hydrochemistry of forested catchmentshydrochemistry of forested catchments 25 figure 1...

P1: MBL/plb P2: MBL/rsk QC: MBL/abe T1: MBL

March 26, 1997 10:24 Annual Reviews 29-02 AR29-02

Annu. Rev. Earth Planet. Sci. 1997. 25:23–59

HYDROCHEMISTRY OF FORESTEDCATCHMENTS1

M. Robbins ChurchUS Environmental Protection Agency, National Health and Environmental EffectsResearch Laboratory, Corvallis, Oregon 97333

KEY WORDS: catchment hydrochemistry, watersheds, streamwater

“The song of a river ordinarily means the tune that waters play on rock, root, and rapid....This song of the waters is audible to every ear, but there is other music in these hills, byno means audible to all. To hear even a few notes of it you must first live here for a longtime, and you must know the speech of hills and rivers. Then on a still night, when thecampfire is low and the Pleiades have climbed over rimrocks, sit quietly and listen for awolf to howl, and think hard of everything you have seen and tried to understand. Then youmay hear it—a vast pulsing harmony—its score inscribed on a thousand hills, its notes thelives and deaths of plants and animals, its rhythms spanning the seconds and the centuries.”Aldo Leopold,Song of the Gavilan[Reprinted in part from Leopold (1940), by permissionof The Wildlife Society.]

ABSTRACT

The pathways that water may take through a catchment and its reactions withorganisms and soils are myriad and ever varying. A promising means to unravel-ing the mystery of watershed hydrochemistry is the study of “small” catchments,yet the hydrochemical function of even the smallest of catchments involves anamazingly intricate web of flowpaths and biogeochemical processes. Monitoringof catchments and comparison of their inputs to their outputs yields clues to theirworkings. Manipulation of catchments offers some means of “controlled” exper-imentation as to their nature. Modeling of catchment hydrochemical responseattempts to put it all together. Every forested catchment is individual in its struc-ture and hydrochemical response, yet a select, carefully studied few often aredrafted to serve as representatives in assessments of responses to environmentalinfluences or perturbations (e.g. acid rain). Many factors must be considered insuch extrapolations. Studies of forested catchments often are driven by environ-mental concerns and thus fluctuate accordingly as to their location and intensity.Despite such fluctuations, the future of the field is clear and bright. Much has been

1The US Government has the right to retain a nonexclusive, royalty-free license in and to anycopyright covering this paper.

23



24 CHURCH

learned as a result of recent studies, not only of what to think about catchmentfunction but, more importantly, how to think about it.

INTRODUCTION

Many fascinating connections exist among the physical,øchemical, and biolog-ical processes (Figure 1) that shape the chemical composition of water flowingfrom forested catchments.2 Where water goes in a forested catchment, howlong it lingers, what materials and organisms it encounters on its journey—allinteract to determine its character. All combine to present a fascinating puzzleto those who study catchment hydrochemistry. As John Muir, the Americannaturalist, wrote, “When we try to pick out anything by itself, we find it hitchedto everything else in the universe.” This certainly seems true in the study ofwatershed hydrochemistry, regardless of whether we are trying to understandnatural systems for our own basic curiosity or in order to better understandand to predict the response of ecosystems to large-scale stresses such as acidicdeposition, non–point source pollution, or changes in climate. As we attemptto understand how substance “A” comes to be at point “B” at concentration“C,” we find ourselves at every turn meeting questions that require us to crosslong-established disciplines—from hydrology to aqueous geochemistry to mi-crobiology and more. We find ourselves everywhere needing to learn moreabout “what the other guy knows.” These efforts challenge us. They make usthink and re-think! They keep us questioning.

HYDROLOGIC PROCESSES AND FLOW PATHS INFORESTED CATCHMENTS

Four primary factors combine to determine the chemical and isotopic charac-teristics of waters draining from forested watersheds. These factors are (a) thechemical composition of incident precipitation, (b) the abiotic materials or thebiota that water contacts as it moves through a watershed, (c) the abiotic reac-tivity of the materials or biotic activity of the organisms contacted, and (d) thelength of contact with those materials or organisms. A key factor in this equa-tion is the pathway that water follows through the catchment. The pathways thatwater follows determine the materials it contacts and the length of that contact,and thus they play a pivotal role in determining drainage water composition.In its travels through a forested catchment water may encounter forest canopyand understory plants, plant stems, litter and organic soil layers, mineral layers

2A catchment (or watershed) is a unit of natural landscape (i.e. including bedrock, soils, biota)delineated by a topographic divide that converges at a downstream terminus, defined by the pointat which flow from the catchment is measured.



HYDROCHEMISTRY OF FORESTED CATCHMENTS 25

Figure 1 Conceptualization of interconnected watershed physical, chemical, and biological pro-cesses. Reprinted from Johnson & Van Hook (1989), by permission of Springer-Verlag.

of soil, coarse tills, and bedrock as well as a host of macro- and micro-biota.All act to alter its composition.

Study of drainage water chemical and isotopic composition helps to unravelthe mysteries of where water has been in a watershed and how long it has beenthere. We can use knowledge of past pathways and residence times under knownantecedent and precipitation (or snowmelt) conditions, to better understand howwatersheds “work” and perhaps to predict their future hydrochemical behavioror condition. The analysis of rainfall/runoff events (stormflow generation) isparticularly useful in the study of the hydrologic behavior of forested catch-ments. It is under these dynamic transitory conditions that watersheds revealtheir workings. Studies of stormflow generation, particularly over the last 30years, have added much to our knowledge and appreciation of the variabilityand complexity of catchment hydrochemistry.



26 CHURCH

A variety of storage zones, transport processes, or flow paths may exist andoperate for watersheds. In general, however, one can characterize stormflowfor a catchment as originating by way of any (or any combination) of fourbasic paths: (a) direct interception of precipitation by the stream channel,(b) surface flow, (c) subsurface flow, or (d) groundwater flow. The relativeimportance of each of these pathways varies both among and within individualwatersheds depending on topography, forest type, soil makeup, and geologicalstructure. Within any particular watershed, pathways may also vary dependingon antecedent conditions of wetness and intensity of rainfall (or snowmelt).

Direct interception of precipitation by the stream channel is self-explanatoryas both a mechanism and a pathway. Although direct channel precipitation maycontribute to small flow peaks at times, its quantitative contribution to overallstormflow is usually slight in headwater forested catchments.

The topic of surface flow in watersheds (commonly termed overland flow) hasa long history in catchment hydrology. Horton (1933) proposed that overlandflow resulted when rainfall rates exceeded soil infiltration rates leading, in turn,to substantial surface runoff causing rapid increases in stream flow. This typeof flow, commonly termed Hortonian overland flow (sometimes sheet flowor infiltration-excess flow), can be very important in arid areas where soilsmay swell and greatly reduce infiltration rates or in areas where soils may bedisturbed or compacted (e.g. croplands or heavily grazed areas). It may alsobe important in watersheds with extensive rock outcrops. Usually it is of minorimportance in forested watersheds.

Other sources of overland flow may exist, however. Musgrave & Holtan(1964) used the term “return flow” to describe the situation in which subsurfaceflow returns to the surface as the water table (or perched water table due toshallow impervious layers) rises to the surface in response to precipitationor snowmelt. As precipitation falls on these saturated areas, it too will flowoverland. Together these components comprise “saturation overland flow”(e.g. see Dunne & Leopold 1978). Such flow may occur adjacent to the streamchannel or further upslope at points of flow convergences within the catchment.Hewlett & Hibbert (1967) proposed that during rainfall or snowmelt events,such areas of flow contribution immediately adjacent to the stream expandlaterally from the stream as well as longitudinally upstream leading to greaterexpanses of directly contributing zones. As rainfall or snowmelt later dissipate,these zones shrink. Hewlett & Hibbert (1967) termed this conceptualization offlow production the “variable source area concept.” Hibbert & Troendle (1988)have provided an authoritative review of the concept and its development fromearlier work (most notably at the Coweeta Hydrologic Laboratory).

Hewlett & Hibbert (1967) were careful to point out that although increases inoverland flow would occur as contributing zones expanded and channel length




extended during storms, “most or all of the water may still be entering thesechannels as subsurface flow.” Building on their own observations and theobservations of others, they hypothesized that in small (for them,<50 km2)temperate forest watersheds of the eastern United States, most of stormflowrunoff comes as subsurface flow. In so doing they directly challenged the rulingidea of the day that overland flow was the dominant component of stormflowgeneration. In their classic publication, Hewlett & Hibbert (1967) posed the keyquestion that in many ways still challenges forest hydrologists: “If we assumethat subsurface flows predominate on most wildland soils, how do direct runoffand base flow get to the channel?”

Hewlett & Hibbert (1967) perceived direct runoff (stormflow) from a rain-storm as composed partly of some of the actual rainfall (often termed “new”water), but mostly of pre-event water (“old” water) displaced by the rain fromthe soil column. They termed the piston-type flow produced by this displace-ment as “translatory flow” and noted that it should be most important in lower-to mid-watershed slopes. Such a “matrix flow,” however, seems incapable ofdelivering flows at rates sufficient to match observed increases in stormflow[e.g. see review discussions by Turner et al (1990), Bonell (1993), and Jenkinset al (1994)].

Sklash & Farvolden (1979) proposed an alternative mechanism of near-stream matrix flow to account for rapid stormflow generation. They combinedhydrometric and isotope studies in forested and partly forested catchments inQuebec to show that groundwater (saturated matrix flow of “old” water) con-tributed between 60% and 80% of stormflow generation in their studies. Theycombined their field observations with computer simulations to build upon priorwork of Ragan (1968) and to propose that “groundwater ridging” was occurringand was sufficient to account for delivery of most of the storm flow. In this pro-cess, newly infiltrating water quickly converts the capillary fringe (just abovethe saturated zone) to a state of saturation (groundwater ridge). This increasesboth the size of the near-stream discharge area and the near-stream head, thusleading to relatively rapid and substantial increases in flows of pre-event waterto the stream channel. This phenomenon would of course be most important inlower valley near-stream zones [discussions by Bonell (1993) and Jenkins et al(1994)]. Although Turner et al (1990) termed this proposed process “the mostwidely accepted hypothesis to explain observations of rapid subsurface responsein forested catchments,” Bonell (1993) noted that some subsequent studies offlow generation in forested catchments failed to demonstrate the mechanismand cautioned that “it is clear that further experimentation is necessary to verifythe generality of this process.”

Water may also move in forested catchments through large pores in the soil(Beven & Germann 1982, Germann 1986, 1990). Hydrologists call such pores



28 CHURCH

macropores [though pores larger than 1 cm (e.g. see Bonell 1993) may be termedpipes]. Decaying of root channels, burrowing of animals (large and small), andcracking of clay soils are all processes that create macropores and pipes. Watermay move through macropores vertically (e.g. through unsaturated soils to thesaturated zone) or laterally (principally through saturated zones). In so doing,macropore or pipe flow can rapidly bypass some areas of relatively slowermatrix flow only to later arrive at and substantially interact with other, perhapsnewly, saturated matrices (McDonnell 1990). In some cases macropores maydischarge water directly back to the surface in some convergence zones of thewatershed or directly into stream channels (e.g. see McDonnell 1990). In mostforested catchments in humid temperate climates, however, the most likely rolefor macropore flow is to transmit water primarily within the soil and regolith.Most often researchers do not (or cannot) know the extent of macropore inter-connections within any particular watershed (Bonell 1993).

Regardless of the particular path or mix of paths that water takes as it flowsthrough catchments to become surface water runoff, one thing now seemsclear—in most forested catchments in most situations stormflow appears tobe composed predominantly of old (pre-event) water. As reviewed by Turneret al (1990) and Bonell (1993), evidence for this conclusion has mounted con-sistently with the ever increasing number of isotopic (see reviews by Sklash &Farvolden 1982, Buttle 1994) and chemical tracer studies of stormflow sourcesin forested catchments. Despite the apparent general applicability of this con-clusion, however, one must be careful not to apply it without evidence to anyparticular catchment at any particular time. The sources of water in stormflowand the types of pathways that water takes through catchments most certainlycan and will vary between (even adjacent) catchments and within catchmentsdepending on local topography, antecedent conditions of wetness, and rainfallor snowmelt intensity. Mulholland et al (1990), for example, have presentedan interesting case study of within-watershed variation of flow generation pro-cesses for a forested catchment.

The pathways that water takes through a forested catchment play a majorrole in determining that water’s physical, chemical, and isotopic composition(e.g. Mulder et al 1991). Conversely, these streamwater characteristics canhelp to tell us where the water has been. In recent years, dramatic increaseshave occurred in the use of various types of tracers to study flow paths andprocesses in forested catchments. Tracers have included temperature (Shanley& Peters 1988), specific conductance (Pilgrim et al 1979), and a host of chemicalconstituents (e.g. SiO2, Ca2+, Mg2+, Cl−, Br−, SO2−

4 ; see Jenkins et al 1994).Use of chemical tracers both benefits from, and is complicated by, the fact thatchemical modifications possibly, but do not necessarily (Hooper & Kendall1994), occur anywhere along the flow path.




Environmental isotopes [those of water itself (e.g.18O) or of solutes (e.g.15N)]may provide useful information on flow paths and processes in forested catch-ments. In recent years there has been a tremendous growth in the use of envi-ronmental isotopes for these purposes (Sklash 1990, Buttle 1994, Kendall et al1995). The isotopic composition of pre-event water held within the watershedcommonly differs from that of rainfall (Kendall et al 1995), which itself canvary seasonally or from storm to storm (Pearce et al 1986, Kendall et al 1995).Therefore, any mixing of new water and old water is likely to change the isotopiccomposition of the resultant stormflow. By solving the mass balance equations(e.g. Buttle 1994) for the fluxes of water and the isotopes, one can compute therelative contributions of the new and old water sources to flow from forestedcatchments. One must always be careful, however, to ensure that importantassumptions of the technique are met (see Buttle 1994, Kendall et al 1995).

To date, the most commonly used environmental isotopes have been the sta-ble isotopes of water—18O and deuterium (2H) (Buttle 1994, Kendall et al 1995,Gat 1996). The use of these isotopes has two significant advantages. First, theyare natural constituents of the water molecule and thus go exactly where thewater goes. Second, they are readily available and “applied” (i.e. in rain andsnowfall). Their use can entail some potential complications, however. Forexample, canopy interception of rainwater can be accompanied by significant(e.g. 20%) evaporation, resulting in fractionation (Saxena 1986, DeWalle &Swistock 1994). Other physical interactions may also lead to fractionation(Kendall et al 1995). Fractionation alters the isotopic composition of the inputwater and complicates the analysis.

Another component of the water molecule that researchers sometimes haveused as a tracer to study flow paths and flow generation processes is tritium (3H).Horton & Hawkins (1965), for example, applied tritium to the top of a saturatedsoil laboratory column and observed that nearly all of the early water leavingthe water column was untritiated pre-event water “pushed ahead” of the tritiumcontaining pulse. Their observations appear to have significantly influencedHewlett & Hibbert (1967) in their proposal of the importance of translatory(piston) flow in catchments. Although the use of tritium in whole-catchmentstudies can be informative (especially in combination with other isotopic orgeochemical tracers; Rose 1996), its use is relatively infrequent now. Its prin-cipal utility seems to have been to date groundwater in relation to precipitationenriched by atmospheric bomb testing of the midcentury (Jenkins et al 1994).

The innovative use of other tracers in catchment hydrochemical researchis increasing at a rapid rate. Recently used tracers have included222Rn (e.g.Genereux & Hemond 1990, Genereux et al 1993),13C (e.g. Kendall et al 1994),as well as isotopes of Sr, Pb, and others of the uranium and thorium series. Forexample, at Sleepers River Research Watershed in northern Vermont a number



30 CHURCH

of hydrogeologic conditions exist that allow innovative site-specific analyses tobe performed (Bullen et al 1994, Kendall et al 1994). At that site, (a) shallowgroundwater and mineral soil water have isotopic signatures for Pb isotopes (re-sulting from decays through the uranium series) that differ from the signatureproduced from atmospherically deposited Pb; (b) shallow groundwater is en-riched in radiogenic87Sr compared with unsaturated soil water; and (c) carbon-ates in the upper mineral horizons (shallow flowpaths) have been largely strippedby weathering relative to bedrock (deeper flowpaths). Bullen et al (1994) andKendall et al (1994) have used these conditions to further identify flowpathsduring snowmelt and stormflow events. Other sites, no doubt, have specific con-ditions that would allow researchers to perform other such innovative analyses.

A relatively recent development in methods for studying catchment hydro-chemical response has been the adoption of three-component flow separations.DeWalle et al (1988) used a three-component tracer model with measurementsof 18O to discriminate among the relative contributions of channel precipita-tion, soil water, and groundwater to stormflow generation in a small watershed(Fish Run) of the Appalachian Plateau. Prior to their work, nearly all chem-ical or isotopic separations had been two-component models of new water vsold water, necessarily lumping soil water and groundwater together as the oldwater component. Now, three-component separations are becoming increas-ingly common in hydrochemical studies of catchments (e.g. Christophersenet al 1990, Hooper et al 1990, Genereux et al 1993, O’Brien & Hendershot1993, Ogunkoya & Jenkins 1993, Bazemore et al 1994, Hinton et al 1994). Aswith two-component separations, care must be taken, of course, when applyingthree-component separations to make sure that all necessary assumptions aremet (see Buttle 1994 and Hinton et al 1994 for discussions).

Researchers have used isotopic or chemical tracers not only in isolationbut also in combination and in association with catchment hydrometric data(Kennedy et al 1986, Obradovic & Sklash 1986, Pearce et al 1986, Sklash et al1986, Genereux et al 1993, Bazemore et al 1994, Buttle 1994, Hinton et al1994). Such integrated use can be especially powerful in studying stormflowsources, pathways, or processes in forested catchments. Such techniques, whenthoughtfully applied, hold promise to increase our understanding of the physicaland chemical ways in which forested watersheds and the surface waters drainingthem are interconnected.

CHEMICAL AND BIOLOGICAL INTERACTIONSIN FORESTED CATCHMENTS

Equally important to the pathways of water movement in determining catch-ment hydrochemistry are the physical, chemical, and biological processes that




act within catchments to determine the composition of drainage waters. Amongthe most important of such processes are (a) the physical processes of erosionand gas exchange; (b) the chemical processes of weathering, chemical precip-itation, cation exchange, and ion sorption; and (c) the biological processes ofuptake, respiration, decomposition, mineralization, and microbiologically me-diated oxidation and reduction. Even the most cursory inclusive treatment ofthese processes is outside of the scope of this review. A number of sourcesprovide excellent descriptions and discussions of these processes within thecontext of catchment hydrochemistry (e.g. Trudgill 1986, Binkley et al 1988,Turner et al 1990, Moldan &Cerny 1994).

More important than a recounting here of the specifics of each process is therealization that it is the sum total of interactions (summed in different ways ondifferent catchments and summed in different ways at different times withinan individual catchment) that determine the chemical composition of drainagewaters. An illustrative discussion of some of the factors that influence catchmenthydrochemical response to the input of a particular pollutant, in this case acidicdeposition, provides a useful example of such interactions.

Combustion of fossil fuels by electric utilities, factories, and motor vehiclesyields emissions of oxides of sulfur and nitrogen that react in the atmosphereto produce acidic deposition (often called acid rain; see Bricker & Rice 1993).The major cationic components of acid rain are hydrogen and ammonium ions.Its major anions are sulfate and nitrate. When this chemical mixture enters aforested catchment it becomes subject to both a host of sequential processes andsimultaneous reactions acting (and interacting) in ways that are not yet fullyunderstood.

For example (and in no particular sequence in space or time), hydrogen ionsin acid rain can increase the dissolution (chemical weathering) of primary orsecondary minerals in the catchment. One effect of such weathering is to in-crease the concentration of “base” cations (Ca2+, Mg2+, Na+, K+) as well asof aluminum in soil solution. Aluminum concentrations also may be increasedin soil solution by the dissolution of amorphous or crystalline aluminum trihy-droxide phases. These dissolved species as well as the H+ (and other cations)in the acid rain engage in cation-exchange reactions with cations held on thesoil exchange complex. In cation exchange the ions of higher valence aredisplaced preferentially from the soil. Thus trivalent aluminum will be dis-placed in greater proportion than divalent or monovalent base cations or H+

(Brady 1974). Often, forest soils that receive substantial acidic deposition al-ready are “acidic”—they have low percentage base saturation (percentage ofsoil exchange capacity occupied by base cations) and their exchange complex isdominated by aluminum and H+. The increased ionic strength of soil solutionsthat results from acidic deposition leads to enhanced levels of cation exchange



32 CHURCH

(Reuss et al 1987, Binkley et al 1988, Turner et al 1990). In this situationsubstantial aluminum previously held on soil exchange sites can be mobilizedto solution. Dissolved aluminum is a weak acid in solution and can decreasethe pH of the soil solution (Binkley et al 1988). Taken together, this seriesof exchange and equilibrium reactions is referred to sometimes as “salt effectacidification” (Reuss & Johnson 1985, 1986, Binkley et al 1988). If soil basesaturation (percentage of soil exchange capacity occupied by base cations) isgreater than roughly 15–20%, the effect of this acidification on soil solution acidneutralizing capacity (ANC) will be minor (Reuss & Walthall 1989). At per-centage base saturation lower than roughly 15% (e.g. acid forest soils) the effectis more pronounced and can be important for resulting water quality. When suchaffected soil solutions leach to streams and lakes the result can be surface wateracidification—the lowering (sometimes dramatic) of ANC of those waters.

The effect of leaching of low pH soil solutions to surface waters can varydepending upon the ANC of the soil solutions and the effects of gas exchange.Because of the effects of root and bacterial respiration in soils, soil atmospheresinvariably have concentrations of CO2 well above ambient atmospheric con-centrations (Brady 1974). When soil waters that have equilibrated with soilatmosphere drain to surface waters, CO2 degasses from the solution. If theleaching soil solution has positive ANC, then the degassing will increase itspH (but will not affect ANC, as no net proton acceptors or donors have beenadded to the solution). If ANC has become negative, however, then pH in-creases associated with CO2 degassing will be negligible (Reuss & Johnson1985, 1986).

Not only can acidic deposition decrease the pH and ANC of soil solutions andsurface waters, it can also cause acidification of soils—the decreasing of soilpercentage base saturation. This occurs when base cations exchanged from thesoil complex leach to surface waters faster than chemical weathering can serveto replace them on the complex. This effect has an important interaction in theoverall process of acidification. As soil base saturation decreases, H+ and alu-minum occupy a progressively higher fraction of the exchange complex and inthe soil solution. This, in turn, will lead to more acidic soil solutions and surfacewaters. In addition, the aluminum leached is toxic to aquatic biota (Baker et al1990, Baker et al 1996, Van Sickle et al 1996). (It was the ultimate toxic effectsof acidic deposition on fish species that has led, via increased research funding,to many recent advances in our understanding of catchment hydrochemistry.)

The greater the loading (concentrationx volume) of acidic deposition tocatchment soils, the greater the loading of sulfur (mostly as sulfate) and nitrogen(mostly in the forms of ammonium and nitrate). Sulfur and nitrogen engagein a variety of reactions that significantly affect catchment hydrogeochemicalresponse to acidic deposition (Galloway 1996).




For sulfur, the inorganic anion adsorption of sulfate onto hydrous oxides ofiron and aluminum (Chao et al 1964) is a major control on its mobility in catch-ment soils. The process not only acts to retain sulfur deposited to catchments(sometimes very large percentages of deposition; Rochelle et al 1987, Rochelle& Church 1987), but it also acts as a net neutralization mechanism either throughanion exchange for hydroxide ions or through creation of a cation exchange site(Rajan 1978). Sulfate adsorption is concentration dependent. Under conditionsof continuing sulfur loadings, progressively smaller fractions of the inputs areadsorbed and retained, and sulfate concentrations in soil waters and receivingsurface waters continue to increase (Reuss & Walthall 1989).

The biological processes of vegetative and microbial uptake, decomposi-tion, mineralization, and oxidation and reduction reactions play major rolesin determining nitrogen cycling in forested catchments (Aber et al 1989). Inrecent years researchers have focused much more attention on the idea thatforests might reach a condition where nitrogen was available in excess of thetotal ecosystem biotic (vegetative and microbial) demand. Aber et al (1989)have termed such a situation nitrogen saturation. [In contrast,Agren & Bosatta(1988) defined nitrogen saturation as the rarer condition whereby nitrogen out-puts from a catchment equaled or exceeded inputs over periods (e.g. a year)sufficiently long to nullify “temporary disturbances.” This latter definition isfavored by some (e.g. Christ et al 1995, Van Sickle & Church 1995) becauseit is more quantitative.] Factors such as levels of atmospheric deposition ofnitrogen, stand age, catchment history (e.g. timbering), and relative supplies ofwater and other nutrients for vegetative and microbial growth seem to play themost significant roles in determining the relative cycling within and leachingof nitrogen from forests (Aber et al 1991). As levels of available ammoniumrise in forest soils, so do levels of microbial nitrification (the biologically me-diated conversion of ammonium to nitrate). Increased nitrification leads toincreased levels of H+ (which will then affect soil exchange reactions), as wellas increased soil solution nitrate concentrations.

In the soils of forested catchments strongly affected by acid rain, the resultof these various different yet highly connected processes can be (or can cometo be) soil solutions dominated by the anions SO2−

4 and (to a much less frequentdegree) NO−3 . Such soil solutions also would have elevated levels of the cationsH+ and Al3+, leading to adverse effects on receiving waters and their biota.

As water moves downward through soil, the dominance of specific reactionsand reaction products vary and shift as a function of the differing characteristicsof the existing soil horizons (Brady 1974, Richter & Markewitz 1995). Thus,the paths that the deposition and resulting soil solutions take through a forestedcatchment can greatly affect the chemistry of drainage waters. Solutions mov-ing through poorly buffered near surface soils may lead to drainage waters with



34 CHURCH

poor water quality, both over long-term periods (Newton & April 1982) andover very brief periods (“episodes”) associated with snowmelt or heavy rainfall(Eshleman 1988, Wigington et al 1990, 1996a). If drainage paths are deeperwithin the watershed, however, neutralization by much more highly bufferedgroundwaters (e.g. resulting from slow weathering over much longer residencetimes) can eliminate much of the adverse effects (Newton & April 1982).

This example presents only a few of the ways in which individual processescan interact with substances entering a forested catchment to influence hydro-chemical response. Many more are known (but cannot be addressed here);numerous interactions likely are unknown. Forested catchments function incomplex ways that can be difficult to observe and extremely difficult to quantify.

APPROACHES TO STUDYING CATCHMENTHYDROCHEMISTRY

The “Small Catchment Approach”A very powerful approach to understanding the hydrochemistry of forestedcatchments is through the study of “small catchments” (Moldan &Cerny 1994),sometimes known as the small watershed approach (Likens et al 1977, 1995).A catchment or watershed is generally considered “small” if it is on the orderof 500 ha or less (e.g. see Moldan &Cerny 1994). Small catchments are often(and sometimes incorrectly) assumed to be hydrologically “closed”—that is,underlain by impermeable bedrock such that all water (excluding water vapor)leaving the catchment flows through the catchment outlet. In the absence ofdramatic wind effects or of unbalanced large scale animal movements, closedcatchments allow for the measurement of losses from the system in the formof chemical outputs in streamwater (Likens et al 1977, 1995). The determi-nation of such a major component of watershed losses is a powerful tool inunderstanding the hydrochemical function of forested ecosystems.

Quite a number of well-integrated small catchment studies now exist, or havequite recently existed, around the world (e.g. see Andersson & Olsson 1985,Jeffries et al 1988, Hauhs 1989, Johnson & Van Hook 1989, Hornung et al1990, Wheater et al 1990, Kirkby et al 1991, Baron 1992, Adams et al 1994,Seip et al 1995). Two of the earliest small catchment studies were the CoweetaHydrological Laboratory (Swank & Crossley 1988b) and the Hubbard BrookEcosystem Study (Likens et al 1977, 1995). These research sites continue tooperate and to produce many important research findings.

COWEETA HYDROLOGICAL LABORATORY The Coweeta Hydrologic Labora-tory encompasses 2,185 hectares (ha) in the Nantahala Mountain Range ofwestern North Carolina in the Blue Ridge Physiographic Province (Swank &




Crossley 1988b). The US Department of Agriculture–Forest Service (USFS)established the site in 1934 (Douglass & Hoover 1988) for the purpose of eval-uating effects of forest practices on soil and water characteristics, with themajor original emphasis being on stream flow and water balance. Workers atCoweeta began measuring precipitation there in 1934 and in the period 1936–1939 established six climatic stations. Stream gaging began on nine streams in1934, and the number of streams being actively gaged has varied over the years(e.g. 28 in 1941, 16 by 1988) (Swift et al 1988). In 1968 Coweeta scientistsbegan measuring stream chemistry, and by 1972 they had established muchof the current system for long-term water chemistry measurements (Swank &Waide 1988). Monitoring of stream chemistry at Coweeta has been continuoussince that time, varying on component tributaries depending on the nature ofactive projects. With the advent of paired water balance and chemical moni-toring at Coweeta the researchers there opened the door to the first integratedhydrochemical research on southern Appalachian forest ecosystems. The hy-drochemical monitoring at Coweeta, especially as performed on both referenceand manipulated catchments and in concert with more detailed process-levelresearch, has proven to be a powerful tool in observing and unraveling catch-ment behaviors (with particular emphasis on responses to disturbance) at theecosystem level (Swank & Crossley 1988a,b).

HUBBARD BROOK EXPERIMENTAL FOREST In 1955 the USFS set aside a 3,037-ha watershed in the White Mountain National Forest in New Hampshire “asa major center for hydrologic research” (USDA Forest Service 1991). Thissite is the now well-known Hubbard Brook Experimental Forest. In the firsteight years of work there, researchers developed a network of stream gagesand precipitation and weather stations as well as sites for monitoring vege-tation and soils on small experimental watersheds. The major emphasis inthese first studies was on evaluating the effects of forest land management “onwater yield and quality and flood flow” (USDA Forest Service 1991). Fund-ing through a cooperative agreement between the USFS and Dartmouth Col-lege and from the National Science Foundation allowed the Hubbard BrookEcosystem Study to begin in earnest in 1963 (USDA Forest Service 1991).A small staff at the site began measurements of precipitation chemistry in1963–1964 and stream chemistry in 1964–1965 (Likens et al 1977, 1995).Thus was begun the first small watershed hydrochemical study of elemen-tal cycling. Since this work began in the early 1960s, researchers at theHubbard Brook have gathered a wealth of monitoring and site-intensive data(Veen et al 1994) and have published many research papers (Likens 1994) onthe structure and function of this northeastern forested system. Integrationof plot and site-specific intensive data with stream hydrochemical records at



36 CHURCH

Hubbard Brook continues to yield answers, offer clues, and prompt questionsregarding ecosystem function (e.g. Likens et al 1996) that otherwise mightelude us.

Catchment MonitoringOne of the most productive means of studying hydrochemistry of forestedcatchments is through a carefully organized and executed program of extendedcatchment monitoring. Especially useful is the joint monitoring of catchmentinputs and outputs (Cerny et al 1994). Monitoring is an important and usefultool (Likens 1988) for several reasons. Certainly it provides the best means“to formulate meaningful, testable hypotheses” (Likens 1983) about catchmentbehavior. Additionally, through the use of withheld data or split data sets, itprovides ways to test hypotheses (e.g. using mathematical models; Cosby et al1996). The obvious utility of the approach has led to the creation of numer-ous site-specific (e.g. seeCerny et al 1995) and integrated (e.g. Johnson &Lindberg 1992, Newell 1993) programs of catchment monitoring and associatedresearch.

A number of recent interesting examples of long-term hydrochemical mon-itoring leading to the formulation of hypotheses on catchment hydrochemicalfunction come from, or involve, work at Hubbard Brook. For example, as a resultof controls on pollutant emissions, the deposition of base cations to forestedwatersheds in certain parts of the world has decreased (Hedin et al 1994).Hubbard Brook lies in one such area (Likens & Bormann 1995). Driscoll et al(1989) examined monitoring data of precipitation inputs and streamwater out-puts of chemical species at Hubbard Brook and noted that during the period1964–1987, both inputs and outputs of base cations declined significantly. Al-though they noted that several mechanisms might be involved in causing thedeclines of base cations in outputs, they favored the hypothesis that decreases ininputs played a major role in causing decreased outputs—a hypothesis furtheremphasized by Hedin et al (1994), Likens & Bormann (1995), and Likens et al(1996). Dillon (1989), however, pointed out that the majority of the declines inbase cations in deposition at Hubbard Brook occurred during the early periodof the record, whereas the decreases in the base cation concentrations of therunoff water changed predominantly in the last part of the record, a circum-stance seemingly not supportive of the Driscoll et al (1989) hypothesis. Inaddition, Holdren & Church (1989) noted that an existing hypothesis of water-shed function (i.e. leaching associated with mobile acid anions; Galloway et al1983, Reuss & Johnson 1986) also could explain the decrease in streamwaterbase cations, a hypothesis strongly supported by Kirchner & Lydersen (1995) inexplaining their observations of base cation decreases in runoff from acidifiedNorwegian catchments.




Extending their analysis of nutrient cation loss at Hubbard Brook, Likens et al(1996) compared data sets of atmospheric deposition, streamwater chemistry,soil chemistry, vegetation chemistry, and forest growth and hypothesized thatthe decreased growth of forests observed at the site could be related to the loss ofbase cations (especially calcium) seen in streamwaters there. Whatever mech-anism may dominate the loss of base cations from the soil exchange complexof this watershed (and likely others in the region), the effect is real and is ofconsiderable importance to the quality of the receiving waters as well as to theforest of the catchment.

A second example of the use of long-term monitoring of catchment hy-drochemistry in unraveling catchment function comes from the study of thephenomenon of watershed nitrogen saturation. Regardless of whether one fa-vors the Aber et al (1989) definition of nitrogen saturation or that ofAgren& Bosatta (1988), concerns about environmental consequences are the same,namely that increased leaching of nitrogen from forested catchments may leadto increased acidification of upland surface waters or to eutrophication of eithercoastal estuaries (Hecky & Kilham 1988) or, more rarely, of fresh waters, wherenitrogen is the limiting nutrient (see discussion by Stoddard 1994).

Driscoll & Van Dreason (1993) observed that from the period 1982–1991,9 of 15 Adirondack drainage lakes studied in a regional monitoring projectshowed increases in nitrate concentrations, whereas none showed decreases.Murdoch & Stoddard (1992) analyzed surface water monitoring data from theCatskill Mountain region of New York state for varying time periods and foundincreasing nitrate concentrations from a number of sites, especially duringthe period from the early 1970s to 1990. Inasmuch as both regions receiveelevated inputs of nitrogen from acidic deposition, the authors all speculatedthat decreased biotic demand within the terrestrial portions of the systems mightbe leading to the increased surface water nitrate concentrations—i.e. that thecatchments were moving towards or experiencing nitrogen saturation.

Subsequent catchment hydrochemical monitoring in the region, however,has shown a more complicated and interesting effect. Driscoll et al (1995) re-ported that since 1991 there has been a notable decline in nitrate concentrationsin Adirondack lakes such that there are now no significant trends in nitrateconcentrations for these waters for the period 1982–1994.

Similarly, Mitchell et al (1996) found that nitrate losses from four forestedcatchments distributed across the northeastern United States declined markedlyin the early 1990s (Figure 2). Losses of nitrate from Hubbard Brook Watershed6 (New Hampshire), East Bear Brook Watershed (Maine), Arbutus Watershedin the Adirondack Mountains (New York), and Biscuit Brook in the CatskillMountains (New York) were notably high during the spring snowmelt of 1990.Following this period of high loss of nitrate from the catchments, loss rates de-



38 CHURCH

Figure 2 Streamwater concentrations (monthly average) of nitrate in four forested northeasterncatchments before and after a period of unusual cold and soil freezing. Reprinted from Mitchellet al (1996), by permission of the American Chemical Society.




creased substantially, leading to a reversal of increases from the preceding years.The high losses of spring 1990 seemed to be associated with an extremely coldand dry period in the preceding winter. A severe cold period in December 1989combined with lack of snow cover most likely led to substantial freezing of soilsacross the region (Mitchell et al 1996). Boutin & Robitaille (1995) showed thatsoil freezing induced by removal of snow from the ground in mature stands ofsugar maple in Quebec induced high leaching losses of nitrogen (as nitrate).Mitchell et al (1996) hypothesize that the combination of deep freezing of soilsin the four catchments they studied followed by wet conditions that facilitatedloss in runoff served to remove labile nitrogen from the watersheds. Theyalso noted that high nitrate concentrations were observed in streamwaters atHubbard Brook following periods of soil freezing during the winters of 1969–1970 and 1973–1974 (Likens & Bormann 1995). Likens (1992) had previ-ously noted the capacitance-like pattern of nitrogen storage and periodic loss atHubbard Brook and had wondered whether this pattern indicated an importantrole in the effects of changing atmospheric inputs of nitrogen in that the trendsin inputs and outputs seemed roughly to parallel each other.

The analysis and hypothesis formulations illustrated above could not havebeen undertaken without the critical baseline information provided by care-ful long-term monitoring of catchments. The resolution of the operant mecha-nisms and their quantification undoubtedly will be tied to continued monitoringstudies.

Hydrochemical monitoring of catchments sheds light on catchment functionnot only over the long term (e.g. years) but also over shorter term intervals.For example, the importance or even dominance of a variety of biogeochemicalfunctions or reactions wax and wane over the course of a year as inputs of heat,energy, and moisture vary through their annual cycles. One very interestingexample is that of the annual cycling of nitrogen in forested catchments. Duringannual growing periods, uptake of available nitrogen by vegetation may resultin very little nitrogen leaching from catchment soils to surface waters. Duringannual periods of vegetation dormancy, however, surface water concentrationsof nitrogen (usually in the form of nitrate) often increase (Stoddard 1994),sometimes quite dramatically. Long-term changes in the status of catchmentnitrogen cycling (e.g. as affected by maturation of forests or elevated inputsfrom atmospheric deposition) may also be reflected not only in long-term mon-itoring data (Kahl et al 1993, Mitchell et al 1996, Norton et al 1994) but alsoin observations of qualitative and quantitative changes in seasonal patterns ofstreamwater concentrations (Stoddard 1994) (Figure 3).

At even shorter time scales, relatively abrupt changes (i.e. hours to days) inwatershed leaching and consequent stream chemistry may be caused by the ef-fects of snowmelts or heavy rainstorms (Wigington et al 1990, 1996a). Patterns



40 CHURCH

Figure 3 Top panels are schematic representations of nitrogen cycle in watershed at progressivestages of watershed nitrogen loss. Sizes of arrows indicate the magnitude of process or transfor-mation. Differences between winter-spring and summer-fall seasons are shown on opposite sides.(a) STAGE 0 of watershed nitrogen loss. (Top panel) At Stage 0, nitrogen transformations aredominated by plant and microbial assimilation (uptake), with little or no NO−3 leakage from thewatershed during the growing season. (Bottom panel) Small amounts of NO−3 may run off duringsnowmelt, producing the typical Stage 0 seasonal NO−

3 pattern. Data in lower panel are from BlackPond, Adirondack Mountains. Reprinted from Stoddard (1994), by permission of the AmericanChemical Society.




Figure 3 (b) STAGE 1 of watershed nitrogen loss. (Top panel) As in Stage 0, uptake dominates thenitrogen cycle during the growing season at Stage 1 and little or no NO−

3 leaks from the watershedduring the summer and fall. The primary difference between Stage 0 and Stage 1 is the delay in theonset of N limitation during the spring season. (Bottom panel) Large runoff events (e.g. snowmeltor rainstorms) during the dormant season can produce episodic pulses of high NO−

3 concentrations,as shown in the typical Stage 1 seasonal NO−

3 cycle. Data in bottom panel are from Constable Pondin the Adirondack Mountains. Reprinted from Stoddard (1994), by permission of the AmericanChemical Society.



42 CHURCH

Figure 3 (c) STAGE 2 of watershed nitrogen loss. (Top panel) Uptake of nitrogen by forest plantsand microbes is much reduced at Stage 2, resulting in loss of NO−

3 to streams during winter andspring and to groundwater during the growing season. Loss of gaseous forms of nitrogen throughdenitrification may also be elevated at Stage 2 if conditions necessary for denitrification are present.Although episodes of higher NO−3 concentrations continue to occur during high-flow events suchas spring snowmelt, the primary difference between Stage 1 and Stage 2 is the presence of elevatedNO−3 concentrations in groundwater. (Bottom panel) The typical seasonal NO−3 pattern at Stage2 includes both high episodic concentrations and high base-flow concentrations. Data in bottompanel are from Fernow Experimental Forest, Control Watershed No. 4, West Virginia. Reprintedfrom Stoddard (1994), by permission of the American Chemical Society.




Figure 3 (d) STAGE 3 of watershed nitrogen loss. (Top panel) At Stage 3, no sinks for nitrogenexist in the watershed and all inputs, as well as mineralized nitrogen, are lost from the system eitherthrough denitrification or in runoff water. Because mineralization supplies nitrogen in excess ofdeposition, concentrations of NO−3 in runoff may exceed those in deposition. (Bottom panel) Typicalseasonal NO−3 pattern at Stage 3 includes concentrations at all seasons in excess of concentrationsattributable to deposition and evapotranspiration. Data are from Dicke Bramke in Germany andrevised from Stoddard (1994) (JL Stoddard, personal communication). Reprinted from Stoddard(1994), by permission of the American Chemical Society.



44 CHURCH

in the ionic chemistry during such hydrologically driven episodes also can shedlight on both processes and flow paths that control catchment hydrochemicalfunction and response (Wigington et al 1996b).

Monitoring across a variety of time scales undoubtedly will continue to pro-vide critical information for both the qualitative and quantitative understandingof how forested catchments operate.

Catchment ManipulationsResearchers in the earth sciences often link effect to cause first through obser-vation. Direct experimental manipulation is, of course, the most direct routeto hypothesis formation and testing. The more complicated the system underinvestigation, the more difficult it is to reach correct interpretations. Forestedcatchments (no matter how small) are still large, complex natural systems.Although manipulation of such systems is difficult physically and interpreta-tion of results challenging intellectually, such efforts can be very rewarding.

The earliest hydrochemical studies of catchment manipulations were thoseof the effects of tree removal (e.g. Likens et al 1977, Swank & Crossley 1988a).More recently, researchers have experimented with chemical manipulations toforested catchments (Adams et al 1993, Rasmussen & Farrell 1994, Wright &Tietema 1995) with interesting and important results. As an example, an inter-esting paired-catchment manipulation has occurred as part of the Bear BrookWatershed Manipulation (BBWM) Project in Maine (Thornton & Wigington1990).

The purpose of this project has been to investigate, using chemical ma-nipulations, the response of whole terrestrial ecosystems to increased load-ing of acidic or acidifying substances. The project has included a wide rangeof activities. Process research there (laboratory and plot studies) has focusedon better understanding the biogeochemistry of aluminum, sulfur, nitrogen,and dissolved organic carbon. Investigators there also have studied controlson cation supply by the mechanisms of chemical weathering and cationexchange.

The focus of the work, however, has been the manipulation of one of apair of contiguous small (∼10 ha) catchments with the second of the pairserving as a reference site (Kahl et al 1993, Norton et al 1994). The pairedcatchments for the BBWM are East Bear Brook (reference) and West BearBrook (experimental). These catchments have flows and hydrologic behaviorsthat are highly similar. Researchers began monitoring of the catchments inearly 1987 and began chemical manipulation of West Bear Brook in November1989. Before manipulation, the chemical compositions of the two streams wereremarkably close across all stages of flow. Consequently, chemical budgets forthe two watersheds were very similar.




Manipulations of the West Bear Brook catchment have consisted of additionsof dry ammonium sulfate (with distinct isotopic signatures for both sulfur andnitrogen) applied bimonthly at a rate of 1,800 equivalents per ha per year. Theadditions have tripled the annual catchment loading of sulfur and have quadru-pled the annual loading of nitrogen. In response, stream concentrations andwatershed fluxes of hydrogen ions, calcium, magnesium, sodium, potassium,sulfate, nitrate, and aluminum have increased relative to the control (Figure 4),while acid neutralizing capacity has declined and chloride has remained nearlyunchanged. After six years of manipulation, average annual sulfate concentra-tion of West Bear Brook increased from about 100 microequivalents per literto about 190 microequivalents per liter (expected steady state sulfate is about300 microequivalents per liter). Concentrations of nitrate have also increasedmarkedly. Total nitrogen output from West Bear Brook more than doubled, duemainly to large increases in the flux of nitrate. Stream nitrate concentrationsin West Bear Brook have increased markedly not only during fall and winterwhen vegetation is dormant but also during the summer growing season. Thisindicates that leaching of nitrogen to deeper soil horizons and to flow paths thatdominate summer base flow may be occurring (Kahl et al 1993, Norton et al1994).

The responses of West Bear Brook explicitly demonstrate that increases inatmospheric inputs can have dramatic (and rapid) effects on stream chemical re-sponse. Furthermore, they provide corroborative evidence that such changes areoccurring in a manner consistent with prevalent theory (e.g. Turner et al 1990)and with simulations from models based on such theory (Cosby et al 1996). Thiswork puts us in a better observational and theoretical position both to under-stand the ways in which acidic deposition affects catchment hydrochemistryand to develop predictive models to postulate potential future effects.

Modeling Catchment Hydrochemical ResponseModeling the response of any system to variations in its environment is anexcellent way to learn more about it or, more often, more about our lack ofunderstanding of the way it operates. The more complex the system, the moredifficult the modeling. At any given moment, there exist myriad pathwaysthat water can follow within a forested catchment and an astonishing varietyof biogeochemical interactions that can influence that composition of that wa-ter. When one considers that the pathways and biogeochemical influences canchange at time scales as short as minutes to hours (e.g. during rainstorm events),the task becomes even more daunting. Add to this the desire of resource man-agers to use models to predict the influence of external pollutants (e.g. acidicdeposition) at times often years to decades in the future, and challenges be-come formidable. In the last decade, however, more and more researchers



46 CHURCH




and modelers have accepted the various challenges of modeling catchmenthydrochemistry.

As with catchment studies themselves, modeling of catchment function beganwith hydrological topics, e.g. flow generation. With the influx of large amountsof research funding dedicated to the tasks of understanding and predictingeffects of acidic deposition, the conflation of models of flow with soil profilegeochemical models began in earnest. The result has been a generation ofmodels of whole-catchment hydrochemical response (see review by Ball &Trudgill 1995). A cursory comparison of review coverage of the topic ofsolute modeling by Trudgill (1986, 1995) ably demonstrates the explosion ofcatchment hydrochemical modeling that has occurred over the past decade.Often these models have built upon recent or co-developing catchment flowgeneration models (e.g. TOPMODEL; Beven & Kirkby 1979, Hornberger et al1985).

Of the numerous catchment hydrochemical models that have come into beingin the last 10–15 years, several deserve mention either because of their seminalimportance in the field or because of their continuing application or evolution.These models include the Birkenes model (Christophersen et al 1982), theIntegrated Lake/Watershed Acidification (ILWAS) model (Chen et al 1983,Gherini et al 1985), the Trickle-Down (Schnoor et al 1984) and ExtendedTrickle-Down models (Nikolaidis et al 1991), and the Model of Acidificationof Groundwater in Catchments (MAGIC) (Cosby et al 1985a,b).

Early models focus on watershed acidification mediated by sulfur deposition(e.g. Cosby et al 1985a). More recent model development efforts are working toincorporate nitrogen cycling into process representations (Liu et al 1992, Posteket al 1995; JD Aber, BJ Cosby, personal communications). The complexity ofthe biologically dominated nitrogen cycle greatly complicates this task.

Linking and applying hydrologic and biogeochemical models at catchmentscales requires ingenuity. An interesting and useful aid in such modelinghas been a technique called end-member mixing analysis (EMMA) (Neal &Christophersen 1989, Christophersen et al 1990, Hooper et al 1990). Instead ofapplying hydrologic models with knowledge of catchment chemical and biolog-ical properties to infer resulting stream chemistry, EMMA approaches the taskof understanding catchment hydrochemical response in an apparent inverse,

←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−Figure 4 Streamwater concentrations of nitrate, sulfate, calcium, and total aluminum in EastBear Brook (reference catchment) and West Bear Brook (manipulated catchment) before and afteradditions of ammonium sulfate to the soils of West Bear Brook. Reprinted in part from Kahl et al(1993), by permission of the American Chemical Society.



48 CHURCH

yet more intuitive, manner. With EMMA one uses the observed chemistry ofrunoff to help determine prior flow paths of the water. This can be accomplishedbecause different soil horizons often have very different chemical properties.Thus water flowing from upper horizons can be very different in chemicalcomposition than water flowing from lower horizons. If one has concurrentmeasurements of chemical species in streamwater as well as from soil watersof several potentially contributing horizons, one may then use least-squaresprocedures to solve for the relative contributions from each of the sources—the end members. Principal necessary assumptions are that the compositionof end members is constant in space and time over the applicable time periodof the applied analysis, that the chemical species mix conservatively (i.e. in anonreactive manner), and that sufficient differences exist in the composition ofend members to make the analysis feasible. Hooper & Christophersen (1992)have used EMMA to provide flow routing for the application of MAGIC to theprediction of both average annual stream chemistry and episodic chemistry fora well-studied catchment in the southeastern United States.

One of the most difficult conceptual and practical problems facing modelersof catchment hydrochemical response is that of scaling. That is, how doesone reconcile the huge variety and variability (in space and time) of fine-scalemeasures of catchment properties to the relatively gross lumping required forrealistic application of physically based simulation models? Hydrologic mod-elers have faced this problem for some time (Beven 1989, Grayson et al 1992,Hornberger & Boyer 1995). The advent of catchment hydrochemical mod-eling and the attendant desire to reconcile measures of catchment chemicaland biological properties (sometimes at many points in space and time) withpossible flow paths and stream hydrochemical response have made the prob-lem even more vexing (Christophersen et al 1993, Wheater et al 1993, Ball &Trudgill 1995, Wheater & Beck 1995). If one feels overwhelmed by the vari-ability in natural systems or has difficulty reconciling current measures (e.g.soil water chemistry) within a catchment to observed stream composition, howcan one ever hope to model future catchment response to changes in inputsaccurately?

On the analytical front there is the possibility that empirical approaches mayoffer real hope for predictive uses in some carefully defined instances amid anabundance of environmental noise (e.g. see Christophersen et al 1993, Gauch1993, Kirchner et al 1993a,b). More philosophically, a resolution perhaps liesin the realization that every tool has its appropriate uses and its limitations.The use of physically based watershed simulation models as heuristic toolsto understand how catchments function thrives on the availability of data fortesting for model invalidation. Such efforts should proceed at every opportu-nity. On the other hand, there is often a need for application of models for




predictive purposes within a management context, especially at regional scales(e.g. Church et al 1989, 1992, Thornton et al 1990, Van Sickle & Church1995). Such applications require data aggregation at a variety of scales, andthus introduce substantial additional uncertainty into a task already replete withuncertainty. An appropriate use of models in this context is to serve in the roleof “scenario testers,” e.g. “if all other factors (climate, land use, etc) remainedunchanged (a condition a priori known to be unrealistic), what might be the rel-ative effects of an increase or decrease in levels of acidic deposition on surfacewater quality in 50 years?” Such applications require that modelers convey toresource managers the uncertainties (quantifiably large, qualitatively huge) as-sociated with such scenario tests and the caution that the projections, althoughthey may seem consistent with other lines of evidence (latest conceptual the-ory, trends analyses, etc), cannot be validated as “truth” (Oreskes et al 1994).Currently they must remain hypotheses at best.

Modeling of the hydrochemical response of forested catchments is a fieldof inquiry confronted by numerous interesting and difficult questions. It willbenefit from continued and iterative efforts at model development, model appli-cation, and model testing (e.g. against whole catchment manipulations; Cosbyet al 1996). Continued efforts spanning spatial scales from plots to regions areneeded to move the endeavor forward. Much exciting work remains to be done.

EXTRAPOLATION OF WATERSHEDHYDROCHEMICAL RESPONSE

Site RepresentativenessOften the intent of studying the hydrochemistry of an individual catchment is tolearn about not only its function or response to a perturbation (e.g. acidic depo-sition), but also about the behavior of other similar catchments under the sameor changing conditions. A key question that investigators often ask themselvesis, “Although my catchment appears similar to others that have been studied (orto which I would like to extrapolate my findings), how similar are they really?”That is, how representative is it of others around it or perhaps, far removed(either in space or time) (Berkowitz et al 1988)? This can be a difficult thingto know. Researchers must consider a variety of factors including bedrock andsurficial geology, soil development, vegetation type and age, climate, and othereffects. Site history, for example, can be important. Watersheds that are nowforested may previously have been burned, timbered, or farmed. Such activitiescan have lasting effects, for example, on catchment hydrochemical response re-garding nitrogen cycling (JD Aber, JE Compton, unpublished data) as well ason other functions (Likens et al 1977, 1995, Swank & Crossley 1988a).



50 CHURCH

Statistical SamplingExtrapolation of observations made at fine scales to coarser scales is a very chal-lenging task (Rastetter et al 1992). One quantitative approach to the problemis to use a specified statistical approach to select catchments for study (or sam-pling) such that each is quantitatively representative of a larger target populationof interest. Such an approach has been used in a few large research and assess-ment programs (e.g. Linthurst et al 1986, Kaufmann et al 1988, Church et al1992, Stevens 1994) as well as in modeling studies based on sampling withinsuch programs (Van Sickle & Church 1995). Such sampling and analysis pro-grams are large, time consuming, expensive activities. Catchment sampling,and thus process-level understanding, in such programs can never be at thelevel attained in longer-term fully integrated studies (e.g. Hubbard Brook),but that often is not the intent of the larger sampling efforts. The coarser-scalesampling often is a key component of assessment activities focused on regional-scale adverse environmental effects (e.g. acidic deposition effects) on forestedcatchments. Its spatial nature is required to attack regional-scale questions, andits quantitative nature is useful in the arena of public policy where effects ofreal (or alleged) individual site bias may be of concern. Regional statisticalsampling can be a very useful complement to finer-scale catchment studies inunderstanding and predicting catchment hydrochemical response.

Development of a Catchment TaxonomyThere may be other ways to attack the problems of determining site represen-tativeness and the appropriate extrapolation of site-specific data. One of theprincipal conclusions to come from the American Geophysical Union ChapmanConference onHydrogeochemical Responses of Forested Catchments(Churchet al 1990) convened in the Fall of 1989 was that integrated watershed re-search, indeed, is hampered by a lack of a structured and efficient means ofcommunication of results among studies. Researchers at one watershed, forexample, may know quite a lot about how water moves through their systemand how it interacts geochemically with their particular rocks and soils. Re-searchers at another watershed also may have the same kind of knowledge ontheir local scale. These researchers have difficulty, however, communicatingtheir results accurately and efficiently with one another within an organizedrelevant context. This is because there is lack of a general uniform schemeor organizational structure for the portrayal of results of watershed studies andexperiments.

Something that would be a great aid to watershed researchers would be thedevelopment of a taxonomic scheme for watersheds based upon not just catch-ment characteristics (e.g. aspect or percentage of conifer coverage), but also oncatchment hydrochemical function (e.g. whether the system exhibits nitrogen




saturation; Stoddard 1994). Such a catchment taxonomy would aid researchersby organizing site-specific knowledge on catchment hydrochemistry into a uni-form structure related to a general paradigm designed to address watershedfunction.

With such a taxonomy and with a compendium of catchments placed within it,researchers for a particular watershed could investigate where their watershedmight be classified within the scheme and they could relate results of theirspecific experiments to experiments performed by researchers at other (perhapsquite distant) sites. If the researchers at different sites had similar types ofwatersheds (based on the taxonomy) and observed similar results from similarexperiments, then this would serve to corroborate their conclusions. If theresearchers had drawn different conclusions from the similar results, the useof the taxonomy could lead to profitable discussions as to why. Similarly, ifresearchers observed different responses in watersheds presumed to be similarwithin the taxonomy, this would also call for reevaluation.

In addition to assisting in analyzing existing data, such a taxonomy also couldaid in organizing experiments yet to be performed. Researchers could benefitby knowing what watersheds are similar or dissimilar (in structure and func-tion) to theirs and thus could contact other researchers directly to discuss moreefficiently the pros and cons of potential experiments. Results of observationsand experiments past and present would lead to constant evaluation of the ex-periments, the watersheds, and the taxonomy. Such an open perspective wouldlead to iterative improvement in both the taxonomy and, more importantly, inthe understanding of catchment hydrochemical function.

RECENT TRENDS AND FUTURE OF WATERSHEDHYDROCHEMICAL RESEARCH

Attacking the complex environmental problem of effects of acidic depositionon soils and surface waters has called for investigation of forest watershed sys-tems in an integrated fashion. Over the past 15 or so years, the intense (andhighly funded) study of the environmental effects of acidic deposition has ledhydrologists, geochemists, and biologists to work together better to integratetheir studies and to think of catchments as integrated units (e.g. Church et al1990, Neal & Hornung 1990). Consequently, a generation of watershed scien-tists has come to appreciate the value of working as part of a multidisciplinaryteam.

Although major questions related to the effects of atmospheric depositionof nitrogen remain, the heavily funded days of acid rain research are over,and funding of fully integrated studies of catchment hydrochemistry has de-creased dramatically. This comes at a time when many studies had established



52 CHURCH

a strong groundwork of specific site knowledge, research capabilities, and, justas important, project infrastructure to tackle additional challenges regardingthe workings of forested catchments. Thus, numerous studies have been dis-rupted or dismembered. Baseline monitoring programs eventually will indicatethe need to better understand catchment processes and their interconnections(Likens 1983). When funding returns for whole-watershed hydrochemical stud-ies there will be a great inefficiency in reestablishing work at prior research sites.Hopefully, key site data and institutional memory will not be lost. More effec-tive than such an oscillation of research driven by legislative or programmaticpriorities would be a more stable long-term combination of monitoring andsite-specific research. The current structure of agencies that fund such researchand of the laws that govern the mechanisms by which such research can befunded, however, present a formidable barrier to achieving that goal.

What specifically the near-term future holds for integrated research in forestedcatchments is unclear. The long-term future, however, is bright and exciting.Regardless of whether it comes through “fits and starts” or whether througha more smoothly functioning continuation of established research sites andteams, research on the hydrochemistry of forested catchments will continueto be a fascinating field. Recent advances, combined with a new and keenerperspective of catchments as integrated wholes, promise to make future workmore productive and stimulating than ever.

CLOSING THOUGHT

“There are men charged with the duty of examining the construction of the plants, animals,and soils which are the instruments of the great orchestra. These men are called professors.Each selects one instrument and spends his life taking it apart and describing its stringsand sounding boards. This process of dismemberment is called research. The place fordismemberment is called a university.

A professor may pluck the strings of his own instrument, but never that of another, andif he listens for music he must never admit it to his fellows or to his students. For all arerestrained by an ironbound taboo which decrees that the construction of instruments is thedomain of science, while the detection of harmony is the domain of poets.” Aldo Leopold,Song of the Gavilan[Reprinted in part from Leopold (1940), by permission of The WildlifeSociety.]

Aldo Leopold wryly yet eloquently argues the case for holistic approaches inthe natural sciences. Multidisciplinary studies promise to help us better to un-derstand and comprehend catchment hydrochemical function. Such approacheswill yield answers to our questions of today and will provide us with more (andever more challenging) questions for tomorrow. Day by day we detect more ofthe harmony of the “great orchestra.” There is much to anticipate.




SUGGESTED READING/SOURCES OF INFORMATION

There exist a number of excellent sources of information on hydrochemistry offorested catchments.

The following journals routinely carry research and review articles in thetopical area and sometimes devote entire volumes to the topic.

Biogeochemistry(Kluwer Academic)Ecological Applications(Ecological Society of America)Global Biogeochemical Cycles(American Geophysical Union)Hydrological Processes(John Wiley & Sons)Journal of Environmental Quality(American Society of Agronomy, Crop

Science Society of America, Soil Science Society of America)Journal of Hydrology(Elsevier Science), e.g. see Special Issue “Transfer of

Elements Through the Hydrological Cycle.” 1990. vol. 116The Science of the Total Environment(Elsevier Science)Water, Air & Soil Pollution(Kluwer Academic), e.g. see special issue devoted

to the conference “Acid Reign ’95?” 1996. vol. 85 nos. 1&2Water Resources Bulletin(American Water Resources Association)Water Resources Research(American Geophysical Union), e.g. see special

section “Catchment Hydrochemistry.” 1990. vol. 26:2947–3100

A number of professional societies and organizations hold annual meetingsor special conferences that either contain special symposia on or are in somecases devoted entirely to, hydrochemical studies. The Fall and Spring AnnualMeetings of the American Geophysical Union (AGU), for example, often con-tain special sessions dealing with topics in hydrochemistry. The AGU alsosponsors special Chapman conferences in the field (e.g.HydrogeochemicalResponses of Forested Catchmentsheld September 1989 andNitrogen Cyclingin Forested Watershedsheld September 1996). Since July 1991 the GordonResearch Conferences also have sponsored a week-long Gordon ConferenceHydrological-Geochemical-Biological Processes in Forested Catchmentsev-ery other year.

Information relevant to hydrochemistry of forested catchments is becomingavailable on the World Wide Web. The following addresses may be particularlyuseful to students of the topic.

American Geophysical Union—http://earth.agu.org/kosmos/homepage.htmlCoweeta Hydrological Laboratory—http://oikos.ecology.uga.edu/Gordon Research Conferences—http://hackberry.chem.niu.edu: 70/0/Confer-

enceListings/GordonConferences/index.htmlHubbard Brook Ecosystem Study—http://www.yale.edu/edex/index2.htmlNational Science Foundation Long-Term Ecological Research—

http://lternet.edu/



54 CHURCH

ACKNOWLEDGMENTS

I thank Carol Kendall for providing a very helpful review of my text on flowgeneration. I thank George Hornberger for his careful review of that samesection as well as his review of my text on modeling of catchment hydro-chemical response. I thank Paul Shaffer for his careful, detailed review of theentire manuscript. Comments from these reviewers substantially improved thiswork.

I thank Robb Turner for providing the line drawing for Figure 1, MyronMitchell for providing data for Figure 2, John Stoddard for providing linedrawings and data for Figure 3, and Steve Norton, Steve Kahl, and John Scofieldfor providing data for Figure 4.

Visit the Annual Reviews home pageathttp://www.annurev.org.

Literature Cited

Aber JD, Melillo JM, Nadelhoffer KJ, Pastor J,Boone RD. 1991. Factors controlling nitro-gen cycling and nitrogen saturation in north-ern temperate forest ecosystems.Ecol. Appl.1:305–15

Aber JD, Nadelhoffer KJ, Steudler P, MelilloJM. 1989. Nitrogen saturation in northern for-est ecosystems.BioScience39:378–86

Adams MB, Edwards PJ, Wood F, KochenderferJN. 1993. Artificial watershed acidificationon the Fernow Experimental Forest, USA.J.Hydrol. 150:505–19

Adams MB, Kochenderfer JN, Angradi JN, Ed-wards PJ. 1994.Forty years of hydrometeo-rological data from the Fernow ExperimentalForest, West Virginia. USDA For. Serv. Gen.Tech. Rep.NE-184

Agren GI, Bosatta E. 1988. Nitrogen satura-tion of terrestrial ecosystems.Environ. Pollut.54:185–97

Anderson MG, Burt TP, eds. 1990.ProcessStudies in Hillslope Hydrology. Chichester,UK: Wiley & Sons. 539 pp.

Andersson F, Olsson B, eds. 1985.LakeGardsjon—An Acid Forest Lake and ItsCatchments. Swed. Res. Counc. Ecol. Bull.37. 336 pp.

Baker JP, Bernard DP, Sale MJ, Christensen SW.1990. Biological effects of changes in sur-face water acid-base chemistry.NAPAP StateSci. Technol. Rep. 13. Washington, DC: Natl.Acid Precip. Assess. Prog. 381 pp.

Baker JP, Van Sickle J, Gagen CJ, DeWalleDR, Sharpe WE, et al. 1996. Episodic acidi-fication of small streams in the northeastern

United States: Effects on fish populations.Ecol. Appl.6:422–37

Ball J, Trudgill S. 1995. Overview of solutemodelling. See Trudgill 1995, pp. 3–56

Baron J, ed. 1992.Biogeochemistry of a Sub-alpine Ecosystem: Loch Vale Watershed.New York: Springer-Verlag. 247 pp.

Bazemore DE, Eshleman KN, Hollenbeck KJ.1994. The role of soil water in stormflow gen-eration in a forested headwater catchment:synthesis of natural tracer and hydrometricevidence.J. Hydrol.162:47–75

Berkowitz AR, Kolasa J, Peters RH, PickettSTA. 1988. How far in space and time canthe results from a single long-term study beextrapolated? InLong-Term Studies in Ecol-ogy, ed. GE Likens, pp. 192–200. New York:Springer-Verlag. 214 pp.

Beven K. 1989. Changing ideas in hydrology—the case of physically-based models.J. Hy-drol. 105:157–72

Beven KJ, Germann PF. 1982. Macroporesand water flow in soils.Water Resour. Res.18:1311–25

Beven K, Kirkby MJ. 1979. A physically-basedvariable contributing area model of basin hy-drology.Hydrol. Sci. Bull.24:43–69

Binkley D, Driscoll CT, Allen HL, Schoen-berger P, McAvoy D. 1988.Acid Depositionand Forest Soils. New York: Springer-Verlag.149 pp.

Bonell M. 1993. Progress in the understandingof runoff generation dynamics in forests.J.Hydrol. 150:217–75

Boutin R, Robitaille G. 1995. Increased soil




nitrate losses under mature sugar maple treesaffected by experimentally induced deepfrost.Can. J. For. Res. 25:588–602

Brady NC. 1974.The Nature and Properties ofSoils. New York: Macmillan. 639 pp. 8th ed.

Bricker OP, Rice KC. 1993. Acid rain.Annu.Rev. Earth Planet. Sci.21:151–74

Bullen TD, Shanley JB, Clark S. 1994. Sr andPb isotopes as surrogate tracers of water flow-paths in a forested catchment.EOS Am. Geo-phys. Union, April 19, 1994. p. 144 (Abstr.)

Buttle JM. 1994. Isotope hydrograph separa-tions and rapid delivery of pre-event wa-ter from drainage basins.Prog. Phys. Geogr.18:16–41

Cerny J, Billett MF, Cresser MS. 1994. Elementbudgets. See Moldan &Cerny 1994, pp. 189–205

Cerny J, Novak M, Paces T, Wieder RK, eds.1995.Biogeochemical Monitoring in SmallCatchments. Dordrecht: Kluwer Acad. 432pp.

Chao TT, Harward ME, Fang SC. 1964. Iron oraluminum coatings in relation to sulfate ad-sorption characteristics of soils.Soil Sci. Soc.Am. Proc. 28:632–5

Chen CW, Gherini S, Hudson RJM, Dean J.1983. The integrated lake/watershed acidifi-cation study, volume 1: model and proce-dures.Elect. Power Res. Inst. Rep. no. EA-3221, Palo Alto, CA.

Christ M, Zhang Y, Likens GE, Driscoll CT.1995. Nitrogen retention capacity of a north-ern hardwood forest soil under ammoniumsulfate additions.Ecol. Appl. 5:802–12

Christophersen N, Neal C, Hooper RP. 1993.Modelling the hydrochemistry of catch-ments: a challenge for the scientific method.J. Hydrol.152:1–12

Christophersen N, Neal C, Hooper RP, VogtRD, Andersen S. 1990. Modelling streamwa-ter chemistry as a mixture of soilwa-ter end-members—A step towards second-generation acidification models.J. Hydrol.116:307–20

Christophersen N, Seip HM, Wright RF.1982. A model for stream water chemistryat Birkenes, Norway.Water Resour. Res.18:977–96

Church MR, Hornberger GM, Sorooshian S.1990. Catchment hydrogeochemistry.WaterResour. Res.26:2947

Church MR, Thornton KW, Shaffer PW,Stevens DL, Rochelle BP, et al. 1989. Di-rect/Delayed Response Project: future ef-fects of long-term sulfur deposition on sur-face water chemistry in the northeast andsouthern Blue Ridge Province.EPA Rep. no.EPA/600/3-89/061a-d, Washington, DC: USEnviron. Protect. Agency. 887 pp.

Church MR, Shaffer PW, Thornton KW, Cas-

sell DL, Liff CI, et al. 1992. Direct/DelayedResponse Project: Future effects of long-term sulfur deposition on stream chemistryin the mid-Appalachian region of the east-ern United States,EPA Rep. no. EPA/600/R-92/186, Washington, DC: US Environ. Pro-tect. Agency. 384 pp.

Cosby BJ, Hornberger GM, Galloway JN,Wright RF. 1985a. Modeling the effects ofacid deposition: assessment of a lumped pa-rameter model of soil water and streamwaterchemistry.Water Resour. Res.21:51–63

Cosby BJ, Hornberger GM, Galloway JN,Wright RF. 1985b. Time scales of catchmentacidification: a quantitative model for esti-mating freshwater acidification.Environ. Sci.Technol.19:1144–9

Cosby BJ, Norton SA, Kahl JS. 1996. Usinga paired-catchment manipulation experimentto evaluate a catchment-scale biogeochemi-cal model.Sci. Total Environ.183:49–66

DeWalle DR, Swistock BR. 1994. Differencesin oxygen-18 content of throughfall and rain-fall in hardwood and coniferous forests.Hy-drol. Proc.8:75–82

DeWalle DR, Swistock BR, Sharpe WE. 1988.Three-component tracer model for stormflowon a small Appalachian forested catchment.J. Hydrol.104:301–10

Dillon PJ. 1989.Environ. Sci. Technol.23:623,648 (Letter)

Douglass JE, Hoover MD. 1988. History ofCoweeta. See Swank & Crossley 1988, pp.17–31

Driscoll CT, Likens GE, Hedin LO, Eaton JS,Bormann FH. 1989. Changes in the chem-istry of surface waters.Environ. Sci. Technol.23:137–143

Driscoll CT, Postek KM, Kretser W, Raynal DJ.1995. Long-term trends in the chemistry ofprecipitation and lake water in the Adiron-dack region of New York, USA.Water AirSoil Pollut.85:583–88

Driscoll CT, Van Dreason R. 1993. Seasonal andlong-term temporal patterns in the chemistryof Adirondack lakes.Water Air Soil Pollut.67:319–44

Dunne T, Leopold LB. 1978.Water in Environ-mental Planning. New York/San Francisco:Freeman. 818 pp.

Eshleman KN. 1988. Predicting regionalepisodic acidification of surface waters us-ing empirical models.Water Resour. Res.24:1118–26

Galloway JN. 1996. Anthropogenic mobiliza-tion of sulfur and nitrogen: immediate anddelayed consequences.Annu. Rev. EnergyEnviron.21:261–92

Galloway JN, Norton SA, Church MR. 1983.Freshwater acidification from atmosphericdeposition of sulfuric acid: A conceptual



56 CHURCH

model.Environ. Sci. Technol. 17:541A–5AGat JR. 1996. Oxygen and hydrogen isotopes

in the hydrologic cycle.Annu. Rev. EarthPlanet. Sci.24:225–62

Gauch HG Jr. 1993. Prediction, parsimony andnoise.Am. Sci. 81:468–78

Genereux DP, Hemond HF. 1990. Naturally oc-curring222Rn as a tracer for streamflow gen-eration: steady state methodology and fieldexperiment.Water Resour. Res.26:3065–75

Genereux DP, Hemond HF, Mulholland PJ.1993. Use of radon-222 and calcium as trac-ers in a three-end-member mixing modelfor streamflow generation on the west forkof Walker Branch Watershed.J. Hydrol.142:167–211

Germann P. 1986. Rapid drainage response toprecipitation.Hydrol. Process. 1:3–13

Germann PF. 1990. Macropores and hydrologichillslope processes. See Anderson & Burt1990, pp. 327–63

Gherini SA, Mok L, Hudson RJ, Davis GF, ChenCW, Goldstein RA. 1985. The ILWAS model:formulation and application.Water Air SoilPollut. 26:425–59

Grayson RB, Moore ID, McMahon TA. 1992.Physically based hydrologic modeling. 2. Isthe concept realistic?Water Resour. Res.26:2659–66

Hauhs M. 1989. Lange Bramke: An ecosys-tem study of a forested catchment. InAcidicPrecipitation, Vol. 1, Case Studies, ed. DCAdriano, M Havas, pp. 275–305. New York:Springer-Verlag. 311 pp.

Hecky RE, Kilham P. 1988. Nutrient limita-tion of phytoplankton in freshwater and ma-rine environments: A review of recent evi-dence on the effects of enrichment.Limnol.Oceanogr.33:796–822

Hedin LO, Granat L, Likens GE, Buishand TA,Galloway JN, et al. 1994. Steep declines in at-mospheric base cations in regions of Europeand North America.Nature367:351–54

Hewlett JD, Hibbert AR. 1967. Factors affect-ing the response of small watersheds to pre-cipitation in humid areas. InInt. Symp. For.Hydrol., pp. 275–290. New York: Pergamon

Hibbert AR, Troendle CA. 1988. Stream flowgeneration by variable source area. SeeSwank & Crossley 1988, pp. 111–127

Hinton MJ, Schiff SL, English MC. 1994. Ex-amining the contributions of glacial till wa-ter to storm runoff using two- and three-component hydrograph separations.WaterResour. Res.30:983–93

Holdren GR, Church MR. 1989.Environ. Sci.Technol.23:1028, 1078–79 (Letter)

Hooper RP, Christophersen N. 1992. Predict-ing episodic stream acidification in the south-eastern United States: combining a long-termacidification model and the end-member mix-

ing concept.Water Resour. Res.28:1983–90

Hooper RP, Christophersen N, Peters NE. 1990.Modelling streamwater chemistry as a mix-ture of soilwater end-members—an applica-tion to the Panola Mountain catchment, Geor-gia, U.S.A.J. Hydrol.116:321–43

Hooper RP, Kendall C. 1994. Can we believewatershed mixing models? A comparison ofisotopic and chemical tracers. p. 144.EOSAm. Geophys. Union, April 19, 1994, 614pp. (Abstr.)

Hornberger GM, Beven K, Cosby J, Sapping-ton DE. 1985. Shenandoah watershed study:calibration of a topography-based, variablecontributing area hydrological model to asmall forested catchment.Water Resour. Res.21:1841–50

Hornberger GM, Boyer EW. 1995. Recent ad-vances in watershed modelling. US Nat.Rep. to IUGG, 1991–1994.Rev. Geophys.July:949–57 (Suppl.)

Hornung M, Roda F, Langan SJ, eds. 1990. Areview of small catchment studies in WesternEurope producing hydrochemical budgets.Air Pollut. Res. Rep. 28. Brussels: Comm.Eur. Communities. 186 pp.

Horton RE. 1933. The role of infiltration in thehydrologic cycle.Am. Geophys. Union Trans.14:446–60

Horton RE, Hawkins RH. 1965. Flow path ofrain from the soil surface to the water table.Soil Sci.100b:377–83

Jeffries DS, Kelso JRM, Morrison IK. 1988.Physical, chemical, and biological character-istics of the Turkey Lakes Watershed, CentralOntario, Canada.Can. J. Fish. Aquat. Sci.45:3–13 (Suppl. 1)

Jenkins A, Peters NE, Rodhe A. Hydrology.1994. See Moldan &Cerny, pp. 31–54

Johnson DW, Lindberg SE, eds. 1992.Atmo-spheric Deposition and Forest Nutrient Cy-cling. New York: Springer-Verlag. 707 pp.

Johnson DW, Van Hook RI, eds. 1989.Anal-ysis of Biogeochemical Cycling Processesin Walker Branch Watershed. New York:Springer-Verlag. 401 pp.

Kahl JS, Norton SA, Fernandez IJ, NadelhofferKJ, Driscoll CT, Aber JD. 1993. Experimen-tal inducement of nitrogen saturation at a wa-tershed scale.Environ. Sci. Technol.27:565–68

Kaufmann PR, Herlihy AT, Elwood JW, MitchME, Overton WS, et al. 1988. Chemical char-acteristics of streams in the mid-atlantic andsoutheastern United States, Volume I: pop-ulation descriptions and physico-chemicalrelationships. EPA Rep. no. EPA/600/3-88/021a. Washington, DC: US Environ. Pro-tect. Agency, 397 pp.

Kendall C, Bullen TD, Shanley JB. 1994. Multi-




isotope approach for the evaluation of thecontributions of shallow and deep flowpathsas a function of catchment scale. p. 144.EOSAm. Geophys. Union, April 19, 1994, 416 pp.(Abstr.)

Kendall C, Sklash MG, Bullen TB. 1995. Iso-tope tracers of water and solute sources incatchments. See Trudgill 1995, pp. 261–303

Kennedy VC, Kendall C, Zellweger GW,Wyerman TA, Avanzino RJ. 1986. Determi-nation of the components of stormflow usingwater chemistry and environmental isotopes,Mattole River Basin, California.J. Hydrol.84:107–40

Kirchner JW, Dillon PJ, LaZerte BD. 1993a.Predictability of geochemical buffering andrunoff acidification in spatially hetero-geneous catchments.Water Resour. Res.29:3891–901

Kirchner JW, Dillon PJ, LaZerte BD. 1993b.Separating hydrological and geochemical in-fluences on runoff acidification in spatiallyheterogeneous catchments.Water Resour.Res.29:3903–16

Kirchner JW, Lydersen E. 1995. Base cation de-pletion and potential long-term acidificationof Norwegian catchments.Environ. Sci. Tech-nol. 29:1953–60

Kirkby C, Newson MD, Gilman K. 1991. Plyn-limon research: The first two decades.Inst.Hydrol. Rep.109:1–187

Leopold A. 1940. “Song of the Gavilan.”J. Wild.Manage.(Correspondence) 4:329–32

Likens GE. 1983. A priority for ecological re-search.Bull. Ecol. Soc. Am.64(4):234–43

Likens GE, ed. 1988.Long-Term Studies inEcology. New York: Springer-Verlag. 214pp.

Likens GE. 1992.The Ecosystem Approach: ItsUse and Abuse. Oldendorf/Luhe: EcologyInstitute. 166 pp.

Likens GE, Bormann FH, Pierce RS, EatonJS, Johnson NM. 1977.Biogeochemistry ofa Forested Ecosystem. New York: Springer-Verlag. 146 pp. 1st ed.

Likens GE, Bormann H. 1995.Biogeochem-istry of a Forested Ecosystem. New York:Springer-Verlag. 159 pp. 2nd ed.

Likens GE, Driscoll CT, Buso DC. 1996. Long-term effects of acid rain: Response andrecovery of a forested ecosystem.Science272:244–46

Likens PC. 1994.Publications of the Hub-bard Brook Ecosystem Study. Millbrook, NY:Mary Flagler Cary Arboretum. 109 pp.

Linthurst RA, Landers DH, Eilers JM, BrakkeDF, Overton WS, et al. 1986. Characteris-tics of lakes in the eastern United States, vol-ume I: population descriptions and physico-chemical relationships.US EPA Rep. no.

EPA/600/4-86/007a. Washington, DC: USEnviron. Protect. Agency. 136 pp.

Liu S, Munson R, Johnson DW, Gherini S, Sum-mers K, et al. 1992. The nutrient cyclingmodel (NaCM): overview and application.See Johnson & Lindberg, pp. 583–609

McDonnell JJ. 1990. A rationale for old waterdischarge through macropores in a steep, hu-mid catchment.Water Resour. Res.26:2821–32

Mitchell MJ, Driscoll CT, Kahl JS, Likens GE,Murdoch PS, Pardo LH. 1996. Climatic con-trol of nitrate loss from forested watershedsin the northeast United States.Environ. Sci.Technol.30:2609–12

Moldan B,Cerny J, eds. 1994.Biogeochemistryof Small Catchments. Chichester, UK: Wiley& Sons. 419 pp.

Mulder J, Pijpers M, Christophersen N. 1991.Water flow paths and the spatial distribu-tion of soils and exchangeable cations in anacid rain-impacted and a pristine catchmentin Norway. Water Resour. Res. 27:2919–28

Mulholland PJ, Wilson GV, Jardine PM. 1990.Hydrogeochemical response of a forestedwatershed to storms: Effects of preferentialflow along shallow and deep pathways.WaterResour. Res.26:3021–36

Murdoch PS, Stoddard JL. 1992. The role ofnitrate in the acidification of streams in theCatskill Mountains of New York.Water Re-sour. Res.28:2707–20

Musgrave GW, Holtan HN. 1964. Infiltration.In Handbook of Applied Hydrology, ed. VChow, pp. 12-1–12-30. New York: McGraw-Hill

Neal C, Christophersen N. 1989. Inorganicaluminum-hydrogen ion relationships foracidified streams: the role of water mixingprocesses.Sci. Total Environ.80:195–203

Neal C, Hornung M. 1990. Transfer of ele-ments through the hydrologic cycle.J. Hy-drol. 116:1–2

Newell A, ed. 1993. U.S. Environmental Pro-tection Agency’s Long-Term MonitoringProject.Water Air Soil Pollut.67:247–478

Newton RM, April R. 1982. Surficial geologiccontrols on the sensitivity of two Adirondacklakes to acidification.Northeast Environ. Sci.1:143–150

Nikolaidis NP, Muller PK, Schnoor JL, Hu H.1991. Modeling the hydrogeochemical re-sponse of a stream to acid deposition usingthe Enhanced Trickle-Down Model.Res. J.Water Pollut. Control Fed.63:220–27

Norton SA, Kahl JS, Fernandez IJ, Rustad LE,Scofield JP, Haines TA. 1994. Response of theWest Bear Brook Watershed, Maine, USA, tothe addition of (NH4)2SO4: 3-year results.For. Ecol. Manage.68:61–73



58 CHURCH

O’Brien C, Hendershot WH. 1993. Separatingstreamflow into groundwater, solum and up-welling flow and its implications for hydro-chemical modelling.J. Hydrol.146:1–12

Obradovic MM, Sklash MG. 1986. An isotopicand geochemical study of snowmelt runoffin a small arctic watershed.Hydrol. Process.1:15–30

Ogunkoya OO, Jenkins A. 1993. Analysis ofstorm hydrograph and flow pathways us-ing a three-component hydrograph separationmodel.J. Hydrol.142:71–88

Oreskes N, Shrader-Frechette K, Belitz K. 1994.Verification, validation, and confirmation ofnumerical models in the earth sciences.Sci-ence263:641–46

Pearce AJ, Stewart MK, Sklash MG. 1986.Storm runoff generation in humid headwatercatchments, 1. Where does the water comefrom? Water Resour. Res.22:1263–72

Pilgrim DH, Huff DD, Steele TD. 1979. Useof specific conductance and contact timerelations for separating flow components instorm runoff.Water Resour. Res.15:329–39

Postek KM, Driscoll CT, Aber JD, Santore RC.1995. Application of PNET-CN/CHESS toa spruce stand in Solling, Germany.Ecol.Model.83:163–72

Ragan RM. 1968. An experimental investiga-tion of partial area contributions.Proc. Gen.Assem. Int. Assoc. Sci. Hydrol., pp. 241–49.Berne: Int. Assoc. Sci. Hydrol.

Rajan SSS. 1978. Sulfate adsorbed on hydrousalumina, ligands displaced, and changes insurface charge.Soil Sci. Soc. Am. J.42:39–44

Rasmussen L, Farrell EP. 1994. Experimentalmanipulations of biota and biogeochemicalcycling in forest ecosystems.For. Ecol. Man-age.68:1–126

Rastetter EB, King AW, Cosby BJ, HornbergerGM, O’Neill RV, Hobbie JE. 1992. Aggregat-ing fine-scale ecological knowledge to modelcoarser-scale attributes of ecosystems.Ecol.Appl.2:55–70

Reuss JO, Cosby BJ, Wright RF. 1987. Chemi-cal processes governing soil and water acid-ification.Nature329:27–32

Reuss JO, Johnson DW. 1985. Effect of soil pro-cesses on the acidification of water by aciddeposition.J. Environ. Qual.14:26–31

Reuss JO, Johnson DW. 1986.Acid Depositionand Acidification of Soils and Waters. Ecol.Series, Vol. 59. New York: Springer-Verlag.119 pp.

Reuss JO, Walthall PM. 1989. Soil reaction andacidic deposition. InAcidic Precipitation:Soils, Aquatic Processes, and Lake Acidifica-tion, Vol. 4, ed. SA Norton, SE Lindberg, ALPage, pp. 1–33. New York: Springer-Verlag.293 pp.

Richter DD, Markewitz D. 1995. How deep issoil? BioScience45:600–9

Rochelle BP, Church MR. 1987. Regional pat-terns of sulfur retention in watersheds of theeastern U.S.Water Air Soil Pollut.36:61–73

Rochelle BP, Church MR, David MB. 1987.Sulfur retention at intensively studied sites inthe U.S. and Canada.Water Air Soil Pollut.33:73–83

Rose S. 1996. Temporal environmental isotopicvariation within the Falling Creek (Georgia)watershed: implications for contributions tostreamflow.J. Hydrol.174:243–61

Saxena RK. 1986. Estimation of canopy reser-voir capacity and oxygen-18 fractionation inthroughfall in a pine forest.Nordic Hydrol.17:251–60

Schnoor JL, Palmer WD Jr, Glass GE. 1984.Modeling impacts of acid precipitation fornortheastern Minnesota. InModeling of TotalAcid Precipitation Impacts, ed. JL Schnoor,pp. 155–73. Boston: Butterworth. 222 pp.

Seip HM, Christophersen N, Mulder J,Taugbøl G. 1995. Integrating field work andmodelling—the Birkenes case. See Trudgill1995, pp. 387–415

Shanley JB, Peters NE. 1988. Preliminary ob-servations of streamflow generation duringstorms in a forested Piedmont watershed us-ing temperature as a tracer.J. Contam. Hy-drol. 3:349–65

Sklash MG. 1990. Environmental isotope stud-ies of storm and snowmelt runoff generation.See Anderson & Burt 1990, pp. 401–35

Sklash MG, Farvolden RN. 1979. The roleof groundwater in storm runoff.J. Hydrol.43:45–65

Sklash MG, Farvolden RN. 1982. The use ofenvironmental isotopes in the study of highrunoff episodes in streams. InIsotope Stud-ies of Hydrologic Processes, ed. EC Perry Jr,CW Montgomery, pp. 65–73. Dekalb, IL: N.Ill. Univ. Press

Sklash MG, Stewart MK, Pearce AJ. 1986.Storm runoff generation in humid headwa-ter catchments. 2. A case study of hillslopeand low-order stream response.Water Re-sour. Res.22:1273–82

Stevens DL Jr. 1994. Implementation of a na-tional environmental monitoring program.J.Environ. Manage. 42:1–29

Stoddard JL. 1994. Long-term changes in wa-tershed retention of nitrogen: Its causesand aquatic consequences. InEnvironmen-tal Chemistry of Lakes and Reservoirs, Ad-vances in Chemistry Series No. 237, ed. LABaker, pp. 223–284. Washington, DC: Am.Chem. Soc.

Swank WT, Crossley DA Jr, eds. 1988a.For-est Hydrology and Ecology at Coweeta. NewYork: Springer-Verlag. 469 pp.




Swank WT, Crossley DA Jr. 1988b. Introductionand site description. See Swank & Crossley1988a, pp. 3–16

Swank WT, Waide JB. 1988. Characterizationof baseline precipitation and stream chem-istry and nutrient budgets for control water-sheds. See Swank & Crossley 1988a, pp. 57–79

Swift LW Jr, Cunningham GB, DouglassJE. 1988. Climatology and hydrology. SeeSwank & Crossley 1988a, pp. 35–55

Thornton KW, Marmorek D, Ryan PF,Heltcher K, Robinson D. 1990. Methods forprojecting future changes in surface wateracid-base chemistry.NAPAP State Sci. Tech-nol., Rep. 14. Washington, DC: Natl. AcidPrecip. Assess. Prog. 162 pp.

Thornton KW, Wigington PJ. 1990.Water-shed Manipulation Project: Rationale forHypothesis Formulation and Testing. NTISPB91-159616, EPA/600/3-91/006, Wash-ington, DC: U.S. Environ. Prot. Agency,59 pp.

Trudgill ST, ed. 1986.Solute Processes. Chich-ester, UK: Wiley & Sons. 512 pp.

Trudgill ST, ed. 1995.Solute Modelling inCatchment Systems. Chichester, UK: Wiley& Sons. 473 pp.

Turner RS, Cook RB, Van Miegroet H, JohnsonDW, Elwood JW, et al. 1990. Watershed andlake processes affecting surface water acid-base chemistry.NAPAP State Sci. Technol.,Rep. 10. Washington, DC: Natl. Acid Precip.Assess. Prog. 167 pp.

USDA Forest Service. 1991.The HubbardBrook Ecosystem Study: Site Description andResearch Activities. Northeast. For. Exp. Stn.,NE-INF-96-91. 54 pp.

Van Sickle J, Baker JP, Simonin HA, BaldigoBP, Kretser WA, Sharpe WE. 1996. Episodicacidification of small streams in the north-eastern United States: Fish mortality in fieldbioassays.Ecol. Appl.6:408–21

Van Sickle J, Church MR. 1995. Methods for

estimating the relative effects of sulfur and ni-trogen deposition on surface water chemistry.EPA Rep. no. EPA/600/R-95/172. Washing-ton, DC: US Environ. Protect. Agency. 121pp.

Veen C, Buso D, Siccama T. 1994. Structure andfunction of the Hubbard Brook data manage-ment system.Bull. Ecol. Soc. Am.75(1):45–48

Wheater HS, Beck MB. 1995. Modelling up-land stream water quality: process identifica-tion and prediction uncertainty. See Trudgill1995, pp. 305–24

Wheater HS, Kleissen FM, Beck MB, TuckS, Jenkins A, Harriman R. 1990. Modelingshort-term flow and chemical response in theAllt a’Mharcaidh catchment. InThe SurfaceWaters Acidification Programme, ed. Sir BJMason, pp. 455–66. Cambridge: CambridgeUniv. Press

Wheater HS, Tuck S, Ferrier RC, Jenkins A,Kleissen FM, et al. 1993. Hydrological flowpaths at the Allt a’Mharcaidh catchment: ananalysis of plot and catchment scale observa-tions.Hydrol. Proc.7:359–71

Wigington PJ Jr, Baker JP, DeWalle DR, KretserWA, Murdoch PS, et al. 1996a. Episodic acid-ification of small streams in the northeasternUnited States: Episodic Response Project.Ecol. Appl.6:374–88

Wigington PJ Jr, Davies TD, Tranter M, Esh-leman KN. 1990. Episodic acidification ofsurface waters due to acidic deposition.NA-PAP State Sci. Technol., Rep. 12. Washington,DC: Natl. Acid Precip. Assess. Prog. 200pp.

Wigington PJ Jr, DeWalle DR, Murdoch PS,Kretser WA, Simonin HA, et al. 1996b.Episodic acidification of small streams in thenortheastern United States: Ionic controls ofepisodes.Ecol. Appl. 6:389–407

Wright RF, Tietema A, eds. 1995. NITREX:nitrogen saturation experiments.For. Ecol.Manage.71:1–169

hydrochemistry of forested catchmentshydrochemistry of forested catchments 25 figure 1...

Documents