Rivers and streams are the major routes of transfer of phosphorus to oceans and to many lakes. Fluvial transport depends on the topography, climate and land uses in catchments and on hydrological and ecological processes in rivers, their floodplains and associated lakes. Rivers are more than conduits and function as ecosystems that store, release and transform dissolved and particulate materials (Likens, 1984; Meyer et al., 1988; Lewis et al., 1990). Rivers and streams integrate the heterogeneous losses of P from their catchments, and processes within flowing water modify the forms of P and the rates of transport. Ecological and hydrological conditions within streams and rivers and within associated lakes and reservoirs can alter the timing, magnitude and chemical form of P fluxes (Meyer et al., 1988; Thornton et al., 1990).
The mining of phosphate and its agricultural, industrial and domestic uses have increased during the last few decades. Other activities of modern societies such as clearing of forests, extensive cultivation and urban waste disposal and drainage systems have enhanced the transport of P from terrestrial to aquatic environments. As a consequence of these activities, concentrations of P in rivers and supply of P to lakes and estuaries have increased (Stumm, 1972; Smith et al., 1987). In most lakes and in some marine habitats enrichment with P increases plant growth (Schindler, 1977; 1978; Howarth, 1988; Hecky and Kilham, 1988) and can lead to reduced water quality and other environmental problems. The role of P in the eutrophication of aquatic systems is reviewed by Fisher et al. and Grobbelaar and House (Ch. 16).
Alterations in the inputs of P to aquatic habitats can have important effects on the chemical cycles of other elements such as carbon, nitrogen, sulphur and iron (Schindler 1985). Increased rates of photosynthesis associated with P-enhanced plant growth can increase CO2 invasion from the atmosphere. Phosphorus enrichment can reduce the N:P ratio which can favour growth of nitrogen fixing cyanobacteria. Greater amounts of plant biomass resulting from P enrichment can lead to augmented respiration rates and development of anoxic waters in deeper portions of lakes. Anoxic conditions favour methanogenesis, sulfate reduction, production of ammonium, and release from sediments of ferrous iron.
This chapter examines the hydrological, chemical and ecological processes operating within fluvial systems and their associated lentic systems that regulate movement of P. Estimates of the transport of P by rivers to inland waters and oceans are reviewed.
Chemical forms of phosphorus
In most studies of rivers and standing waters, the forms of P are operationally defined based on analytical practicality. The distinction between particulate and dissolved P is made by the porosity of the filter used to separate the two fractions; filters with porosities around 0.5 µm are commonly used. Total dissolved P (TDP) is often divided into soluble reactive P (SRP) which can sometimes be considered dissolved inorganic P, and dissolved organic P. Similarly, total particulate P (TPP) is determined as particulate inorganic P and particulate organic P (POP). Recently, considerable effort has been devoted to improve analytical capabilities and understanding of the physical, chemical and biological meaning of the measurable P fractions; much of this work is reviewed in Boström et al. (1988b), Broberg and Pettersson (1988), Broberg and Persson (1988), Pettersson et al. (1988), Froelich (1988) and Engle and Sarnelle (1990).
Discharge-concentration relations
Transport of P in streams and rivers depends on concentration and discharge, and this relation varies among flowing waters and for different forms of P. In a comparison of concentration versus discharge relations for streams ranging in size from first to ninth order and in location from the tropics to the Arctic, Meyer et al. (1988) found that soluble reactive P concentrations increased, decreased or remained constant and that total P consistently increased as discharge increased. However, because a large fraction of the total P is associated with particles, P transport occurs disproportionately during high flows.
A well documented example of concentration versus discharge relationships is found in first-order streams located in the forested catchments of the Hubbard Brook watershed (New Hampshire, USA). As discharge varied over four to five orders of magnitude, concentrations of SRP remained relatively constant; hence, transport was directly proportional to discharge (Likens, 1985). In contrast, concentrations and transport of particulate P increased exponentially with discharge and about 82% of the P in the streams was in the particulate fraction (Meyer et al., 1981). Hence, transport of particulate P was strongly related to the occurrence of storms which were irregular in frequency and magnitude. For example, Bormann et al. (1974) calculated that 86% of the total particulate matter was exported in 1.6% of the time and with 23% of the water. Meyer and Likens (1979) found that export of P varied from year to year as a function of streamflow. In years with large storm flows exports of P exceeded inputs to the stream, but in average years with few or small floods, inputs were stored or balanced outputs.
Concentration versus discharge relations for a montane stream receiving seasonal snowmelt runoff (Como Creek, Colorado, USA) were similar among three years with low, moderate and high discharge (Lewis and Grant, 1979). Dissolved organic P had no relation to discharge, while SRP increased with increased discharge. Yield, the product of concentration and discharge, increased faster than discharge for both SRP and dissolved organic P.
In regions with strongly seasonal rains such as southern and sahelian Africa, almost all the export of P can occur in brief periods often at the onset of the rainy season (Thornton, 1986). Discharge versus concentration relations in the Gambia River draining western African savanna differ among the forms of P (Lesack et al., 1984). The clockwise hysteresis (advance peak) of particulate P as river discharge rose and fell was similar to that for total particulates. In contrast, the increase in concentration of total dissolved P with increasing discharge is perplexing. The simple explanation that the organic matter on the river's edge was mobilized during the initial stages of the annual flood is inconsistent with the counter-clockwise hysteresis (lag) with discharge for dissolved organic carbon. The Apure River of northern South America also drains a savanna, and both particulate P and total dissolved P increased in concentration as discharge rose (Saunders and Lewis, 1988). Over the full range of discharge, SRP and total dissolved P were described by hyperbolic functions of discharge, but particulate P had no consistent relation to discharge.
Retention in streams and rivers and associated lakes and reservoirs
On an annual basis streams usually retain only a small fraction of the dissolved and particulate nutrients that enter from the catchment. Temporary or seasonal retention and recycling of nutrients occurs by several physical, chemical and biological mechanisms and is critical to the development of lotic ecosystems (Meyer et al., 1988; Triska et al., 1989a; Svendsen and Kronvang, 1993). Solutes can be retained in side channels and in the hyporheic zone (Bencala, 1984, Triska et al., 1989b) and sorbed to surfaces and taken up by epilithon (Meyer, 1979). Uptake can be enhanced by grazing (Mulholland et al., 1983). Storage of particles in freely flowing streams depends on accumulations of terrestrial organic debris (Bilby and Likens, 1980, Webster et al., 1988). The overall result of these processes is influenced by a mosaic of patches composed of different substrata with different biota (Pringle et al., 1988). Retention is controlled by factors including discharge, current, temperature, solute concentration, light, lithology of sediments and riparian vegetation (Triska et al., 1989a, Munn and Meyer, 1990).
To quantitatively account for the simultaneous processes of transport and recycling of nutrients in streams, a measure called the spiralling length has been developed (Newbold et al., 1981, Ellwood et al., 1983). Spiralling length is defined as the downstream distance traveled by a molecule as it completes one cycle of uptake and transformations from the dissolved inorganic form to various organic forms and back to the overlying water. Comparative measurements of spiralling lengths for P are available for only a few streams.
The first analysis of spiralling was conducted in Walker Branch, a first-order woodland stream (Tennessee, USA), and was based on the release of 32PO4 into the stream (Newbold et al., 1983). The uptake of the radioactive P from the water and its dynamics in fine particulate organic matter (FPOM), coarse particulate organic matter (CPOM), epilithon, invertebrate consumers and predators was followed for six weeks. The spiralling length was determined to be 190 m and was partitioned into transport in the water (165 m), transport in FPOM and CPOM (25 m), and transport associated with drift of aquatic animals (0.05 m). In a later study, Mulholland et al. (1985) determined seasonal variations in P spiralling in Walker Branch and found that the total spiralling length ranged from 23 m in November, just after the peak in leaf fall, to 99 m in August. Almost all the transport was in the dissolved P fraction, and the greatest increase in spiralling length occurred in late autumn or winter after storms reduced CPOM. For short periods during storms, the average distance traveled by an atom of P in the particulate fraction may increase one to two orders of magnitude above the more typical distances of 1 to 3 m.
There are often large differences in spiralling length along the length of a stream and between streams. Munn and Meyer (1990) found large differences in uptake or spiralling lengths in stream reaches with different habitats in two second-order mountain streams (Hugh White Creek, North Carolina, USA; WS2 stream, Oregon, USA). Uptake lengths of P ranged from 32 m in a cobble reach to 188 m in a gravel reach of Hugh White Creek, and from 188 m at a rock outcrop to 666 m in a cobble reach of WS2 stream. Average uptake lengths weighted for the relative abundance of each habitat were 85 m (Hugh White Creek) and 687 m (WS2) and primarily reflected differences in geology. Hugh White Creek drains deeply weathered granitic bedrock while WS2 drains exposed volcanic bedrock known to have high P export. Munn and Meyer (1990) also estimated uptake lengths for two habitats in Bear Brook (Hubbard Brook Experimental Forest, New Hampshire, USA), and found lengths ranging from 5 to 21 m on a rock outcrop with abundant moss and from 10 to 67 m for a cobble-riffle site. The mechanism of P uptake was primarily abiotic and was controlled by sediment particle size and Al content (Meyer, 1979).
In contrast to the northern temperate streams, streams in Australia experience leaf fall year-round with maximum leaf fall when temperatures are highest and flows are lowest in summer. However, the mean spiralling length of 84 m measured in a second-order stream (Myrtle Creek, Victoria, Australia) was similar to lengths in northern temperate streams (Hart et al., 1992). Almost all of the P retained in the creek was taken up by the microbes and physico-chemical sorption in the sediments rather than by storage in the hyporheic zone.
Spiralling lengths have not been calculated for large rivers where floodplains may retain P as particles decant from flood waters and as organisms strip P from the water. Lewis et al. (1990) showed negligible net export of P from the Orinoco floodplain to the river. Fisher et al. (1991) found that a floodplain lake on the Amazon River was a net source of P to the river in one year. However, the net, long-term effect of floodplains on downstream transport in major rivers is largely unknown.
Impoundment of rivers by man-made dams reduces downstream transport of suspended sediments and of P associated with the seston. As of 1971 about 12,000 dams exceeding 15 m in height had been built world-wide and were impounding 4000 km3 of water (Petts, 1984). Since 1971, world-wide dam construction has been about 700 per year (Mermel, 1981) and has included structures on large tropical rivers. By the year 2000 it is likely that the majority of the world's stream-flow will be controlled by dams (Petts, 1984). Although the proportion of sediment input released through dams varies considerably, on average only about one quarter of the sediment yield from the catchment to the reservoir is released downstream (Petts, 1984). Because in many rivers the upper basin provides the majority of the sediment load, dams can trap most of the sediments leaving a catchment. Furthermore, higher sedimentation rates usually occur in reservoirs than in natural lakes (Canfield and Bachman, 1981; Higgins and Kim, 1981). However, erosion of channels below dams can be increased by flow variations and greater scouring by the sediment-depleted waters (Petts, 1984). In addition, biological production of organic matter within reservoirs can add to the particulate matter in rivers below dams.
Net retention of P in lakes and reservoirs has been determined from detailed measurements of mass balances in intensively studied lakes (e.g. Schindler et al., 1977; LaBaugh, 1985; Kleiner and Stabel, 1989; Caraco et al., 1992) and from input-output estimates for numerous lakes in a region (e.g. Canfield and Bachmann, 1981). Although a number of factors are known to influence sedimentation and retention of P, the most widely used empirical relations are functions of flushing rates expressed several ways (Vollenweider, 1976; Ahlgren et al., 1988; Prairie, 1989). While the statistical significance of the empirical relations may be high at least for a particular region, they do have considerable variance and must be applied cautiously to individual lakes. For example, north-temperate lakes with anoxic hypolimnia retain less P than lakes with oxic hypolimnia (Nürnberg, 1984), and calcareous lakes have lower P retention than soft-water lakes in Wisconsin (Stauffer, 1985).
Phosphorus-particle interactions
Dissolved inorganic and organic P interacts with particles suspended in the water and with sediments residing on the bottom. Grobbelaar and House (Ch. 16) review two approaches to the investigation of P-particle interactions: (a) mechanistic, laboratory studies of co-precipitation of P with calcite and of iron-P compounds, and (b) empirical studies using laboratory microcosms or chambers inserted into sediments in the field.
Transfer of P to lake sediments through deposition of particulate matter usually exceeds the release of soluble P. However, the seasonal release of SRP can be an important source of P, particularly in shallow lakes (Ryding and Forsberg, 1977; Ryding, 1985; Jensen and Andersen, 1992) and in lakes with anoxic hypolimnia (Nürnberg, 1984).
Numerous environmental controls on P release from sediments have been identified. Temperature, turbulence or bioturbation, dissolved oxygen concentration, nitrate concentration, sulfate concentration, salinity, pH or alkalinity, dissolved organic carbon, calcium concentration, and sediment concentrations of aluminium, manganese, particulate organic carbon, iron or the Fe to P ratio have all been reported to increase or decrease the release of P (reviewed in Boström et al., 1982; Boström et al., 1988a; Baccini, 1985; Caraco et al., 1991). Among these factors, the onset of seasonal anoxia in the water overlying the sediments has been most consistently linked to an increase in P release from the sediments (Einsele, 1936; Mortimer, 1941, 1942; Caraco et al., 1991). Diel variations in P release may be attributed to daily formation and breakdown of an oxidized microlayer at the sediment-water boundary (Carlton and Wetzel, 1988).
In a comparative analysis of P release from sediments in 23 lakes, Caraco et al. (1989, 1991) found that the sulphate concentration explained 30% of the variation in P release, and they proposed that lakes with low sulfate concentrations have a higher capacity to retain P than lakes with high sulfate concentrations because of interactions of the sulphur and iron cycles in sediments. In 15 Danish lakes, the ratio of Fe to P in surficial sediments explained 58% of the variation in the rates of aerobic release of SRP from the sediments, and the ratio was well correlated with total P concentrations in the water in both summer (101 lakes) and winter (81 lakes) (Jensen et al., 1992).
Several important roles for bacteria in P release from sediments are becoming evident (Gächter and Meyer, 1993). Direct bacterial uptake and release of P can occur (Gächter et al., 1988) with bacteria out-competing algae for P (Currie and Kalff, 1984). Because uptake and release of SRP by bacteria and precipitation and dissolution of FeOOH occur at about the same redox potential, microbial activities may be overlooked in favour of the classical abiotic model of Mortimer (1941). Nitrate-reducing bacteria can catalyze the reduction of extracellular FeOOH-PO4 complexes (Jansson, 1987).
Transport and transformations in large river systems
The Amazon River accounts for about 20% of the total annual riverine discharge of water to the oceans (Richey et al., 1986) and slightly less than 10% of the total riverine transport of sediments (Milliman and Meade, 1983). An extensive floodplain with thousands of lakes borders the river and its tributaries (Sippel et al., 1992), and up to 30% of the discharge of the river contacts the fringing floodplain (Richey et al., 1989), which is a mosaic of open water, floating meadows and flooded forests. Longer storage times, more circuitous routing of water, adsorption by abiotic processes and biotic uptake and sedimentation of particulate P all occur on the floodplain and slow the downstream transport of P in a manner analogous to the nutrient spiralling concept that has been applied to small streams (Melack and Fisher, 1990; Fisher et al., 1991). Considerable local variation occurs in the extent to which suspended sediments enter the floodplain, and in regions with low current speeds only the clay fraction of riverine particulates penetrates more than a few hundred meters (Mertes, 1990). More than 90% of the suspended material carried into Lake Calado, an Amazon floodplain lake, was lost from the upper water within 2 km of the river during periods of channelised and over-bank flow (Engle and Melack, 1993). The decline in suspended particulate P was linearly related to total suspended solids and unaffected by increases in phytoplankton. In portions of the floodplain with sustained flows, periphyton attached to the submerged roots of floating grasses is especially important in the retention of riverine P (Engle and Melack, 1993).
Devol et al.'s (1991) study of temporal and spatial variations of P concentrations and fluxes along a 1700 km reach of the Amazon River spanned the river's hydrograph and is an example of the kind of study needed for other large rivers. Average concentrations of total P were 9.2 µM in the mainstream of the Amazon River, 4.1 µM in the main tributaries and 2.2 µM in floodplain lakes. Although they did not analytically distinguish inorganic and organic fractions of the particles, they provided indirect evidence that most of the total particulate P was inorganic.
To examine the roles of processes within the river and of mixing with tributary and floodplain waters on the seasonal and downstream patterns of concentration of the P fractions, Devol et al. (1991) calculated mass balances for the entire reach sampled for each cruise. The mass balances for fine particulate P during the three early falling-water cruises indicated that about twice as much material was leaving the reach as was entering and implicated sediment resuspension within the channel or erosion of banks as a source. Conversely, mass balances on the other sampling dates indicated net deposition within the channel. In contrast, mass balances of dissolved organic P showed no evidence of net consumption or production, but those for SRP indicated an excess of export over the sum of inputs on all cruises. Based on the stoichiometry of C, N and P in various fractions, Devol et al. (1991) conclude that the excess SRP is derived mainly from decomposition of organic matter and that dissolution of primary and secondary minerals or desorption of SRP from clay minerals and metal oxides are probably of secondary importance.
The integration of all the within-channel, floodplain and catchment processes and fluxes finally produces the yield of P from the Amazon basin to the ocean. By combining the average concentrations measured at Obidos with the discharge, Devol et al. (1991) calculated export by the Amazon River to be about 7% of the global riverine transport of particulate P, about 10% of the global transport of dissolved organic P and about 30% of the global transport of SRP when compared with Meybeck's (1982) estimate for natural rivers.
Riverine export of phosphorus
Several approaches can be used to estimate the riverine flux of materials into lakes, estuaries or oceans (Meybeck, 1982; Meybeck, 1988; Schlesinger and Melack, 1981). The larger the scale of interest, the greater the extrapolation usually required and the more uncertain the estimate. The most direct method is to combine measurements of discharge and concentrations made at the mouths of all rivers that enter the body of water of interest. Such data are not available for most inland waters or for whole ocean basins. A modification of this method is to extrapolate from a subset of rivers within a region based on the transport per unit volume of river discharge and the total discharge. A second approach is to determine the fluvial loss per unit area of land for each ecosystem or land use within a region, and then extrapolate to the whole region based on the area covered by each landscape category. Refinements of this approach could consider the spatial organisation of the landscape units in relation to the river. For example, the presence or absence of riparian vegetation is known to strongly influence elemental transport into rivers (Kedziora et al., Ch. 13). The extent to which catchments can be subdivided and the subunits spatially distributed is limited by both the lack of information available on land use in most catchments and the difficulty of calculating the transport of P among subunits and to the river. Usually, the smallest landscape unit that can be used for extrapolation is a headwater catchment.
Phosphorus transport to the oceans
Early estimates of P transport by rivers to the world's oceans
were often simple but clever by necessity because so few data
were available. Stumm (1972) derived an average dissolved phosphate
concentration for rivers from the solubility of major phosphate-rich
minerals, assumed a similar amount in organic P and used a value
about ten times higher for industrialised countries. When his
global average P concentration of 1.8 µM was multiplied
by a total annual riverine discharge of 33,000 km3
y-1, he obtained a value of 0.6 x 10 11 moles P y -1 or 1.86
Tg P
y-1. Lerman et al. (1975) computed a total P export to the oceans
of 20 Tg y-1 as the product of the combined rates of mechanical
and chemical denudation of continents and mean P content of crustal
material.
Two previous SCOPE projects included estimates of riverine
transport of P. Pierrou (1976) used Holeman's (1968) tabulation
of the amount of suspended sediments carried to the oceans and
Emery et al.'s (1955) value for the P content of the suspended
sediments to calculate riverine export of 13.7 Tg P y-1.
In addition, Pierrou used the product of a rough estimate of
dissolved P in rivers (100 µg l-1) and world-wide riverine
runoff of 37,000 km3 y-1 to calculate riverine transport of dissolved
P of 3.7 Tg y-1, giving a total P export of 17.4 Tg y-1. Richey
(1983) offered up-dated values for dissolved (1.5 to 4 Tg P y-1)
and particulate (17 Tg P
y-1) riverine export to the oceans but did not document clearly
the basis for these numbers. Richey also tabulated estimates
of fluvial fluxes for ten geopolitical regions based on total
discharge for each region and a range of P concentrations in
the rivers of each region, but he did not include a summary of
data for each region which would allow evaluation of the estimates.
Wollast (1983) derived a global flux for dissolved P of 2.15
Tg y-1 based on a relation between P concentration and the number
of inhabitants in a catchment per unit discharge which was probably
biased toward over-estimation by inclusion of several polluted
rivers in industrialised areas.
The first well documented, systematic analysis of the concentrations and transport of P by rivers of the world was done by Meybeck (1982). Although he tried to use annual mean values, not all were discharge-weighted, and few data were available for several important tropical rivers (i.e. Amazon, Zaire, Orinoco and Niger). Concentrations of SRP and TDP were obtainable from 19 and 12 rivers, respectively, and 13 and 9 streams or groups of streams, respectively, judged to be uncontaminated by human activities. The rivers with SRP values had an annual discharge of 10,200 km3 y-1 and an average SRP concentration of 10 µg l-1 (Table 6 of Meybeck, 1982, but stated to be 12.5 µg l-1 on page 420). Based on a SRP to TDP ratio of 0.4, the average TDP concentration was 25 µg l-1. Concentrations of SRP and TDP were available from an additional 29 and 23 rivers, respectively, judged to be contaminated by human activities. Martin and Meybeck (1979) provided numerous determinations of TPP, but few of POP, and they calculated an average of 1.15 mg P per g of suspended sediment.
Total fluvial transport of P to the oceans was calculated by Meybeck (1982) using Baumgartner and Reichel's (1975) value of 37,400 km3 y-1 for runoff from exorheic basins and his own average concentrations for the different P fractions. Hence, Meybeck reported transport of TDP by natural rivers to be 1.0 Tg y-1 and double that value when rivers judged to be contaminated by human activities were included. To determine transport of TPP, he combined his estimate of the flux of total suspended material ( i.e. 17.5 x 1015 g y-1) with that for P content per gram of sediment to obtain a value of 20 Tg y-1. Although Meybeck (1982) did not consider the influence of human disturbance on particulate transport, his estimate of the flux of total suspended material was higher than the 14.5 to 15.5 x 1015 g y-1 determined by Milliman and Meade (1983) for suspended sediments and bedload in disturbed and undisturbed rivers.
Other recent estimates of total P transport to the oceans range from 24 to 39 Tg y-1 (Froelich et al.,1982; GESAMP, 1987). Howarth et al. (Ch. 19) have assessed the estimates of Meybeck (1982) and GESAMP (1987) and suggest 22 Tg P y-1 as the best estimate for total riverine flux to the oceans. However, all these estimates remain uncertain for several reasons (Meybeck, 1988). The lack of measurements of particulate P in a sufficiently large number of rivers with consistent techniques and with adequate sampling of the non-uniform distribution of particulates within a river are the largest sources of error. Furthermore, impounding rivers reduces transport of P by increasing sedimentation of particulate P within reservoirs, but may also augment transport of P below dams by increasing erosion of channels, floodplains and deltas deprived of sediment-laden waters. The net result of impoundments on P inputs to the world's oceans has not been quantified, but estimates for several large rivers and many reservoirs indicate the likelihood of a net decrease in P transport from land to oceans (Petts, 1984; Meade, 1988).
In the decade since Meybeck (1982) appeared, the SCOPE project entitled "Transport of carbon and minerals in major world rivers" has stimulated research and synthesised information on riverine elemental fluxes (Degens et al., 1991). Unfortunately, P was not emphasised and fewer new data on P than on carbon resulted. Important improvements in information on P did result from independent studies on the Amazon River (Richey et al., 1991; Devol et al., 1991) and the Orinoco River (Lewis and Saunders, 1989). The new data for the Amazon River double the average concentrations used by Meybeck (1982) for SRP and TDP, and increase slightly the annual discharge. The new data for the Orinoco River are about 50% higher than Meybeck's SRP value and, more importantly, include measurements of TDP and TPP, all made over four complete years.
Application of yield coefficients
Estimates of yields of P per unit area (e.g. kg ha-1 y-1) for different ecosystems or land uses are valuable for calculating P export from catchments of widely varying sizes and landscape complexity. This approach, in principle, can be applied to catchments of individual lakes, climatic regions or even the earth as a whole (e.g. Reckhow et al., 1980; Schlesinger and Melack, 1981). Unit-area yields of P are available for a variety of land uses and regions in North America and Europe (e.g. Dillon and Kirchner, 1975; Reckhow et al., 1980; Prairie and Kalff, 1986; Harper and Stewart, 1987). However, these yields should not be applied casually to other climatic, hydrological or geological conditions. For example, yields from semi-arid South African and Australian catchments appear to be more variable than those in more humid, north temperate climates (Thornton and Walmsley, 1982; Cullen et al., 1988), and few data on P export are available with which to evaluate tropical catchments (Lesack, 1993; Lewis, 1986).
A general problem with the application of unit-area yields to catchments of different sizes is the extent to which differing hydrological paths and transit times and associated chemical and biological processing of P change the yields (see Hillbricht-Ilkowska et al. Ch. 12 and Sharpley et al., Ch. 11). For example, some empirical models of P export have found a reduction in P delivery per unit area with increasing catchment size (Prairie and Kalff, 1986). Moreover, the whole catchment usually does not contribute runoff, and the proportion that does varies with discharge. Riparian vegetation also influences the amount of P that reaches a stream (Peterjohn and Correll, 1984). Because the composite export from a basin is not necessarily the simple sum of the yields of each subunit within a catchment, a number of more sophisticated models have been developed to account for the hydrological and biogeochemical complexity of catchments of varying sizes (Jolankai, 1992).
As human activities continue to modify the earth's terrestrial and aquatic ecosystems, prediction of riverine transport of P will rely increasingly on analyses that incorporate combinations of hydrology and yields of P for different landscape units. To test these predictions will require long-term monitoring of discharge and concentrations of several fractions of P in rivers of various sizes. A major challenge will be to improve models of P transport and transformations to be able to make meaningful forecasts of water quality and trophic status of inland waters, estuaries and coastal oceans in the coming decade.
ACKNOWLEDGMENTS
I thank SCOPE for the invitation to participate in the discussions in Budapest that stimulated the preparation this chapter, and I appreciate the comments on drafts by S. MacIntyre, T.R Fisher, S. Hamilton and D. Engle. Partial funding for preparation of the manuscript was provided by a grant from NASA's EOS program.
|
Phosphorus in the Global Environment. Edited by H. Tiessen © 1995 SCOPE. Published in 1995 by John Wiley & Sons Ltd. |
|
Back to table of contents
Last updated: 30.06.2001