., 2005), the majority of research hasbeen conducted in temperate river systems, whilelittle attention has been devoted to beaver withinmore arid environments (see Supplementarymaterial, Table S1). Regions with an arid or semi-arid climate, formally defined by the ratio ofprecipitation to potential evapo-transpiration, arecollectively described as ‘drylands’ (MEA, 2005).Given the limited contemporary presence of theEurasian beaver in drylands, owing to widespreadextirpation centuries ago (Nolet and Rosell, 1998),this review focuses on the North American beaver(but see Supplementary Material, Appendix 1). InNorth America, drylands make up most of the landarea of the western United States and Mexico(Figure 1). Major ecoregions of this vast areainclude arid warm deserts (Sonoran, Mojave, andChihuahuan); cold desert sagebrush steppes of theGreat Basin and the Columbia–Snake RiverPlateau; tablelands of the Colorado Plateau;grasslands of the more arid western half of theGreat Plains; and dry shrublands, grasslands, and

forests of southern California and northern Mexico(CEC, 1997). The aim of this paper is to synthesizecommonalities among the effects of beaverinhabiting the broad range of habitats characterizedby aridity, but it should be emphasized that there isgreat ecological variability within drylands, as wellas between dryland and temperate regions.

The return of beaver poses new challenges andopportunities for conservation in dryland streamsand wetlands. Biodiversity of drylands tends to bedisproportionately concentrated in aquatic andriparian ecosystems: in dryland regions of westernNorth America riparian areas are estimated tocover less than 2% of total land area, yet theysupport species diversity comparable to uplandareas (Knopf et al., 1988). Water demand tosupport growing urban centres and agriculturaldevelopment is already high, and water withdrawals

Figure 1. Approximate historical range of North American beaver inwestern North America and reported locations of beaver dams indrylands. Climate zones are based on an aridity index (MEA, 2005);regions with a hyperarid, arid, or semi-arid climate (i.e. aridityindex≤ 0.5) are considered drylands in this paper. Sources:Trabucco and Zomer (2009) (aridity data) and Patterson et al. (2007)

(beaver range map).

P. P. GIBSON AND J. D. OLDEN392

Copyright # 2014 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. 24: 391–409 (2014)

threaten riparian ecosystems and vulnerable aquaticspecies (Stromberg, 2001). These stresses areexpected to grow more acute in the near future:climate change forecasts predict more drought,reduced rainfall, and reduced stream flow for thedesert South-west (Seager et al., 2007), whilepredicted human population increases will onlyincrease demand for water (Sabo et al., 2010).Extensive loss and alteration of wetland habitat dueto human disturbance make these systems a priorityfor conservation and restoration.

There is a growing sentiment among scientists andresource managers that beaver engineering offers oneapproach to counteract some conservation threats todryland streams. Beaver dams increase waterstorage, raising the local water table and potentiallysupplementing low stream flows during dry seasons.Beaver ponds can provide habitat for rare species orpromote growth of riparian vegetation (Pollocket al., 2003). In present times, beaver dams mayperform additional ecological functions such asfiltration of toxins or nutrients and control ofexcessive erosion resulting from human land-usepractices (Pollock et al., 2007; Gangloff, 2013).Beavers are returning to ecosystems that bear littleresemblance to historical conditions: widespreadriver regulation has fundamentally altered thefunction of dryland river ecosystems (Stromberg,2001), declines of entire native biota and spread ofnon-native species has altered community dynamics(Pool et al., 2010), and development has shiftedpriorities for managing river systems. The challengeis to understand the ecological function of beaver inthis contemporary dryland landscape.

Opinions and beliefs are strongly divided on theecological value of beaver in contemporary landscapes.Current approaches to dryland beaver managementrange from beaver promotion (e.g. reintroduction aspart of stream restoration; Fredlake, 1997) to activeremoval (e.g. to protect riparian vegetation;Mortenson et al., 2008), and there is disagreementover whether the benefits of using beaver in drylandriparian management outweigh the costs (see ‘Birds’below). Scientific knowledge and consensus areurgently needed to guide decision making aboutmanaging beaver in drylands, but the availablescience is scattered and often limited in scope. Astandard reference on North American beaver notes

that ‘[a]n important research need is to developindependent lines of evidence about how beaveraffect ecosystem structure and function over thefull range of ecological conditions inhabited bythe beaver, especially in the less well-knowncommunities such as southeastern forests, westernshrub-steppe, and desert grasslands’ (Baker and Hill,2003; p. 306). In response to this need, the objectivesof this paper are: (1) to synthesize published researchrelated to the ecology and management of beaverin drylands; (2) to highlight the ways in whichbeaver ecological function may differ between aridand semi-arid (‘dryland’) and humid (‘temperate’)regions; and (3) to identify the most importantknowledge gaps and propose a new researchagenda for dryland beaver ecology. Our hope isthat current and future science will guidemanagement of beaver in order to conservevaluable dryland stream and riparian ecosystems.

METHODS

Systematic literature review

A formal review of the peer-reviewed literature wasconducted according to a search protocol thataimed to maximize transparency and repeatabilitywhile minimizing bias. Results from standardizedkeyword searches in ISI Web of Knowledge andGoogle Scholar were screened by title, abstract,and, when necessary, full text (see SupplementaryMaterial, Table S2 for search terms) to identifypapers meeting the following criteria: (1) publishedin English-language, peer-reviewed journals, inyear 2012 or earlier; (2) relevant to the ecologyand management of beaver (papers dealing solelywith physiology, palaeontology, epidemiology, orproviding no information beyond beaver presence,were excluded); (3) study was conducted in adryland region (Figure 1); and (4) study reportedempirical data or original analyses. Reference listswere used to identify additional eligible sourcesnot located by the original search, but no attemptwas made to locate all such sources systematically.Grey literature sources are discussed in the textwhen relevant, but they are not included inthe formal results list (Supplementary Material,Table S3).

CONSERVATION IMPLICATIONS OF NORTH AMERICAN BEAVER IN DRYLAND STREAMS 393


Dam locations and density

All sources explicitly reporting presence of beaverdam(s) in a dryland location (including locationsprovided by unpublished data or casual reports) werecollected opportunistically, to provide an indicationof the full range of beaver dam occurrence. Whereavailable, reports of beaver dam density from drylandstreams were also collected. Eligible dam densityreports included (1) a longitudinally continuoussurvey for all beaver dams present in a streamsegment; and (2) reporting of both number ofdams (rather than number of colonies) found andtotal stream length surveyed. In some cases thisinformation was obtained by contacting authors.A similar opportunistic search for density reportsfrom temperate streams provided a comparisonwith dryland streams.

Data fromOregonDepartment of Fish andWildlifeAquatic Inventories Project surveys (1990–2011) ofboth dryland and temperate streams throughout thestate of Oregon permitted a more systematic,broad-scale assessment of beaver dam density.Survey data reported the total number of beaverdams present in wadeable stream segments (ODFW,1997). For this paper each stream survey segmentwas classified as either dryland or temperate(Figure 1). Survey segment length and beaver damdensity were compared between dryland andtemperate streams using Student’s t test. The analysiswas restricted to streams with at least one reportedbeaver dam (all survey reaches from the same streamwere aggregated into a single survey segment) toensure that all habitat considered was at leastbroadly suitable for beaver dam construction. Allanalyses were performed using R version 2.13.1 (RDevelopment Core Team, 2011).

DISTRIBUTION AND STATUS

History and status of North American dryland beaverpopulations

During the 19th century, British and Americantrappers ventured west through North America insearch of beaver pelts; although the majority oftheir effort was concentrated on temperate montaneregions, the desert South-west also supported a

thriving fur trade (Weber, 1971). Accounts by earlytrappers suggest that beaver were present and oftenabundant on most perennial dryland streams andwetlands. James Pattie, for example, describedtrapping beaver through southern Arizona in the1820s (Pattie, 1905), and Peter Skene Ogdenreported large beaver populations along parts of theHumboldt River in northern Nevada in 1829: ‘Inno part have I found beaver so abundant. … Thetrappers now average 125 beaver a man and aregreatly pleased with their success’ (Ogden, 1971).

By the end of the 19th century, this intensive trappingeffort had drastically reduced or extirpatedmanybeaverpopulations throughout North America (Naiman et al.,1986). In Arizona, for example, beaver were entirelyextirpated from the San Pedro, Santa Cruz, and lowerSalt and Gila rivers (Hoffmeister, 1986). A 1931 reportfrom New Mexico notes the relative absence ofbeaver from numerous locations where they wereformerly abundant:

In 1903 [the author] also visited the headwaters of thePecos River and found that [beaver] were stilloccupying some of the streams in that region….Therewere old cuttings along many of the other streams, butin most cases the beaver had been entirely trappedout… (Bailey, 1931; p.215).

Loss of beaver was widely recognized as a problemby the turn of the century, and legislation protectingbeaver was followed by deliberate reintroductions inmany areas, beginning in the early 20th century(Baker and Hill, 2003). In general beaver numbershave increased over the past 60 years, and currentlybeaver occupy much of their historical NorthAmerican range (Baker and Hill, 2003; Pollocket al., 2003). However, population densities maybe low, and beaver remain absent from many areasof former occupation, especially areas withurban or agricultural development (McKinstry andAnderson, 2002; Baker and Hill, 2003; Carrilloet al., 2009). Currently, numerous conservationgroups and government agencies advocateincreasing dryland beaver populations as part of ariparian conservation strategy (Fredlake, 1997;Wild, 2011).

The San Pedro River, Arizona, provides a casestudy for the history of beaver management in adryland river. Nineteenth century accounts describedextensive open marshlands and abundant beaver



(Webb and Leake, 2006), as in James Pattie’sdescription of the San Pedro during his 1825trapping expedition: ‘[the river] being veryremarkable for the number of its beavers, we gave itthe name of Beaver River. At this place we collected200 [beaver] skins;…’ (Pattie, 1905). However, heavytrapping, supplemented by dynamiting of beaverdams in an attempt to reduce mosquito-bornemalaria, effectively extirpated beaver from the SanPedro River by the early 20th century (Johnson,2011). Around the same time, probably due to acombination of climatic conditions and land-usechange (including loss of beaver dams), rapiddowncutting and arroyo formation drained riparianwetlands (Webb and Leake, 2006). More recently,as one of Arizona’s only fully free-flowing riversand an important site for migratory birds, the SanPedro River has become an important site forriparian conservation in the desert South-west(Stromberg et al., 1996; Johnson, 2011). In the1990s, reintroduction of beaver was proposed as ameans to increase perennial surface water, reduceerosion, and improve habitat heterogeneity forother wildlife; in particular, beaver activity couldpromote development of wetlands more closelyresembling historical conditions along the river(Fredlake, 1997). Since reintroduction in 1999–2001,beaver populations have increased, spread, andbuilt dams (Johnson, 2011). Continuing monitoringof ecosystem responses to this reintroductionprovides an opportunity to improve understandingof the historical consequences of widespreadbeaver removal.

Distribution of beaver in dryland streams

The historical range of the North American beaverincludes most of the drylands of western NorthAmerica (Figure 1). Establishing whether beaverwere historically present in a region can play animportant role in decisions about beavermanagement (Longcore et al., 2007), but the rapiddecimation of beaver populations from westernNorth America makes it difficult to determine theprecise limits of historical beaver distribution.Borders of the widely cited beaver native rangemap (Jenkins and Busher, 1979; Figure 1) havebeen called into question by recent research:

although it is widely believed that beaver werehistorically absent from large areas of Californiaand Nevada (Jenkins and Busher, 1979; USFWS,2009), Lanman et al. (2012) suggest that beaverwere in fact present in the dryland Carson andWalker river basins of western Nevada, and asimilar review of the evidence indicates that beavermay also be native to arid southern California(M. Pollock, pers. comm.; Lanman et al., 2012).In Mexico, Gallo-Reynoso et al. (2002) provideevidence for historical and current presenceof beaver extending farther south into theSierra Madre Occidental than is usually includedin the beaver native range. Natural variation inbeaver populations over time (Baker and Hill,2003) adds to the difficulty of establishinghistorical distribution.

Beaver occupy a wide range of aquatic habitats,including streams and rivers, lakes, and wetlands(Baker and Hill, 2003). In drylands, many flowingwaters are ephemeral and perennial streams arerelatively rare (Levick et al., 2008); otherimportant habitat types occupied by beaver indryland North America include large rivers (Brecket al., 2001), the sloughs, backwaters, sidechannels, and other riparian wetlands in riverfloodplains (Billman et al., 2013), and isolatedspring-fed wetlands (Kindschy, 1985). Historicalsources document abundant beaver in the marshesand sloughs of the lower Colorado River andDelta, for example (Mellink and Luévano, 1998).In addition, beaver are highly adaptable and makeuse of developed and novel ecosystems includingreservoir shores (Tallent et al., 2011), urbanenvironments (Nolte et al., 2003) and, especially,the irrigation canals that are common in drylandagricultural landscapes (Hoffmeister, 1986;Demmer and Beschta, 2008). Development ofcanal networks in arid regions such as California’sImperial Valley has allowed beaver populations toexpand into formerly unsuitable territory (Tappe,1942; Mellink and Luévano, 1998). However,beaver damming of canals frequently leads to theirremoval as nuisance animals (McKinstry andAnderson, 2002). Most dryland studies focus onbeaver in streams and rivers, and informationabout beaver ecology in other habitat types,especially springs and wetlands, is limited.



Several models have been developed to characterizepotential beaver habitat within dryland NorthAmerica. The Southwest Regional Gap AnalysisProject broadly estimates range-wide potentialbeaver habitat within the south-western UnitedStates, based on availability of perennial water,land slope (<15%), and land use (excluding denseurban development; USGS, 2007). A similar model(excluding non-stream habitat types) for the stateof New Mexico suggests that large areas ofpotential beaver habitat are currently unoccupied(Wild, 2011). However, ‘relatively little descriptivework (on beaver-habitat associations) has beendone in the Southwest’ (Wild, 2011; p. 7), andmodel parameters are of necessity based primarilyon studies from temperate regions. Data areneeded to evaluate the performance and utility ofthese and other dryland habitat suitability models.

Availability of perennial water is probably themost important factor governing beaver distributionin dryland streams and wetlands. A survey ofbeaver occupancy along an eastern Washingtonstream found that beaver were present only in thelower, perennial reaches (Lind, 2002), but numerousauthors report beaver presence on intermittentstreams (Ffolliott et al., 1976; Mellink andLuévano, 1998; Albert and Trimble, 2000;McKinstry and Anderson, 2002). It is clear thatbeaver require a year-round source of water;however, even when the channel is not flowing,stream reaches classified as ‘intermittent’ may stillcontain permanent (‘perennial’) pools of water thatcan support beaver. Thus habitat models that limitpossible habitat to perennial stream reaches may betoo conservative, although it is likely that the bulkof dryland beaver occupancy is concentrated alongtrue perennial streams and wetlands.

Availability of riparian vegetation is alsobelieved to influence the distribution of beaver indrylands, where vegetation is often restricted to anarrow riparian corridor (Hoffmeister, 1986;Andersen and Shafroth, 2010) and generally morelimited than along temperate streams (MacFarlaneand Wheaton, 2013). Riparian vegetation mayshape reach-level distribution of beaver in drylandstreams: studies have shown close associationsbetween beaver presence and density of willow(see ‘Effects of herbivory on native plants’ below).

Anecdotally, beaver absence or failure to re-establishis often attributed to lack of vegetation (Mellink andLuévano, 1998; Albert and Trimble, 2000). Inparticular, loss of riparian vegetation owing tolivestock or other ungulate grazing, another commonfeature of North American drylands, may preventbeaver establishment (Baker et al., 2005; White andRahel, 2008). At a larger scale, however, thehypothesis that availability of vegetation limitsrange-wide beaver distribution in drylands has notbeen formally tested.

River regulation by large dams, a primary form ofhuman alteration to dryland river ecosystems, has acomplex relationship with beaver ecology.Dewatering downstream of dams can sometimesreduce habitat available to beaver (Mellink andLuévano, 1998); however, flow regulation preventsthe large floods that might displace beaverand destroy their dams, and it often increasesdownstream perennial flow (Andersen andShafroth, 2010). These hydrologic effects alsopromote riparian vegetation close to the active riverchannel, which in turn supports a greater number ofbeavers (Breck et al., 2001). Construction of GlenCanyon Dam, for example, is thought to haveincreased the beaver population of the GrandCanyon as result of increased availability of stableriparian habitat (Hoffmeister, 1986). Moreover,flow regulation can permit construction andmaintenance of beaver dams at a much greaterdensity than would have been possible historically;this has occurred on the Bill Williams River(Arizona), which provides a compelling example ofthe relationship between beaver and flowregulation. The extent of historical beaveroccupancy of this remote desert river is uncertain,but intermittent surface flow and lack of riparianvegetation are likely to have limited permanentbeaver presence, and large floods would haveremoved any dams with some regularity. However,on the present-day river, dam regulation hasmaintained perennial downstream flow, mostlyeliminated large floods, and promoted denseriparian vegetation; these conditions are highlyfavourable for beavers and beaver dams. Andersenand Shafroth (2010) calculated that construction ofbeaver dams over seven flood-free years convertedlotic habitat to lentic at a rate of approximately 3%



per year. Current management goals for the riverinclude periodic removal of beaver dams (bycontrolled floods; Figure 2) in order to maximizehabitat diversity and promote establishment ofnative cottonwood–willow riparian vegetation(Shafroth et al., 2010).

Distribution of beaver dams in dryland streams

Most of the ecosystem engineering abilities attributedto beaver result from construction of beaver dams.Dam-building primarily occurs on small streams,and beaver may also dam secondary channels andbackwaters within the floodplain of large rivers(Naiman et al., 1986; Billman et al., 2013). Becauseof its ecological importance, this review focusesprimarily on beaver dam-building in streams.However, not all beavers construct dams: alonglakes and large rivers, for example, beavers digbank dens and make use of the existing deep water(Mortenson et al., 2008). A relatively largeproportion of dryland beavers are bank-dwellersrather than dam-builders (Breck et al., 2001), inpart because much of the perennial water in aridenvironments is concentrated in large rivers.

The hydrology of many dryland catchments mayalso limit persistence of beaver dams. High inter- andintra-annual variation in runoff is characteristic ofdryland streams generally, and intense flash floodscommon in low desert streams would presumablydestroy any existing beaver dams (Andersen andShafroth, 2010). Thus it seems likely that floods andflow variability may limit the distribution, density,or longevity of beaver dams in dryland streams.Beaver dams can be found throughout the range of

beaver occupancy, including arid, low-desertstreams, although reports from higher-elevation,semi-arid locations are more common (Figure 1).Within broad regional requirements, local channelgeomorphology determines specific locations wheredam construction is possible (McComb et al., 1990;MacFarlane and Wheaton, 2013). Because somany of the ecological effects of beaver activitydepend on dam construction, predicting theconsequences of a beaver introduction orpopulation increase will require an accurateprediction of where and in what densities beaversare capable of building and maintaining dams.

Recently, Macfarlane and Wheaton (2013)developed a model specifically to estimate thepotential extent of beaver dam-building activityacross a landscape, emphasizing conditions typicallyfound in dryland streams. In addition to availabilityof perennial water and riparian vegetation, themodel incorporates stream power at base flows andat flood levels to predict the effects of streamhydrology on dam construction and persistence.Other studies have found that dam presence indryland streams is strongly associated with lowstream gradient (as in temperate streams; Bakerand Hill, 2003), and also with alluvial substrate,gentle bank slopes, and presence of hardwood orriparian vegetation (McComb et al., 1990; Lind,2002). When riparian trees are not available,beavers may construct dams of willow (Salix sp.;Call, 1970), cattails (Typha sp.; Andersen andShafroth, 2010), or even sagebrush (Artemisia sp.;Apple et al., 1985), but authors suggest that suchdams are likely to be less stable than thoseconstructed with large wood.

Figure 2. Two beaver dams on the Bill Williams River: intact dam (left); and breached dam during an experimental flood (right). Photos: Julian Olden.



The typical hydrology and relatively limitedvegetation of dryland streams suggests thehypothesis that beaver dams in dryland streamswill be less abundant than in temperate streams.However, dam densities reported in the literaturedo not support this hypothesis: there is noconsistent difference in reported densities betweendryland and temperate streams at either small orrelatively large scales (Figure 3(a)). However, thiscollection drew heavily on studies from high-elevation, semi-arid Wyoming, where very highdensities have been reported, and the relativelyfewer values from fully arid locations areconsistently low (Supplementary Material, TableS4). Similarly, data from standardized streamsurveys throughout the state of Oregon show nosignificant difference in mean beaver dam densitybetween dryland (1.07 dams km-1; n = 44) andtemperate stream (1.50 dams km-1; n = 329)segments (Student’s t-test, P=0.117, df = 62;Figure 3(b)), despite a significantly greater meansurvey length for dryland (16.2 km) than fortemperate (9.0 km) stream segments (Student’st-test, P=0.002, df = 49). Together these resultsindicate that, where dams are built, beaver damsin dryland streams can achieve densitiescomparable to those found on average intemperate streams, even over relatively large scales

(>100 km). However, the few available large-scalebeaver dam surveys reported from dryland streamsdo suggest a tendency for uneven, patchydistribution of dams (McComb et al., 1990;Andersen and Shafroth, 2010; Gibson, unpub. data;but see Call, 1970) compared with a morehomogeneous distribution in temperate streams(Johnston and Naiman, 1990). Additional researchon beaver dam abundance in dryland streams isneeded for a better description of distributionpatterns and to quantify the range of dam densitiesto be expected across a variety of habitat settings.

Beaver modification of dryland streams is adynamic process that varies in time as well as inspace. In eastern North America, beaver dams tendto be stable landscape features, some persisting aslong as a century (Burchsted et al., 2010). Indryland streams, however, floods and variablestream discharge usually preclude such longevity; inaddition, variable discharge can produce widefluctuations in dam density over time. Over 17 yearsof biannual beaver dam census in a small centralOregon stream, the total number of dams presenton 32 km of stream ranged from 9 to 103. Noindividual dam persisted more than 7 years, andmost were breached within 2 years or less (Demmerand Beschta, 2008). Dam longevity varies withenvironmental setting: in high-elevation, semi-arid

Figure 3. Comparisons of beaver dam densities reported for dryland and temperate streams. Panel (a) shows beaver dam densities reported in theliterature, as a function of total stream length surveyed. Light-grey triangles indicate densities from dryland streams and dark circles representtemperate streams. Not shown is an extreme dryland value of 25.6 dams km-1 over 145 km (Call, 1970). Density sources are listed inSupplementary Material, Table S4. Panel (b) shows a boxplot of beaver dam densities reported for standardized stream segment surveys fromdryland (n= 44) and temperate (n= 329) streams distributed throughout Oregon, USA. Stream length of survey segments ranges from 2–200 km(dryland streams) and from 2–275 km (temperate streams). Boxplots show median, interquartile range (IQR), 3 times IQR (whiskers), and outliers;

box widths are proportional to the square root of the number of observations in each group. Data source: ODFW (1997).



Wyoming, for example, most beaver dams werefound to be between 5 and 19 years old (Call, 1970),while dams in the desert San Pedro River (Arizona)usually wash out each year in seasonal monsoonfloods (Johnson, 2011). However, dam longevityalso depends on channel morphology, which maychange over time in response to climatic oranthropogenic factors (Cluer and Thorne, 2012; seealso ‘Geomorphology’ below). Historical marsheson the San Pedro River, for example, may havebeen capable of supporting much more stablebeaver dams than do contemporary entrenchedchannels (see ‘History and status’ above).

Beginning to quantify the relationship betweenstream discharge and dam failure, Andersen andShafroth (2010) found that flood pulses of about60m3 s-1 (relative to base flow of ~1m3 s-1) damagedat least 50% of monitored beaver dams belowAlamo Dam on the Bill Williams River (Arizona).Fewer dams were damaged in a 37m3 s-1 flood pulse– but ‘significant’ damage to beaver dams wasobserved at discharge as low as 5m3 s-1, suggestingthat even relatively small managed floods may affectbeaver dam function. This study also showed thatdams were quickly rebuilt even following flushingfloods large enough to obliterate all dams.

ABIOTIC EFFECTS

Geomorphology

Beaver dams play a significant role in shaping themorphology of river channels (Pollock et al.,2003). Fundamentally, construction of a beaverpond increases the extent of surface water andlentic wetland habitat; this function may haveparticular ecological significance in drylands,where wetland habitat is rare (McKinstry et al.,2001). Beaver ponds sometimes maintain perennialsurface water or wetlands in otherwise intermittentstream channel segments (Albert and Trimble,2000; McKinstry and Anderson, 2002), which canpromote a stable riparian community and providea water source for wildlife and livestock during thedry season (Call, 1970; Demmer and Beschta,2008). In many temperate streams, beaver pondstypically fill with sediment over time andeventually develop into beaver meadows or other

wetland landforms (Burchsted et al., 2010), but thislongevity is unlikely for beaver dams in flood-pronedryland streams (see ‘Distribution of beaver dams’above). Despite high rates of dam failure, however,former beaver dams on a central Oregon streamwere associated with increased channel sinuosity,diverse wetland habitats, and pool–riffle complexesresulting from sediment deposition within formerponds (Demmer and Beschta, 2008). Westbrooket al. (2011) documented a similar effect in atemperate montane stream characterized byfrequent dam failure during high flows: bothpresence and breaching of beaver dams increasedheterogeneity in sediment deposition and riparianvegetation on floodplain terraces. Microhabitatcomplexity associated with even short-lived beaverdams may promote aquatic biodiversity (Billmanet al., 2013).

The ability of beaver ponds to trap and retainsediment has proved useful in restoration of incisedstream channels, a common problem for drylandstreams. Channel incision, typically following land-use change such as development or introduction oflivestock grazing, is associated with rapid erosion,lowering of the water table, severed connectivitywith the floodplain, elimination of riparianvegetation, and loss of fish habitat (Pollock et al.,2007). Encouraging beaver dam construction can bea technique to restore functioning of these channels.Pollock et al. (2007) studied the process in detailon Bridge Creek in semi-arid central Oregon;sediment accumulation behind beaver damsindicated relatively rapid aggradation of the streamchannel and reduction in channel slope associatedwith the dams. In a similar study from semi-aridIdaho, DeVries et al. (2012) documented increasedfrequency of overbank flows (i.e. hydrologicconnectivity with the floodplain) around artificialstructures constructed to imitate beaver dams.These studies demonstrate that beaver dams caneffectively speed up the relatively slow process ofaggrading incised streams sufficient to reconnectthem with abandoned terraces.

However, the geomorphic effects of beaver damsvary depending on channel morphology, which mayfluctuate over time (Cluer and Thorne, 2012). Damsconstructed within incised channels are less likely tocreate overbank flooding and geomorphic complexity



such as braided channels than are dams in anunimpaired, low-gradient floodplain (Johnson, 2011;Pollock et al., 2012). Furthermore, there is a positivefeedback relationship between dams and channelform: stable beaver dams promote aggradation andoverbank flooding, which spreads and dissipatesflood energy across the floodplain; within incisedchannels, however, concentrated flood energytypically washes out beaver dams in their first year(Demmer and Beschta, 2008; Johnson, 2011; Pollocket al., 2012), thus preventing development of thestable beaver colonies that might counteract theincision. Artificially stabilizing beaver dams or dam-like structures within incised channels has beenproposed as a management technique to break theincision cycle and enhance the restoration potential ofbeaver dams (Apple et al., 1985; Pollock et al., 2012).

Hydrology

A number of studies indicate the importance ofdryland beaver ponds in groundwater processes thatshape patterns of discharge and riparian vegetation.Groundwater monitoring along a central Oregonstream showed increases in groundwater surfaceelevation (i.e. water table), groundwater storagepotential, and aquifer recharge surrounding abeaver dam (Lowry, 1993), and anecdotal reportsindicate a similar increase in water table elevationassociated with construction of new beaver dams onthe San Pedro River, Arizona (Johnson, 2011). In asemi-arid Wyoming stream, where beaver damswere associated with increased hyporheic exchange,Lautz et al. (2006) also found that the effect ofdams varied with geomorphic setting: in gainingreaches, water diverted to the subsurface by abeaver dam re-entered the main channel shortlybelow the dam, but in losing reaches, water wasdiverted to deeper, longer-term flow paths. Theseeffects are generally consistent with studies fromtemperate streams (Rosell et al., 2005), but maytake on a new importance in the different ecologicalcontext of water-limited dryland stream ecosystems.

The most intriguing aspect of the relationshipbetween beaver ponds and groundwater dynamicsin dryland streams is the potential for the elevatedwater storage to increase stream flow during dryseasons, potentially converting downstream

hydrology from intermittent to perennial. Chroniclow-flow conditions are a common feature ofdryland streams, and increasing loss of surfaceflow to climate change and human use is aprimary conservation concern for many drylandstreams (Seager et al., 2007; Levick et al., 2008).Advocates for beaver reintroduction frequently citemore stable, perennial flow as a benefit to beprovided by beaver dams (Fredlake, 1997;Wild, 2011); unfortunately, very few data areavailable to evaluate this hypothesis. Someanecdotal reports, from both temperate and drylandstreams, suggest instances where beaver dams didconvert intermittent streams to perennial flow(reviewed in Pollock et al., 2003). Studiesmonitoring beaver reintroductions in drylands havereported that beaver ponds maintained perennialstanding water (see ‘Geomorphology’ above), butno data are available to evaluate downstreameffects on discharge. In addition, construction ofbeaver ponds may affect stream water budgets bychanging evaporative processes (Andersen et al.,2011). Research is needed to quantify therelationships between beaver dams and hydrologicprocesses in dryland streams.

In contrast to their ability to maintain flow duringdrought conditions, beaver dams may also reducestream velocity and erosive power during peakflows. Somewhat limited empirical evidence fromtemperate streams supports this hypothesis (reviewedin Pollock et al., 2003; Rosell et al., 2005), but fewdata are available from dryland streams. Wheremanagement goals include reducing erosion orsedimentation, promotion of beaver dams may be aneffective strategy (DeVries et al., 2012). However, inmany dryland river systems, large peak flows areimportant in conservation of native fish (Rinneand Miller, 2006) or plant (Stromberg, 2001)communities. In this case dam-building activity maycounteract management goals, although beaverdams are unlikely to have a significant effect on largemagnitude, infrequent floods.

Water quality and chemistry

High water temperature is a primary managementconcern for water quality in many drylandstreams, in particular with respect to endangered



salmon and trout. In general beaver dams arethought to increase water temperatures owing toincreased water surface area, longer residencetime, and decreased shading (Rosell et al., 2005),although cooler water temperature due to deeppools and increased willow shading has also beencited as a potential benefit to be provided bybeaver activity (Wild, 2011). Evidence for thermaleffects of beaver dams in dryland streams is mixed.Water temperatures in a south-east Oregon streamwere consistently slightly warmer within beaverponds than in neighbouring unimpounded reaches(Talabere, 2002), but in the cool tailwaters belowAlamo Dam on the Bill Williams River (AZ),Andersen et al. (2011) found no consistent trend inwater temperature within beaver ponds. It isinteresting that both Lowry (1993) and Pollock et al.(2007) observed relatively cooler water temperatureimmediately below beaver dams in central Oregon,presumably an effect of groundwater upwelling. Inaddition to beaver dams creating thermal effects,beaver foraging may also influence water temperatureby reducing riparian canopy shade (Rosell et al.,2005), a potential effect that has not been examined.

Several studies have shown effects of beaver damson input and retention of organic matter andnutrients in dryland streams that are largelyconsistent with results from temperate streams.Harper (2001) confirmed that benthic sediments frombeaver ponds in an arid Nevada stream containedhigher levels of particulate organic matter than didsediments from unimpounded reaches. Based on thisstudy it seems probable that beaver ponds increasenet ecosystem retention of nitrogen and, thus, overallproductivity of ponds and downstream waters(Coleman and Dahm, 1990). Altered nutrientretention has conservation implications for waterquality in dryland streams: concentrations ofnutrients and suspended solids decreased in aWyoming stream after passing through severalbeaver ponds, indicating that promotion of beaverdams in tributaries may be an effective strategy toreduce downstream nutrient export (Maret et al.,1987). Beaver dams can also affect nutrient retentionthrough increased hyporheic exchange (Lautz et al.,2006); this may be particularly important wherechannel incision and livestock grazing add pollutionor limit ecosystem uptake of nutrients.

BIOTIC EFFECTS

Riparian vegetation

Considerable management effort has been devotedto conservation and restoration of riparian plantcommunities in dryland environments, especiallythe globally endangered cottonwood–willow foresttype (Stromberg, 2001), and there is clear evidencethat beaver activity alters the riparian communityboth directly through herbivory and indirectlythrough dam construction. Beaver are unique intheir ability to cut mature trees and thus alter theriparian canopy cover (Baker and Hill, 2003); inaddition, beaver foraging activity is concentratedalong the water’s edge (McGinley and Whitham,1985). Unlike a majority of the ecosystem effectsassociated with beaver activity, herbivory willoccur wherever beaver are present, not limited todam-building sites. Managers are frequentlyconcerned that beaver foraging will damagedesired riparian vegetation (Mortenson et al.,2008), and numerous studies have sought to assessthe net effects of beaver foraging on drylandriparian communities.

Effects of herbivory on native plants

Feeding trials and field observations indicate thatwillow (Salix spp.) and cottonwood (Populus spp.)are preferred woody forage plants for drylandbeaver (Harper, 2001; Kimball and Perry, 2008).Several studies have documented a closeassociation between beaver presence anddistribution of willow (Mortenson et al., 2008;Tallent et al., 2011), but there is little evidence fora negative population-level response of willow tobeaver foraging. For example, despite concern thatbeaver foraging may contribute to observed declinesof now-rare Goodding’s willow (S. gooddingii) standsin Grand Canyon National Park, findings from asurvey of the spatial distribution of beaver andwillow did not support this hypothesis (Mortensonet al., 2008). Along the shores of Lake Mojave,Tallent et al. (2011) showed a significant positiveassociation between beaver herbivory andpercentage willow cover, and they suggested thatbeaver foraging promoted willow ‘coppicing’ andregrowth into dense stands, thus increasing total



willow cover. Kindschy (1985) found that stems ofred willow (S. lasiandra) grew faster after being cutby beaver, compared with growth in unbrowsedplants. At a small scale, however, beaver mayconsume most readily available willow plants withinreach of a beaver colony (Hall, 2005).

Like willow, cottonwood browsed by beaver cansometimes resprout from stumps or roots, typicallyproducing a ‘shrubbier’ (short, more branches)growth form, with unknown consequences forreproductive success (McGinley and Whitham,1985). However, unlike willow, substantial negativepopulation-level effects of beaver herbivory oncottonwood have been documented. In combinationwith flow regulation, beaver herbivory mayeffectively prevent establishment and thereforepersistence of cottonwood (Lesica and Miles, 1999).In studies on the Green and Yampa Rivers (Utahand Colorado), Breck et al. (2001) showed that flowregulation promoted growth of willow close to thewetted channel, which allowed beaver populationsto increase. In addition, where cottonwoodabundance and density was lower (i.e. along theregulated Green River), the probability of beaverdamage to any individual plant was much higher,and therefore beaver herbivory had a greater overallimpact on the cottonwood population (Breck et al.,2003). In general these results suggest thatcottonwood populations already stressed by riverregulation or dewatering are more vulnerable tobeaver herbivory. Restoration strategies such asmodifying dam releases on regulated rivers topromote cottonwood establishment may increasethe resilience of cottonwood populations to beaverherbivory. Alternatively, more direct managementinterventions such as protecting vulnerable youngtrees with wire, or even trapping and removal ofbeaver, may be necessary where promotion ofvulnerable riparian cottonwood populations is apriority (Crawford and Umbreit, 1999) or whererestoration projects attempt to plant newcottonwood stands (Nolte et al., 2003).

Effects of herbivory on non-native plants

The spread of non-native riparian plants, especiallysalt cedar (‘tamarisk’; Tamarix spp.), andreplacement of native plant communities represents

a major challenge for dryland riparian management(Mortenson et al., 2008). There is some evidencethat beaver herbivory can promote the spread ofnon-native plants such as tamarisk at the expenseof native communities. Food choice experimentsindicate that high tannin and salt levelsphysiologically limit beaver consumption oftamarisk (Kimball and Perry, 2008), andobservational studies have found that beaver rarelyforage on tamarisk (Lesica and Miles, 2004;Mortenson et al., 2008; Tallent et al., 2011),although they may cut tamarisk shoots for use indam construction (Harper, 2001). In semi-arideastern Montana, Lesica and Miles (2004) foundthat tamarisk and non-native Russian olive(Elaeagnus angustifolia) both grew significantlyfaster under an open canopy created bybeaver foraging than under the shade of intactcottonwood forest. In the Grand Canyon, a positiveassociation between beaver presence and tamariskcover was consistent with the hypothesis thatbeaver herbivory promotes tamarisk dominance(Mortenson et al., 2008). These results suggest thatunder some circumstances, especially riverregulation, the arrival of beaver in streams at risk ofinvasion by tamarisk or other non-native plantsmay compromise conservation efforts for nativeplant communities.

Indirect effects on vegetation

Beaver dams raise and stabilize the surroundingwater table, which creates ideal conditions forsome riparian plants. The strong interdependencebetween beaver dams, groundwater elevation, andwillow has been extensively studied in temperateYellowstone National Park, where restoration oftall riparian willow communities was dependenton restoration of the hydrologic conditions createdby beaver dams (Marshall et al., 2013). Beaverpopulations, in turn, cannot persist withoutabundant willow, thus preventing re-establishmentof beaver once they have been lost (Baker et al.,2005). Management interventions designed tobreak this positive feedback cycle includeconstructing imitation beaver dams in order toencourage willow growth (Marshall et al., 2013),or planting or importing vegetation to provide an



initial food supply for beaver until ponds are re-established (Albert and Trimble, 2000).

The importance of these hydrological effects forvegetation is likely to be even greater within drylandstreams, where lowered water tables caused by waterdiversion and groundwater extraction is considered aserious threat to cottonwood–willow forest(Stromberg et al., 1996). Cooke and Zack (2008)found a positive association between beaver damdensity and width of riparian vegetation cover insemi-arid Wyoming, and several studies anecdotallyreport a general increase in abundance of willow andother vegetation over time following the return ofbeaver dams to semi-arid streams (Apple et al., 1985;Demmer and Beschta, 2008), although in each casethese results are confounded by concurrent exclusionof livestock grazing. Reductions in non-nativetamarisk due to flooding behind beaver dams havealso been reported anecdotally (Albert and Trimble,2000; Baker and Hill, 2003; Longcore et al., 2007),and in contrast to the more positive effects of beaverherbivory (see ‘Effects of herbivory on non-nativeplants’ above). On small streams, promoting beaverdams may be an effective strategy for increasingriparian vegetation cover.

Beaver dam modifications to channel shape andsediment dynamics also have consequences forriparian vegetation. Landforms associated withbeaver dams (secondary channels, breached beaverponds) are known to be favourable for willowestablishment (Cooper et al., 2006). Several authorshave observed that, following the return of beaver tosemi-arid streams, mud bars deposited behind damsand newly exposed sediments of drained ponds weredensely colonized by riparian vegetation (Appleet al., 1985; Demmer and Beschta, 2008).Construction of beaver ponds also promotes growthof aquatic macrophytes, rushes, and sedges, whichmay increase the habitat value for other wildlife oreven produce desirable forage for livestock (Call,1970; Hall, 2005). However, in addition to willowgermination, bare soils of breached beaverponds probably also promote colonization byopportunistic, ‘weedy’ non-native plants (Zedler andKercher, 2004). A dynamic cycle of pond creation,abandonment, and breaching may be the mosteffective regime to promote establishment andpersistence of a native riparian vegetation community.

Wildlife

Mammals

Research indicates that dryland beaver activity canenhance habitat for aquatic and riparian-associated mammals, including some species ofconservation concern. River otter (Lontracanadensis), which have suffered particularly steeppopulation declines in the desert South-west, areknown to make use of beaver ponds and dens;within the upper Colorado Basin, Depue andBen-David (2010) documented an associationbetween beaver sign and otter presence, and theysuggested that otter reintroduction efforts shouldfocus on locations where beaver are present. Freyand Malaney (2009) proposed that beaver pondsare likely to provide ideal riparian habitat for therare meadow jumping mouse (Zapus hudsoniusluteus) in New Mexico. More generally, Medinand Clary (1991) found that a semi-arid Idahobeaver pond supported a greater abundance and adifferent assemblage of riparian small mammalsthan an adjacent unimpounded stream.

Birds

Dryland cottonwood–willow riparian forestssupport a high richness and density of breedingsongbirds (Knopf et al., 1988; Johnson, 2011),which highlights the conservation importance ofbeaver impacts to these forests. Considerableresearch in temperate regions has found that thewetland habitat and altered riparian vegetationstructure surrounding beaver ponds promotesspecies richness and abundance of birds (reviewedin Rosell et al., 2005); similar patterns appear inthe more limited studies from dryland streams.Density, biomass, and species richness of riparianbirds were all higher surrounding a beaver pondthan along an unimpounded reach of a semi-aridWyoming stream (Medin and Clary, 1990). Cookeand Zack (2008) further showed that riparian birdabundance and diversity in similar Wyomingstreams were positively related to the density ofbeaver dams in a stream reach. Followingreintroduction of beaver to the San Pedro River(see ‘History and status’ above), Johnson (2011)found that abundance and species richness of



obligate riparian birds were correlated withpresence and age of beaver dams even aftercontrolling for covariates such as presence ofsurface water and density of riparian vegetation.Waterfowl presence in dryland streams is alsostrongly associated with beaver ponds (Brownet al., 1996; McKinstry et al., 2001). Collectivelythese studies indicate strong and positiverelationships between presence of dryland beaverponds and the overall abundance and speciesrichness of riparian-associated bird communities.

Considerable conservation attention in the desertSouth-west is focused on the federally endangeredsouth-western willow flycatcher (Empidonax trailliiextimus). The flycatcher breeds in dense riparianvegetation, including willow and cottonwood(Finch and Stoleson, 2000). Managers speculatethat beaver dam-building activity may benefitflycatcher populations by creating desirablebackwater habitat, and beaver reintroduction hasbeen proposed as a restoration technique forflycatcher conservation plans (USFWS, 2002).However, there is also concern that beavers‘damage habitat by removing vegetation’ and thusbeaver activity may be detrimental to flycatchers(Finch and Stoleson, 2000). This concern has evenprompted active beaver removal efforts (Longcoreet al., 2007), although no published studies haveaddressed the issue empirically. Further research isneeded to clarify the relationships between beaveractivity and habitat for flycatchers and otherriparian bird species of concern.

Fish

The effects of beaver dam-building activity on fishpopulations have received extensive study intemperate streams (Pollock et al., 2003). Withindryland streams, most research addressing beaver–fish relationships has focused on trout species;these studies generally conclude that, consistentwith temperate stream findings, trout populationsbenefit from beaver ponds (Jakober et al., 1998;Talabere, 2002). For example, White and Rahel(2008) showed that beaver ponds could provideimportant refuge habitat for native cutthroat trout(Oncorhynchus clarkii utah) during extendeddrought conditions. However, beaver ponds also

provide excellent conditions for non-native brooktrout (Salvelinus fontinalis) (Call, 1970).Ultimately the conservation value of beaver pondsfor native trout will depend on the trade-offbetween benefits in habitat for native fish andcosts in promotion of undesired non-native species.

Surprisingly little information is available todescribe associations between beaver activity andnon-salmonid fishes in dryland streams. From aconservation perspective, native fishes of theColorado River basin are of particular interest: thisfish community is both highly endemic and highlyendangered, due in part to widespread introductionsof predatory non-native fishes. Many of thesenon-native fishes prefer pool habitats and deep,slow-moving water, and increased abundance ofpool habitats has been associated with greaterdensity of non-natives (Rinne and Miller, 2006).This suggests that construction of beaver pondsmay enhance the success of non-native fishes, to thedetriment of native fish communities.

Other taxa

For several aquatic and riparian animal taxa,including amphibians and invertebrates, searchesfound no published studies addressing theinfluence of dryland beaver activity. Availablenatural history information suggests potentialrelationships between these groups and beaverpond habitats, but relationships are speculativeonly. In all cases, particularly the effects of beaveractivity on populations of undesired non-nativespecies, research is needed to investigate theserelationships empirically.

Amphibians. By constructing perennial wetlands instream channels that might otherwise go dry –

especially as climate change and water withdrawalsincrease the threat of stream drying – beaverdam-building activity could provide valuablehabitat for dryland amphibians. For example, thefederally threatened Chiricahua leopard frog(Lithobates chiricahuensis), native to Arizona andNew Mexico, requires perennial water for successfulreproduction. However, creation of rare perennialwetland habitat may also benefit the invasivebullfrog (Lithobates catesbeianus), which iswidespread throughout the desert South-west.



Bullfrogs are associated with significant impacts onnative aquatic communities, including native frogsand fishes; bullfrogs also require perennial water forbreeding, and they are generally associated withlentic rather than lotic habitats (Maret et al., 2006),suggesting that beaver ponds could provide idealhabitat. Beaver ponds are believed also to pose athreat to federally endangered arroyo toad(Anaxyrus californicus) populations in Californiabecause the ponds inundate favourable breedinghabitat and support non-native crayfish, Africanclawed frogs (Xenopus laevis), and bullfrogs, all ofwhich prey on the toads (USFWS, 2009).

Invertebrates.Very little is known about relationshipsbetween beaver activity and dryland invertebratecommunities. Perhaps the most urgent researchneed is for studies of how beaver ponds mayinfluence crayfish populations. Historically nocrayfish were present in the Colorado River basin,but several non-native species (notably Orconectesvirilis and Procambarus clarkii) are now well-established, and continuing spread of these andother species poses an immediate threat to nativeaquatic communities (Moody and Taylor, 2012).

CONCLUSIONS

Growing interest in using beaver in streamconservation plans has outpaced research on theconsequences and effectiveness of this approach indryland streams. This systematic review of theliterature revealed that a majority of studies aresmall-scale and observational, and in many caseslack the replication needed to draw strong inferences.Many hypotheses are supported only by anecdote orspeculation. Despite these limitations, workcompleted to date indicates that (1) in general, dam-building behaviour is less likely in dryland than intemperate streams, and stream hydrology probablyplays an important role. (2) Beaver activity,including both herbivory and dam-building, can be apowerful force in structuring the riparian vegetationcommunity. In some cases beaver herbivory mayinhibit regeneration of vulnerable cottonwoodpopulations and/or promote spread of non-nativeplants. (3) Beaver dams have strong effects on localgeomorphology, promoting diverse and perennialwetland habitat; it has been demonstrated that

promotion of beaver dams can be an effectivetechnique for restoration of incised stream channels.(4) Beaver ponds have been implicated in promotingthe spread of a variety of problematic non-nativeanimal species, but this hypothesis has not been tested.

Better knowledge of the role of beaver in drylandstreams and wetlands will improve understanding ofthe function of these ecosystems and how they arelikely to respond to change. In addition, we hopethat good science will be available to guide decisionson when management interventions in beaverpopulations will be most effective, appropriate, andethical. This review highlights that ecological effectsof beaver are wide-ranging and complex, especiallyin the context of varying management goalsand human alteration of dryland riparianecosystems. As indicated by feedback relationshipsbetween beaver activity and environmentalconditions, the contemporary ecological landscapedoes not necessarily represent the full range ofpotential effects of beaver. Rather than attemptingto classify beaver activity as ‘good’ or ‘bad’, froma management perspective the role of thisiconic species in the conservation of dryland streamecosystems could best be viewed as a seriesof trade-offs involving both challenges andopportunities for conservation.

RESEARCH AGENDA

Results of this systematic review of the literaturesuggests several research topics that are most inneed of further study in dryland streams andwetlands. First, where possible, researchers shouldseek to quantify the ecological effects of beaveractivity. The literature is replete with anecdotalreports of ecological patterns observed in relation tobeaver ponds, yet the quantitative data needed forcost-benefit analyses in complicated managementdecisions are scarce (Shafroth et al., 2010). Anexample of this quantitative approach is providedby the model to estimate the potential density ofbeaver dams across a landscape (Macfarlane andWheaton, 2013). Second, the majority of drylandbeaver research has been conducted in highelevation, semi-arid rangeland systems; more studyis needed within lowland, arid desert landscapes. In



particular, the distribution and density ofdam-building behaviour in warm desert streamsremains largely unknown. Third, in the face ofchallenges from increasing drought and demand forwater in dryland environments, empirical study ofthe influence of beaver activity on stream hydrology(including surface water, groundwater, evaporation,and total stream water budget) and hence itspotential as a climate adaptation strategy is needed(Wild, 2011). Fourth, the spread of many non-nativespecies poses a significant threat to conservation ofdryland stream ecosystems, and research is needed toassess the effects of beaver ponds on populations ofnon-native fishes, bullfrogs, and crayfishes.

Ongoing beaver reintroductions represent large-scaleecological experiments and provide an unparalleledopportunity to advance scientific knowledge of beaverecology in dryland streams while concurrentlyinforming on-the-ground restoration practices. Weurge scientists and managers to work together todevelop clear hypotheses, define robust controls (i.e.Before-After-Control-Impact designs), and implementmonitoring programmes where hydrologic,physical and ecological responses are tracked overtime. This approach has been implemented incontinuing research on the effects of beaverdam structures at Bridge Creek, Oregon, wheredata collection includes monitoring of streamdischarge, groundwater level, channel morphology(gradient, sinuosity, and lateral connectivity),water temperature, abundance of riparianvegetation, and abundance of steelhead trout(Oncorhynchus mykiss), associated with numbersand locations of beaver dams over time (Pollocket al., 2012). We cite the need for long-termmonitoring of ecosystem responses in order tounderstand the broader or longer-term success ofbeaver reintroductions as a restoration strategy.

ACKNOWLEDGEMENTS

Funding for PPG was provided by a University ofWashington Top Scholar Graduate Fellowship andby the H. Mason Keeler Endowment for Excellencethrough the School of Aquatic and FisherySciences. JDO was supported in part by anH. Mason Keeler Endowed Professorship and by the

Department of Defense - Strategic EnvironmentalResearch and Development Program (RC-1724).M. Pollock and two anonymous reviewers providedvaluable comments that improved the manuscript.We also thank W. Anderson for enlighteningdiscussion and H. Cooke, P. Gurche, G. Johnson,J. Smith, A. Talabere, and S. White for generouslysharing their data.

REFERENCES

Albert S, Trimble T. 2000. Beavers are partners in riparianrestoration on the Zuni Indian Reservation. EcologicalRestoration 18: 87–92.

Andersen DC, Shafroth PB. 2010. Beaver dams, hydrologicalthresholds, and controlled floods as a management tool in adesert riverine ecosystem, Bill Williams River, Arizona.Ecohydrology 3: 325–338.

Andersen DC, Shafroth PB, Pritekel CM, O’Neill MW. 2011.Managed flood effects on beaver pond habitat in a desertriverine ecosystem, Bill Williams River, Arizona USA.Wetlands 31: 195–206.

Apple LL, Smith SB, Dunder DJ, Baker BB. 1985. The use ofbeavers for riparian–aquatic habitat restoration of colddesert gully-cut stream systems in southwestern WyomingUSA. In Investigations on Beavers, Pilleri G (ed.). BrainAnatomy Institute: Berne, Switzerland; 123–130.

Bailey V. 1931. Mammals of New Mexico. United States Dept.of Agriculture, Bureau of Biological Survey.

Baker BW, Hill EP. 2003. Beaver (Castor canadensis). InWild Mammals of North America: Biology, Management,and Conservation, Feldhamer GA, Thompson BC,Chapman JA (eds). John Hopkins University Press:Baltimore, MD; 288–310.

Baker BW, Ducharme HC, Mitchell DC, Stanley TR,Peinetti HR. 2005. Interaction of beaver and elkherbivory reduces standing crop of willow. EcologicalApplications 15: 110–118.

Billman EJ, Kreitzer JD, Creighton JC, Habit E, McMillan B,Belk MC. 2013. Habitat enhancement and native fishconservation: can enhancement of channel complexitypromote the coexistence of native and introduced fishes?Environmental Biology of Fishes 96: 555–566.

Breck SW, Wilson KR, Andersen DC. 2001. The demographicresponse of bank-dwelling beavers to flow regulation: acomparison on the Green and Yampa rivers. CanadianJournal of Zoology 79: 1957–1964.

Breck SW, Wilson KR, Andersen DC. 2003. Beaver herbivoryand its effect on cottonwood trees: influence of flooding alongmatched regulated and unregulated rivers. River Researchand Applications 19: 43–58.

Brown DJ, Hubert WA, Anderson SH. 1996. Beaver pondscreate wetland habitat for birds in mountains ofsoutheastern Wyoming. Wetlands 16: 127–133.

Burchsted D, Daniels M, Thorson R, Vokoun J. 2010. Theriver discontinuum: applying beaver modifications tobaseline conditions for restoration of forested headwaters.BioScience 60: 908–922.



Call MW. 1970. Beaver pond ecology and beaver–troutrelationships in southeastern Wyoming. PhD thesis,University of Wyoming, WY.

Carrillo C, Bergman D, Taylor J, Nolte D, Viehoever P, DisneyM. 2009. An overview of historical beaver management inArizona. USDA National Wildlife Research Center – StaffPublications, Fort Collins, CO.

CEC (Commission for Environmental Cooperation). 1997.Ecological regions of North America. Commission forEnvironmental Cooperation, Montreal, Quebec.

Cluer B, Thorne C. 2012. A stream evolution model integratinghabitat and ecosystem benefits. River Research andApplications. DOI: 10.1002/rra.2631.

Coleman RL, Dahm CN. 1990. Stream geomorphology:effects on periphyton standing crop and primaryproduction. Journal of the North American BenthologicalSociety 9: 293–302.

Cooke HA, Zack S. 2008. Influence of beaver dam density onriparian areas and riparian birds in shrubsteppe ofWyoming. Western North American Naturalist 68: 365–373.

Cooper DJ, Dickens J, Hobbs NT, Christensen L, Landrum L.2006. Hydrologic, geomorphic and climatic processescontrolling willow establishment in a montane ecosystem.Hydrological Processes 20: 1845–1864.

Crawford CS, Umbreit NE. 1999. Restoration and monitoringin the Middle Rio Grande Bosque: current status of floodpulse related efforts. In Rio Grande Ecosystems: LinkingLand, Water, and People, Finch DM, Whitney JC, KellyJF, Loftin SR (eds). USDA Forest Service: Albuquerque,NM; 151–163.

Demmer R, Beschta RL. 2008. Recent history (1988–2004) ofbeaver dams along Bridge Creek in central Oregon.Northwest Science 82: 309–318.

Depue JE, Ben-David M. 2010. River otter latrine site selectionin arid habitats of western Colorado, USA. Journal ofWildlife Management 74: 1763–1767.

DeVries P, Fetherston KL, Vitale A, Madsen S. 2012.Emulating riverine landscape controls of beaver in streamrestoration. Fisheries 37: 246–255.

Ffolliott PF, Clary WP, Larson FR. 1976. Observations ofbeaver activity in an extreme environment. TheSouthwestern Naturalist 21: 131–133.

Finch DM, Stoleson SH. 2000. Status, ecology, andconservation of the southwestern willow flycatcher. USDAForest Service, Rocky Mountain Research Station, Odgen,UT.

Fredlake M. 1997. Re-establishment of North American beaver(Castor canadensis) into the San Pedro Riparian NationalConservation Area. US Bureau of Land Management,Environmental Assessment EA no. AZ-060–97-004, Tucson,AZ.

Frey JK, Malaney JL. 2009. Decline of the meadow jumpingmouse (Zapus hudsonius luteus) in two mountain ranges inNew Mexico. The Southwestern Naturalist 54: 31–44.

Gallo-Reynoso J-P, Suárez-Gracida G, Cabrera-Santiago H,Coria-Galindo E, Egido-Villarreal J, Ortiz LC. 2002. Statusof beavers (Castor canadensis frondator) in Rio Bavispe,Sonora, Mexico. The Southwestern Naturalist 47: 501–504.

Gangloff MM. 2013. Taxonomic and ecological tradeoffsassociated with small dam removals: Editorial. AquaticConservation: Marine and Freshwater Ecosystems 23:475–480.

Hall F. 2005. Emigrant Creek cattle allotment: lessons from30 years of photomonitoring. USDA Forest Service, PacificNorthwest Research Station, Portland, OR.

Harper BJ. 2001. The ecological role of beavers (Castorcanadensis) in a Southwestern desert stream. MS thesis,University of Nevada, NV.

Hoffmeister DF. 1986. Mammals of Arizona. University ofArizona: Tucson, AZ.

Jakober MJ, McMahon TE, Thurow RF, Clancy CG. 1998.Role of stream ice on fall and winter movements andhabitat use by bull trout and cutthroat trout in Montanaheadwater streams. Transactions of the American FisheriesSociety 127: 223–235.

Jenkins SH, Busher PE. 1979. Castor canadensis. MammalianSpecies 120: 1–8.

Johnson G. 2011. Bird abundance and richness in a desertriparian area following beaver re-introduction. PhD thesis,University of Arizona, AZ.

Johnston CA, Naiman RJ. 1990. Aquatic patch creation inrelation to beaver population trends. Ecology 171: 1617–1621.

Kimball BA, Perry KR. 2008. Manipulating beaver (Castorcanadensis) feeding responses to invasive tamarisk (Tamarixspp.). Journal of Chemical Ecology 34: 1050–1056.

Kindschy RR. 1985. Response of red willow to beaver use insoutheastern Oregon. The Journal of Wildlife Management49: 26–28.

Knopf FL, Johnson RR, Rich T, Samson FB, Szaro RC. 1988.Conservation of riparian ecosystems in the United States.The Wilson Bulletin 100: 272–284.

Lanman RB, Perryman H, Dolman B, James CD, Osborn S.2012. The historical range of beaver in the Sierra Nevada: areview of the evidence. California Fish and Game 98: 65–80.

Lautz LK, Siegel DI, Bauer RL. 2006. Impact of debris damson hyporheic interaction along a semi-arid stream.Hydrological Processes 20: 183–196.

Lesica P, Miles S. 1999. Russian olive invasion intocottonwood forests along a regulated river in north-centralMontana. Canadian Journal of Botany 77: 1077–1083.

Lesica P, Miles S. 2004. Beavers indirectly enhance the growthof Russian olive and tamarisk along eastern Montana rivers.Western North American Naturalist 64: 93–100.

Levick LR, Goodrich DC, Hernandez M, Fonseca J, SemmensDJ, Stromberg JC, Tluczek M, Leidy RA, Scianni M,Guertin DP. 2008. The ecological and hydrologicalsignificance of ephemeral and intermittent streams in thearid and semi-arid American southwest. US EnvironmentalProtection Agency, Office of Research and Developmentand USDA/ARS Southwest Watershed Research Center.

Lind JM. 2002. A habitat suitability model for beavers (Castorcanadensis) in an arid environment using GeographicInformation Systems (GIS). MS thesis, Central WashingtonUniversity, WA.

Longcore T, Rich C, Müller-Schwarze D. 2007. Managementby assertion: beavers and songbirds at Lake Skinner(Riverside County, California). Environmental Management39: 460–471.

Lowry MM. 1993. Groundwater elevations and temperatureadjacent to a beaver pond in central Oregon. MS thesis,Oregon State University, OR.

MacFarlane WW, Wheaton JM. 2013. Modeling the capacityof riverscapes to support dam-building beaver – case study:



Escalante River Watershed. Ecogeomorphology andTopographicAnalysis Lab,Utah State University, Logan,UT.

Maret TJ, Parker M, Fannin TE. 1987. The effect of beaverponds on the nonpoint source water-quality of a stream insouthwestern Wyoming. Water Research 21: 263–268.

Maret TJ, Snyder JD, Collins JP. 2006. Altered drying regimecontrols distribution of endangered salamanders andintroduced predators. Biological Conservation 127: 129–138.

Marshall KN, Hobbs NT, Cooper DJ. 2013. Stream hydrologylimits recovery of riparian ecosystems after wolfreintroduction. Proceedings of the Royal Society B:Biological Sciences 280: 20122977.

McComb WC, Sedell JR, Buchholz TD. 1990. Dam-siteselection by beavers in an eastern Oregon basin. GreatBasin Naturalist 50: 273–281.

McGinley MA, Whitham TG. 1985. Central place foraging bybeavers (Castor canadensis) – a test of foraging predictionsand the impact of selective feeding on the growth form ofcottonwoods (Populus fremontii). Oecologia 66: 558–562.

McKinstry MC, Anderson SH. 2002. Survival, fates, andsuccess of transplanted beavers, Castor canadensis, inWyoming. Canadian Field-Naturalist 116: 60–68.

McKinstry MC, Caffrey P, Anderson SH. 2001. Theimportance of beaver to wetland habitats and waterfowl inWyoming. Journal of American Water ResourcesAssociation 37: 1571–1577.

MEA (Millennium Ecosystem Assessment). 2005. Drylandsystems. In Ecosystems and human well-being: current stateand trends: findings of the Condition and Trends WorkingGroup of the Millennium Ecosystem Assessment, HassanRM, Scholes RJ, Ash N (eds). Island Press: Washington,DC; 623–662.

Medin DE, Clary WP. 1990. Bird populations in and adjacentto a beaver pond ecosystem in Idaho. USDA ForestService, Ogden, UT.

Medin DE, Clary WP. 1991. Small mammals of a beaver pondecosystem and adjacent riparian habitat in Idaho. USDAForest Service, Ogden, UT.

Mellink E, Luévano J. 1998. Status of beavers (Castorcanadensis) in Valle de Mexicali, Mexico. Bulletin SouthernCalifornia Academy of Sciences 97: 115–120.

Moody EK, Taylor CA. 2012. Red swamp crayfish(Procambarus clarkii) discovered in the San Pedro River,Arizona: a new invader in a threatened ecosystem.Southwestern Naturalist 57: 339–340.

Mortenson SG, Weisberg PJ, Ralston BE. 2008. Do beaverspromote the invasion of non-native Tamarix in the GrandCanyon riparian zone? Wetlands 28: 666–675.

Naiman RJ, Melillo JM, Hobbie JE. 1986. Ecosystemalteration of boreal forest streams by beaver (Castorcanadensis). Ecology 67: 1254–1269.

Nolet BA, Rosell F. 1998. Comeback of the beaver Castorfiber: an overview of old and new conservation problems.Biological Conservation 83: 165–173.

Nolte DL, Lutman MW, Bergman DL, Arjo WM, Perry KR.2003. Feasibility of non-lethal approaches to protectriparian plants from foraging beavers in North America.USDA National Wildlife Research Center, Fort Collins, CO.

ODFW (Oregon Department of Fish and Wildlife). 1997.ODFW Aquatic Inventories Project stream habitatdistribution coverages. Oregon Department of Fish andWildlife, Corvallis, OR.

Ogden PS. 1971. Peter Skene Ogden’s Snake Country Journals,1827–28 and 1828–29, Vol 28. Hudson’s Bay Record Society:London.

Patterson BD, Ceballos G, Sechrest W, Tognelli MF, BrooksT, Luna L, Ortega P, Salazar I, Young BE. 2007. Digitaldistribution maps of the mammals of the WesternHemisphere, version 3.0. NatureServe, Arlington, VA.

Pattie JO. 1905. The personal narrative of James O. Pattie ofKentucky. The Arthur H. Clark Company: Cleveland, OH.

Pollock MM, Heim M, Werner D. 2003. Hydrologic andgeomorphic effects of beaver dams and their influence onfishes. In The Ecology and Management of Wood in WorldRivers, Gregory SV, Boyer K, Gurnell A (eds). AmericanFisheries Society: Bethesda, MD; 213–233.

Pollock MM, Beechie TJ, Jordan CE. 2007. Geomorphicchanges upstream of beaver dams in Bridge Creek, anincised stream channel in the interior Columbia Riverbasin, eastern Oregon. Earth Surface Processes andLandforms 32: 1174–1185.

Pollock MM, Wheaton JM, Bouwes N, Volk C, Weber N,Jordan CE. 2012. Working with beaver to restore salmonhabitat in the Bridge Creek Intensively MonitoredWatershed: design rationale and hypotheses. NationalOceanic and Atmospheric Administration, Seattle, WA.

Pool TK, Olden JD, Whittier JB, Paukert CP. 2010.Environmental drivers of fish functional diversity andcomposition in the Lower Colorado River Basin. CanadianJournal of Fisheries and Aquatic Sciences 67: 1791–1807.

R Development Core Team. 2011. R: A language andenvironment for statistical computing. R Foundation forStatistical Computing, Vienna, Austria.

Rinne JN, Miller D. 2006. Hydrology, geomorphology andmanagement: implications for sustainability of nativesouthwestern fishes. Reviews in Fisheries Science 14: 91–110.

Rosell F, Bozser O, Collen P, Parker H. 2005. Ecologicalimpact of beavers Castor fiber and Castor canadensis andtheir ability to modify ecosystems. Mammal Review 35:248–276.

Sabo JL, Sinha T, Bowling LC, Schoups GH, Wallender WW,Campana ME, Cherkauer KA, Fuller PL, Graf WL,Hopmans JW. 2010. Reclaiming freshwater sustainability inthe Cadillac Desert. Proceedings of the National Academy ofSciences 107: 21263–21269.

Seager R, Ting M, Held I, Kushnir Y, Lu J, Vecchi G, HuangH-P, Harnik N, Leetmaa A, Lau N-C, et al. 2007. Modelprojections of an imminent transition to a more arid climatein southwestern North America. Science 316: 1181–1184.

Shafroth PB, Wilcox AC, Lytle DA, Hickey JT, Andersen DC,Beauchamp VB, Hautzinger A, McMullen LE, Warner A.2010. Ecosystem effects of environmental flows: modellingand experimental floods in a dryland river. FreshwaterBiology 55: 68–85.

Stromberg JC. 2001. Restoration of riparian vegetation in thesouth-western United States: importance of flow regimesand fluvial dynamism. Journal of Arid Environments 49:17–34.

Stromberg JC, Tiller R, Richter B. 1996. Effects ofgroundwater decline on riparian vegetation of semiaridregions: The San Pedro, Arizona. Ecological Applications 6:113–131.

Talabere AG. 2002. Influence of water temperature and beaverponds on Lahontan cutthroat trout in a high-desert



stream, southeastern Oregon. MS thesis, Oregon StateUniversity, OR.

Tallent N, Nash M, Cross CL, Walker LR. 2011. Patterns inshoreline vegetation and soils around Lake Mohave,Nevada and Arizona: implications for management.Western North American Naturalist 71: 374–387.

Tappe DT. 1942. The status of beavers in California. State ofCalifornia, Department of Natural Resources, Division ofFish and Game, Sacramento, CA.

Trabucco A, Zomer RJ. 2009. Global aridity index(Global-Aridity) and global potential evapo-transpiration(Global-PET) geospatial database. CIGAR Consortium forSpatial Information. http://csi.cgiar.org/aridity/index.asp/[23 October 2012]

USFWS (US Fish and Wildlife Service). 2002. Southwesternwillow flycatcher recovery plan. US Fish and WildlifeService, Albuquerque, NM.

USFWS (US Fish and Wildlife Service). 2009. Arroyo toad(Bufo californicus), 5-year review: summary and evaluation.US Fish and Wildlife Service, Ventura, CA.

USGS National Gap Analysis Program. 2007. Digital animal-habitat models for the Southwestern United States. NewMexico Cooperative Fish and Wildlife Research Unit, NewMexico State University, Las Cruces, NM.

Webb RH, Leake SA. 2006. Ground-water surface-waterinteractions and long-term change in riverine riparian

vegetation in the southwestern United States. Journal ofHydrology 320: 302–323.

Weber DJ. 1971. The Taos Trappers: the Fur Trade in the FarSouthwest, 1540–1846. University of Oklahoma Press:Norman, OK.

Westbrook CJ, Cooper DJ, Baker BW. 2011. Beaver assistedriver valley formation. River Research and Applications 27:247–256.

White SM, Rahel FJ. 2008. Complementation of habitats forBonneville cutthroat trout in watersheds influenced bybeavers, livestock, and drought. Transactions of theAmerican Fisheries Society 137: 881–894.

Wild C. 2011. Beaver as a climate change adaptation tool:concepts and priority sites in New Mexico. SeventhGeneration Institute, Santa Fe, NM.

Zedler JB, Kercher S. 2004. Causes and consequences ofinvasive plants in wetlands: Opportunities, opportunists,and outcomes. Criticial Reviews in Plant Science 23: 431–452.

SUPPORTING INFORMATION

Additional supporting information may be found inthe online version of this article at the publisher’sweb site.



http://csi.cgiar.org/aridity/index.asp/

ecology, management, and conservation implications of north...

Documents