SCOPE 42 - Biogeochemistry of Major World Rivers

14.1 INTRODUCTION

In the following study we shall consider the elements carbon, phosphorus, nitrogen and sulphur as biologically significant species taken up by the receiving waters, released either by natural processes or by anthropogenic impacts. These elements can be characterized according to their chemical speciation, that is the chemical compounds in which they occur, which in turn influence their relative abundance in the atmosphere, hydrosphere and pedosphere and their availability to the biota.

Natural or anthropogenic impacts can change the chemical speciation of the element considered or can imply an actual transfer from one subcompartment to another of the system considered. The river input of chemical species is usually the result of chemical and physical (erosion) forces acting on terrestrial ecosystems resulting in a transfer from the terrestrial pools to the riverine system. Within the SCOPE world river programme a considerable body of information has been collected with respect to both dissolved and particulate matter transported by rivers.

In this chapter we try to establish the relations between the processes which make the terrestrial ecosystems a source of material inputs and the measured loads in the rivers themselves. In a statistical analysis using one factor correlation analysis the question is asked which properties of the ecosystems correlate best with the observed river loads in dissolved and particulate organic carbon (DOC and POC). From the various ecological, climatic and geomorphological elements considered we shall be able to show that the best correlation is obtained with the parameters of the hydrological cycle, namely precipitation and river runoff.

Man has had a decisive impact on the biosphere on a global scale within the past 125 years. The most important direct effect on ecosystems has certainly been the change in land use, converting natural vegetation to intensively used systems either for resource exploitation or for agricultural use or living space. There have been also indirect impacts on ecosystems, such as pollution (air, water and soil pollution) or fertilization (directly applied fertilizers or involuntary fertilization as a result of industrial processes). The discharge of materials through directly or indirectly changed ecosystems can be predicted in detail only if the causing processes are known. Distinction and care should be taken to separate the processes which lead to river inputs and those transformations which occur in a river itself. In order to describe the complete spectrum of man's activities with regard to river source terms we identified the following major topics:

• Tropical deforestation
• Fossil fuel emissions
• Food and fertilizer production
• Industrial production with emphasis on phosphates

14.2 VEGETATION AND SOILS: THE INTERACTION OF THE HYDROLOGICAL CYCLE WITH UNDISTURBED SYSTEMS AND POSSIBLE INFLUENCES OF LAND USE CHANGES

14.2.1 A MODEL FOR DOC AND POC IN RIVERS

This model is based on a watershed analysis considering climatic, topographic, ecological and anthropogenic impacts. A major source for either dissolved organic carbon, DOC, or particulate organic carbon, POC, is the carbon pools of the terrestrial biosphere. These pools include the living biomass, the dead but not yet decomposed biomass (litter), and the soil organic carbon, mainly phenolic compounds which result from the decomposition of litter. These humic substances often form complexes with the clay minerals of the soil.

The carbon input into the biota is represented by the flux net primary production, NPP, which is the net flux from the atmospheric CO₂ to the phytomass of green plants. More than 95% of the carbon flux NPP is passed along to the litter pool where it is decomposed by the detritus food chain forming mainly CO₂ and to a smaller extent complex organic substances, summarized as humic and phenolic substances. Some of the material is leached from litter and soils forming DOC or is washed out from the terrestrial pools by erosion processes contributing to the POC.

DOC and POC loads for many of the larger river systems in the world have been determined in the past years, as well as further information about drainage and mineral composition (Degens 1982; Degens et al. 1983, 1985).

Table 14.1 Characteristics of 18 major rivers of the world and their watersheds. Mean max. elevation is the mean of the maximum elevations of all grid elements of a 2.5 degree grid which fall into the watershed

The information on DOC and POC freights may be coupled now to certain results of models of the terrestrial biosphere in order to investigate the possible mechanisms and driving forces for DOC and POC formation.

The authors used the Osnabrück Biosphere Model (OBM, Esser 1986, 1987) to analyse correlations between the carbon freights of 18 selected major world rivers (see Table 14.1) the carbon pool litter, the flux NPP, and land use characteristics in the watersheds of those rivers as well as some climatic and geomorphological features.

In Table 14.1 certain geomorphological features of the watersheds, the annual carbon load, and the annual river discharge of the rivers under consideration are listed.

Corresponding values for the biospheric litter pool and the flux NPP were calculated by several model runs of the OBM using a standard set of data for climate, soils and agricultural use (see Esser 1986,1987). Since the OBM rests upon a global 2.5 degree grid as regional structure the values for the watersheds are calculated as weighted means of the values of the grid elements covering the respective watershed (Table 14.2).

To make the area-related variables comparable with those not related to surface area like temperature and land-surface elevations the area-related variables were recalculated for the unit area of watershed (m²). Table 14.3 includes the results.

Table 14.2 Litter pool and net primary productivity (in carbon equivalents) calculated by the OBM as totals for the watersheds of the rivers in Table 14.1 and the averaged climatic data on the standard data set of the OBM

In Table 14.4 the land-use characteristics of the 18 watersheds are listed. The calculations are based on the data of the files of the OBM. The distribution of agricultural area in these files follows the World Atlas of Agriculture (Istituto Geografico de Agostini 1969, 1971, 1973). The changes in the reference period 1950-78 are on country basis from a data pool supplied by Richards et al. (1983). For South America it was supplemented with data derived from the evaluation of 934 Landsat scenes for the period 1971-81 by Esser and Lieth (1986). The Landsat scenes cover 72% of the area of South America.

The DOC and POC forming processes may be understood as part of a system, which consists of an input flux NPP, a source pool litter, the output fluxes DOC and POC, and the control variables: temperature, precipitation, discharge, effective evapotranspiration, mean maximum elevation of a 2.5 degree grid, its standard deviation, and the four land-use characteristics. The mean maximum elevation stands for the elevation above sea level of the watershed, its standard deviation for the slope of the terrain.

Table 14.3 Area related values of Tables 14.1 and 14.2 recalculated per unit area of watershed. Discharge, DOC and POC mean export per m² of land surface

As a first approach, the correlation coefficients between certain variables were used to investigate the probable importance of these system variables for DOC and POC formation. For data pools with two variables x(i) and y(i), i = 1, ...,I the correlation coefficient -1 < g(x,y) < +1 is defined as

                                      
      S(x,y) is the covariance, S(x) and S(y) are the standard deviations and and the arithmetic mean of either data pool:
                         
              

(14.2)

Table 14.4 Land-use characteristics of the watersheds. Data are from the files of the OBM and are based on the World Atlas of Agriculture (Istituto Geografico de Agostini 1969, 1971, 1973) and on a data collection by Richards et al. (1983)

With Equations (14.1)-(14.6) the g(x,y) for several system variables and either DOC or POC export values were calculated. The values of the variables recalculated per unit area of the watersheds (Table 14.3) were used for the calculations. Figures 14.1 and 14.2 display both systems and each correlation coefficient.

There is no fundamental difference between the behaviour of DOC and POC. As a consequence, both show a positive correlation and a g_DOC,POC of 0.82. The strongest correlation exists between the watershed discharge and the DOC and POC values (0.94 and 0.92 respectively). This result strongly suggests that leaching and soil surface erosion processes may be the most important mechanisms for DOC and POC formation. Since discharge mainly depends on precipitation there is a strong correlation of DOC and POC values with precipitation too (0.81 and 0.66 respectively). The correlation with the litter pool is weak. The reason for that may be that the annual export through dissolved and particulate organic carbon rarely exceeds 0.1% of the source pool litter and 0.2% of the biospheric input NPP (see Table 14.3).

Figure 14.1 Pools. fluxes and control variables of the system 'DOC export from land surfaces' and the related correlation coefficients on the basis of 18 watersheds

The effects of land use on DOC and POC appear insignificant according to this analysis. Even if one considers that the comparison of data for entire watersheds is a crude method, the correlation may be negative rather than positive. This is not an unexpected result since it is well known that agricultural use of an area may reduce both NPP and litter pool in comparison to uninfluenced systems (Esser et al. 1982; Lieth and Aselmann 1983). On the other side it has been widely accepted in the past, that the great reduction in phytomass which accompanies the land-use changes at least partially should show up as carbon and nutrient input into the river systems. In contast to this expectation, the correlation of DOC and POC river freights with either changes or relative changes of land use is zero or slightly negative (Figures 14.1 and 14.2).

Figure 14.2 The POC export from land surfaces and its correlation with variables of the system

There is no significant influence of mean maximum elevation and of its standard deviation, which represents steepness of the terrain, except that POC export is correlated with the slope (y = 0.67).

The plots of either DOC or POC values against discharge suggest linear relationships (Figures 14.3 and 14.4). The respective linear correlations were given by

Figure 14.3 Data points for DOC plotted against the discharge of the watershed. A linear relationship is obvious (see Equation (14.7))

Figure 14.4 Values for POC plotted against the discharge of the watershed. For the values see Table 14.3. The line is due to Equation (14.8)

The variables in Equations (14.7) and (14.8) mean: 

Equations (14.7) and (14.8) were used in the OBM to calculate the regional distribution of DOC and POC export from land surfaces. The results were mapped and are shown in Figures 14.5 and 14.6.

The total global transport calculated on a regional basis by use of Equations (14.7) and (14.8) is 304 Tg C/year, 226 Tg C/year are leached as DOC and 78 Tg C/year are transported as POC. These figures include natural and potential anthropogenic sources since the river freights as measured were used for calibration of the equations.

This submodel has certain weaknesses and has to be improved in the future. First, the data situation is poor. The number of rivers under investigation may be improved but it is much more important that smaller water- sheds with well-defined climate, soils and land-use characteristics should be included. A certain shortcoming is the correlation of river data measured within a few distinct years with long-time means of climatic and land-use variables. This was necessary since values of those variables for the time period under consideration were not available. Similarly, most of the river data exist only from a few selected stations, which alone cannot represent the dynamic processes in rivers.

14.2.2 ASSESSMENT OF THE IMPACTS OF TROPICAL DEFORESTATION ON CARBON AND MINERAL (N, P, S) EXPORT THROUGH WATERSHEDS ON THE BASIS OF FAO STATISTICS

In the previous section the dominating role of the hydrological cycle with respect to the export of organic carbon has been demonstrated. Surprisingly enough, an increase in the extent of agricultural area or intensity of agricultural activity did not show up in the DOC and POC freights of the respective rivers. The reason for this is not entirely clear, perhaps one explanation may be that land-use changes often affect only a small portion of the watershed to a major extent, such that the total watershed will hide the anthropogenic impacts occurring behind the dominating natural processes. Although many efforts were made over the past decade (Myers 1981; Fearnside 1982; Lanly 1982) to document the role of tropical deforestation as a major land use change, its effect as a source function of carbon and other elements to rivers lacks still the corresponding data sets. Most research groups have concentrated to collect river data of the watershed, not distinguishing between natural and anthropogenic processes. The qualitative effects of deforestation on air, soil and water have been discussed by Vitousek (1983). A quantitative estimate of the possible range of the river source function from tropical deforestation is difficult. We believe, however, that one reliable basis to estimate the concerned areas will be the land-use changes given by the F AO statistical data, as summarized by Lanly (1982).

Figure 14.5 Regional distribution of the DOC export from land surfaces. Each plot sign represents a grid element of the 2.5 degrees grid of the Osnabrück Biosphere Model

Figure 14.6 Regional distribution of the POC export from land surfaces. For further explanation see legend of Figure 14.5

Figure 14.7 Global estimates of mean annual areal transfers in tropical deforestation according to FAO data (Lanly 1982). The upper numbers in the boxes mean: 1, undisturbed forests; 2, logged-over forests; 3, forest fallow; 4, first year shifting cultivation; 5, second year shifting cultivation; 6, permanent crops and pasture; 7, degrading land. The lower numbers in the boxes give land areas (10⁶ ha)

In a previous study (Kohlmaier et al. 1985) we have established an area transfer diagram, based on Lanly's data, which distinguishes between the open and closed tropical forests of Africa, Asia and America, which are either converted to logged-over forests or to shifting cultivation or to permanent agriculture and pasture. In Figure 14.7 we summarize these processes, indicating that the present estimated deforestation rate of 11.1 x 10⁶ ha/year (7.3 x 10⁶ ha/year of virgin closed tropical forests and 3.8 x 10⁶ ha/year of open tropical forests) amounts to 0.64% per year of the total present tropical forests of 1724 x 10⁶ ha. The shifting cultivation cycle plays an important role in the deforestation, where 5.1 x 10⁶ ha are added each year to the forest fallow stands of 409 x 10⁶ ha (34.9 x 10⁶ ha of which are harvested annually). 

Table 14.5 Estimated carbon densities in phytomass and soils in tropical forests and their anthropogenic substitutes (all figures in kg/m² carbon) 

With the densities in phytomass and soil carbon of the forests and their anthropogenic substitutes, as listed in Table 14.5, one can make some first order estimates about the organic carbon involved by determining the corresponding differences in the carbon densities and multiplying them by the areal transfers. In Table 14.5 we show the still existing great uncertainty in standing phytomass density, which differs by more than a factor of two, when using either the ecological estimate, based on the life zone concept of selected measured plots, or the total phytomass derived from timber volumes (Brown and Lugo 1982, 1984); corresponding uncertainties remain for the soil carbon densities which are not listed here explicitly. Table 14.6 gives a summary of the estimated carbon and nutrient transfers during tropical deforestation, based on the FAO data (Lanly 1982), the model of Kohlmaier et al. (1985) and the ecological phytomass data of Brown and Lugo (1982). Processes 'A'  are phytomass processes, which are distinguished from the processes 'B', the soil processes. The different contributions Aa, Ab and Ac refer to logging and forest clearing for permanent agriculture, new clearings in shifting cultivation and processes involved in the actual shifting cultivation cycle of the fallow forests, respectively. In processes of clearing, the immediate effect of the controlled burning and logging must be distinguished from the delayed effects of deterioration of dead material above and below ground and erosion. The immediate effects, described by processes A1 to A5, add up to 100%, where the percentage distribution varies highly with ecosystem type, and the weighted average presented here is uncertain. The immediate release of CO₂ has been estimated to contribute 24%, while the delayed effects, often occurring over a time span of many years, contribute an estimated 40% of the originally involved phytomass, assuming that 75% of the dead biomass is contributing to a CO₂ release.

Table 14.6 Estimated carbon and mineral (N, P, S) transfers during tropical deforestation (based on Kohlmaier et al. (1985). Values are in Tg/year respective fraction of the pool)

Erosion of particulate and wash-out of dissolved organic carbon (A2) has been estimated to contribute up to 12% of the originally affected phytomass, with contributions from standing phytomass, dead material and charcoal. Erosion from the soils (process Bd2) has been previously estimated to amount up to 30% (Kohlmaier et al. 1985), while the delayed CO₂ release process (Bd1) has been assessed with 70% of the soil carbon difference. Both the release of CO₂ from soils as well as erosion from phytomass and soils are considered delayed processes, though often immediately after clearing a first burst of release of organic and soil material can occur (Ewel et al. 1981). In a first-order estimate the time delays in carbon and nutrient release do not need to be considered explicitly, especially under the assumption that the clearing processes have been occurring at a quasi steady-state rate. Esser et al. (1982) used a dynamic litter depletion model and estimated the average global turnover time for litter to be 3 years. The corresponding values for the continents were 3.8 years for North and Central America, 1.9 years for South America, 3.9 years for Europe, 2.4 years for Africa, 4.0 years for Australia, and 3.7 years for Asia. Considering time delay, actual annual contributions either to the release of CO₂ to the atmosphere or the release of carbon and minerals to rivers are somewhat smaller (88.6% for a 2% increase in deforestation and 79.8% for a corresponding 4% increase) than the ones determined without time delay.

Table 14.6 summarizes that erosion derived from phytomass (230 Tg C/ year) and from soils (240 Tg C/year) amounts to c. 470 Tg C/year, accepting the perhaps conservative FAO deforestation estimate of 11.1 x 10⁶ ha/year. Allowing for a possible deforestation rate of 14 x 10⁶ ha/year an upper bound for erosion is given by about 600 Tg C/year. Considering on the other hand the lower phytomass estimate derived from timber volumes (about 43% of the ecological estimate) and the fact that perhaps not all but only 50- 75% of the deforestation release of organic carbon should reach the rivers, a lower level of about 200 Tg C/year is given. Our best estimate of 400 ± 200 Tg C/ year as DOC and POC river input can be compared with an ecosystem type analysis, given by Schlesinger and Melack (1981), who estimated that the tropical forest contribution to the total river load of organic carbon, measured at the mouth of the rivers, should amount to 120 Tg C/year. If we assume, with no better knowledge available, that about 50% of the carbon load estimated by Schlesinger and Melack (1981) corresponds to the natural contribution, we then should compare 60 Tg C/year of organic carbon of anthropogenic origin entering the world oceans from tropical ecosystems to our calculation, where 400 ± 200 Tg C/year are released from tropical ecosystems. Thus, we are led to assume that 10% to 30% (on the average 15%) of the total organic carbon released is reaching the ocean, while 70% to 90% should be oxidized during transport or redeposited in the terrestrial watersheds and neighbouring lands.

A first-order estimate of the total nutrients released into the soil/water space and into the atmosphere at the deforestation sites can be derived from the elemental composition ratios by weight as given by Melillo and Gosz (1983), namely P : S : N : C = 1: 1: 10: 1500 for forest biomass and 1 : 1 : 9.4: 120 for forest soils, which again has great uncertainties attached to it, because of the great variation with ecosystem type. We distinguish between the immediate release of nutrients (process A5) in the controlled burning of forest clearing, the delayed release of nutrients by oxidation on site (process Ad1 from phytomass and Bd1 from forest soils) and finally the release of nutrients into river systems by the partial decay of eroded material (derived from processes Ad2 and Bd2). In the first fire clearing process carbon, nitrogen (3.1 Tg N/year) and sulphur (0.3 Tg S/year) become air-borne as oxides, while c.0.3 Tg P/year remain as phosphate ash on the ground. In addition we need to include here all NO_x compounds (c. 1 to 10 Tg N/year) (Crutzen 1983), which are formed from N₂ and O₂ during the combustion process.

Delayed oxidation on site of both phytomass and forest soils should lead to a release of 49 Tg N/year in the soil/water space, which in part may be lost to the atmosphere by denitrification (N₂ and N₂O) and in part may be transferred into the riverine systems, the percentage of which is difficult to quantify. In the corresponding release of 5.2 Tg P/year as phosphates and 5.2 Tg S/year as sulphates (sulphites) again the soil component plays the dominant role. For processes A5 + Ad1 + Bd1 we estimate that c. 25% of the N-compounds and c. 10% of the P- and S-compounds are discharged into rivers. The estimate of an additional nutrient release into the rivers is based on the assumption that on the average c. 85% of the nutrients given in processes Ad2 and Bd2 will be released into the river systems, corresponding to 17.4 Tg N/year, 2.1 Tg P/year and 2.1 Tg S/year .

In conclusion, we state that the deforestation in the tropics could add significantly to the input of carbon and nutrients into tropical rivers, if we assume this conservative approach to be realistic. In contrast, the direct comparison of measured river data carried out in the previous chapter did not support this assumption. However, if we assume that the sites of river data acquisition are far removed from the sites of deforestation, processes going on in the rivers may hide the processes which lead to the exports from land originally.

Input over the entire watershed and load at the river mouths are not equivalent, because of the processes occurring along the rivers and their associated floodplains. Our total estimated DOC and POC input of 400 ± 200 Tg C/year over an estimated watershed of 15 x 10⁶ km² of tropical rivers corresponds to an average release of 27 g/m²/year, which exceeds substantially the estimate of the average river load data given in Section 14.2.1 and Schlesinger and Melack's (1981) estimate of an average tropical river load of c. 8 g/m²/year. The input of nitrogen, derived from deforestation, is an even more complex problem, both because of the different nature of immediate (mostly airborne) and delayed (in the soil/water space) release at the ecosystem site and the co-occurring effects of denitrification. The delayed release of humus oxidation should be the most significant contribution with an upper level of 49 Tg/year, which, however, should be reduced eventually through denitrification, soil absorption and plant uptake. Similarly the phosphate and sulphate contributions stem mainly from the delayed release of soils with upper values of 5.3 Tg P/year and 5.3 Tg S/year, respectively. Release of nutrients from organic compounds within the river systems seems quite reasonable, especially since according to our estimates about 85% of the organic river input is decomposed. Based on these assumptions, an input of about 17.4 Tg N/year, 2.1 Tg P/year and 2.1 Tg S/year is calculated.

14.3 FOSSIL FUEL EMISSIONS: WET AND DRY DEPOSITION OF S, N AND P AND THE RUNOFF FROM RIVERS

With fossil fuels still being the predominant energy source, the most important long-term perturbation of natural systems is the steady increase of atmospheric CO₂. Rotty (1987) has documented the fossil fuel production since the beginning of the industrialization (1860) up to the year 1984, in which the emission totalled 5.2 Gt C/year. Coal and oil contain a number of impurities, including the plant nutrients nitrogen, phosphorus and sulphur.

The amount of nitrogen emitted as NO_x exceeds the amount of nitrogen contained in the fossil fuels, since in high temperature combustion processes N₂ and O₂ of the air are combined to nitrogen oxides as well. Ehhald and Orummond (1982) estimate the production of NO_x, connected with the burning of fossil fuels, to amount to 8.2 and 18.5 Tg N/year while Logan (1983) gives a higher estimate of 18 to 24 Tg N/year. The additional production of ammonia from coal burning is given by Warneck (1988) to be equal to or less than 2 Tg N/year, an estimate which lies considerably below that of Crutzen (1983) who estimated 4-12 Tg N/year. On a regional scale about 180 Tg C/year are consumed in the Federal Republic of Germany and approximately 1 Tg N/year is produced as nitrogen oxides, with traffic being responsible for approximately 50% of the total emission. If the FRG (about 1% of the total world population) is compared in its CO₂ and NO_x emissions with the global emission data, it is seen that it shares 3.5% of the total CO₂ output and approximately 6% of the total NO_x output, the latter figure being proportionally high because of the large share in high temperature combustion in motor traffic with still very little fuel emission control.

Phosphorus is a minor constituent of fossil fuels and practically only contained in coal. Bernhardt (1978) estimated the average phosphorus content in fossil fuels to approximately 0.015%, amounting to approximately 0.8 Tg/year in the worldwide consumption of fossil fuels. This figure is compatible with the estimate of Melillo and Gosz (1983), who propose a release of about 1 Tg P/year derived from fossil fuel combustion.

Finally the amount of sulphur released both from combustion of fossil fuel and metal smelting has been estimated by Freney et al. (1983) to amount to 169 Tg SO₂-sulphur/year, while Crutzen (1983) estimates a total emission of 90 Tg S/year from the burning of fossil fuel alone. If the latter number is set into proportion to the total carbon emission of 5200 Tg C/year, it is seen that on the average about 1.7% of S is contained and released from fossil fuels. A look at the FRG shows that with the pollution control in power plants and with the use of low sulphur oil SO₂ pollution can be reduced. Despite these measures, however, approximately 1.8 Tg S/year are released as SO₂, corresponding to c. 1% of the fossil fuels consumed.

It should be emphasized that both nitrogen and sulphur compounds enter the atmosphere as gaseous compounds but may recombine in the atmosphere to some extent to aerosols, while the phosphorus enters the atmosphere as small fly ash particles, since there is no equivalent gaseous phosphorus  compound.

The amount of river runoff derived from the fossil fuel combustion may be estimated only in a very crude way by considering the relative proportions of wet and dry deposition of the corresponding compounds. We can perhaps assume that most of the dry deposition on vegetation and soil will stay at the site, while certainly a fraction of the dissolved compounds in wet deposition will be found entering the receiving waters. Georgii (pers. comm.) has estimated that about one-third to two-thirds of the sulphur emissions enter the environment by dry deposition, the fraction of the nitrogen compounds in dry deposition is estimated to be somewhat smaller. Lacking precise data we assume that both the sulphur and nitrogen compounds will have an equal share of wet and dry deposition.

From the worldwide water balance we know that the continental precipitation amounts to 111 X 10¹² m³/year with a river runoff of 39.7 X 10¹² m³/year or 35.7% of the total precipitation. An estimated lower bound of river input of initially airborne nitrogen and sulphur compounds is then given by the product of the percentage of wet deposition multiplied by the percentage of river runoff, or 18% of all airborne nitrogen and sulphur compounds. With a mean value of airborne N-compounds of c. 16 Tg/year and S-compounds of c. 130 Tg/year, we estimate a minimum river input of 2.9Tg N/year and of 23.4 Tg S/year. Freney et al. (1983) estimate the river input of sulphur compounds from fossil fuel combustion and metal smelting to amount to 47 Tg S/year, out of the total of 169 Tg S/year, amounting to approximately 28% of the total anthropogenic release.

It is expected that very little phosphorus of the total phosphorus emission from fossil fuel combustion (10% to 25% or 0.1 to 0.2Tg P/year) will reach the fresh water systems, since essentially all of the phosphorus is in particulate form, released predominantly as dry deposition over land.

 14.4 FOOD PRODUCTION AND FERTILIZERS: ENHANCED NUTRIENT FLOW, CAUSED BY HOUSEHOLDS, INTO THE RECEIVING WATERS

 Increase in agricultural productivity is achieved both by intensive measures (application of fertilizer, herbicides and pesticides, irrigation) and extensive measures (increase in agricultural area, including shifting cultivation). Food production in developed countries and increasingly in developing countries makes use of nitrogen, phosphorus and potassium fertilizers as well as calcium and sulphur in particular cases. Industrial fixation of nitrogen leads to nitrogen-containing fertilizers in the form of NH₃, NH₄⁺ as well as NO₃^-. The UN data for 1981, as collected by Rosswall (1983), suggest an industrial fixation of 60 Tg N/year, the world production of which has increased from 7 Tg/year in 1955 to 19 Tg/year in 1965/66 and up to 44 Tg/year in 1975/76, as cited by Finck (1979).

While the application is about 15 kg N/ha/year on the world average, about 90 kg N/ha/year are applied in the FRG and c. 220 kg N/ha/year in the Netherlands. In some intensively used tropical areas with two or three harvests per year up to 1000 kg N/ha/year are used. Because of the high solubility of most nitrogen fertilizers considerable leaching is expected especially of nitrate, which in contrast to ammonium is not adsorbed to soil clays and therefore may show up in the river runoff as well as in the groundwaters. Fields which are fertilized with liquid ammonia will also be a source of airborne ammonia as well as soils which show a high pH value. Aside from these two processes, denitrification must be considered as well, resulting in molecular nitrogen, N₂, or in dinitrogen oxide, N₂O. Lacking detailed knowledge about the fraction of nitrogen reaching the rivers we shall assume here that approximately 25% of the total fertilizer input of 60 Tg N/ year or approximately 15 Tg N/year will reach the receiving waters.

According to Bernhardt (1978) the world phosphate fertilizer amounted to 14.2 Tg P/year in 1975, extrapolated to 17.3 Tg P/year in 1980, and 22.2 Tg P/ year in 1985, assuming an annual increase of approximately 4%. In contrast to the nitrogen compounds, all introduced phosphates are adsorbed to soil particles readily, such that its percentage of leaching should be lower than for the corresponding nitrogen compounds, perhaps only 5-10% of the total input, resulting in a release of 1-2 Tg P/year. The lower value has been substantiated by the extensive phosphorus study in the FRG (Bernhardt 1978).
  
Freney et al. (1983) estimated the sulphur fertilizer contribution to be 28 Tg S/year, assuming a near steady state in the soils, thus implying that the leaching of fertilizers will be equal to the input, which is certainly an upper value. We assume that the agricultural soil systems are nearly everywhere in a non-steady state, accumulating nutrients over the time of anthropogenic impact. In that sense the discharge from sulphur fertilizers into the riverine systems is perhaps better approximated to 25% or 7 Tg S/year .

Based on the average daily caloric intake of man of c. 2400 kcal/day (including c. 400 kcal/day animal protein which corresponds to 4000 kcal/day of animal feed) the food and feed production and consumption involve approximately 1300 Tg C/year, if only the direct consumption by man and animal is considered. The productive systems, however, involve a C turnover, which is perhaps four times as high, resulting in 5200 Tg C/year.

From the element ratios in food and feedstuffs the estimated uptake of N, P and S may be calculated. The C : N : P : S weight ratio for food and feed biomass, as derived from Finck (1982), is assumed to be given by 330: 12: 2: 1. Using these ratios we calculate an annual uptake of 47 Tg N/ year, 8.7 Tg P/year and 4.0 Tg S/year, by directly consumed food and feed.

About 10% to 20% of the plant food production enter households, resulting in a corresponding sewage and solid refuse production. The remaining 80% to 90% enter the animal food chain, resulting in the corresponding animal food production, with an energy efficiency of approximately 10% and a remaining animal excretion of 90% .Accounting for both plant and animal food in man's household, about 20% of the elements in food and feed production are converted into sewage and solid refuse in the ratio of approximately 1 : 1. Neglecting any precipitation in sewage treatment plants, about 10% of the total nitrogen, phosphorus and sulphur should end up in riverine systems, i.e. 4.7 Tg N/year, 0.9 Tg P/year and 0.4 Tg S/year.

The remaining 80% of the total food production will be recycled through the feed and animal food chain with perhaps an additional leaching from the fields fertilized with slurry containing a maximum of 20% for nitrogen compounds and 10% for phosphorus and sulphur compounds of the total food and feed production, resulting in an additional discharge of 7.5 Tg N/year, 0.7 Tg P/year, and 0.3 Tg S/year with a corresponding fraction of 130 Tg DOC.

In summing up, we state that both the leaching of fertilizers and the cycling of materials through man's household contribute to the food production related nutrient input into the receiving waters. Fertilizer nitrogen (15 Tg N/ year) and food consumption (12.2 Tg N/year) add up to nearly 30 Tg N/year (to be reduced by a corresponding fraction for denitrification). Phosphorus from fertilizer leaching (1.5 Tg P/year) and from food consumption (1.6 Tg P/ year) result in a nutrient load of approximately 3 Tg P/year, while the corresponding sulphur input adds up to perhaps 10 Tg S/year.

14.5 INDUSTRIAL PRODUCTION: PHOSPHA TE-CONTAINING DETERGENTS AND THEIR DISCHARGE INTO RIVERINE SYSTEMS

About 90% of the worldwide rock phosphate production (estimated 24.4 Tg P/year in 1985) enters into the mineral fertilizer production (22.2 Tg P/year), while the remaining 10% (2.5 Tg Plyear) goes into detergents and cleansers, and a small fraction of industrial chemicals. In the industrialized countries like the FRO the share of detergents and chemicals is higher, with about 16% used for detergents and about 8% for industrial chemicals, leaving about 76% for agriculture. Practically all detergents on a phosphate basis (polyphosphates) are highly soluble in water and thus enter the sewage system after usage. Only a small fraction of approximately 20% is held back in the sewage treatment plants, such that still 80-90% will end up in the receiving waters. A maximum input can be calculated on the basis that all of the detergent phosphorus or approximately 10% of the total phosphate production and consumption are transfered into riverine systems, amounting to 2.2 Tg P/year.

14.6 SUMMARY

Table 14.7 summarizes our estimated discharges of organic carbon and minerals (N, P, S) into rivers from terrestrial sources.

A conservative estimate on the basis of land-use changes occurring during deforestation and assumptions for related changes in phytomass, litter and organic soil carbon, leads to a source value of 400 ± 200 Tg C/year to the riverine system as DOC and POC. Agreement with the data collected through the world river transport programme carried out by Degens (1982) and Degens et al. (1983, 1985) is only obtained if one accepts that the major fraction of released organic carbon is oxidized, deposited or transferred in the upper courses of the rivers. Otherwise the hypothesis of tropical deforestation being a source for DOC and POC is not supported, as pointed out in Section 14.2.1. The annual global freight of 226 Tg C/year as DOC and 78 Tg C/year as POC calculated by our model thus should be due to natural processes only. At the moment we are not yet in a position to rate the two contradicting hypotheses. Further data are essential which should be sampled on the river sites with well-defined watersheds. On the other side, a lot of evidence is available that during deforestation minerals are lost to the receiving waters which were components of the cleared biomass. It is well known that cleared sites on tropical soils loose their fertility within 2-7 years after deforestation, depending on the soil type and the amount of biomass that was burned. It appears likely that the reason is the release of a considerable amount of minerals, which partly reach the river systems because of ineffective absorption by cultivated plants and by tropical soils because of their low exchange capacity. Denitrification may also be insignificant on 'terra firme' sites since tropical soils are generally well drained and not anaerobic, which is an essential condition for effective denitrification. In contrast, on floodplains and in river sediments denitrification may well be effective.

Table 14.7 Discharge of organic carbon and minerals (N, P, S) into rivers and the ocean from terrestrial systems

We assumed a decomposition of up to 85% of the organic material within the rivers, the remaining load of organic carbon at the mouth of the river should have been substantially reduced, while at the same time minerals from the decomposed material are released in the form of inorganic ions which may undergo further reactions within the river system.

The consumption of fossil fuels again will introduce only a small contribution of organic carbon into the river systems while through atmospheric washout the concomitant nutrient will partially end up in the river systems.

The process of food production including the introduction of nitrogen and phosphorus fertilizers into the fields results in a fairly sizeable nutrient contribution to the river systems, both from the remainders of food and feed  and fertilizer runoff. 

Finally, industrial production of detergents can be a major source of phosphates in industrial countries. Although the production numbers are relatively small compared with the fertilizer industry, a very high fraction of the highly soluble phosphates will end up in the river systems since detergents are introduced directly into the runoff systems.

Adding up all four contributions a maximum discharge of nutrients released due to man's activity is obtained because it neglects all reactions within the river system (with the exception of the assumed decomposition of organic material in the river course). One may compare these calculated freights with measured freights at the river deltas including natural sources. It is then seen that the organic carbon and the inorganic sulphur are in excess at the river deltas in comparison to the above calculated contributions. In contrast, nitrogen and phosphorus freights are smaller at the deltas than the calculated inputs. It may be concluded that phosphate and nitrogen compounds undergo major transformations within the river systems such that these compounds become diminished with respect to their input.

REFERENCES

Bernhardt, H. (Ed.) (1978) Phosphor. Wege und Verbleib in der Bundesrepublik Deutschland, Verlag Chemie, Weinheim, 285pp.

Brown, S. and Lugo, A. E. (1982) The storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 14, 161-87.

Brown, S. and Lugo, A. E. (1984) Biomass of tropical forests: a new estimate on forest volumes. Science 223, 1290-3.

Crutzen, P. J. (1983) Atmospheric interactions¾homogeneous gas reactions of C, N and S containing compounds. In: Bolin, B. and Cook, R. B. (Eds) The Major Biogeochemical Cycles and their Interactions, Wiley, New York, pp. 67-112.

Degens, E. T. (Ed.) (1982) Transport of Carbon and Minerals in the Major World Rivers, Pt 1. Mitt. Geol.-Paläont. Inst. Univ. Hamburg, SCOPE/UNEP Sonderbd. 52, 764 pp.

Degens, E. T., Kempe, S. and Soliman, H. (Eds) (1983) Transport of Carbon and Minerals in Major World Rivers, Pt 2. Mitt. Geol.-Paläont. Inst. Univ. Hamburg, SCOPE/UNEP Sonderbd. 55, 535 pp.

Degens, E. T., Kempe, S. and Herrera, R. (Eds) (1985) Transport of Carbon and Minerals in Major World Rivers, Pt 3. Mitt. Geol.-Paläont. Inst. Univ. Hamburg, SCOPE/UNEP Sonderbd. 58, 645 pp.

Ehhald, D. H. and Drummond, J. W. (1982) The tropospheric cycle of NO_x. In: Georgii, H.-W. and Jaeschke, W. (Eds) NATO ASI Series Vol. C 96. Reidel, Dordrecht, pp. 19-251.

Esser, G. (1986) The carbon budget of the biosphere¾structure and preliminary results of the Osnabrück Biosphere Model (in German with extended English summary). Veröff. Naturforsch. Ges. zu Emden von 1814, Vol. 7, 160 pp. and 27 figures.

Esser,G. (1987) Sensitivity of global carbon pools and fluxes to human and potential climatic impacts. Tellus 39B, 245-60.

Esser, G. and Lieth, H. (1986) Evaluation of climate relevant land surface characteristics from remote sensing. Proc. ISLCP Conference, Rome, ESA Publication SP- 248, pp. 205-11.

Esser, G., Aselmann, I. and Lieth, H. (1982) Modelling the carbon reservoir in the system compartment 'litter'. In: Degens, E. T. (Ed.) Transport of Carbon and Minerals in Major World Rivers, Pt. 1. Mitt. Geol.-Paläont. Inst. Univ. Hamburg, SCOPE/UNEP Sonderbd. 58, pp. 39-58.

Ewel, J. C., Berish, B., Brown et al. (1981) Slash and burn impacts in a Costa Rican wet forest site. Ecology 62, 816-29.

Fearnside, P. M. (1982) Deforestation in Brazilian Amazon; how fast is it occurring? Intersciencia 7, 82-8.

Finck, A. (1982) Fertilizer and Fertilization. Verlag Chemie, Weinheim, 439pp. Freney, J. R., Ivanov, M. V. and Rodhe, H. (1983) The sulphur cycle. In: Bolin, B. and Cook, R. B. (Eds) The Major Biogeochemical Cycles and their Interactions, SCOPE Report 21, Wiley, New York, pp. 56-61.

Istituto Geografico de Agostini (1969, 1971, 1973) World Atlas of Agriculture. Novara (Italy).

Kohlmaier, G. H., Bröhl, H., Stock, P., Plöchl, M., Fischbach, U., Janecek, A. and Fricke, R. (1985) Biogenic CO₂ release and soil carbon erosion connected with changes in land-use in the tropical forests of Africa, America, and Asia. In: Degens, E. T., Kempe, S. and Herrera, R. (Eds) Transport of Carbon and Minerals in Major World Rivers, Pt. 3. Mitt. Geol.-Paläont. Inst. Univ. Hamburg, SCOPE/UNEP Sonderbd. 58, pp. 123-36.

Lanly, S.-P. (1982) Tropical Forest Resources. F AO, Rome, 106 pp.

Lieth, H. and Aselmann, I. (1983) Comparing the primary productivity of natural and managed vegetation. An example from Germany. In: Holzner et al. (Eds) Man's Impact on Vegetation. Dr W. Junck Publ. The Hague, Boston, London.

Logan, J. A. (1983) Nitrogen oxides in the troposphere: global and regional budgets. J. Geophys. Res. 88, 10785-807.

Melillo, J. M. and Gosz, J. R. (1983) Interactions of biogeochemical cycles in forest ecosystems. In: Bolin, B. and Cook, R. B. (Eds) The Major Biogeochemical Cycles and their Interactions, Wiley, New York, pp. 177-222.

Myers, N. (1981) The Hamburger connection: how Central America's forests become North America's Hamburgers. Ambio 10, 3-8.

Richards, J. F., Olson, J. S. and Rotty, R. M. (1983) Development of a data base for carbon dioxide releases resulting from conversion of land to agricultural uses. Inst. Energy. Anal., Oak Ridge Associated Universities, ORAU/IEA-82-10(M), ORNL/ TM-8801.

Rosswall, T. (1983) The nitrogen cycle. In: Bolin, B. and Cook, R. B. (Eds) The Major Biogeochemical Cycles and their Interactions, Wiley, New York, pp. 46-50.

Rotty, R. M. (1987) Estimates of seasonal variations in fossil fuel CO₂ emissions. Tellus 39B, 184-202.

Schlesinger, W. H. and Melack, J. M. (1981) Transport of organic carbon in the world's rivers. Tellus 33, 172-87. .

Vitousek, P. M. (1983) The effects of deforestation on air, soil, and water. In: Bolin, B. and Cook, R. B. (Eds) The Major Biogeochemical Cycles and their Interactions, Wiley, New York, pp. 223-45.

Warneck, P. (1987) Chemistry of the Natural Atmosphere. In: Donn, W. L. (Ed.) Intern. Geophys. Ser., Vol. 41, Academic Press, San Diego, 757pp.