Ch24 - University of Michigandallan/pdfs/allan_benke.pdf · INTRODUCTION THE VARIETY OF RIVERS FEW RIVERS ARE PRISTINE NORTH AMERICA’S RIVERS IN THE TWENTY-FIRST CENTURY LITERATURE

Ch24.qxd 3/24/05 2:21 PM Page 1086

1087

24

OVERVIEW ANDPROSPECTS

a time of repair and restoration of the rivers of NorthAmerica?

THE VARIETY OF RIVERS

The twenty-two chapters describing individual riverbasins or geographical regions are rich with detailedinformation about their regions and include one-pagesummaries for 218 rivers, representing most of thelargest rivers on the continent. Although the major-ity of rivers are relatively large within their respec-tive regions, river size exhibits wide variation acrossand within chapters. In addition, river basins varygreatly in their mean temperatures and precipitation,the fractions of precipitation that flow into rivers, thediversity of landscape types drained by rivers, andtheir natural biological diversity.

Variation in Physical CharacteristicsThe major rivers of North America, the focus of thisbook, typically are large rivers, whether assessed byriver order, drainage area, or discharge. Of the fiveto twelve rivers described for each basin or region,some individual rivers are substantially smaller, oforder as low as 3 (e.g., the Virgin and Bill Williamsin the Colorado basin) or 4 (the Octonagon andAuSable of the St. Lawrence basin); their inclusion

INTRODUCTION

THE VARIETY OF RIVERS

FEW RIVERS ARE PRISTINE

NORTH AMERICA’S RIVERS IN THE TWENTY-FIRST CENTURY

LITERATURE CITED

INTRODUCTION

This concluding chapter provides an overview ofinformation in the previous chapters and addressessome of the major challenges facing rivers in thetwenty-first century. It is clear from previous chap-ters that the rivers of North America exhibit analmost bewildering variety of natural features anddegrees of human impact. Our initial purpose, there-fore, is to convey a sense of their variety and status.They differ greatly in their natural features, depend-ing on physical, climatic, and biological factors. Fewrivers can be called pristine, and the extent and typeof human influence adds additional layers of com-plexity and variation according to how humans haveused these rivers for water supply, power, navigation,waste disposal, and other purposes. Thus, the ob-served variability among rivers is ultimately a com-bination of natural variation and changes broughtabout by human activities. Both of these types ofvariation differ greatly across the continent, and thisis reflected throughout the chapters.

A second purpose of this chapter is to examinethe major challenges today and into the future facingNorth America’s rivers. How will the diverse pres-sures from human society further alter NorthAmerica’s rivers? What might we anticipate about thefuture challenges that rivers will experience? Whatare the opportunities to make the twenty-first century

J. DAVID ALLAN ARTHUR C. BENKE

© 2005, Elsevier Inc. All rights reserved.

Ch24.qxd 3/24/05 2:21 PM Page 1087

reflects their occurrence in very arid regions or theirregional significance. In contrast, the lower mainstems of the largest rivers are of order 9 (Mackenzie,Ohio, Missouri, St. Lawrence, Yukon, Nelson) or 10(Mississippi). In rare instances, rivers of substantiallysmaller basin size (the Moisie in Atlantic Canada)have been considered as order 9. In spite of this widerange, the median value for river order in this bookis generally 6 or 7, which usually represents amedium to large river.

Another way of looking at the variety of riversizes is that whereas the median drainage area of individual basins is approximately 25,000km2, basinareas range from as small as 217km2 (the Dunk Riveron Prince Edward Island) to 3,270,000km2 (theentire Mississippi basin, about 42% of the land areaof the 48 coterminous states). A drainage area over100,000km2 provides an arbitrary criterion for “verylarge”; after excluding the lower main stems of the largest river systems, some 33 individual riversmeet this criterion. However, because many basins>100,000km2 are in arid regions, they do not neces-sarily have the highest discharges.

Because of their spatial extent, most river systemsencompass considerable physical heterogeneity, andthis is especially true of the largest river basins. Thenumber of physiographic provinces through whichrivers run influences potential diversity in geology,gradient, channel morphology, and habitat types. Thenumber of ecoregions not only reflects this geologi-cal diversity, but climate and vegetation as well.Thus, plant communities in particular should influ-ence the amount and type of organic matter inputsfrom a basin’s smallest tributaries to its main stem(Vannote et al. 1980, Webster and Meyer 1997), andclimate influences precipitation and the fraction thatbecomes runoff.

A typical river drains two ecoregions and twophysiographic provinces (median values1), but thelargest river basins encompass much more hetero-geneity. For example, the St. Lawrence drains eightphysiographic provinces and nine ecoregions; thesenumbers are four and seven for the Saskatchewan,seven and thirteen for the Missouri, and six and eightfor the Ohio. In some cases, relatively small basinscan have high heterogeneity (e.g., five ecoregions andfive physiographic provinces for the Potomac). Basin

relief (from highest peak to river mouth) also varieswithin and among rivers. The median value for basinrelief for individual rivers is approximately 1300m,but vertical relief in some “flatland” rivers isminimal. For example, basin relief is <120m for theMaumee River of Ohio, the Illinois and Minnesotarivers of the Upper Mississippi, and the St. Johns andSatilla rivers of the southeastern Coastal Plain. Incontrast, median vertical relief for Pacific Coast riversof Canada and Alaska is 2628m, despite a relativelymodest average basin area of about 46,000km2, andmedian relief for Mexican rivers is 3000m, with anaverage basin area of about 49,000km2. Given suchwide variation in river basin relief, the variety ofhabitats and ecological conditions within a river canbe expected to vary accordingly.

The rivers of North America differ greatly in discharge and runoff. Of the 218 rivers described,two-thirds have a mean annual discharge exceeding100m3/s, 15% exceed 1000m3/s, and two exceed10,000m3/s. Mean annual discharge in m3/s is anindication of how much water is exported by a riverbasin. It is higher for rivers with large drainage areasand wet climates (Table 24.1). The lower Mississippi(which receives inflows from the Upper Mississippi,the Missouri, and especially the Ohio) ranks 9th inthe world (Leopold 1994). A number of northernriver systems, including the St. Lawrence, Macken-zie, Columbia, Yukon, and Fraser, discharge verylarge quantities of water. River basins in arid climateshave markedly lower mean annual discharges. In aridregions, the majority of rivers have discharges below100m3/s. For an individual example, consider theBrazos River on the Gulf Coast of the southwesternUnited States and the Tennessee River in the Ohiobasin. Both have slightly over 100,000km2 indrainage area but annual discharge of the Tennesseeis ten times higher than the more arid Brazos.

Discharge (Q) is influenced by basin area, precip-itation (PPT), and the amount of PPT that becomesevapotranspiration (ET), as can be observed by com-paring rivers and regions (see Table 24.1). Table 24.1includes the largest river for each region or basin (leftside), as well as the median values for the remainingrivers (right side). Annual runoff (RO, cm/yr) isanother way to represent water yield (1cm/yr =100m3 ha-1 yr-1) or the amount of water that annu-ally runs off a unit area of basin. We estimatedannual runoff directly from monthly runoff data pro-vided by chapter authors (see figures in one-pagesummaries of each chapter). Median annual runofffor rivers of each region is shown in Table 24.1. Itreveals the low water yields of such arid-land basins

24 Overview and Prospects

1088

1 Most values reported are medians rather than averagesbecause of occasional extreme values, and the subset of riversincluded within a basin is not truly a random sample. Hence, allcomparisons reported here should be interpreted as broadly indica-tive but values given are approximate.

Ch24.qxd 3/24/05 2:21 PM Page 1088

as the Missouri (6cm/yr), the Colorado (4cm/yr), theGreat Basin (7cm/yr), the Gulf Coast of the south-western United States (4cm/yr), and the Nelson–Churchill (7cm/yr). Runoff values also identify riverswith high yields, typically draining areas of high pre-cipitation and low evapotranspiration. The highestrunoff regions are the Pacific Coast rivers of Canadaand Alaska (90cm/yr), and the Atlantic Coast riversof Canada (73cm/yr). At least one river of Mexico(Usumacinta–Grijalva), however, has an annualrunoff of at least 70cm/yr in spite of high evapo-

transpiration, having the highest annual precipitationof any basin in North America. Several rivers of thePacific United States have high runoff, including theRogue and Eel, but the median for rivers of thatregion is lower due to the inclusion of some very aridriver basins in southern California.

Another advantage of converting discharge torunoff is that it can be compared directly to precipi-tation (both expressed as cm/yr) to provide a roughestimate of evapotranspiration, assuming no losses orgains to groundwater or interbasin transfers. For all

The Variety of Rivers

1089

TABLE 24.1 Some vital statistics of the rivers of North America based on data presented in the 22 chaptersummaries. The name, basin area, and discharge are given for the largest single river of each basin or region,which is the main stem when all rivers drain into a single basin. Median values are given for major attributesof the rivers of each region, but main-stem rivers were excluded so that median values are not influenced bythe much higher main-stem values.

Largest River Median Values for Rivers of Region

MedianBasin Basin Basin annual Runoff/ No.

Major Basin or Discharge Area Discharge Area relief runoff Precip of Region Name (m3/s) (km2) (m3/s) (km2) (m) (cm) (%) rivers

Lower Mississippi Lower Mississippi 18,400 3,270,000 98 7,773 452 44 35 9St. Lawrence St. Lawrence 12,600 1,600,000 150 16,458 510 42 40 9Mackenzie Mackenzie 9,020 1,743,058 404 68,134 2,605 18 45 9Ohio Ohio 8,733 529,000 420 30,300 690 54 45 11Columbia Columbia 7,730 724,025 225 22,667 2,353 28 58 11Yukon Yukon 6,340 839,200 670 67,250 2,458 29 99 6Fraser Fraser 3,972 234,000 231 13,500 2,625 36 77 8Upper Mississippi Upper Mississippi 3,576 489,510 157 25,929 163 27 30 10Mexico Usumacinta/ 2,678 112,150 65 47,124 3,000 10 9 10

GrijalvaNelson Churchill Nelson/Churchill 2,480 1,093,442 164 148,900 350 7 13 6Missouri Missouri 1,956 1,371,017 51 32,600 1,182 6 13 11Gulf Coast SE Mobile 1,914 111,369 289 20,400 220 54 37 11

StatesPacific Canada Kuskokwim 1,900 124,319 1,214 43,149 2,562 90 100 10Atlantic Canada Churchill 1,861 93,415 246 7,860 490 73 67 11Arctic Thelon/Kazan 1,380 239,332 408 40,363 1,464 26 86 6Atlantic Coast Susquehanna 1,153 71,432 361 25,707 1,358 58 55 10

NE StatesSouthern Plains Arkansas 1,004 414,910 68 20,230 714 23 23 11Pacific Coast Sacramento 657 72,132 171 11,158 2,847 42 72 10

StatesColorado Colorado 550 642,000 17 24,595 2,600 4 15 11Atlantic Coast SE Santee 434 39,500 227 25,326 372 31 31 11

StatesGulf Coast SW Brazos 249 115,566 79 46,540 720 4 5 9

StatesGreat Basin Bear 71 19,631 26 7,925 2,341 7 13 7

Ch24.qxd 3/24/05 2:21 PM Page 1089

rivers in this volume, the median annual precipita-tion and runoff were 84 and 31cm, respectively, indi-cating that roughly 37% of precipitation becomesrunoff, and the remainder is lost to evapotranspira-tion (or possibly groundwater). Our estimate ofmedian precipitation is higher than values of 67cm(Hornberger et al. 1998) and 76cm (Shiklomanov1993) reported for the continent as a whole (seeChapter 1). This might suggest that the river datafrom this book are biased toward regions whererivers are concentrated (i.e., in higher-precipitationareas). On the other hand, the median runoff of 31cm is similar to reported continentwide values of 29cm (Hornberger et al. 1998) and 34cm (Shiklomanov 1993).

Table 24.1 gives median estimates for the indi-vidual (excluding main stem) rivers of runoff as a per-centage of precipitation, where low values usuallyindicate high ET. The percentage of precipitation thatbecomes runoff varies greatly among rivers andregions and is strongly dependent on both precipita-tion and temperature. Not surprisingly, northernrivers with their cold climates and low plant pro-duction have very high RO/PPT ratios, approaching100% (values for Pacific Canada likely are inflatedbecause precipitation data are available mainly forthe lower basins whereas runoff reflects higher pre-cipitation in the upper basins). In comparison,median annual precipitation for the rivers of Mexicoincluded in this volume is approximately 100cm, andrunoff only 10cm, indicating that many of these low-latitude basins are relatively wet but experiencehigh evapotranspiration (although at least one, theCandelaria, loses much of its water through ground-water seepage to the sea).

The effect of temperature on percentage runoff isparticularly clear from examination of rivers drain-ing into the Atlantic Ocean, where mean annual precipitation for 32 river basins (Chapters 2, 3, and21) falls within the relatively narrow range of 92 to147cm. In these eastern rivers, mean air temperatureexplains 72% of the variation in percentage runoff(Fig. 24.1). In the southeast, where mean air tem-peratures range from 15°C to 20°C, percentagerunoff is near 30%. But in eastern Canada, wheremean air temperatures are <6°C, percentage runoff istypically >60%. The influence of low precipitation onpercentage runoff in regions of relatively high tem-perature can be illustrated for the Gulf Coast riversof the southwestern United States. The westernmostPecos and Rio Grande have <30cm precipitation and£1% runoff, whereas the easternmost rivers (Sabineand Neches) have >125cm of precipitation and 16%

to 20% runoff. Of course, these estimates can bestrongly influenced by human withdrawals, whichalso reduce runoff, and it is often impossible to sep-arate this from natural evapotranspiration losses.

In addition to the annual water balance describedhere, seasonal patterns of PPT versus ET and theextent of intra- and interannual variability in bothare primary determinants of a river’s flow regime(Poff et al. 1997; Chapter 1). Mean and seasonal temperatures also are important, as they affect howmuch precipitation falls as snow and may be storeduntil spring thaws result in a rise in runoff. Theextent of agriculture, types of crops, and reliance onirrigation drawn from surface waters (deep ground-water would not be part of the normal water budget)will further affect flows, principally during thegrowing season, and human withdrawals for otherpurposes, including municipal use, may have notice-able impacts.

In Chapter 1, several examples were given to illus-trate the major factors affecting seasonal patterns ofPPT and RO (see Fig. 1.4). Similar graphs of monthlyPPT and RO presented throughout this book (withET inferred from the difference) provide multipleexamples that clearly distinguish the seasonal flowsignatures of different regions. For example, seasonalrunoff patterns of the southeastern Atlantic andEastern Gulf drainages appear largely influenced byseasonal patterns of evapotranspiration. The north-eastern Atlantic United States, the upper Mississippi,and the Ohio basin also are influenced by seasonalpatterns of evapotranspiration, but in addition theyare affected to varying degrees by spring snowmelt,depending on their latitudes and monthly air tem-

24 Overview and Prospects

1090

0

20

40

60

80

100

-5 0 5 10 15 20 25

Mean Air Temperature (oC)

Ru

no

ff /

Pre

cip

itat

ion

(%

)

FIGURE 24.1 Annual runoff as a percentage of precipita-tion versus mean annual air temperature for rivers drain-ing into the Atlantic Ocean (from Chapters 2, 3, and 21).R2 = 0.72. This graph excludes the St. Lawrence River,which drains a much larger area, including the Great Lakesand their tributaries.

Ch24.qxd 3/24/05 2:21 PM Page 1090

peratures. On the other hand, runoff patterns ofnorthern rivers and those draining large westernmountains are dominated by spring snowmelt. Alongthe Pacific Coast of the coterminous United States,however, the pattern of runoff is most obviouslyinfluenced by the distinct winter precipitation. Incontrast, the runoff in most rivers of Mexico followsthe strong pattern of summer precipitation.

Variation in Biological and Ecological CharacteristicsThe foregoing comparisons emphasize the variety of rivers in physical terms and landscape context, but what about biological and ecological variationamong rivers? Chapter accounts amply documentsubstantial diversity in the number of fish speciesamong rivers and regions. Rivers of the Ohio basinhave a reported median number of 120 fish speciesper individual river basin. Similarly, the mediannumber of fish species in the southeastern Atlantic,the Gulf Coast of the southeastern United States, the upper Mississippi, and the lower Mississippi all exceed 100 per river. The Tennessee River andMobile River each have a staggering >225 fishspecies. These Midwestern and southern basinssupport far more fish species per river basin thanrivers of the west or far north. For example, themedian number of native fish species reported byindividual river basin was 23 for rivers of the Colum-bia basin, 18 for rivers of the Pacific United States,and 10 for rivers of the Colorado basin (although theentire basin of the Columbia and Colorado includesconsiderably more native fishes than the median forindividual rivers). The Nelson–Churchill system ofsouthern and central Canada supports a median of62 fish species (the extensive lake habitat of theNelson–Churchill likely contributes to this speciescount), and the rivers of the Yukon basin support 27species. Physiographic and habitat variation, glacialhistory, and dispersal opportunities are some of thefactors that underlie this enormous zoogeographicvariation (Hocutt and Wiley 1986). In general, thenumber of fish species increases with area of drainagebasin, is markedly higher in the east than the west,and is reduced in far northern rivers relative to theeastern United States. Mexican rivers have somewhatfewer species per river than the eastern United States,but they are not as well sampled. The fraction ofendemic species is very high in Mexican rivers,however, and when scaled to account for differencesin area by country rather than river basin, the

number of Mexican fish species per km2 exceeds boththe United States and Canada (Chapter 23).

Biogeographic information for other taxa is scant.There was insufficient information available on ariver-specific basis to describe patterns for any taxaother than fish. Molluscan diversity (approximately>300 species in North America; Master et al. 1998,Abell et al. 2000) is well known to be highest in thesoutheastern United States, which is globally rich infreshwater mollusks, less diverse in the west, anddepauperate in the far north. This statement alsoapplies to decapod crustaceans (nearly 300 species in North America). However, much more work isneeded before species diversity trends can be identi-fied for the abundant aquatic insects of rivers, which,based on a few well-studied systems, likely includeseveral hundred species per river. Many studies ofaquatic insects use genus-level taxonomy, species dis-tributions are largely unknown, and large rivers areunderstudied. Certain genera, including Baetis andStenonema (mayflies), Hydropsyche (caddisflies),Simulium (black flies), and Polypedilum (nonbitingmidges) are widely reported in rivers throughout thisbook, indicating that species in these genera (andothers) may often be important components of themajority of river communities.

Few rivers of order 6 and higher have receivedintensive investigation of ecosystem processes, andeven description of the biota often is restricted largelyto surveys of fishes, owing to the tendency ofrunning-water ecologists to focus on smaller streamsthat are easier to study. In addition, our interpreta-tion of current knowledge is complicated by the vari-able extent of anthropogenic disturbance. Relative totheir presettlement state, many rivers have elevatedturbidity and nutrients, reduced connectivity bothlaterally and longitudinally, and altered waterbudgets and water-residence times, and receive acomplex brew of industrial, agricultural, and phar-maceutical chemicals.

Based on studies of ecosystem metabolism from amodest number of North American rivers, low ratiosof gross primary production to respiration (P/R)appear to be common in large rivers. Among the bestexamples of low P/R ratios are careful studies in theHudson, Ogeechee, and Ohio rivers (Chapters 2, 3,and 9, respectively). Of these three, the Ogeechee isarguably the least altered because it is unregulated by dams along its entire length and has a large intactfloodplain. Given the tea-colored waters and exten-sive floodplain of this low-gradient blackwater river,which receives much allochthonous organic matterfrom the floodplain swamp and has relatively low

1091

The Variety of Rivers

Ch24.qxd 3/24/05 2:21 PM Page 1091

primary production in the water column, theobserved dominance of heterotrophic processes isexpected. The carbon budget of the Hudson (Table2.1), although also providing strong evidence of het-erotrophy, nonetheless suggests a substantial role forautochthonous production, which perhaps was evengreater prior to a number of human impacts.

Primary production within large rivers is expectedto be limited by some combination of environmentalfactors, including light, nutrients, and downstreamexport of algal cells (Allan 1995). Light limitation isfrequently reported. In the lower Missouri River(Chapter 10), reported photic depths averaged 0.78m and mixed depth/photic depth ratios averaged10.2. Because river water columns are well mixedand most of the water column experiences light levelstoo low to support algal growth, light limitation canbe significant. Summer primary production by riverphytoplankton for the middle and lower Ohio riversections also showed evidence of light limitation,although in this system large impoundments havenotable effects by lengthening water residence time,and so phytoplankton production in river sectionsslowed by impoundments likely are higher thanwould otherwise be expected. The well-studiedHudson River (Chapter 2) illustrates several ways in which human actions may influence observedprimary production. High turbidity due to silt result-ing from intensive land use and reduced water resi-dence times during summer due to human alterationof hydrology both reduce algal production. Theintroduced zebra mussel has had profound impacts,including through its suspension-feeding, and likelyhas contributed to a reduction in phytoplankton andan increase in macrophytes.

Although nutrients may seldom be limiting in the rivers just mentioned, northern rivers frequentlyhave low background nutrient levels and thereforerespond strongly to nutrient enrichment. The WapitiRiver, a tributary of the Smokey and Peace rivers, has low background nutrients but receives highpoint-source inputs of N and P from a pulp mill anda municipality. Periphyton biomass below inputsincreased more than tenfold, resulting in highermacroinvertebrate densities and increased fish condi-tion in a bottom-up trophic cascade.

Despite the dominance of ecosystem metabolismby heterotrophy, presumably reflecting the extent ofinputs of allochthonous carbon, autochthonousprimary production can be important to large-riverfood webs (Chapter 1; Thorpe and DeLong 2002).Some studies suggest that high system respirationlikely is due to the metabolism of allochthonous

carbon by microbes, and little of this carbon isthought to reach the metazoan food chain. Stableisotope and other evidence indicates that most meta-zoan production is derived from autochthonous production, often phytoplankton. Thus, the river isheterotrophic overall, with a P/R ratio well below 1,but the metazoan food webs may be primarilyautotrophic. Reports for the Ohio River, where much of the work leading to this hypothesis wasdone, and for the Hudson River provide supportingevidence.

Snags and large wood play an important role inthe ecology of large rivers, creating stable substratefor producers and invertebrates and cover for fishes(Benke et al. 1985). Snags and backwater areassupport the most diverse and productive assemblageof invertebrates, as the fine substrate of main chan-nels commonly is too unstable for invertebrates. Veryhigh secondary production of invertebrates occurs onsnags, and, exported as drift, contributes to fishbiomass and production. Many large rivers had anabundance of snags in their presettlement condition,as exemplified by descriptions of the Cape Fear Riverin the early 1700s (Chapter 3). Clearing of snags,which was common in the 1800s and early 1900s,improved river navigation and provided access forsteamboats but likely had substantial effects onecosystem processes. Blockage of the main channel ofthe Red River in Louisiana with driftwood was soextensive that a major effort by the U.S. governmentbeginning in the 1870s was undertaken to open theriver to boat traffic (Chapter 7). Detailed studies of snag habitat in the Satilla and Ogeechee rivers(Chapter 3) reveal what has been minimized or lostin many other rivers. The main channel of theunchannelized Missouri River provides anotherexample, where woody habitats were shown to con-tribute over two-thirds of total insect productioncompared with mud and sand substrates and back-waters (Chapter 10).

In summary, ecosystem processes have beenstudied for only a modest number of North American rivers, which strongly argues for theimportance of further study. The lack of informationon ecosystem-level processes for many of the largerivers of North America represents not only a signif-icant frontier for further investigations but also aserious shortage of knowledge on which to base man-agement decisions. Because these large rivers differ intheir natural settings and extent of anthropogenicdisturbance, the comparisons and generalizationsoffered here should be viewed with considerablecaution, and much remains to be learned.

24 Overview and Prospects

1092

Ch24.qxd 3/24/05 2:21 PM Page 1092

FEW RIVERS ARE PRISTINE

Many rivers of North America have been affected byhuman actions since at least European settlement andsome were influenced much earlier, as described forancient cultures in Mexico (Chapter 23). Over timethe variety and magnitude of impacts has grown dra-matically, although we should recognize that therehave also been successes in river management. At thebeginning of the twenty-first century we cannot accurately describe the condition of North America’srivers, because presently we lack effective systems of national assessment and large rivers are under-studied. We do know that many of our rivers sufferfrom pollution, habitat degradation, fragmentationby dams, colonization by nonnative species, andmore, and that climate change, water withdrawals,the continued spread of nonnative species, and theongoing expansion of human activities will posegreater threats in the future.

Chemical contamination and water quality inrivers continue to be of major importance, reflectingthe public’s legitimate concern for drinking water andpublic health, as well as the ecological consequencesof polluted waters. Although the past 30 or moreyears of regulatory activities focused on improve-ments in water quality have achieved notable suc-cesses, many rivers struggle to overcome a legacy ofsevere pollution, particularly from industrial pollu-tion on the lower portions of rivers. This has beenespecially true in high-population areas of theAtlantic Coast rivers of the northeastern UnitedStates (Chapter 2). Water quality of rivers such as theDelaware, Hudson, and Connecticut have improved,but these rivers and others still have sedimentsheavily contaminated with heavy metals, PCBs, andother chemicals.

The Delaware River in the vicinity of Philadelphiareminds us that water-quality deterioration has along history (Chapter 2). Significant pollution wasreported as early as 1799; low values of dissolvedoxygen, presumably due to the decomposition oforganic waste, were reported in 1915; and dockworkers along the river during World War II report-edly were nauseated by the stench. In the era afterWorld War II, new, potentially highly harmful chem-icals were added to the brew being discharged intorivers. The Connecticut River contains a substantialburden of contaminants, including trace metals(chromium, copper, lead, mercury, nickel, and zinc)and organic compounds (chlordane, DDT, polychlo-rinated biphenyls, and various polycyclic aromatichydrocarbons). Fish and shellfish advisories have

been issued in Massachusetts and Connecticutbecause of high levels of polychlorinated biphenyls.PCB contamination in the middle Hudson likewise isresponsible for restrictions on commercial and sportfisheries and an ongoing battle over the methods and financial responsibilities for their removal(Chapter 2).

Other rivers have been greatly affected by thetimber industry due to pulp mill wastes and associ-ated factors. The Saguenay, a very large tributary ofthe St. Lawrence (Chapter 21), has seen considerablepollution from the timber industry and industrial pollution associated with paper manufacturing(mercury, PCBs, and PAHs) despite its relativeremoteness, low population density, and sparse agri-cultural activity in its watershed.

Confined animal feeding operations on anunprecedented scale add a new dimension to chemi-cal contamination of large rivers. The numerousindustrial hog and poultry operations located in andnear the floodplain of the Cape Fear River (Chapter3), a region prone to hurricanes and extreme flood-ing, illustrate the risk that river ecosystems willreceive high volumes of organic wastes.

Nutrient enrichment is widespread in most majorriver systems that drain landscapes of significanthuman population and disturbed land use. Heavilyagricultural rivers such as the Platte (>90% agricul-ture, Chapter 10), the Great Miami (80% agricul-ture, Chapter 9), and the Minnesota (>95%agriculture, Chapter 8) typically have high concen-trations of nutrients (phosphates and nitrates) andpesticides. For example, the Minnesota River, as wellas other rivers of the Upper Mississippi basin, typically has concentrations of NO3-N and PO3-Papproaching or exceeding 3mg/L and 0.1mg/L,respectively, each of which is substantially higherthan natural background levels (Chapter 1).Although the impact of elevated nitrogen levels onprimary production and the resultant hypoxia withinthe Gulf of Mexico is now well established (Chapter6), the potential effect of nutrient enrichment onlarge-river food webs is not well studied and may bemasked by other factors, including turbidity andaltered water residence times. However, in somenorthern rivers that have experienced little distur-bance and have very low nutrient concentrations theeffects of elevated nutrients can be very pronounced,as in the case of the Wapiti River.

Dams and impoundments, levees, and channel-ization have altered the physical dimensions anddynamics of many kilometers of large rivers. Evenbefore the twentieth-century era of large-scale river

Few Rivers Are Pristine

1093

Ch24.qxd 3/24/05 2:21 PM Page 1093

engineering, North America’s major rivers werealtered for transport. An extensive canal system wasdug in the seventeenth century to provide access andtransport of goods into the Delaware valley (Chapter2). Broad floodplains of backwaters and wetlandswere transformed by the simultaneous activities ofnarrowing and deepening the main channel, snagremoval, and drainage of adjacent bottomlands forfarming. The considerable loss of channel complex-ity and lateral connectivity in the upper freshwatertidal Hudson, where deepening of the channel andfilling of wetlands and backwaters resulted in anarrow and simplified river system, is illustrated inFig. 2.11. In the mid-1800s, less than half a centuryafter Lewis and Clark ascended the Missouri, similarchannel modification processes began, followed inthe early 1900s by the construction of a system ofwooden pile dikes to create a single, self-scouringnavigation channel. Then came flood-control leveesby the mid–twentieth century, and a series of majordams. As Chapter 10 reports, flooding historicallywas essential in maintaining the natural character of the river–floodplain complex. Subsequent to thischain of channel-modifying events the channelizedsection of the Missouri River underwent major geomorphic changes, including an 8% reduction inchannel length, a 50% reduction in channel watersurface area, a 98% reduction in island area, and an

89% reduction in the number of islands (see Fig.10.2).

Using information on size, number, and distribu-tion of dams in main-stem rivers we have attemptedto place rivers into one of three categories: rivers withfew or no dams along main stems or on tributaries;rivers moderately fragmented, with at least onesizable main-stem dam; and rivers strongly frag-mented, with multiple or large dams along the mainstem. Figure 24.2 shows the percentage of riversystems within a basin or region that were recorded ineach category. Over half of the river basins or regionsof North America and ten of thirteen basins orregions primarily located in the United States had atleast 50% of their main rivers scored as strongly frag-mented. Of the eight primarily Canadian basins andregions, the Nelson–Churchill and the St. Lawrenceare highly impacted by dams, as are rivers of theAtlantic Canada region. The remaining Canadianbasins and regions experience low dam fragmentationoverall, although this conceals some specific rivershighly influenced by dams, including the Peace Riverin the Mackenzie basin and the Nechako River in theFraser basin. Although the most fragmented Mexicanrivers are found in the arid north, less fragmentedrivers in the tropical south, particularly the large andfree-flowing Usumacinta, are endangered due to pro-posed hydropower projects.

24 Overview and Prospects

1094

0

20

40

60

80

100

Nelson C

hurchill

St. Law

rence

Atlantic

Coas

t NE S

tate

s

Gulf Coas

t SW

Sta

tesOhio

Great

Bas

in

Colora

do

Souther

n Plai

ns

Upper M

ississ

ippi

Pacifi

c Coas

t Sta

tes

Colum

bia

Lower M

ississ

ippi

Missouri

Atlantic

Coas

t SE S

tate

s

Atlantic

Can

ada

Mexico

Gulf Coas

t SE S

tate

s

Arctic

Macke

nzie

Frase

r

Pacifi

c Can

ada

Yukon

Fra

gm

enta

tio

n (

%)

FIGURE 24.2 The percentage of river systems within a region (based on 5 to 12 rivers, median of 10) that wererecorded as strongly fragmented, with multiple or large dams along their main stems (black bar); moderatelyfragmented (gray bar); or essentially unfragmented (unshaded bar).

Ch24.qxd 3/24/05 2:21 PM Page 1094

The extent of dams in North America has beenwell-documented by previous publications (Dynesiusand Nilsson 1994, Graf 1999), and there is a largeliterature on their consequences (Ward and Stanford1979, Hart and Poff 2002, Stanley and Doyle 2003).The imposition of lentic conditions can eliminatebenthic species, including many freshwater mussels,and generally promotes a shift to plankton-basedfood webs. With nine main-stem reservoirs, very littleof the Tennessee River main stem remains freeflowing (Chapter 9), the density and diversity of themain-stem mussel populations have declined dra-matically, and gizzard shad and threadfin shad, whichare dependent on the abundant plankton in thesystem, have become the principle forage fishes. Theinfluence of dams on any given river can vary greatly,but alterations to the river’s natural flow regime are widespread. In some cases, regulation by a dam largely flattens the monthly hydrograph (e.g.,Churchill Falls dam on the Churchill River,Labrador; Hoover Dam on the Colorado River). Inmany other cases, hydropower facilities cause enor-mous hourly and daily fluctuations in discharge andwater height that are hidden by plots of meanmonthly discharge or runoff values. Perhaps the mostextreme example is on the Nelson River (Chapter19), where dams can cause hourly summer dischargeto range between 0 and 7200m3/s. In yet other cases,dams and diversions can reduce downstream flows to

zero for very long periods (e.g., Rio Grande, Chapter5; Gila River, Chapter 11; Rio Conchos, Chapter 23).By reducing the magnitude and frequency of floodevents, a large reservoir on the Peace River, a majortributary of the Slave River, is causing the Slave Riverdelta to shrink (Chapter 18). Loss of wetland habitatin the Slave River Delta and in many such wetlandsthroughout North America has devastating impactson birds and other wildlife.

Human population density can be a useful proxymeasure of human impacts, including nutrientloading (Cole et al. 1993). It may also serve as a veryrough proxy for land-use change and urbanization,although the current phenomena of urban sprawl anddepopulation of agricultural areas complicate thisrelationship significantly (Meyer and Turner 1994).Although human density estimates reported in thisbook may be approximate, they nonetheless reflectvery substantial regional differences. River basins inthe Atlantic states of the northeastern United Statessupport much higher population densities than anyothers in North America (Fig. 24.3). Within thecoterminous 48 states the Missouri and Coloradobasins and river basins of the southern plains are leastdensely settled. The St. Lawrence and Nelson–Churchill basins support human populations com-parable to less dense U.S. basins, whereas morenorthern river basins have comparatively very lowpopulations.

Few Rivers Are Pristine

1095

0

20

40

60

80

Atlantic

Coas

t NE S

tate

s

Atlantic

Coas

t SE S

tate

s

Ohio

Gulf Coas

t SE S

tate

s

Mexico

Upper M

ississ

ippi

Gulf Coas

t SW

Sta

tes

Pacifi

c Coas

t Sta

tes

St. Law

rence

Lower M

ississ

ippi

Missouri

Souther

n Plai

ns

Great

Bas

in

Colum

bia

Nelson C

hurchill

Colora

do

Frase

r

Pacifi

c Can

ada

Yukon

Macke

nzie

Atlantic

Can

ada

Arctic

Po

pu

lati

on

(p

eop

le/k

m2 )

FIGURE 24.3 Median values of human population density (people/km2) for regions and river basins. In someinstances within-basin variation varied widely; for instance, the Fraser basin has a density of 11.7/km2 for theFraser itself, compared to <1/km2 in most of its tributary river basins. Densities for individual Atlantic Coastrivers of Canada show a similarly wide range of values.

Ch24.qxd 3/24/05 2:21 PM Page 1095

The extent of urban and agricultural land withinriver basins provides another indication of humanimpact. It is difficult to assess the influence ofbetween-basin differences in the extent of urban land,because urban land use commonly is a low percent-age of total area yet exerts a disproportionately largeinfluence both proximately and over distance (Pauland Meyer 2001). The river basins with the greatestpercentage of urban land include Atlantic Coastrivers of the northeastern United States, Gulf Coastrivers of the southwestern United States, the St.Lawrence, and the Colorado. Agricultural land usevaried considerably among basins, from near zero insome Canadian basins to 66% of the Upper Missis-sippi basin. Six major river basin of the United Stateshad over 40% of their areas in agriculture: the LowerMississippi, Upper Mississippi, Southern Plains,Ohio, Missouri, and Colorado.

The combined data on dams, population, andland use argue that all river basins of the cotermi-nous United States experience considerable humandisturbance. However, all or nearly all of these basinscontain tributaries and river segments of very highquality that are suitable for designation under federaland state river protection programs. Indeed, thesedata, and the insightful analyses of individual chap-ters, clearly demonstrate the urgency and importanceof protecting least-disturbed river segments wheneverpossible. Major river basins of the far north includemany that appear little changed according to themeasures reported here (which do not, however,include mineral extraction, timber harvest, or roadand recreational development, which are significantthreats to boreal regions; Schindler 1998), offeringthe potential for protection of even larger and moreintact watersheds and river systems.

Nonnative species pose another, less-visible threatto river ecosystems, and in a number of instances thenative fauna is largely displaced, at least in terms ofbiomass, energy flow, and community structure. Thisis especially so in western rivers, where the numberof nonnative species can equal or exceed the totalspecies richness of native fishes (Fig. 24.4). Com-monly in such situations, native species persist atreduced abundances and in isolated locales, and oftenwarrant endangered species status. From the view-point of species richness, the fish assemblage mayappear little changed in the number of native speciesbut be greatly augmented with nonnative species.From the perspective of the biological community,however, biomass and energy flow may be totallydominated by the nonnative component (e.g., asurvey of the Colorado River in Canyonlands

National Park, Utah, by Valdez and Williams [1986],reported 83% percent of individuals to be nonna-tive), and so ecologically the system is even morealtered than might be inferred from consideration ofa species list.

Based on information presented in river sum-maries, the median number of native fish species forthe 11 river basins of the Colorado system exclusiveof the main stem is 10 (Chapter 11), and a medianof 4 species are listed as endangered. Today the

24 Overview and Prospects

1096

Colorado Basin

0

20

40

60

80

100

Colora

do

Green Gila

Little

Colora

do

Yampa

Gunnison

Verde

San Ju

an

Virgin

Salt

Bill W

illiam

s

Black

TotalNative

Pacific-US Basins

0

20

40

60

80

Sacra

men

to

San Jo

aquin

Klamat

h

Santa

Ana

Russia

n

Salinas

Umpqua

Eel

Rogue

Santa

Mar

garita

Columbia Basin

0

20

40

60

80

100

120

140

Colum

bia

Willa

met

te

Owyhee

Snake

Grande R

onde

Methow

Clearw

ater

Flathea

d

John D

ay

No

. of

Fis

h S

pec

ies

FIGURE 24.4 Native fish species and total fish species inthree river basins or regions of the western United States(from Chapters 11, 12 and 13). Note that the limited nativefish fauna has been substantially augmented by nonnativespecies.

Ch24.qxd 3/24/05 2:21 PM Page 1096

median number of all fish species for these 11 basinsis 29, and so the addition of nonnatives has roughlytripled the species count, whereas much of the nativefauna is in need of protected status. The 10 riverbasins of the Pacific Coast rivers of the United Statescontain 18 native species and a total of 39 fish species(median values). Within the Columbia system, basedon eight rivers (excluding the main stem), chapterdata indicate a median number of 23 native fishesand a total of 38 species.

The influence of nonnative fishes may be less dramatic elsewhere compared to the western UnitedStates, perhaps because high native species richnessof some regions has helped to limit invasions andperhaps in far north river systems because fewer non-native species are preadapted to these environments,and there have been fewer invasion opportunities.Nonetheless, the influence of nonnative fishes hasbeen profound in many other rivers. For example, in several rivers of the northeastern (Chapter 2; Connecticut, Hudson, Delaware, and Susquehannarivers) and southeastern United States (Chapter 3;James River), approximately one-third of all speciesare nonnatives, raising the total number of species to>100 in many cases. Nonnatives sometimes dominatefish collections in these rivers (e.g., Chapter 3; CapeFear River, North Carolina), with unknown impactson ecosystem function.

Finally, a changing climate driven by greenhousegasses is likely to have profound impacts on fresh-water ecosystems, altering hydrologic cycles, flowregimes, and riparian vegetation and likely facilitat-ing range expansions and the proliferation of non-native species. These climate-related effects are as yet poorly understood and so pose one of the mostimportant emerging threats to rivers in comingdecades.

NORTH AMERICA’S RIVERS IN THE TWENTY-FIRST CENTURY

During the twenty-first century the rivers of NorthAmerica will experience growing human pressures.Water use will increase due to the growth of afflu-ence and population, and in many areas surface-water supplies are nearly fully appropriated (NRC2001). Reliance on groundwater will increase, withuncertain but serious consequences for river flows(Postel and Richter 2003). Some areas of the westernUnited States are expected to exhaust their ground-water sources during this century. Surely some will call for transbasin water diversions on a scale

presently unknown in North America, potentiallyaltering river flows and facilitating the spread ofnonnative species. Conflicts between human andenvironmental needs for water are certain to increase.Meanwhile, all of the known threats remain,although some, such as point-source pollution, arediminished in many areas, whereas climate changeadds a new challenge, likely to interact with otherthreats in complex ways. Opportunities exist to haltthe decline in the condition of the rivers of NorthAmerica and even to improve their status. However,those undertakings call for better knowledge to guidemanagement and restoration and a stronger resolvethat consumption of water resources cannot be dic-tated by uncontrolled growth.

Climate ChangeRivers and other freshwater ecosystems will beaffected by projected climate change in multipledirect and indirect ways, of which only the moststraightforward can be identified with high confi-dence (Firth and Fisher 1992, Meyer et al. 1999,Allan et al. 2005). A warmer climate will result ingreater evaporation from water surfaces and greatertranspiration by plants. However, whether rainfallwill increase or decrease in a particular region isuncertain, and it also is difficult to predict whetherthe change in PPT or ET will be greater. Thus,increased precipitation might be accompanied byeven greater evapotranspiration, leading to reducedrunoff. General circulation models (GCMs) are notyet able to reliably predict how precipitation andwater supplies will change at the local or regionallevels (NAST 2000), and there is still a great deal ofuncertainty in climate-change forecasts (Forest et al.2002, Elzen and Schaeffer 2002). For example, Frederick and Gleick (1999) examined runoff for 18water-resource regions of the United States using twocontrasting GCMs and found that predictions oftenwere in disagreement. The two models predicted thesame direction of change in runoff in only 9 of the18 regions, and where the direction was similar oftenthe magnitude was not. Perhaps the greatest singlechallenge in evaluating aquatic ecosystem response tofuture climate change is the considerable uncertaintyregarding the local and regional responses of thehydrologic cycle.

Despite this uncertainty, the potential for climatechange to alter total runoff and the seasonality offlow regimes is considerable. Some seasonal shifts arecertain to take place, because warming winter tem-peratures will cause some areas to experience more

North America’s Rivers in the Twenty-First Century

1097

Ch24.qxd 3/24/05 2:21 PM Page 1097

winter rain and episodes of rain on snow, trans-forming predictable spring melt runoff into highlyvariable winter runoff. Rivers of the far north maybe most vulnerable because they will experience dis-proportionately greater warming, along with associ-ated dislocations of the hydrologic cycle (Poff et al.2001); for example, the Mackenzie basin may warmby as much as 5°C by the middle of the century(Chapter 18).

In addition to affecting the hydrologic cycle,warmer temperatures are expected to result in higherecosystem metabolism and productivity. The biota ofrivers are dominated by cold-blooded organisms, andthese ectotherms generally increase their metabolismwith each degree increase in temperature until verynear their upper temperature tolerances. Rates gen-erally increase by a factor of 2 to 4 with each 10°Cincrease in water temperature, up to about 30°C(Regier et al. 1990). In a review of over 1000 estimates of macroinvertebrate production, Benke(1993) estimated a 3% to 30% increase in biomassturnover rates for each 1°C increase in temperature.Thus, although there may be complex and unpre-dictable changes in species composition, an overallincrease in system productivity is a potential responseto climate warming.

Poleward range shifts by those taxa able to dis-perse are highly probable under warmer climate scenarios. Sweeney et al. (1992) estimated that a 4°Cwarming would result in a 640km northward latitu-dinal shift in thermal regimes for macroinvertebrates,and several authors have estimated comparable dis-tances for the range displacement of particularspecies of fishes. The opportunity to disperse, and thepresence of corridors versus barriers, can be highlyvariable among river basins, however. For example,fishes in the southern Great Plains and the desertSouthwest cannot move northward because thosestreams and rivers tend to run west and east. Becausesummer water temperatures now approach the upperlimit for a number of fish species, just a few degreesof warming poses serious risk of extinction for nativefishes in these regions (Chapter 7; Matthews andZimmerman 1990). Fishes of the far north may beespecially vulnerable, as warming will favor speciesthat presently are only marginally successful. Arcticcoregonids (broad whitefish, least cisco, Arctic cisco)may be particularly at risk. Temperature also sets thenorthern range limit for harmful nonnative speciessuch as the zebra mussel (Strayer 1991), and so anorthward range expansion seems highly probable.Given the well-established negative impacts of non-native species on freshwater ecosystems (Allan and

Flecker 1993), native biodiversity may be adverselyaffected by such range shifts.

Rivers draining forested landscapes and withforested riparian and headwater zones derive muchof their energy as organic matter inputs from the ter-restrial environment (Chapter 1). Climate change islikely to bring about shifts in terrestrial vegetationand changes in leaf chemistry and affect the process-ing of detritus and functioning of the microbial-shredder food web linkage in complex ways. Alteredcarbon/nitrogen ratios of the leaves likely will reducepalatability, temperature changes will affect leaf-processing rates, and floods may export leaf matterbefore it can be processed (Rier and Tuchman 2002,Allan et al. 2005). These interactions are complexand potentially offsetting, making the overall impactof climate on this important energy supply difficultto predict.

Other ThreatsAlthough the potential impact of a changing climatedeservedly is the focus of much current concern,more familiar threats may prove to be of equal orgreater importance. Pollution from point and non-point sources continues, and legacies of past con-tamination are numerous. Rivers continue to befragmented by dams, which modify flow, alter waterbudgets, disrupt migrating species, and interfere withecosystem connectivity and function. Levees andother flood-control measures remain largely in place.The spread of nonnative species is virtually a cer-tainty. Although heartening examples of improve-ments can be found and some of the most harmfulpollution practices have largely been halted, manyrivers experience a mix of stressors that includelegacy effects, long-standing threats, and newlyemerging challenges.

Water withdrawals and transfers compriseanother potential threat to rivers but are associatedwith so much uncertainty they are difficult to assess.Due to the combined influence of population andeconomic growth, freshwater demand is expected to grow significantly during the twenty-first century(NRC 2001). On the other hand, past projections ofdemand have invariably proven to be overestimatesdue to unanticipated advances in efficiency. Nonethe-less, because surface waters are largely appropriatedin many regions, it seems highly likely that ground-water withdrawals will increase, with uncertaineffects on hydrologic budgets and river ecosystems.Some areas of the Great Plains and the southwesternUnited States that now rely heavily on nonrenewable

24 Overview and Prospects

1098

Ch24.qxd 3/24/05 2:21 PM Page 1098

groundwater will almost assuredly desire to importwater via canals and interbasin transfers.

Examples of existing out-of-basin transfers aredescribed in several chapters. The Sacramento River innorthern California (Chapter 12) has approximately50% of its flow diverted via the California Aqueductand Delta–Mendota Canal to meet urban and agricul-tural water demands in the southern part of the state.The Colorado River experiences transfers from itsheadwaters via the Colorado–Big Thompson projectto provide water to Denver and other cities of the Mis-sissippi drainage, whereas a diversion from the LowerColorado below Lake Havasu to Phoenix and California sends almost 40% of the virgin river flowout of basin (Chapter 11). Water is siphoned off fromthree reservoirs in the Delaware River headwaters aspart of the water supply of New York City and dis-charged into the Hudson estuary (Chapter 2). Amassive diversion of water occurs from the ChurchillRiver in northern Manitoba in order to increasehydroelectric generating capacity through dams in thenearby Nelson River. Some 75% (or >700m3/s) of theflow of the Churchill River was diverted into theNelson system during the 1970s, causing severe eco-logical and social impacts on both the lower Churchilland the lower Nelson, where hydroelectric releasesalso have extreme daily fluctuations (Chapter 19).Large diversions are a common practice in Canada,where Dynesius and Nilsson (1994) report total out-of-basin transfers of 4400m3/s. Based on various newsmedia accounts, interbasin water transfers have beenconsidered from the Great Lakes to surroundingareas, between the Upper Missouri and southern trib-utaries of the Hudson Bay drainage via the proposedGarrison River diversion, and from the Pacific North-west to California. Although no transfer of such mag-nitude has yet taken place, pressure may intensify aswater shortages become more severe.

RestorationWe should aspire to make the twenty-first century aperiod of restoration and repair of damaged ecosys-tems, including rivers, for many reasons. First, we areincreasingly aware of the extent of environmentaldegradation and have the capacity to make improve-ments. Second, society increasingly values biologicaldiversity, the variability of nature, and the psycho-logical and social benefits we derive from natural sur-roundings. Third, all of us benefit from a number ofecosystem goods and services, including clean water,recreation, and harvestable fishes. Healthy rivers areof direct value. Sometimes restoration will be costly,

such as in the greater Florida Everglades ecosystem,where large sums were spent to channelize theKissimmee River and even larger sums are currentlybeing spent to restore it. This is a call to learn frompast mistakes. In other instances restoration will be cost effective, as with the purchase of riparian land to preserve water quality in the Catskill andDelaware watersheds supplying water to New YorkCity (City of New York Department of Environmen-tal Protection 2002 http://www.nyc.gov/html/dep/html/fadplan.html), a far cheaper solution than thealternative, building treatment plants to removeexcess sediments. This is a call to recognize the effi-ciency of sound environmental policy.

Restoration has been defined as “returning asystem to a close approximation of its condition priorto disturbance, with both the structure and functionof the system recreated” (NRC 1992). However, itmay not be possible to know the predisturbance con-dition, and it may not be practical to achieve it.Various authors and scientific working groups havewrestled with the definition of restoration and thesetting of goals and targets (Bradshaw 1988, Palmeret al. 1997, SER 2004 http://www.ser.org/content/ecological_restoration_primer.asp). In our opinion,although we should make every effort to be aware ofthe historical state of the system, restoring systems topresettlement condition will very rarely be practical.In practice, restoration today seems to encompasseverything we might do to repair or improve anecosystem. Using reference sites and historical infor-mation when available, and incorporating an under-standing of the dynamic processes that govern riversand their biota are key ingredients of thoughtfulrestoration practice. Recognizing the importance ofstakeholders, it may be sensible to consider a rangeof options, with public input determining the desiredoutcome (Hobbs and Harris 2001).

A great deal of river restoration presently takesplace under the auspices of federal, state, and localgovernments, as well as by private citizens. Unfortu-nately, we do not have very good information con-cerning the extent and types of restoration practicesor their effectiveness, and little or no accountabilityfor the large sums being spent (Palmer et al. 2003).In fairness, river restoration is a relatively new area,and recent program initiatives such as CALFED (apartnership of California and federal governments,which is investing substantially in San FranciscoBay–area restoration) are requiring external scientificreview, explicit hypotheses, and other key elementsthat should help to ensure sound science and theapplication of adaptive management.

North America’s Rivers in the Twenty-First Century

1099

Ch24.qxd 3/24/05 2:21 PM Page 1099

The type of river restoration depends greatly onlocation, prioritization of threats, and how the riveris valued. In addition, there may be several ways to address a particular problem. For example, in agricultural regions and wherever bank erosion isserious, maintenance of a vegetated buffer zone alongriver margins, which may be accompanied by plant-ing trees, is common practice. Gabions (wire basketsof rocks) and rip-rap (individual large stones or concrete blocks, even old automobiles, referred to as“Detroit rip-rap”) are less attractive means to stabi-lize banks. In some cases, recognizing that river channels naturally migrate, it may be appropriate todo nothing at all. Clearly, identifying the problem,understanding the fundamental processes at work,and clarifying the desired outcome are all importantin arriving at the optimal solution.

Flow restoration is increasingly becoming anobjective of water managers, particularly thoseresponsible for dam releases and water withdrawals.How much water a river needs is being reevaluatedin light of the scientific consensus that rivers (andmost ecosystems) are naturally variable and dependupon that natural variability for full geomorphic,hydrologic, and ecological function (Richter et al.1997, Poff et al. 1997). River channels are shapedlargely by the interplay among slope, sedimentsupply, and water supply and maintained mainly bybankfull flood events; if human actions change oneof these variables, adjustments will occur in one ofthe others. For example, regulation of the ColoradoRiver below Glen Canyon Dam after 1963 restrictedthe supply of sediments, and in response the river cutdownward, lowering its slope. How much water isneeded for fish populations may be difficult to deter-mine, or perhaps society has not yet agreed to includea margin of error in its estimates. For example, in anextreme conflict between human use (for agriculture)and ecosystem use (for fishes), water-supply man-agers for the Klamath lake and river system experi-enced massive protests from farmers whose cropswere threatened, and then only months later oversawthe largest die-off of Pacific salmon yet recorded, ina case marked by conflicting reports of investigatingcommittees (Service 2003). This may indicate a needfor more and better science to guide decisions andmore documentation of successes and failures inorder to build the case-based knowledge necessaryfor management of conflicts with such potentiallyhigh costs on both sides (Poff et al. 2003).

Few examples of river restoration are reported inthe chapters of this book, and this likely can be taken

as rough evidence that few well-documented casescan be found. This may reflect the fact that riverrestoration is a new agenda. In addition, restorationof large rivers is expensive and often encounters real conflicts of interest and financial constraints. Inseveral instances, including the Columbia, Missouri,and Willamette, recovery and restoration plans datefrom 2000 or later and are not yet implemented.Recovery plans for anadromous fishes, includingsalmon and shad, typically are based on hatcheriesand fish passage around dams and occasionally ondam breaching, and are perhaps most frequentlycited. Examples include largely successful efforts torestore shad to the Connecticut River and largelyunsuccessful efforts to restore salmon runs on theColumbia River.

Flow restoration through modifications of damoperations is at least not uncommon as a restorationtactic, although costs of lost hydropower and othercommercial benefits pose a constant challenge to suchefforts. In some instances, including sections of theMissouri, the Flaming Gorge Dam on the GreenRiver of the Colorado drainage, and the YakimaRiver, which enters the only free-flowing section ofthe Columbia and is where most natural salmonreproduction occurs, biological opinion is clear, butaction has yet to be taken.

Recovery proposals recommended by theNational Research Council show that current sciencecan guide actions to benefit ecological services as wellas societal benefits through a combination of acquir-ing floodplain lands, increasing main-channel habitatcomplexity, and modifying reservoir water manage-ment to restore more natural river flows (Chapter10). However, it is unfortunately worth noting thatefforts to resolve environmental and human uses ofthe Missouri River have been in political and legalgridlock for years and continue to be in controversyas of this writing.

If there is a lesson to be drawn from these fewrestoration examples from large rivers, it may be thatthe effort is at an early stage, and the extent of soci-etal valuation of river ecosystems for their naturalvalues and their ecosystem services remains to bedetermined.

In conclusion, whether North America’s riverswill be in better condition at the end of the twenty-first century than at the beginning is uncertain. Butthey can be. Many rivers have improved due to areduction of careless resource extraction followed bysufficient time for some natural recuperation. Addi-tional improvement has resulted from reductions in

24 Overview and Prospects

1100

1

Ch24.qxd 3/24/05 2:21 PM Page 1100

point-source pollution and other targeted actions,including habitat improvement, dam removals, andactive river restoration. Moreover, rivers have signif-icant restorative capacity: The flow of water andmovement of sediments can cleanse pollutants, andperiodic floods can restore dynamism to river chan-nels and allow natural processes to dominate. Somerecovery has occurred passively and some due tosound management.

But there is discouraging news as well. At present,some good restoration plans are struggling to beimplemented due to conflicting pressures and finan-cial constraints. Water demand will increase with thegrowth of population and affluence, climate warmingand the spread of nonnative species may pose evengreater threats over the present century, and all of the familiar threats associated with human activitiesremain with us. We find ourselves between, on theone hand, improved knowledge, stakeholder support,and the capacity to manage and restore, and on theother hand, an array of familiar and unfamiliarthreats. Although improvements in the science andpractice of river restoration clearly have much to contribute, they matter little unless major strides aremade to grow awareness, political will, and the poli-cies and institutions necessary to advocate soundriver management.

In this light, the growing influence of citizengroups and nongovernmental organizations con-cerned with river health is of great importance. Byjoining the dialogue on the future of our rivers andcontributing scientific expertise where it is appropri-ate, whether through giving public lectures, servingon panels, sharing scientific knowledge with riveradvocates, or in any other way, scientists as individ-uals can help to see that knowledge is used. It will berare (and probably should be) for decisions affectingriver health to be made by a few specialists. In thelong run, involvement of all sectors of society likelyholds the greatest promise for the future condition of the river ecosystems so ably discussed in thisvolume.

LITERATURE CITED

Abell, R. A., D. M. Olson, E. Dinerstein, P. T. Hurley, J. T.Diggs, W. Eichbaum, S. Walters, W. Wettengel, T.Allnutt, C. J. Loucks, and P. Hedao. 2000. FreshwaterEcoregions of North America: A conservation assess-ment. Island Press, Washington, D.C.

Allan, J. D. 1995. Stream ecology: Structure and functionof running waters. Kluwer Academic, The Netherlands.

Allan, J. D., and A. S. Flecker. 1993. Biodiversity conservation in running waters. BioScience 43:32–43.

Allan, J. D., M. A. Palmer, and N. L. Poff. 2005. Fresh-water ecology. In T. E. Lovejoy and L. Hannah (eds.),Climate change and biodiversity. Yale University Press,New Haven.

Benke, A. C. 1993. Concepts and patterns of invertebrateproduction in running waters. Verhandlungen derInternationalen Vereinigung für Theoretische undAngewandte Limnologie 25:15–38.

Benke, A. C., R. L. Henry III, D. M. Gillespie, and R. J.Hunter. 1985. Importance of snag habitat for animalproduction in southeastern streams. Fisheries 10:8–13.

Bradshaw, A. D. 1988. Restoration: An acid test forecology. In W. R. Jordan III, M. E. Gilpin, and J. D.Aber (eds.), Restoration ecology: A synthetic approachto ecological research, pp. 23–29. Cambridge Univer-sity Press, Cambridge.

Cole, J. J., B. L. Peierls, and N. F. Caraco. 1993. Nitrogenloading of rivers as a human-driven process. In M. J.McDonnell and S. T. A. Pickett (eds.), Humans as com-ponents of ecosystems: The ecology of subtle humaneffects and populated areas, pp. 141–157. Springer-Verlag, Berlin.

Covich, A. P., S. C. Fritz, P. J. Lamb, R. D. Marzolf, W. J.Matthews, K. A. Poiani, E. E. Prepas, M. B. Richman,and T. C. Winter. 1997. Potential effects of climatechange on aquatic ecosystems of the Great Plains of North America. Hydrological Processes 11:993–1021.

Dynesius, M., and C. Nilsson. 1994. Fragmentation andflow regulation of river systems in the northern thirdof the world. Science 266:753–762.

Elzen, D. M., and M. Schaeffer. 2002. Responsibility forpast and future global warming: Uncertainties inattributing anthropogenic climate change. ClimaticChange 54 (1–2): 29–73.

Firth, P., and S. G. Fisher (eds.). 1992. Global climatechange and freshwater ecosystems. Springer-Verlag,New York.

Forest, C. E., P. H. Stone, A. P. Sokolov, M. R. Allen, andM. D. Webster. 2002. Quantifying uncertainties inclimate system properties with the use of recent climateobservations. Science 295:113–117.

Frederick, K. D., and P. H. Gleick. 1999. Water and globalclimate change: Potential impacts on U.S. waterresources. Pew Center on Global Climate Change,Arlington, Virginia.

Graf, W. L. 1999. Dam nation: A geographical census ofAmerican dams and their large-scale hydrologicimpacts. Water Resources Research 35:1305–1311.

Hart, D. D., and N. L. Poff. 2002. A special section ondam removal and river restoration. BioScience 52:653–655.

Literature Cited

1101

2

3

Ch24.qxd 3/24/05 2:21 PM Page 1101

Hobbs, R. J., and J. A. Harris. 2001. Restoration ecology:Repairing the earth’s ecosystems in the new millen-nium. Restoration Ecology 9:239–246.

Hocutt, C. H., and E. O. Wiley (eds.). 1986. Zoogeogra-phy of North American freshwater fishes. Wiley Inter-science, New York.

Hornberger, G. M., J. P. Raffensperger, P. L. Wiberg, andK. N. Eshleman. 1998. Elements of physical hydrology.Johns Hopkins University Press, Baltimore.

Lake, P. S. 2001. On the maturing of restoration: Linkingecological research and restoration. Ecological Man-agement & Restoration 2:110–115.

Leopold, L. B. 1994. A view of the river. Harvard Univer-sity Press, Cambridge.

Master, L. L., S. R. Flack, and B. A. Stein (eds.). 1998.Rivers of life: Critical watersheds for protecting freshwater biodiversity. The Nature Conservancy,Arlington, Virginia.

Matthews, W. J., and E. G. Zimmerman. 1990. Potentialeffects of global warming on native fishes of the southern Great Plains and the Southwest. Fisheries15:26–32.

Meyer, J. L., M. J. Sale, P. J. Mulholland, and N. L. Poff.1999. Impacts of climate change on aquatic ecosystemfunctioning and health. Journal of the American WaterResources Association 35:1373–1386.

Meyer, W. B., and B. L. Turner (eds.). 1994. Changes inland use and land cover: A global perspective. Cambridge University Press, New York.

National Assessment Synthesis Team (NAST). 2000.Climate change impacts on the United States: Thepotential consequences of climate variability andchange. U.S. Global Change Research Program, Washington, D.C.

National Research Council (NRC). 1992. Restoration ofaquatic systems: Science, technology, and public policy.National Academy Press, Washington, D.C.

National Research Council (NRC). 2001. Envisioning theagenda for water resources research in the twenty-first century. National Academy Press, Washington,D.C.

Palmer, M. A., R. F. Ambrose, and N. L. Poff. 1997. Eco-logical theory and community restoration ecology.Restoration Ecology 5:291–300.

Palmer, M. A., D. D. Hart, J. D. Allan, E. Bernhardt, andthe National Riverine Restoration Science SynthesisWorking Group. 2003. Bridging engineering, ecologi-cal and geomorphic science to enhance river restora-tion: Local and national efforts. World Water andEnvironmental Resources Congress 2003 and RelatedSymposia. Proceedings of the World Water and Envi-ronmental Resources Congress, Philadelphia, Pennsyl-vania, June 24–26, 2003.

Paul, M. J., and J. L. Meyer. 2001. Streams in the urbanlandscape. Annual Reviews of Ecology & Systematics32:333–366.

Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C.

Stromberg. 1997. The natural flow regime: A paradigmfor river conservation and restoration. BioScience47:769–784.

Poff, N. L., J. D. Allan, D. D. Hart, M. A. Palmer, B. D.Richter, J. L. Meyer, and J. A. Stanford. 2003. Riverflows and water wars: Emerging science for environ-mental decision-making. Frontiers in Ecology and theEnvironment 1:298–306.

Poff, N. L., P. A. Angermeier, S. D. Cooper, P. S. Lake, K. D. Fausch, K. O. Winemiller, L. A. K. Mertes, M.W. Oswood, J. Reynolds, and F. J. Rahel. 2001. Fish diversity in streams and rivers. In F. S. Chapin III, O. E. Sala, and E. Huber-Sannwald (eds.), Global biodiversity in a changing environment: Scenarios forthe 21st Century, pp. 315–350. Springer-Verlag, NewYork.

Poff, N. L., and D. D. Hart. 2002. How dams vary andwhy it matters for the emerging science of damremoval. BioScience 52:659–668.

Postel, S., and B. Richter. 2003. Rivers for life: Managingwater for people and nature. Island Press, Washington,D.C.

Regier, H. A., J. A. Holmes, and D. Pauly. 1990. Influenceof temperature change on aquatic ecosystems: An interpretation of empirical data. Transactions of theAmerican Fisheries Society 119:374–389.

Richter B. D., J. V. Baumgartner, J. Powell, and D. P. Braun.1996. A method for assessing hydrologic alterationwithin ecosystems. Conservation Biology 10:1163–1174.

Richter B. D., J. V. Baumgartner, R. Wigington, and D. P.Braun. 1997. How much water does a river need?Freshwater Biology 37:231–249.

Rier, S. T., and N. C. Tuchman. 2002. Elevated-CO2-induced changes in the chemistry of quaking aspen(Populus tremuloides Michaux) leaf litter: Subsequentmass loss and microbial response in a stream ecosys-tem. Journal of the North American BenthologicalSociety 21:16–27.

Schindler, D. W. 1998. A dim future for boreal waters andlandscapes. BioScience 48:157–164.

Service, R. F. 2003. NRC backs ecosystem-wide changes tosave Klamath fish. Science 302:765.

Shiklomanov, I. A. 1993. World water resources. In P. H.Gleick (ed.), Water in crisis: A guide to the world’s freshwater resources, pp. 13–24. Oxford University Press,New York.

Stanley, E. H., and M. W. Doyle. 2003. Trading off: Theecological effects of dam removal. Frontiers in Ecologyand the Environment 1:15–22.

Strayer, D. L. 1991. Projected distribution of the zebramussel, Dreissena polymorpha, in North America.Canadian Journal of Fisheries and Aquatic Sciences48:1389–1395.

Sweeney, B. W., J. K. Jackson, J. D. Newbold, and D. H.Funk. 1992. Climate change and the life histories and biogeography of aquatic insects in eastern NorthAmerica. In P. Firth and S. G. Fisher (eds.), Global

24 Overview and Prospects

1102

4

5

6

7

Ch24.qxd 3/24/05 2:21 PM Page 1102

climate change and freshwater ecosystems, pp. 143–176. Springer-Verlag, New York.

Thorpe, J. H., and M. D. Delong. 2002. Dominance ofauthochthonous autotrophic carbon in heterotrophicrivers. Oikos 96:543–550.

Tuchman, N. C., R. G. Wetzel, S. T. Rier, K. A. Wahtera,and J. A. Teeri. 2002. Elevated atmospheric CO2 lowersleaf litter nutritional quality for stream ecosystem foodwebs. Freshwater Biology 8:163–170.

Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R.Sedell, and C. E. Cushing. 1980. The River Continuum

Concept. Canadian Journal of Fisheries and AquaticSciences 37:130–137.

Valdez, R. A., and R. D. Williams. 1986. Cataract Canyonfish study. Final Report for U.S. Bureau of Reclama-tion Contract 6-CS-40-03980. U.S. Bureau of Recla-mation, Salt Lake City, Utah.

Ward, J. V., and J. A. Stanford (eds.). 1979. The ecologyof regulated rivers. Plenum, New York.

Webster, J. R., and J. L. Meyer. 1997. Organic matterbudgets for streams: A synthesis. Journal of the NorthAmerican Benthological Society 16:141–161.

Literature Cited

1103

8

Ch24.qxd 3/24/05 2:21 PM Page 1103

Ch24.qxd 3/24/05 2:21 PM Page 1104

AUTHOR QUERY FORM

Dear Author,

During the preparation of your manuscript for publication, the questions listed below have arisen. Pleaseattend to these matters and return this form with your proof.

Many thanks for your assistance.

Query Query RemarksReferences

1. AU: source callout here?2. AU: update? page range?3. AU: pls. call out in text or delete from this list.4. AU: pls. call out in text or delete from this list.5. AU: is this a presentation at the congress? a paper? a book of the congress proceedings? If a book, pls.

provide city and publisher.6. AU: pls. call out in text or delete from this list.7. AU: pls. call out in text or delete from this list.8. AU: pls. call out in text or delete from this list.

BRN24

AUTHOR QUERY FORM (BRN24) 3/24/05 2:20 PM Page 1