18.1 Introduction

18.1.1 Rationale for using water and solute isotopes as tracers in catchments

Forested catchments provide a unique hydrogeologic setting in which to study the effects of processes such as climate change, acid deposition, deforestation, and encroaching urbanization. However, major uncertainties evident from attempts to model the hydrology and chemistry of forested catchments are the quantification of water delivered from various hydrologic flowpaths to stream flow, and the determination of weathering reactions that mobilize solutes along those flowpaths. Reliable tools for the assessment of both water and solute sources are essential if we are to fully understand the impacts of both natural and anthropogenic perturbations on catchment function.

Oxygen and hydrogen isotopes (i.e., "water isotopes") are now commonly used tools for determining water sources in catchments. The main use of O and H isotopes in catchment research is for determination of contributions of "old" and "new" water to stormflow (Sklash et al., 1976; Sklash and Farvolden, 1979; Bishop, 1991), where "old" water is defined as the water that existed in the catchment prior to a particular storm or melt event, and "new" water is defined as the water that triggered the event. The water isotopes are such effective tracers of water sources ("time-sources" in the terminology of Sklash et al., 1976) because oxygen and hydrogen are the constituents of, and thus move with water molecules. Under favorable circumstances, waters flowing along a particular flowpath may have distinctive O- and/or H-isotopic compositions; hence, the water isotopes can sometimes be viewed as tracers of water flowpaths. The processes that affect water isotope composition and the application of water isotopes to catchment studies are addressed in several chapters of this book, and the reader is referred to those chapters for discussion of the theoretical bases of the oxygen and hydrogen isotope systems.

In contrast, the solute isotopes (e.g., strontium, carbon, lead) are as yet under-utilized in catchment research compared to the water isotopes. Detailed analysis of the isotopic composition of Sr in stream waters, organic and mineral soil horizons, biomass, and atmospheric input at a handful of sites has provided insight into the cycling of Sr, and by analogy Ca and related nutrients in catchments (Graustein, 1989; Aberg et al., 1989; Gosz and Moore, 1989; Miller et al., 1993; Bailey et al., 1996). The stable C isotopes provide unique information on the origin of carbonate alkalinity and the relative importance of processes such as carbonate mineral dissolution, silicate hydrolysis, oxidation of organic matter and methano-genesis (Mills, 1988; Kendall, 1993; Katz et al., 1995). Pb isotopes can be used to distinguish shallow flowpaths, dominated by atmospheric Pb, and deep flowpaths, dominated by Pb from catchment minerals. However, Pb isotopes have been employed in only a handful of catchment studies (Sirahata et al., 1980; Erel et al., 1991; Bullen et al., 1994) and are not generally recognized as useful tracers in catchments.

Solutes do not necessarily move conservatively with the water, and thus solute isotopes can trace only solute sources, not water sources. They can, however, serve as useful tracers of flowpaths in the sense that different geologic units and their constituent minerals ("geographic-sources" in the terminology of Sklash et al., 1976) encountered along those flowpaths may have contrasting isotopic compositions. For example, flowpaths through the soil zone may be characterized by different chemical and solute-isotope compositions compared to those that typify flowpaths through the bedrock. In addition, solute isotope tracers are often used to test hypotheses developed from other criteria, such as interpretations of water chemistry or hydrology. In particular, they can be used to substantiate or eliminate possible geochemical reaction paths (Plummer et al., 1983). Because solute isotope ratios are usually affected by a smaller number of processes than chemical concentrations, interpretations of changes in isotopic composition are often less ambiguous than the simultaneous changes in water chemistry (Kendall et al., 1995).

The purpose of this chapter is to demonstrate that by synthesizing the information provided by several isotopic tracers, one can realize greater understanding at the process level than mere hypothesis confirmation. A "multi-isotope approach" can take advantage of the fact that unreactive tracers (such as H and O isotopes) and reactive tracers (such as Sr, Pb and C isotopes) provide very different information about how a catchment "works". For example, the water isotopes can be used to test the validity of a flowpath predicted from hydrologic measurements, and analysis of solute isotopes in waters collected along the flowpath can provide information about the minerals that are contributing to the water chemistry. In the absence of the solute isotope analyses, attempts to uniquely define the reactive mineral assemblage from water chemistry would likely be confounded by the mineralogic complexity of the catchment, even if both mass balance and reaction path chemical models are simultaneously applied (see Chapter 8). Likewise, in the absence of the water isotopes it would be difficult to constrain samples to a single flowpath, due both to the non-conservative nature of most solute species and to the inherent uncertainties resulting from assumptions in hydrologic models.

We view the Sr, Pb and C isotopes to be potentially the most widely applicable solute isotope tracers for catchment research, particularly when considered in light of constraints provided by the water isotopes. In this chapter, we present a discussion of the theoretical bases of the Sr, Pb and C isotope systems, establish the criteria for their successful application, and discuss the factors that influence solute isotope composition in catchment waters. We then present an overview of three catchment studies that employ a multi-isotope approach in addressing specific reaction path or flowpath issues. Finally, we discuss the theoretical bases and potential applications of several other solute isotope tracers.

18.1.2 Theoretical bases of the strontium, lead and carbon isotope systems

A general discussion of the geochemistry of isotopic tracers in catchment systems can be found in Chapter 2. In addition, an introduction to the application of Sr, Pb and other "lithogenic" isotopes to catchment studies is presented in Chapter 8. In this chapter, we present a more detailed synopsis of the fundamentals of the Sr, Pb, and C solute isotope systems, followed by a section on case studies using these isotopes. For the radiogenic Sr- and Pb-isotope systems, we stress the importance of the differential distribution of radiogenic Sr and Pb and their parent isotopes in catchment minerals and atmospheric/anthropogenic sources. For the stable C-isotope system, the discussion centers on processes affecting isotope compositions within both carbonate-bearing and carbonate-free catchments.

Rubidium (Rb) - strontium (Sr)
The alkali-earth metal Sr has four stable, naturally-occurring isotopes: ⁸⁴Sr, ⁸⁶Sr, ⁸⁷Sr, and ⁸⁸Sr. Only ⁸⁷Sr is radiogenic (i.e., produced by radioactive decay of another isotope) and gradually increases in minerals due to beta-decay of the radioactive alkali metal ⁸⁷Rb, which has a half-life of 48.8 x10⁹ years. Thus, there are two sources of ⁸⁷Sr in any material: that formed during primordial nucleo-synthesis along with ⁸⁴Sr, ⁸⁶Sr and ⁸⁸Sr, as well as that formed by radioactive decay of ⁸⁷Rb. The ratio ⁸⁷Sr/⁸⁶Sr, which provides a measure of the relative proportion of radiogenic to stable Sr, is the parameter typically reported in geologic investigations. The present-day ⁸⁷Sr/⁸⁶Sr of a Sr-bearing mineral is a function of the initial ⁸⁷Sr/⁸⁶Sr, the ratio of Rb to Sr (⁸⁷Rb/⁸⁶Sr), and the age of the mineral:

⁸⁷Sr/⁸⁶Sr _present-day = ⁸⁷Sr/⁸⁶Sr _initial + ⁸⁷Rb/⁸⁶Sr _present-day (e^lt -1)   (Eq. 18.1)

where l =1.42 x 10^-11 (ln2/half-life). ⁸⁷Sr/ ⁸⁶Sr in minerals and rocks ranges from about 0.7 to greater than 4.0 (Faure, 1972). As a general rule, carbonates and plagioclase feldspars contain the least radiogenic Sr, whereas K-feldspars and micas contain the most radiogenic Sr of typical reactive minerals in catchments.

Differences in the absolute proportion of radiogenic ⁸⁷Sr among mineral and other sources provide the diagnostic tracer of those source materials. Therefore, measured differences in ⁸⁷Sr/⁸⁶Sr in different materials due to stable isotope fractionation (primarily due to thermal effects in the mass spectrometer) must be corrected for. Fortunately, the stable isotope pair ⁸⁸Sr/⁸⁶Sr can be used as an internal monitor of both natural and analytical fractionation. In the absence of stable isotope fractionation, ⁸⁸Sr/⁸⁶Sr should be identical in all materials, with an accepted value of 8.37521 (IUGS: International Union of Geological Scientists). When the Sr (and other) isotopes fractionate, the amount of fractionation between pairs of isotopes is proportional to their mass ratios. Therefore, as a standard analytical procedure, ⁸⁷Sr/⁸⁶Sr is measured simultaneously with ⁸⁸Sr/⁸⁶Sr, and the measured ⁸⁷Sr/⁸⁶Sr is adjusted by an amount proportional to that required to correct the measured ⁸⁸Sr/⁸⁶Sr to the accepted value. Consequently, the reported ⁸⁷Sr/⁸⁶Sr is "fractionation-corrected".

The utility of the Rb-Sr isotope system results from the fact that different minerals in a given geologic setting can have distinctly different ⁸⁷Sr/⁸⁶Sr values as a consequence of different ages, original Rb/Sr values, and initial ⁸⁷Sr/⁸⁶Sr. For example, consider the case of a simple igneous rock such as a granite that contains the Sr-bearing minerals plagioclase feldspar, K-feldspar, hornblende, muscovite and biotite. If these minerals crystallized from the same magma, each mineral had the same ⁸⁷Sr/⁸⁶Sr_initial as the parent melt, but different relative proportions of Rb to Sr depending on the lattice-site preference for these two chemically contrasting elements. Typically, Rb/Sr increases in the order plagioclase, hornblende, K-feldspar, muscovite, biotite. Therefore, given sufficient time for production of radiogenic ⁸⁷Sr/⁸⁶Sr/86Sr_present-day will be different in the minerals, increasing in the same order. The evolution of variable ⁸⁷Sr/⁸⁶Sr in the granite minerals is shown schematically in Figure 18.1, a traditional Rb-Sr "isochron" diagram. Similar differences in ⁸⁷Sr/⁸⁶Sr should likewise occur among catchment minerals derived from multiple lithologic sources (e.g. in catchments nested in glacial deposits), although the order of increasing ⁸⁷Sr/⁸⁶Sr in the minerals may differ from that for the simple granite system. For example, the aggregate of hornblendes derived from numerous sources may have greater ⁸⁷Sr/⁸⁶Sr than the K-feldspar aggregate. Rigorous mineral separate work is required to confirm the order of increasing ⁸⁷Sr/⁸⁶Sr among minerals from any catchment.

The important concept for isotope tracing is that Sr derived from any mineral through weathering reactions will have the same ⁸⁷Sr/⁸⁶Sr as the mineral. Therefore, differences in ⁸⁷Sr/⁸⁶Sr among catchment waters require either differences in mineralogy along contrasting flowpaths, or differences in the relative amounts of Sr derived from the different minerals. The latter situation can arise in several ways. First, because waters approach saturation with respect to different minerals at different rates, progressive changes in water chemistry along flowpaths will cause parallel changes in the relative dissolution rates of, and the amount of Sr derived from the minerals (Plummer et al., 1983). Differences in factors such as pCO₂ and organic acid content of catchment waters may enhance the contrasts in dissolution rates. Secondly, differences in the relative mobilities of water at scales ranging from inter-grain pores to the catchment level may also profoundly affect ⁸⁷Sr^/86Sr (Bullen et al., 1996). For example, the chemical and Sr-isotope compositions of immobile waters at a plagioclase-hornblende grain boundary will be different from those at a quartz-mica grain boundary. Third, differences in the relative "effective" surface areas of minerals due to selective "poisoning" of reactive surfaces by organic and/or metal-hydroxide coatings will also cause differences in water chemistry and Sr-isotope composition.

In a fundamental sense, it is unrealistic to expect that waters sampled along contrasting flowpaths through a multi-mineralogic catchment should have similar ⁸⁷Sr/⁸⁶Sr. Rather, the waters moving along specific flowpaths react with the minerals in a characteristic manner and gradually approach a distinctive chemical equilibrium over long time-periods. In fact, the chemical composition of stream water in several catchments can be largely explained by the simple mixing of two or more chemically-distinct end members, each having constant composition (Hooper et al., 1990). For example, at Panola Mountain in the Georgia Piedmont, U.S.A. over 80% of the variability in Ca, SO₄, Na, Mg, SiO₂, and alkalinity in stream water can be explained by mixing of three waters, each having constant and distinct compositions: organic horizon water, hillslope water, and groundwater (Hooper et al., 1990). Given the range of water residence times typical of small catchments, waters within specific catchment units may have reached some sort of steady-state chemical composition, and thus might be expected to have steady-state and hopefully distinct Sr-isotope compositions.

For catchments nested in multi-mineralic materials, ⁸⁷Sr/⁸⁶Sr in any water parcel usually represents a mixture of Sr from several sources, and thus the exact contributions from individual minerals are difficult to determine with the Sr-isotope data alone. However, when considered in conjunction with water chemistry, the Sr isotopes provide a powerful tool for distinguishing among solute sources. For example, when attempting to develop a reaction model to explain the progressive increase of Mg in groundwater along a flowpath by the dis-solution of either hornblende or biotite, the Sr isotopes should provide an effective discriminant because Sr derived from hornblende will be substantially less radiogenic than that from biotite.

Once in solution, Sr should behave chemically in a manner similar to that of the alkaline-earth element Ca (Elias et al., 1982). As a cautionary note, however, we point out that ⁸⁷Sr/⁸⁶Sr is not in itself a sure indicator of Ca sources in a catchment because Sr can be contributed by both Ca-rich and Ca-poor minerals. For example, sodic plagioclase and K-feldspar contain little Ca but substantial Sr, and because of their relatively-high Rb/Sr contain more radiogenic Sr than calcic plagioclase. The isotopic composition of Sr derived from the simultaneous weathering of all three types of feldspars would have an intermediate value that could be similar, for example, to that in hornblende. Based on the Sr isotopes alone, one might erroneously conclude that the bulk of Ca in solution was derived from hornblende, even though dissolution of hornblende may have played only a minor role in water chemical evolution. Clearly the Sr isotopes should be closely integrated with water chemistry in order to identify the important contributing minerals along reaction paths.

A brief discussion of reporting notation is warranted here. Throughout the geologic literature, the isotopic composition of Sr in natural materials has traditionally been reported in terms of a four- to six-place decimal notation for ⁸⁷Sr/⁸⁶Sr (eg., 0.71024). For consistency within this volume, we use a "delta" notation for reporting isotope composition whenever possible. As with the standard notations for the oxygen and hydrogen isotope systems, the d-value for a sample relative to a standard is given by the equation:

d⁸⁷Sr = 1000 * [(⁸⁷Sr/⁸⁶Sr_sample) / (⁸⁷Sr/⁸⁶Sr_standard) -1]   (Eq. 18.2)

The problem with using the d-notation for the Sr-isotope system is that there is no consensus concerning what standard to use. Consequently, other researchers have chosen standards such as present-day sea water (Miller et al., 1993), local precipitation (Graustein, 1989), and "bulk earth" to suit their particular needs. Obviously the use of different standards makes inter-study comparisons difficult, although one could argue that the range of isotopic composition at any study site is generally the critical parameter. Regardless, in order to provide a more uniform basis on which to compare Sr-isotope composition between study sites, we propose here that the Sr-standard SRM987 be used as the standard for the delta calculation. This widely-distributed metal from the National Institute of Science and Technology (N.I.S.T.) is used by virtually every isotope laboratory in the world as the replicate standard. The decimal value obtained for the standard over the long term is typically reported in technical documents; in our lab, the value is 0.71024. Therefore, the general use of SRM987 as the basis for the delta calculation will allow direct comparison of both absolute values and ranges of Sr-isotope composition among study sites.

Uranium (U) - thorium (Th) - lead (Pb)
Pb has four stable, naturally-occurring isotopes: ²⁰⁴Pb, ²⁰⁶Pb, ²⁰⁷Pb, and ²⁰⁸Pb. ²⁰⁶Pb, ²⁰⁷Pb and ²⁰⁸Pb are all radiogenic, and are the end products of complex decay chains that begin at ²³⁸U, ²³⁵U and ²³²Th, respectively. The corresponding half-lives of these decay schemes vary markedly: 4.47 x10⁹ years, 7.04 x10⁸ years and 1.4 x10¹⁰ years, respectively. In geologic investigations, each radiogenic isotope is typically reported relative to ²⁰⁴Pb. Growth (decay) equations analogous to those for the Rb-Sr system can be written for the U-Pb and Th-Pb systems. The ranges of isotope ratio values for the majority of geologic materials are 14.0 to 30.0 for ²⁰⁶Pb/²⁰⁴Pb, 15.0 to 17.0 for ²⁰⁷Pb/²⁰⁴Pb and 35.0 to 50.0 for ²⁰⁸Pb/²⁰⁴Pb (Doe, 1970), although numerous examples of values outside these ranges are reported in the literature. Because of the three-component nature of the Pb isotope system, there is no simple means of converting the decimal values for isotope composition into a meaningful d-notation. Therefore, throughout this chapter Pb isotope compositions are reported in the decimal notation.

As with the Sr isotopes, corrections must be made for stable isotope fractionation both in nature and during analysis. However, there is no stable, non-radiogenic pair of Pb isotopes that can be used as the basis for a fractionation correction. To circumvent this problem, Pb isotope analyses are typically corrected for fractionation during analysis using factors determined by repeated analyses of some Pb standard, such as the N.I.S.T. SRM 981, 982 and 983 reference materials. However, there is no obvious means of correcting for possible fractionation in nature; thus, we assume that the Pb isotopes do not fractionate perceptibly in nature due to their high mass values. Although probably valid, the uncertainty associated with correcting for analytical fractionation renders Pb isotope analyses inherently less precise than those for Sr. Fortunately, the range of Pb isotope composition observed in nature is substantial.

The utility of the U-Th-Pb isotope system lies in the fact that in a single rock, individual minerals attain diagnostic Pb-isotope signatures due to long-lived differences in U/Pb, Th/Pb, and Th/U. An example is shown in Figure 18.2, which compares the uranogenic and thorogenic Pb-isotope compositions of primary alumino-silicate minerals separated from soils developed on granitoid alluvium near Merced, California (Bullen et al., 1997). Because feldspars incorporate Pb but do not have lattice positions for U and Th, they have relatively unradiogenic Pb-isotope compositions compared to those of hornblende and micas that both have lattice positions for the transuranics and incorporate U- and Th-rich trace phases. Hornblende tends to have greater Th/U, and thus greater ²⁰⁸Pb/²⁰⁴Pb at given ²⁰⁶Pb/²⁰⁴Pb than that in micas because of the preferential incorporation of trace phases such as Th-rich monazite in hornblende and U-rich zircon in biotite. Although resistant to weathering themselves, these trace phases pro-vide radiogenic nuclides to the lattices of their ferromagnesian hosts through the alpha-particle recoil processes that accompany spontaneous decay of ²³²Th and ²³⁸U. Because U, Th and Pb behave so different geochemically, the fact that the Pb isotope composition of any material is the composite of the three independent decay chains creates the potential for greater differences in isotope values between minerals of a single rock relative to that for the Rb-Sr system.

Carbon
Carbon has two stable, naturally-occurring isotopes: ¹²C and ¹³C. Neither isotope is radiogenic, and therefore differences in ¹³C/¹²C among minerals, waters and gases are the result of isotopic fractionations in various biogeochemical environments. Carbon isotope compositions are reported in a d-notation where:

d¹³C = 1000 * [(¹³C/¹²Csample) / (¹³C/¹²Cstandard) -1]   (Eq. 18.3)

The standard for computation of d¹³C is the Peedee Belemnite, abbreviated PDB. In general, analytical precision for d¹³C is about 0.1‰.

Different carbon-bearing materials in catchments can have characteristic d¹³C values. Carbonate rocks typically have d¹³C values of 0 ± 5‰; hydrothermal carbonates can be outside this range. Plants convert atmospheric carbon (d¹³C = -7‰) to organic compounds via photosynthesis. There is a bimodal distribution of the d¹³C values of terrestrial plants resulting from differences in the photosynthetic reaction utilized by the plant. C₃ plants (e.g. pine and apple trees) have d¹³C values of about -25‰ (range: -22 to -33‰), whereas C₄ plants (e.g. corn) have d¹³C values of about -12‰ (range: -10 to -20‰) (Bender, 1971; Deines, 1980). d¹³C of dissolved inorganic carbon (d¹³C_DIC) in catchment waters is generally in the range of -5 to -25‰. More negative values usually indicate the presence of oxidized methane.

The primary reactions that produce DIC are: (1) weathering of carbonate minerals by acidic rain or other strong acids; (2) weathering of silicate minerals by carbonic acid produced from the dissolution of biogenic soil CO₂ by infiltrating water; and, (3) weathering of carbonate minerals by carbonic acid. The first and second reactions produce DIC having identical d¹³C to that in the reacting carbonate or carbonic acid, respectively, whereas the third reaction produces DIC with a d¹³C value exactly intermediate between the compositions of the carbonate and the carbonic acid. Consequently, without further information, DIC produced solely by the third reaction is identical to, and cannot be distinguished from DIC produced in equal amounts from the first and second reactions.

If the d¹³C values of the reacting carbon-bearing species are known and the d¹³C_DIC of the stream determined, in theory we can calculate the relative contributions of carbonate minerals and carbonic acid to the production of stream DIC and carbonate alkalinity, assuming that: (1) carbonate mineral dissolution occurs under closed-system conditions (i.e. isolated from poten-tial reservoirs of CO₂ in the soil zone or atmosphere); and (2) there are no other sources or sinks for carbon (Kendall et al., 1992). With additional chemical or isotopic information, d¹³C values can be used to estimate proportions of DIC derived from the three reactions listed above. However, other processes that may complicate the interpretation of stream d¹³C values include CO₂ degassing, carbonate precipitation, exchange with atmospheric or soil CO₂, carbon uptake by aquatic organisms, methanogenesis, and methane oxidation (Kendall, 1993). Correlation of variations in d¹³C with changes in chemistry and other isotopes such as Sr, factors that will not be affected by these processes, may provide evidence that such processes are insignificant.

18.1.3 Geological/environmental factors leading to successful tracing with solute isotopes

Isotope hydrogeologists are often asked whether isotopic techniques can help answer specific research questions. Here we present some basic criteria that researchers can apply to their own catchments to determine whether solute isotopes will provide important information. First and foremost, for solute isotopes to be effective tracers of weathering reactions and water flowpaths, there must be significant contrasts in isotopic composition among catchment minerals and weathering solutions. The C isotopes are inherently excellent tracers of weather-ing reactions involving carbon-bearing rocks and/or acids. For Sr and Pb isotopes, age of the minerals and distinctiveness compared to atmospheric/anthropogenic inputs are the important considerations. For example, Sr isotopes would be a poor tracer of differential mineral weathering in a catchment set in a volcanic terrain so young that insufficient time has passed to allow development of significant differences in d⁸⁷Sr among catchment minerals. At the same catchment, however, d⁸⁷Sr could be a useful tracer for distinguishing contributions from weathering and precipitation. Secondly, the weathering solutions must be capable of carrying the tracer nuclides away from the weathering site. For example, even for a catchment set in an old granitic terrain, Pb isotopes would be an ineffective tracer if the weathering solutions were unable to mobilize the Pb liberated from the lattices of the weathered phases.

For Sr, C and Pb isotopes to be useful as tracers of water flowpaths, the rock-water reactions along these flowpaths must impart distinctive isotopic compositions to the waters. Isotopic compositions of waters developed along shallow and deep flowpaths are likely to be distinct because of the consequent differences in water chemistry and residence times. For example, in variably-weathered soil profiles, chemical evolution along shallow pathways is dominated by cation exchange and dissolution of remnant resistant phases such as K-feldspars and perhaps clays. In contrast, chemical evolution along deep pathways, perhaps at the saprolite-bedrock interface, is dominated by dissolution of more abundant reactive phases such as ferromagnesian minerals, plagioclase feldspar and calcite. A similar isotopic contrast between shallow and deep pathways could be established if the atmospheric/anthropogenic input to the catchment is iso-topically distinct from that due to weathering, and shallow flowpaths transport solutes derived primarily from atmospheric sources. Differences in mineral soil thickness could also promote variability in radiogenic isotope signatures developed along different flowpaths.

Differential access of water to reactive minerals along flowpaths is an additional consideration. For example, waters having different chemical composition can develop in response to variability of their residence times in a given portion of the catchment. Infiltration waters traveling slowly by tortuous matrix flow may inherit solutes from a different mineral assemblage than water that is flushed through the system more quickly via macropores, even though the paths of both waters are relatively shallow or traverse the same distance (Kendall, 1993). Likewise, different water compositions can develop in fractured rock, where the fractures might act as preferential conduits and may be lined by distinctive minerals such as calcite. The combined use of Sr and C isotopes may provide an especially powerful tracer of flow through this type of fracture system.

In order to assess the applicability of any solute isotope system as a tracer, each researcher must consider the issues being addressed in their particular catchment study in light of hydrologic and geochemical factors such as those above. Factors such as age of minerals and heterogeneity of lithologic sources, extent of development of the weathering profile, mobility of fluids and solutes through various portions of the catchment, and sources of atmospheric/anthropogenic inputs to the catchment are all important. Moreover, the researcher must recognize that conclusions based on application of a single isotope system are likely to be ambiguous, and thus use of the multi-isotope approach in concert with available chemical and hydrologic data is necessary to reduce the ambiguity.
 

18.2 Influences on Isotopic Composition of Sr, Pb and C in Catchment Waters

18.2.1 Lithologic controls on the isotopic composition of strontium and lead

From a solute perspective, catchments can be viewed as "reaction vessels" through which meteoric waters are processed. In the absence of gross lithologic variability, it is tempting to think that the weathering contribution of a catchment might have a unique, characteristic "signal", determined primarily by the particular mineralogy. However, even in catchments characterized by uniform mineralogic distribution, numerous lithologically-controlled factors such as mineral dissolution rate, cation exchange capacity, fluid mobility, and reaction kinetics at mineral surfaces together determine the net solute inventory added to an evolving water parcel along a flowpath (Bullen et al., 1996). In this section, we discuss how these factors affect the isotopic composition of Sr and Pb in catchment waters. The most fundamental observation is that a range of both Sr and Pb isotope compositions can be attained from a multi-mineralic source material because of isotopic variability among the constituent minerals. Therefore, one should not expect a single, characteristic isotope signal from weathering processes in a catchment. This point is particularly critical when attempting to estimate atmospheric/ anthropogenic inputs to the catchment.

Differential weathering rates of minerals
It is now widely recognized that minerals dissolve or react at different rates, depending on a variety of factors such as mineral composition and structure, and temperature and chemistry of weathering solutions. In a multi-mineralic system, these differential weathering rates should result in progressive changes in mineral proportions and net solute load with time and flowpath length. In theory, one could predict the isotopic composition of Sr and Pb provided by weathering based on the knowledge of the constituent minerals and their rates of dissolution. Unfortunately, estimation of weathering rates at individual catchments is problematic at best due to the assumptions inherent to the reaction models traditionally used to make those estimates (e.g., the estimates are no better than the assumptions made about parameters not easily measured). Moreover, there continues to be a nagging discrepancy between rate estimates calculated for weathering in the field compared to those determined in laboratory experiments (Paces, 1983; Velbel, 1985; White and Peterson, 1990). Although the reaction models are clearly becoming more powerful, they are still too simplistic. Likewise, although the laboratory experiments are becoming more complex, they still fail to accurately simulate the effects of long-term weathering under field conditions. A key unknown in all discussions of weathering is how dissolution rates of minerals change with progressive weathering and evolution of weathering solutions in a multi-mineralic catchment.

In an attempt to assess weathering in a simple granitoid system, White et al. (1996) and Bullen et al. (1997) reported on mineralogic, chemical and isotopic studies of a soil chronosequence from the Merced River drainage in central California (Harden, 1987). The soils range in age from 10 Ka to 3 Ma, and are developed on successive deposits of glacial alluvium derived from a geographically-restricted group of Sierra Nevada granitoid plutons. These alluvial deposits probably had similar mineralogies and bulk compositions, and their constituent minerals clearly had similar Sr- and Pb-isotope compositions prior to soil development. Hydrologically the system has probably been dominated by vertical infiltration of precipitation over the 3 Ma span of soil development.

Based on progressive changes in observed mineral proportions in six members of the chronosequence, mineral weathering rates apparently decrease in the order biotite, hornblende, plagioclase and K-feldspar. This order of mineral weathering is consistent with that determined in other studies of weathering in granitoid terrains (Lasaga, 1984). The calculated average dissolution rate for plagioclase (10^-19.9 mol/cm²/s) is actually greater than that calculated for hornblende (10^-20.1 mol/cm²/s), but the far greater reactive surface area on hornblende grains results in its greater effective weathering rate. K-feldspar is clearly the most resistant phase to weathering, and is the only primary alumino-silicate mineral remaining in the 3 Ma soils (White et al., 1996). The inferred order of mineral weathering is likewise supported by trends in the isotopic composition of Sr that is exchangeable from the soils using pH-buffered ammonium acetate. Exchangeable Sr is most radiogenic in the youngest soils, consistent with a preferential contribution of relatively radiogenic Sr from biotite, and becomes less radiogenic with soil age, reflecting a progressively greater contribution of relatively unradiogenic Sr from plagioclase and hornblende (Bullen et al., 1997).

On the other hand, the mineralogic and isotopic data suggest a more complex interpretation for the origin of Sr in the weathering solutions. For example, although biotite abundance decreases in the first 40 Ka of weathering, its abundance remains essentially constant through the remainder of the 3 Ma history. Furthermore, although biotite separated from the soils is the most radiogenic of the original granitoid phases, its d⁸⁷Sr is substantially less than, and its Sr content is greater than that in biotites separated from the granitoids themselves (compared to analyses reported by Kistler et al. (1986)). These observations suggest that biotite "weathering" actually involves uptake of relatively unradiogenic Sr from the weathering solutions. Further confounding the issue is the fact that although exchangeable Sr is most radiogenic in the youngest soils, its d⁸⁷Sr is consistently nearly identical to but slightly less than that in K-feldspars separated from the soils. One could argue that K-feldspar, the granitoid mineral most resistant to weathering based on observed mineral proportions, actually supplies the greatest proportion of Sr to weathering solutions in the youngest soils, perhaps due to leaching of defect sites on K-feldspar grain surfaces in the presence of dilute solutions (Bullen et al., 1997).

Another weathering scenario in which a presumably resistant mineral phase provides a greater solute contribution than supposedly more reactive phases occurs during the evolution of dilute groundwaters in a sandy silicate aquifer in northern Wisconsin (Bullen et al., 1993; Bullen et al., 1996). In this case, the early chemical evolution of groundwaters emanating from a dilute seepage lake is dominated by contributions from plagioclase, as evidenced by strong progressive increases in dissolved Na, Si and particularly Sr. However, after approximately 15 years of evolution along the flowpath, the rate of Na and Sr increase slows dramatically whereas Ca and Mg concentrations continue to rise, presumably as dissolution of relatively Sr-deficient calcic ferromagnesian phases begins to dominate. The parallel behavior of Na and Sr in this situation argues against a simple cation exchange mechanism to control their concentrations in the evolving waters. Furthermore, the progressive decrease in plagioclase contribution in favor of the ferromagnesian phases along the flowpath is consistent with both mass balance and reaction path models. Clearly the simple rules for the behavior of reactive phases in granitoid weathering systems must be applied with caution, particularly when attempting to interpret isotopic variations along a flowpath.

Reaction kinetics at mineral surfaces
Traditionally, the susceptibility of a given mineral to weathering has been estimated by its presumed or calculated extent of disequilibrium with specific weathering solutions. With respect to conditions at the earth's surface, the ferromagnesian silicate minerals are more out of equilibrium than the feldspars, and the plagioclase feldspars are more out of equilibrium than the K-feldspars. However, an increasing number of studies of reaction kinetics at mineral surfaces have documented that factors other than mineral composition are critical to the determination of mineral dissolution rates (Schott and Petit, 1987; see also reviews by White, 1990 and Blum, 1994). Of key importance is the concept that weathering solutions must be able to access the soluble ions within the mineral lattice or structure, and factors that either limit or enhance such access will have a pronounced effect on measured dissolution rates.

Based on calibrations using simple mineral-fluid systems, the degree of undersaturation of a mineral phase with respect to a complex fluid chemistry can be estimated using any of a number of geochemical computer codes (e.g., WATEQ (Truesdell and Jones, 1974), SOLMINEQ.88 (Kharaka et al., 1988), NETPATH (Plummer et al., 1991)). The mineral dissolution rate is presumably correlated with the degree of undersaturation, and is generally assumed to slow in an exponential manner as saturation is approached. However, recent experimental work in the system albite-H₂O has demonstrated that feldspar dissolution is more closely approximated by a step function, such that dissolution slows abruptly at some critical degree of undersaturation. The step decrease in dissolution rate is presumably related to a decreased ability of the fluid to form etch pits on the surfaces of feldspar grains (Burch et al., 1994). This sort of step function in part explains the fact that laboratory dissolution experiments using fresh mineral grains and distilled water progress at rates two to three orders of magnitude greater than those observed in field studies of weathering (White and Peterson, 1990). The obvious implication is that for a given catchment mineralogy, the relative contributions of solutes from the various minerals, and thus the isotopic composition of those solutes can change considerably as a result of subtle changes in the chemical composition of the weathering solutions, particularly in dilute systems.

Another factor that probably limits access of weathering solutions to a particular mineral phase is the formation of secondary minerals that use the surfaces of the primary minerals as growth platforms. In most silicate weathering systems, solutions quickly become saturated with respect to clay minerals and, under oxidizing and neutral to alkaline conditions, iron oxyhydroxides. Intuitively, if "poisoning" or "armoring" of mineral surfaces by these secondary minerals is to occur, the likely substrate for the initiation of secondary mineral formation will be the surfaces of the relatively unreactive primary minerals. In contrast, secondary mineral growth nuclei that attempt to attach to more reactive phases should be quickly disrupted as vigorous etch pitting of the surfaces of those reactive phases progresses (Velbel, 1993). Clearly, detailed optical and chemical examination of mineral surfaces along specific flowpaths in catchments is required in order to document the importance of preferential mineral surface "poisoning" or "armoring" as means of restricting access of weathering solutions to certain mineral phases.

Cation exchange
The cation exchange pool at any location along a flowpath reflects contributions from both mineral weathering reactions and atmospheric and anthropogenic inputs to the watershed. Presumably, mobile elements such as Sr are efficiently exchanged at mineral surfaces, and the ratio of a cation such as Sr to other exchangeable cations (e.g., Sr/Ca, Sr/Mg, Sr/Na) in the exchange pool remains relatively constant for given clay mineralogy, water pH and water chemistry; see Davis and Kent (1990) for a thorough review of the theoretical bases of cation exchange and mineral surface complexation processes. Once established, the exchange pool effectively buffers the water chemistry against minor fluctuations. Strong perturbations to the system, such as changes in pH caused by acid rain input, produce changes in both the concentration of Sr and other exchangeable cations, as well as the ratio of Sr to those cations in the exchange pool. Regardless, given sufficient time for a steady-state situation to develop, the fluid and the exchange pool should have identical d⁸⁷Sr. Unfortunately, there is currently little data on the time required to attain a steady state.

Departures from steady-state, such as the rapid movement of chemically-diverse infiltration waters into the unsaturated zone via macropores, can provide a situation in which the fluid and exchange pool will be out of exchange equilibrium. The system will then respond to restore a new equilibrium by mineral dissolution and ion exchange, although the time factor for Sr isotopes has not been determined experimentally. Regardless, catchment waters transported to a certain depth along macropores may have substantially different d⁸⁷Sr than waters that have migrated to the same depth via matrix flow, and may not be in Sr-isotopic equilibrium with the exchange pool at that depth. For example, our unpublished data for two clay-dominated catchments in Georgia and Puerto Rico show that the soil waters are not in Sr-isotopic equilibrium with the soil cation exchange pool at the sampling depths, as measured by ammonium acetate exchange. Moreover, a model for the temporal evolution of d⁸⁷Sr of the cation exchange pool in the Merced (CA) chronosequence soils discussed previously requires that exchange efficiency for Sr must decrease significantly as soil age increases (Bullen et al., 1997). Clearly, laboratory experiments to determine the time required for isotopic equilibration of Sr between fluids and the exchange pool in various soil types are necessary.

In contrast to Sr, Pb is strongly sorbed onto mineral surfaces and bound by transition metal-oxyhydroxides. Pb is transported as a solute only in low-pH solutions; in higher pH environments, it may be transported as organo-Pb complexes (Erel et al., 1991). Therefore, even though a relatively Pb-rich phase such as a feldspar may be reacting along a flowpath, the Pb from that feldspar has little tendency to enter solution unless pH is low or total organic carbon (TOC) is high; the Pb otherwise becomes immediately bound at surface sites or is incorporated in precipitating secondary phases. These characteristics make Pb difficult to use as a tracer of weathering reactions and water flowpaths in young systems in which solution pH tends to be near neutral and the cation exchange pool is developing rapidly in response to clay formation, unless TOC is particularly high. On the other hand, the Pb isotopes can be an especially powerful tracer for older, clay-dominated weathering systems in which the pH of soil solutions is relatively low, or for systems in which transport of Pb as organo-Pb complexes is an important process.

Fluid mobility in multi-mineralic matrices
As a dilute fluid migrates through a multi-mineralic matrix, solute concentrations increase in response to dissolution reactions. The thermodynamic stability of any mineral with respect to the chemistry of the fluid in large part determines the rate at which that mineral reacts. Any parcel of the fluid in contact with a given mineral will continue to scavenge solutes from that mineral as long as undersaturation persists. Stagnant pore fluids in contact with only a single reactive mineral such as a feldspar grain in a quartz-rich matrix can evolve chemically only to a limited extent determined by the saturation indices. On the other hand, more mobile fluids have a high probability of coming into contact with all reactive phases of the assemblage, and thus water chemistry will reflect contributions of solutes from several minerals. With progressive chemical evolution, the water soon reaches saturation with respect to minerals such as clays and iron oxyhydroxides, forcing their precipitation and removing solutes inherited from the reactant minerals. The formation of secondary minerals thus stimulates continued dissolution of the reactant minerals. Chemical evolution can thus continue to form relatively concentrated solutions.

In general, model calculations suggest that catchment waters are substantially more under-saturated with respect to ferromagnesian silicates than they are with respect to feldspars (Garrels and Mackenzie, 1967; Schott et al., 1981; Siegel and Pfannkuch, 1984), which supports the observation that feldspars are the most resistant reactive phases in silicate weathering systems over time. Thus, under stagnant conditions and neglecting potentially important factors such as mineral surface poisoning by secondary precipitates, fluid in contact with ferromagnesian minerals should evolve to a greater extent than fluid in contact with feldspars. Furthermore, the chemistry and solute isotope composition of the bulk stagnant fluid should reflect a greater contribution from the ferromagnesian phases than would a fluid formed under more mobile conditions. An additional complexity might arise if the catchment consisted lithologically not only of individual mineral grains but multi-mineralic rock fragments as well. Even under stagnant conditions, the multi-mineralic fragments would act as minute reactive centers around which protracted chemical evolution could occur (Bullen et al., 1996). When this stagnant fluid is later mobilized, the bulk fluid would have chemistry and solute isotope composition reflecting a greater relative contribution from the multi-mineralic fragments than would a fluid formed under mobile conditions. A case study that describes this phenomenon is presented in a later section.

18.2.2 Atmospheric/anthropogenic inputs of Sr, Pb, and C

Several investigators have reported Sr concentrations and isotopic compositions in precipitation at catchments (e.g. Gosz and Moore, 1989; Graustein, 1989; Miller et al., 1993; Bullen et al., 1996, 1997). Precipitation has measurable Sr, with concentrations on the order of several ppb and d⁸⁷Sr close to that of present-day seawater (d⁸⁷Sr = -1.4‰). Our current database of approximately 50 samples of precipitation from numerous watersheds throughout the United States and Puerto Rico show a range of d⁸⁷Sr from -3.50‰ to +1.50‰. Dry deposition also has measurable Sr, with d⁸⁷Sr reflecting that of particulates possibly derived from great distances from the catchment. The relative importance of external and internal contributions of Sr will obviously vary between catchments, and may be difficult to determine if the isotopic composition of atmospheric Sr overlaps that of internal catchment minerals.

Throughfall provides the main Sr input to the catchment surface, and is typically dominated by recycled Sr that has been taken up by plant roots and exuded at leaf surfaces (Bailey et al., 1996). In most catchments, Sr mobilized from the soil zone should be well-averaged, with d⁸⁷Sr reflecting the local extent of mineral weathering. For catchments in which weathering extent is spatially variable, d⁸⁷Sr of root uptake and thus of throughfall will likewise be variable. External sources of Sr are generally of minor importance for catchment waters, although significant contributions from aerosols have been documented; the extent of those contributions is dependent primarily on canopy type (Graustein and Armstrong, 1983; Gosz and Moore, 1989; Miller et al., 1993; Clow et al., 1997).

In contrast to Sr, Pb is both highly particle reactive and easily complexed by organic species (Davis and Kent, 1990). Dissolved Pb concentrations in catchment waters are thus extremely low, typically less than 1 ppb, and highly susceptible to contamination. Although necessitating extreme care during sample collection and laboratory preparation, these characteristics make Pb isotopes an especially sensitive tracer of atmospheric/anthropogenic inputs to a catchment. Pb derived from external sources enters the catchment either as dry deposition or in precipitation, and is effectively bound in the organic-rich, upper few centimeters of soil (Erel et al., 1991). Waters mobilized from this shallow zone will carry a small proportion of this Pb as organo-Pb complexes. If sufficiently distinct isotopically from catchment minerals and thus from waters developed in the mineral weathering zone, the atmospheric/anthropogenic Pb becomes a unique tracer of both infiltration waters and waters mobilized along shallow flowpaths through the O-horizon, as demonstrated for a headwater catchment in Vermont by Bullen et al. (1994). This application is analogous to the use of DOC as a tracer of shallow flowpaths (Hornberger et al., 1994).

Erel et al. (1991) took advantage of the different Pb isotope compositions of atmospherically-derived and rock weathering-derived Pb sampled at a Sierran stream to demonstrate that Pb in stream water during snow melt is supplied primarily from the organic soil accumulation reservoir. They further pointed out that whereas rock weathering-derived Pb should be relatively constant in isotopic composition over time, the isotopic composition and amount of atmospherically-derived Pb has changed markedly over time. This change is due primarily to the effective elimination of Pb additives to automobile gasoline since the 1960's. The success of efforts to reduce this source of Pb pollution is clearly recorded as progressively-lessened Pb concentrations in growth rings of corals from the western Atlantic Ocean (Patterson and Settle, 1984). Therefore, in addition to thorough analysis of potential lithologic sources of Pb, understanding the atmospheric-Pb deposition history at any catchment is essential to successful interpretation of the Pb isotope variations in catchment waters. Detailed isotopic analysis of Pb in throughfall and open-air precipitation at any catchment should help to differentiate Pb derived from local sources (i.e., dust from agricultural operations or nearby industrial emissions) from generally well-averaged regional/global "atmospheric" Pb.

The contribution of C as DIC from precipitation is negligible for most catchment systems. For rain with a pH of 4 and temperatures in the range of 10-25^oC, the equilibrium carbonic acid content is ~ 10-20 µmolar (Stumm and Morgan, 1980). If one makes the reasonable assumption that the carbonic acid is in isotopic equilibrium with atmospheric CO₂ (-7‰), it would have a d¹³C value of -8‰ (Deines et al., 1974). Because very little rain falls directly on the stream channel, as the rain flows over and through the soil the DIC has ample opportunity to exchange isotopically with soil CO₂. Therefore, the contribution of DIC from precipitation to streamflow is probably much less than the total atmospheric input. The effect of the atmospheric source of DIC is likely to be more significant in the winter rainy season than during the summer because of lower average DIC values of the stream, lower rates of CO₂ production in the soil, and higher ratios of runoff to infiltration in the winter and spring (Kendall, 1993).

On the other hand, relative contributions of DIC from carbonate minerals in dry deposition may be considerable, as has been postulated for several alpine catchments such as Loch Vale, Colorado (Clow et al., 1993). Unfortunately, it is unlikely that carbonates in dry deposition will have sufficiently different d¹³C than that of carbonates in catchment lithologies for d¹³C to be useful as a tracer of atmospheric input to the catchment. However, because carbonates typically contain considerable Sr, the combined use of C and Sr isotopes may provide a powerful means of distinguishing internal and external sources of carbonate contributions to catchment waters (Clow et al., 1997).

18.2.3 Effects of organic and inorganic cycling on isotopic composition of carbon

The biosphere, and particularly a number of soil processes have a tremendous influence on the d¹³C of DIC in catchment waters. Soil CO₂ is comprised mainly of a mixture of atmospherically derived (d¹³C= -7‰) and microbially-respired CO₂. Respiration is a type of biologic oxidation of organic matter, and as such produces CO₂ with approximately the same d¹³C value as the organic matter (Park and Epstein, 1961); hence, areas dominated by C₃ plants should have soil CO₂ with d¹³C values around -25‰. The d¹³C of soil CO₂ can also be affected by fermentation which produces methane ranging in composition from -52 to -80‰ (Stevens and Rust, 1982). As fermentation progresses, the compositions of the CO₂ or DIC byproducts become increasingly enriched in d¹³C (Carothers and Kharaka, 1980). The carbon produced would be more enriched in ¹³C than calcite, with values greater than +10‰ not uncommon. Oxidation of methane produces CO₂ with approximately the same composition as the original methane. Reduction of sulfate during degradation of organic matter having d¹³C of about -25‰ produces DIC of about -20‰ (Presley and Kaplan, 1968).

If DIC in soil water or groundwater exchanges isotopically with other carbon-bearing species, then the isotopic signatures may be blurred and conservative mixing of two distinctive end-member compositions cannot be assumed. The evolution of the isotopic compositions of carbon-bearing substances in uncontaminated systems where carbon is derived from carbonate minerals and soil CO₂ is controlled by two limiting cases: open systems where carbonate reacts with water in contact with a gas phase with a constant pCO₂, and closed systems where the water is isolated from the CO₂ reservoir before carbonate dissolution (Deines et al., 1974). The conditions under which carbonate is dissolved fall between these two extremes, both of which assume water residence times long enough for significant isotope exchange between the gas and the aqueous phase to take place.

The predominant carbon species at typical soil pH values of about 5 to 6 is carbonic acid. The equilibrium isotope fractionation between CO₂ and carbonic acid is 1‰ (Deines et al., 1974). DIC produced by the dissolution of calcite (d¹³C = 0‰) by carbonic acid (d¹³C = -22‰) has a d¹³C = -11‰. If this dissolution occurs under open-system conditions, the DIC would exchange with the soil CO₂ reservoir (-21‰) and reach a d¹³C value of about -22‰, thus eliminating any carbon isotopic evidence that half the DIC was derived from dissolution of calcite. Of course, if the d⁸⁷Sr of calcite were distinctive relative to the d⁸⁷Sr of other catchment minerals, dissolution of the calcite may have left its "signature" in the d⁸⁷Sr of the water.

The carbon in subsurface waters that flow into streams is not in chemical and isotopic equilibrium with the atmosphere. For example, CO₂ concentrations in the soil zone are often as high as 5%. Therefore, because the atmospheric concentration is about 0.03%, CO₂ is rapidly lost as soil water seeps into a stream bed. Laboratory experiments performed by Mook (1968) indicate that the d¹³C of DIC rapidly increases by about 0.5‰ during degassing. Furthermore, isotopic exchange between DIC and atmospheric CO₂ is inevitable. In streams with pH values of from 5 to 6 and temperatures of 20^oC, the equilibrium d¹³C of stream DIC should be around -8‰. Hence, if the residence time of water in the stream is long enough, the d¹³C of DIC will gradually approach -8‰. However, because there is usually no evidence of any increase in d¹³C downstream besides the 0.5‰ caused by degassing (Kendall, 1993) isotopic exchange between stream DIC and atmospheric CO₂ does not appear to be a problem in the first and second-order streams of forested catchments.

Additional in-stream processes can affect the d¹³C of DIC. Assimilation of DIC by aquatic organisms produces organic material with a composition about 30‰ depleted in ¹³C relative to the composition of carbon utilized (Rau, 1979), resulting in an increase in the d¹³C of the remaining DIC. In contrast, precipitation of calcite will cause a decrease in the d¹³C of the remaining DIC, due to the equilibrium fractionation between calcite and DIC of about 2‰.

If any of these soil-zone or in-stream processes are significant sources or sinks of carbon, they may complicate the interpretation of stream d¹³C values (Kendall, 1993). However, correlation of variations in d¹³C with changes in hydrology, chemistry, or other isotopes such as Sr may provide evidence that such processes are insignificant. For example, lack of any systematic increase in d¹³C downstream, particularly in the summer when flow is slow, would argue against significant exchange of stream DIC with atmospheric CO₂. Similarly, low pH of stream water would rule out precipitation of calcite as a means to decrease d¹³C of stream DIC. Finally, a strong positive correlation between d¹³C and DIC of stream water, and a typically negative correlation between these parameters and d⁸⁷Sr together argue convincingly for calcite dissolution.

18.3.4 Synthesis: an isotopic view of a catchment

The above studies demonstrate the kinds of isotopic variability to be expected in waters contributing to streamflow at catchments. The water isotopes have the potential to be as variable as the "new" waters delivered to the catchment during events, although some degree of groundwater averaging is likely to occur. In addition, evaporative enrichment of the water isotopes will contribute to the isotopic diversity of groundwater in catchments where lakes are a significant component. The important controls on solute isotope composition appear to be residence time of water in and thickness of the soil zone, length of flowpath to the stream and, in the case of Sr and C isotopes, the extent of silicate and carbonate mineral weathering. Figure 18.11 summarizes the expected variations in water and solute isotope composition for various generalized flowpaths through a hypothetical catchment.

In our view of the catchment, assumed here to contain a carbonate component, strong-acid weathering is expected to dominate where soils are thinnest, resulting in relatively heavy DIC and radiogenic dissolved Sr at the water table. As soils thicken, strong-acid weathering continues to dominate in the shallow soil horizons from which carbonate has largely been stripped, resulting in relatively light DIC and radiogenic dissolved Sr in infiltration waters. However, carbonic-acid weathering becomes important in the deeper carbonate-bearing mineral soils, resulting in intermediate DIC and less radiogenic dissolved Sr at the water table. In terms of Pb isotopes, shallow soil waters should be dominated by atmospheric-derived Pb, whereas deeper groundwater should be dominated by rock weathering-derived Pb. As soils thicken, downward transport of atmospherically-derived Pb should provide a mechanism to stratify groundwater with respect to Pb isotopes. Finally, contributions from bedrock are probably highly variable isotopically, depending primarily on the kinetics of water transport through the fracture system. In general, however, we suggest that groundwaters derived along bedrock pathways should contain relatively radiogenic Sr and Pb, and heavy C.

If we make the assumption that catchment waters develop and maintain characteristic isotopic "signatures" along the various flowpaths, then the water and solute isotope compositions observed in streamflow can be viewed as a function of contributing volume. For example, under baseflow conditions the characteristic isotopic signature of the riparian environment should dominate. However, in response to an event such as a rain storm or snowmelt, additional water will be delivered to the stream along flowpaths determined by the hydrogeologic characteristics of the catchment. Water and solute isotope compositions of the resulting streamflow will differ from those of baseflow depending on the relative mass fluxes from the various catchment "volumes" such as the organic soil horizon, shallow groundwater, and the bedrock fracture system. Although admittedly simplistic, this view of isotopic behavior in catchments is clearly testable by a variety of hydrologic, chemical and isotopic techniques. We view hypothesis testing as a critical component of any study of catchment hydrology.

18.4 Additional Solute Isotope Tracers: Li, B, Fe

From the material presented above, it is clear that Sr, Pb, and C isotopes can provide meaningful constraints for catchment studies, due to both their versatility as hydrologic tracers as well as the relatively straightforward nature of their analysis. It is worth pointing out that the vast majority of solid-source mass spectrometers currently in operation are or could be routinely used to analyze the isotopic compositions of Sr and Pb in rocks, and the transition from analyses of rocks to analyses of dissolved components in water requires only the development of laboratory procedures for separation and purification of the Sr and Pb for loading on filaments. Numerous academic and research facilities currently have the capability to analyze the isotopic composition of C in dissolved species, solids, liquids, and gases. Other solute isotope systems have been advantageously applied in catchment studies, notably the uranium- and thorium-series radionuclides (see review, Chapter 20), the sulfur isotopes (see review, Chapter 15), and the nitrogen isotopes (see review, Chapter 16). The considerable geologic and biologic literature on the behavior of these isotope systems in nature provides a strong foundation for interpretation of variations in isotopic composition observed in catchments with diverse geology and ecosystems.

In the near future, additional solute isotope systems that are currently under-utilized due to both analytical difficulties and lack of information concerning the causes and extent of isotopic variations in nature may potentially find wide applicability in catchment studies. As examples, in this section we present the theoretical bases for the Li, B, and Fe stable isotope systems. These systems have not yet been employed in catchment studies, but should provide important information particularly when applied through the multi-isotope approach. We propose that the Li and B isotopes should behave as relatively conservative tracers of water flowpaths, and should prove valuable for distinguishing contributions from weathering in the soil zone from those due to weathering at the saprolite-bedrock interface. The Fe isotopes may provide a sensitive tracer of biogeochemical redox-sensitive processes occurring within the catchment.

Lithium isotopes
Lithium has two naturally-occurring stable isotopes, ⁶Li and ⁷Li. Unlike most other stable isotope systems, the heavy isotope ⁷Li is the dominant species, such that average ⁷Li/⁶Li in nature is approximately 12.0. In the limited literature that exists for the Li isotopes, data have typically been reported relative to the N.I.S.T. standard L-SVEC in common delta notation. Considerable variability of d⁷Li has been reported for natural materials, although many of the analytical data produced prior to the 1980s are now suspect. Because Li is such a light element, analysis in solid-source mass spectrometers has been difficult due to apparently uncontrollable thermal fractionation effects. However, recently developed alternative analytical procedures (e.g. Chan and Edmond, 1988) have made the precise analysis of the Li isotope composition in both rocks and waters relatively routine. As with any stable isotope system, however, the only true means of assuring accuracy of measurements is by running duplicates of each sample through the total analytical procedure and confirming a set level of external precision.

In nature, layer silicate minerals cause the greatest Li-isotope fractionation, as ⁶Li preferentially substitutes into the octahedral sites usually occupied by Mg and Fe²⁺. This fractionation mechanism was demonstrated for high-temperature systems by Chan and Edmond (1988) who reported isotopic compositions of Li in mid-ocean ridge thermal waters and associated solid phases. In their study, hydrothermally-produced secondary minerals were calculated to have d⁷Li as much as 19‰ lighter than the fluids with which they equilibrated. Continual hydrothermal processes at the ocean ridges and preferential accumulation of ⁶Li into secondary phases over geologic time have caused a progressive heavying of Li of seawater such that it has the highest d⁷Li of natural materials yet analyzed.

Bullen and McMahon (1992) used the anticipated divergence in d⁷Li between marine and layer-silicate environments to track the chemical evolution of groundwater in a clastic coastal aquifer. Along a regional flow path, dilute recharge waters ultimately evolve to NaHCO₃-compositions, presumably first by dissolution of aragonite in the sandy portion of the aquifer and later by Ca-for-Na exchange on clays. Consistent with this scenario, Li is relatively light in the recharge zone, becomes substantially heavier as (HCO₃)--concentrations increase, and becomes markedly lighter in the clay-rich portion of the aquifer where Na/Ca of groundwater climbs rapidly. Based on analyses of solid-phase materials from cores taken along the flow path, Li in aragonite is ~ 35‰ heavier than that leachable from the clay-rich aquifer material.

Based on the conclusions of these studies, some simple predictions concerning the behavior of Li isotopes in catchments can be made. Obviously, clay minerals will play an important role in the establishment of Li isotopic diversity. Clays will contain the lightest Li of catchment minerals, and catchment waters will have the most Li isotope diversity when the effect of interactions with clay minerals differs among flowpaths. For example, infiltration waters passing through an established clay-rich soil should inherit substantially lighter Li than will groundwaters traveling along and reacting at the bedrock-saprolite interface. Furthermore, dissolved Li should become progressively heavier along flow paths where precipitation of Mg-Fe²⁺-Li-bearing clay minerals is actively occurring. An additional Li source particularly relevant to coastal catchments is marine-dominated precipitation, which will have substantially heavier Li than the catchment minerals. Assuming that isotopically-distinct sources or environments can be identified within a catchment, then Li in waters derived therefrom should behave as a relatively conservative solute tracer of those waters.

Boron
Boron has two naturally-occurring stable isotopes, ¹⁰B and ¹¹B. Unlike most other stable isotope systems but in common with the Li isotopes, the heavy isotope ¹¹B is the dominant species, such that average ¹¹B/¹⁰B in nature is approximately 4.0. An extensive literature exists for the B isotopes, particularly for geothermal and marine environments. In most studies, data have typically been reported relative to the N.I.S.T. standard NBS-951 in common d-notation. As with the Li isotopes, considerable variability of d¹¹B has been reported for natural materials (e.g. Bassett, 1990). However, many of the analytical attempts at precise determination prior to the 1980s are now suspect for the same reasons as for the Li isotopes. The fairly recent development of two alternative analytical procedures has made the analysis of B isotopes in water fairly routine; the analysis of B isotopes in silicate rocks is rather difficult and a highly-specialized technique. The two methods involve measurement of either positive ions (Spivack and Edmond, 1986) or negative ions (Vengosh et al., 1991). The advantage of the negative ion analytical method, particularly for dilute systems such as catchments, is that the required sample size (~1 ng B) is substantially less than that for positive ion generation (~1 µg B). On the other hand, reproducibility of measurements is presumably worse with the negative ion analytical method, but with careful sample preparation technique can be better than 1‰.

In natural waters, B exists in two states, boric acid (H₃BO₃) and borate ion (H₄BO₄)^-. The coordination environment for the B atom is affected by the difference in number of hydroxyl groups in these boron species, causing isotopic fractionation between coexisting boric acid and borate ion. Under low temperature conditions such as the marine environment, B in the borate ion is on the order of 20‰ lighter than that in boric acid (Hemming and Hanson, 1992). As an uncharged species, boric acid is essentially conservative; in contrast, the charged borate ion can be sorbed onto mineral surfaces and incorporated as an impurity into minerals such as carbonates. As with the Li isotopes, the progressive development of layer silicate-rich hydrothermal mineral assemblages at mid-ocean ridges and the abundance of clay minerals in the marine environment has resulted in the progressive heavying of B in seawater over geologic time. The range of d¹¹B in natural materials is on the order of 100‰ (Bassett, 1990).

The distribution of B-species as a function of pH may be particularly relevant to catchment studies. At pH less than 6, dissolved B exists entirely as boric acid. With increasing pH, the isotopically-lighter borate ion becomes progressively more abundant and heavier (Kakihana et al., 1977). Thus, there could be considerable variability in d¹¹B of dissolved B along flowpaths through different mineral assemblages, or along flowpaths characterized by different pH and geochemical evolution. The main control on d¹¹B of catchment waters is probably the effectiveness of borate ion sorption as a function of pH and chemistry. Additional sources of isotopically unique B that may be important are carbonates from within the catchment and atmospherically-deposited salts of either marine or anthropogenic origin (i.e., boron-rich soil amendments, fertilizers).

Iron
Iron has four naturally-occurring stable isotopes, ⁵⁴Fe, ⁵⁶Fe, ⁵⁷Fe and ⁵⁸Fe. The relative abundances of the Fe isotopes in nature are approximately ⁵⁴Fe (5.8%), ⁵⁶Fe (91.7%), ⁵⁷Fe (2.2%) and ⁵⁸Fe (0.3%). Much of the past work on measuring the isotopic composition of Fe has centered on determining variations due to processes accompanying nucleosynthesis (i.e., meteorite studies) and ore formation (Volkening and Papanatassiou, 1989; Maeck, 1992). However, primarily for analytical reasons, there is a paucity of information on stable isotopic variations of Fe in low-temperature systems.

Two recent studies using improved analytical techniques are pertinent to potential work with Fe isotopes in catchments. In a reconnaissance study of Fe isotope compositions, Dixon et al. (1992) demonstrated isotopic variations induced by microbial processes at the Loihi seamount. Their results suggest that microbially-mediated Fe reduction favors the breakage of bonds involving the lighter iron species. In another study of Fe reduction and transport at a gasoline spill site, Bullen and McMahon (1997) demonstrated with both field and experimental evidence that microbially-mediated Fe reduction behaves as a Rayleigh distillation process, with Fe²⁺ consistently 5‰ lighter than the coexisting Fe³⁺. An important conclusion of this study is that the Fe isotope composition of water leaving a system can be used to estimate the amount of Fe reduction that has occurred in the system, with increasingly heavy values reflecting greater amounts of Fe reduction.

The potential application of Fe isotopes to catchment studies lies in the assumption that Fe mobilized inorganically from minerals under either reducing or low-pH conditions will have a different isotopic composition than microbially-reduced Fe. To the extent that certain zones or flowpaths in the catchment can be characterized by microbial cycling of labile Fe, the Fe isotopes may provide an effective tracer of contributions from these pathways. For example, characterization of the biogenic processes along water pathways providing the Fe responsible for staining of stream beds may be one application of Fe isotopes to catchment studies. However, further experimental work is required on both development of analytical procedures for measurement of the Fe isotopes at low concentrations as well as determination of the relative fractionation efficiency of various microbial agents.

18.5 Summary

Water isotopes are very useful tools for determining water sources (time-sources) in catchments. Chemical tracers are very useful for understanding the reactions along flowpaths. However, neither isotopic nor chemical tracers are very useful for quantifying amounts of water from different flowpaths or rock/soil units--the former because water sources need not correspond to water flowpaths, and the latter because chemical constituents need not behave conservatively. Clearly, attempts to assess relative contributions from flowpaths must consider a variety of hydrologic, geochemical and isotopic tools in concert, and may not be successful.

Fortunately, reactive solute isotopes such as ¹³C, ³⁴S, ¹⁵N, ⁸⁷Sr, and Pb isotopes can provide valuable information about flowpaths for geochemical and hydrologic modeling precisely because they reflect the reactions that are characteristic of specific flowpaths (Kendall et al., 1992; Kendall et al., 1995; Bullen et al., 1996). First, reactive solute isotopes can serve as additional thermodynamic constraints for testing of possible geochemical reaction paths (Plummer et al., 1983). Second, the waters flowing along mineralogically-distinctive horizons are sometimes uniquely labeled by the isotopic compositions of their solutes. For example, waters flowing through the soil zone often have d¹³C values that are depleted in ¹³C relative to deeper groundwaters because of biogenic production of carbonic acid in organic soils (Kendall, 1993). Similarly, shallow soil waters should transport Pb that is predominantly atmospheric in origin, whereas deeper groundwaters will transport rock weathering-derived Pb (Bullen et al., 1994). Third, a multi-isotope approach can be particularly useful in tracing the reactions specific to a flowpath (Krabbenhoft et al., 1994). Because solute isotope tracers are usually affected by a smaller number of processes than chemical constituents, interpretations of changes in isotopic composition are less ambiguous than the simultaneous changes in solute concentrations (Kendall et al., 1995).

In this chapter we have stressed the importance of using isotopes in a complementary manner, primarily to constrain and enrich models developed from hydrologic and chemical data. Clearly isotopes are powerful tracers of hydrologic and geochemical processes, but often conclusions based on consideration of a single isotope system are ambiguous at best. One strong benefit of the multi-isotope approach discussed herein is that the strengths of one isotope system may compensate for the weaknesses of another. This is not to say that all catchment studies will benefit from analysis of every analytically-available isotope system, and each researcher must carefully determine the kinds of information desired from isotopic data. Isotopes are probably best viewed as tools for testing rather than developing hypotheses, particularly in studies operating under tight budgetary constraints. However, our experience is that isotopes consistently expand our vision of catchment function, and commonly lead to unanticipated avenues of research.

References