Phosphorus has been studied extensively in aquatic systems because it is considered important in the control of eutrophication (Vollenweider, 1981). Eutrophication (eutrophic = well nourished) results from increases in such plant nutrients as P, N, Fe, C, S and Si, and the consequent stimulation of plant production and growth. In aquatic environments, almost all P is present as organic P in living and dead material. A small fraction is found as dissolved organic or inorganic P. In very turbid waters with large concentrations of suspended material, a significant portion of P can exist adsorbed to particles (Grobbelaar, 1983). The analytical separation between particulate and dissolved P is based on physical and chemical properties and partly on their origin (Broberg and Persson, 1988). In nature, P generally occurs as ortho-, poly- or metaphosphates (Van Wazer, 1958), all of which constitute dissolved inorganic P (DIP). Orthophosphate is released from weathering rocks, whilst the other forms are products of biological metabolism (Olsen, 1967). Aquatic chemists often divide orthophosphate into dissolved or soluble reactive P (DRP or SRP), dissolved unreactive P (DUP), total dissolved P (TDP), and particulate P (PP) (Sharpley et al., Ch. 11). One aim of chemical analysis is to quantify the bio-availability of P. Bioassays to determine P availability have been widely applied, but Boström et al. (1988b) concluded that these quantify P poorly in freshwater.

Numerous examples of relationships between P concentration and aquatic productivity have been published. Vollenweider et al. (1974) derived the following relation between annual phytoplankton primary production (Prod) and annual P loading (TPl), for the Laurentian Great Lakes:

 

[1]

 

After analysing the summer total P, total N and epilimnetic chlorophyll in 493 lakes, McCauley et al. (1989) concluded that the relationship between nutrients and chlorophyll is sigmoid. Their analysis also showed that total N accounted for a significant proportion of the variability in chlorophyll content, especially at high summer total P concentrations. These characteristics have been used to place lakes into different trophic categories (Table 1).

Nitrogen occurs in the aquatic environment as dissolved molecular N2, ammonia (NH4), nitrate (NO3), nitrite (NO2), and various organic compounds, such as amino acids, amines, proteins and humic compounds. Nitrogen is also an essential nutrient. The cycles of N and P are very different: N undergoes gaseous transformations, its major reservoir is in the atmosphere, and the majority of flows and exchanges are mediated by organisms. The major P reservoir, on the other hand, is in soils and sediments, and physical-chemical processes are responsible for the major flows. Because P is predominantly found in the earth's crust, it is more easily influenced by man, than N.

 

Table 1. General ranges of phytoplankton production, total P, total nitrogen and chlorophyll a of lakes of different trophy (modified from Wetzel, 1983).

Trophic Type	Primary Productivity	Total P	Total N	Chlorophyll a
	mg C m-2 d-1	---------------- µg l-1 ----------------
Ultraoligotrophic	<50	<5	<250	<0.5
Oligotrophic	20-100	5-10	250-700	0.3-3
Mesotrophic	100-300	10-30	500-1000	2-15
Eutrophic	>300	10-50	500-2500	10-500
Hypereutrophic	>1000	30-5000	500-15000	>100

Phosphorus is usually limiting productivity in freshwater ecosystems (Likens, 1972; Schindler, 1977), whereas N generally limits primary production in marine systems. Phosphorus has been shown to contribute to the eutrophication of many freshwaters (Lean, 1973) and its control is most probably the best strategy for lake management and limiting eutrophication (Vollenweider et. al., 1974; Toerien, 1977). Controlling the N supply is not considered to be a viable option, since algae are known to fix atmospheric N2. This input of N into aquatic environments has only been shown to be important in the last two decades (Hardy et al., 1973), when N fixation has been correlated with the presence of heterocyst containing cyanobacteria (blue-green algae) (Fogg, 1974). Schindler (1977) showed that the long-term phosphate availability is important to bloom-forming N2 fixing species in nitrogen-depleted waters.

 

Uptake of P and N

Most aquatic systems are resource-limited, where P and N are often the primary limiting nutrients. To ensure survival a competitor must be able to maintain net population growth at resource levels less than those required by other species. Algae are particularly adapted to scavenge their environments for resources, be it through structural changes, storage or increased resource utilisation efficiency. Internal adjustments by algae involve biochemical and physiological adaptations, whilst they can also excrete substances to enhance nutrient availability. Algae excrete extracellular phosphatases almost immediately upon the onset of P limited conditions (Healy, 1973). Algae can also excrete other compounds and change the pH of their surroundings, which in turn can render adsorbed P available (Grobbelaar, 1983).

In addition, algae can store resources like P in excess of their immediate needs. This excess or "luxury" uptake is clearly distinct from the Michaelis-Menten or Monod (1950) nutrient uptake kinetics which are based on external resource concentrations. Epply and Strickland (1968) concluded that the growth rate of phytoplankton is more closely related to the cellular nutrient content than to external concentrations. It is, therefore, necessary to establish a relationship between the cell quota of a nutrient and the growth rate of an alga. Such a relationship was given by Droop (1968, 1983) and in a generalised form it is:

 

[2]

 

where µ = specific growth rate, µmax = maximum specific growth rate, kq = the minimum cell quota for the limiting resource or the subsistence quota and Q = cell quota for the limiting resource. This model has been applied to a number of species and nutrients such as; P, N (NO3, NH4 and urea), Si, Vitamin B12 and Fe (Droop, 1983). There have, however, been cases where the model did not work, notably with NH4 limited growth of Monochrysis and Dunaliella (Caperon and Meyer, 1972; Laws and Caperon, 1976).

Nutrient storage capacity is defined in terms of the cellular quota necessary for maintenance and growth. The ratio of kq/Qmax has been termed the luxury storage coefficient (Qmax = maximum cell quota). The greater the difference between Qmax and kq, the greater is the organisms’ quota flexibility and therefore, the potential to adapt to nutrient limitation. In situations of decreasing nutrient concentration, the growth rate of an organism with a constant cell quota will decrease accordingly. When an organism can decrease its cellular requirements for a nutrient, it will be able to offset much of the decrease in resource availability (high quota flexibility), thus minimising the effects of nutrient depletion on the growth rate.

In terms of the steady-state nutrient assimilation, equation (2) can be written as (Droop, 1983):

 

[3]

 

where [S] = steady state substrate concentration and Ks = the half-saturation constant for steady-state nutrient uptake. The hyperbolic relationship between the substrate concentration (Monod kinetics of equation 3, Figure 1), cell quota (equation 2, Figure 2) and specific growth rate is clearly seen. The variation in growth response to different Ks values (Figure 1) at a specific µmax shows increased growth rates at low substrate concentrations with low Ks values and vice versa. Different degrees of quota flexibility at a specific Qmax, with different minimum cell quotas (kq), show that the smaller µmax (i.e. the greater the quota flexibility) becomes, the steeper the initial slope of µ (Figure 2). The slope has a direct influence on the half-saturation constants, being high for low quota flexibility and low for high quota flexibility. Low half-saturation constants are typical of P and N, especially NH4 (Gilbert et al., 1982), whereas high half-saturation constants are typical for carbon (Turpin et al., 1985).

 

 

 

Figure 1. A hypothetical example showing the specific growth rate (µ) of an alga against substrate concentrations (S) and the concept of algal quota flexibility adaptation. Monod kinetics are similar to equation 3 and Ks vary from 0.1 to 1.5.

If an alga can adapt its quota flexibility, it might out-compete competitors in a nutrient-limited environment. High quota flexibility is common for resources found at low concentrations (P and N), whilst a low quota flexibility is found for resources that are present at high concentrations. If an alga can adapt its quota flexibility by lowering its half-saturation constant, this means a higher initial slope of the nutrient-growth curve. A higher initial slope implies a high growth rate at low resource concentrations, and thus a competitive advantage of a particular organism (Figure 1).

 

Competition for limiting resources

Resource limitation may refer to the yield attainable from a given limiting nutrient, usually the primary limiting nutrient, or the rate at which the final yield is attained. Phosphorus might be the primary limiting nutrient in a particular system, which would determine the final yield, but the available light energy would determine the rate at which this final yield is reached, being low under low light and vice versa. Models such as the Monod (1950) and Droop (1968) equation (2) describe both the final yield and the rate at which this final yield is reached, for a single limiting resource.

 

 

Figure 2. Specific growth rate against the cell quota (Q), using equation 2 and varying kq over a range of 0.1 to 5 for a hypothetical alga.

Three scenarios, which cover most of the possibilities for a single limiting resource are shown in Figure 3. As nutrient availabilities wax and wane between the seasons, phytoplankton with different µmax and Ks values will respond differentially. In example A (Figure 3) the maximum growth rates (µmax) of the two competing species are the same, but the nutrient half-saturation constant (Ks) of species 2 is greater than that of species 1. The growth rates of the two species will be similar after the spring overturn, when the nutrient content in the water is high, and no one will dominate the other. As the nutrients become depleted in the euphotic zone, species 1 will progressively become the dominant alga in the system. When the µmax of species 2 is greater than that of species 1 (Figure 3B) and the Ks values are similar, species 2 will always dominate the phytoplankton population and this domination will be progressively greater, the higher the nutrient concentration. In example C, both the µmax and Ks of species 2 are greater than those of species 1. At high nutrient concentrations, which prevail after the spring overturn in a water body, species 1 will dominate the phytoplankton population. As the nutrients become depleted, species 2 will become the dominant alga in the system.

In the above, rather simplistic scenarios many other factors, such as light and temperature, or pH and the availability of CO2, can influence the competition for resources. The relationships of Figure 3 can also be expressed in terms of the mortality rate (D), where a population can only maintain itself if the growth rate (µ) is equal or greater than D. The resource concentration (R) at which the net growth rate equals zero (Tilman, 1977) can be expressed in terms of the Monod equation:

[4]

Figure 3. Three scenarios of the interaction of two species at different maximum specific growth rates and half saturation constants.

 

Many resources can of limit phytoplankton growth rates. Liebig’s law of the minimum states that only a single growth factor can be limiting at any given time. The threshold model of Droop (1974) similarly states that the growth rate of an organism may be limited only by a single resource and interactions with other potentially limiting resources are minimal. Since different organisms in a population could be limited by different limiting resources, it is necessary to identify the limiting factors for the various populations of a community.

An important concept, particularly for N and P supply concentrations, is the optimum nutrient ratio. The optimum nutrient ratio is the ratio at which a transition from one nutrient limitation to another occurs (thus both could be limiting), or where the cellular ratio of resources required is such that the resource is not in short supply relative to another (Rhee and Gotham, 1980). Since the internal concentration of nutrients is important in determining uptake rates, it is possible to determine both the limiting resource concentration and the consumption rate at the transition point where limitation occurs. If the optimum N:P ratios for two species are 20 and 10 respectively, then both will be P limited when the ratio is >20. However, the second species will be more P-limited than the first. If they have similar µmax values, the first species will eliminate the second species at N:P ratio's >20 (e.g. species 1 in Figure 3A). Since a limiting nutrient can be defined as the one with the smallest Q:kq ratio (Droop, 1974), transition between N and P limitation occurs when:

 

[5]

 

According to (Rhee and Gotham, 1980) the relationship between QN:QP and the optimum ratio holds true only when µmaxN = µmaxP. The optimum N:P ratio is therefore the ratio kqN:kqP where µ H 0, or the ratio QN:QP at very low growth rates. The maximum growth rates are not always equal for N and P, because of different storage pool sizes or because of different Q:kq ratios (Goldman and McCarthy, 1978). The result of this is that the QN:QP ratio deviates considerably from the optimum ratio, especially at high growth rates (µ). Rearranging equation (2) for Q, it becomes:

 

[6]

 

The optimum ratio for N:P, showing the dependence of QN:QP on relative growth rates (Figure 4) (Turpin, 1988), can then be written as:

 

[7]

 

On either side of the curve, either N or P limits growth. Note that the higher the growth rate, the more N pro rata is required and vice versa. Experimental support of this growth rate dependence of the optimum nutrient ratio was obtained by Terry et al. (1985) and Turpin (1986). An important detail is that the optimum N:P ratio varied between species and over the diurnal cycle (Rhee and Gotham, 1980). Ahlgren (1985) showed that algae were able to adapt to different N:P ratios at lower growth rates and that the ratio becomes more fixed at higher growth rates.

 

 

Figure 4 Growth rate dependence for the optimal N:P ratio of an alga, showing the P- and N-limited growth regions.

 

It has been shown that optimum N:P ratios vary only slightly (Terry et al., 1985), but large variations have been found for C:P (Turpin, 1986). These optimum ratios are important for the competition and coexistence between species, and for the stability of the system. Since growth rate is influenced by the optimum N:P ratio for a given species, optimum ratio curves for different species could cross. At low growth, one species might be P limited, another N limited. At growth rates higher than the crossover point, the situation would be reversed, which would influence the competition and dominance between species. At the crossover point, also termed the optimum ratio equivalence point, neither one of the species has an advantage over the other (Turpin, 1988).

 

Nutrients, their ratios and species dominance

Complex interactions between chemical, physical and biological factors in water bodies determine which algal species dominates. The structure of a community is particularly dynamic in responding to changes in environmental conditions including more subtle effects associated with competition between species, competition with macrophytes and grazing pressures by zooplankton and other invertebrates. As discussed above, species generally differ in their ability to utilise nutrients: through adaptation to capitalise on available supplies, utilisation of nutrients at chronically depleted concentrations, and development of intracellular storage mechanisms to optimise growth under fluctuating nutrient concentrations. In principle, the concentration of a nutrient that is limiting, can be estimated from the Redfield ratio (106C:16N:1P) for the key nutrients. Reynolds (1992) has estimated that the P requirements of common algae are satisfied at SRP concentrations of ca. 0.07 µmol l-1. When concentrations are lowered, a delayed response is expected as the internal reserve is depleted before growth is affected. Hence P concentrations may become exceedingly small before becoming limiting. In this situation, the ratio of N:P can not serve as an indicator of any deficiency although abrupt changes in the ratio near the P limiting conditions are diagnostic of such conditions. If N, P and Si are in excess of requirements of the algae, the ratios are irrelevant.

In lentic systems N:P ratios are sometimes useful because plankton activities in the summer can drive nutrient concentrations to levels where specific growth rates are less than their maximum. High N:P ratios in rivers indicate that algae are potentially P limited if the ratio remains constant with decreasing nutrient concentrations. Most large rivers have a sufficiently high P load to maintain algal growth without appreciable effects on the SRP concentration. In rivers and lakes the concentrations of the key nutrients, together with other factors influencing algal physiology, determine the relative growth rates and structure of the community. Although N and C are key nutrients for most algae, the availability of a third nutrient may limit the growth of particular species, e.g. when non-diatoms dominate diatoms under low dissolved silicon concentrations. Such nutrient "successions" are particularly important in lakes where the growing season is long enough for the various community changes to be expressed.

The interactive effects between nutrients and trace components, e.g. vitamins or trace metals, raise difficult problems for the prediction of community structures. Studies have either emphasised an essentially non-interactive threshold effect, or multiplicative effects where growth is determined by sub-optimal concentrations of the micro-nutrient (Talling, 1979). The threshold effect is perhaps best illustrated by the macro-nutrient examples given by Ahlgren (1988) from the data of Rhee (1978) and Rhee and Gotham (1980). For green algae, Scenedesmus sp., at molar N:P ratios of <30, growth was determined solely by N limitation and at ratios >30 solely by P limitation. There was no multiplicative effect of the two nutrients. This is consistent with a sharp transition between N and P limitation. Other studies have shown a less sharp transition and a critical ratio of N:P dependent on the growth rate (Ahlgren, 1988; see also Figure 5).

The chemical form of N and P is also important. Experiments with Nitzschia palea and Cyclotella striata have shown that N was preferred in the form of NaNO3 and P in the form of hydrogen phosphate (Giri and Chowdary, 1992). Such chemical preferences also influence species dominance. Cyanobacteria (Microcystis sp.) blooms, which sometimes occur for prolonged periods in shallow eutrophic lakes, may reflect the preference of these algae for dissolved NO3 rather than NH4. As the total N decreases, the riverine flux of NO3 is sufficient to support Microcystis but the NH4 concentration is very low and limits the production of coexisting algal species.

There are many examples of the changes in algae and macrophyte communities following increased nutrient loadings to water bodies. The Everglades National Park is a recent example showing the initial signs of the impact of increasing P concentrations caused by intensive agriculture to the north of the park. At one documented site, P loads from canals are causing changes in periphyton and macrophyte communities along the periphery of the canals. Periphyton diatom diversity and taxon numbers are increasing in association with increased amounts of sediment P (Raschke, 1993). Similarly in some large slow-flowing rivers in Europe, eutrophication has caused increased problems with phytoplankton densities. Many large nutrient rich rivers, e.g. Thames, Rhine, Danube and Meuse, have similar phytoplankton populations generally dominated by diatoms and chlorophyceae. There is evidence that sections of the lower Rhine receiving coastal waters are deficient in Si (Admiraal et al., 1993). In such systems diatoms could deplete dissolved Si, resulting in a surplus of N and P, which in turn causes non-diatom blooms in coastal waters. In the absence of nutrient limitation in these rivers, the main factors controlling the biomass and structure of the phytoplankton are the hydrodynamic conditions, light penetration, temperature of the water and the grazing by zooplankton. Temporal and spatial changes in community structure caused by changing pollution of rivers by toxins is also possible (Descy, 1987). It is not always clear whether changes in community structure in downstream sections of rivers are a result of a local evolution of the population or whether upstream factors are dominant (Lack, 1971). In the past, upstream sections usually had low P concentrations which limited the development of phytoplankton. Increased P inputs subsequently have led to community establishment in upstream sections so that a net increase in phytoplankton populations is also observed downstream (Descy, 1992).

 

The role of sediments in P-cycles

Dissolved P in both inorganic and organic forms usually interacts strongly with suspended and bed sediments. Many of these interactions, including those biologically mediated, are heterogeneous in nature and it is therefore likely that the kinetics of the processes rather than chemical equilibria determine the water composition. This is particularly true of the interactions of dissolved inorganic phosphate with sedimentary minerals and biological uptake by algae and macrophytes in rivers and lakes. The variety of sedimentary minerals poses special problems because P interacts with surfaces by formation of specific inorganic surface complexes. The non-specific hydrophobic interaction, which dominates the energetics of sorption of many organic pollutants, does not appear to be important for dissolved phosphates. The nature of the specific interactions for many systems is still uncertain, because:

 

• the wide range of affinities of P for sediments combined with the uncertainties in the composition of the sedimentary materials make it difficult to identify the key processes.

• dissolution/precipitation, adsorption/desorption and biological uptake and release are difficult to separate. Adsorption/desorption usually produces relatively weak interactions which are rapidly reversible unless transfer is diffusion controlled.

• the transformations of organic P to inorganic P are not well known. Mineralisation reactions mediated by microbes are particularly important in sediments containing P associated with organic material.

 

Some of the fundamental questions about the role of sediments in the cycling and fate of P in freshwaters and marine ecosystems are still not answered. At the moment we have information for specific systems and locations, such as the uptake and release of P under defined conditions. These data are useful in predicting the range of P fluxes in specific environments; they are, however, of limited value in providing information about the key sedimentary processes. Two approaches have provided information on P interactions with sediments:

 

• mechanistic studies of particular phosphate-mineral interactions in the laboratory with applications to understanding processes in rivers and lakes. Two main processes fall in this category: in hard waters, the heterogeneous nucleation of calcite and co-precipitation of P in rivers and lakes; and in soft waters, the formation of co-precipitates of FePO4 and Fe(OH)3 in suspended solids and bed sediments.

• empirical approaches using laboratory microcosms to simulate processes in the field, e.g. use of intact lake-sediment cores to measure P release into water at various redox/pH conditions.

 

The former approach provides reasonable detail of the mechanisms in some systems that can be applied over a wide range of environmental conditions. The latter provides information about particular sediments with a relatively poor understanding of the mechanisms in operation. A great deal of effort is often needed to separate the important abiotic and biotic processes, and usually the studies produce only net release or uptake rates of P.

Mechanistic studies aim to understand P-sediment interactions in detail and need to consider the composition of the water, particularly the chemical speciation of the inorganic phosphate. In hard waters it is essential to consider the formation of calcium phosphate ion-pairs such as CaHPO40, which typically contribute 30% of the SRP in a hard water (~3mM Ca and 0.6 mM dissolved inorganic phosphate). In some natural waters, diurnal changes in CO2 and O2 caused by photosynthesis and respiration, lead to changes in the pH and Ca saturation of the water. This is particularly important near active cell surfaces (Leadbeater and Callow, 1992) where the pH may increase to above 10 because of CO2 depletion and associated chemical concentration gradients at the interface. Many reports have documented the precipitation of calcite in different lakes at various trophic levels, see for example the comprehensive summary of Kuchler-Krischun and Kleiner (1990). Reports of lake "whiting" (Murphy et al., 1983) indicated that the process is mainly due to photosynthetically induced increases in pH. Formation of a precipitate causes a decrease in light transmission, self-flocculation of suspended matter and a reduction in dissolved nutrients such as phosphate and low-molecular weight organic solutes, and has been described as a "self-cleaning" mechanism particularly during periods of intense photosynthesis (Rossknecht, 1980; Koschel et al., 1987; Raidt and Koschel, 1988). A concentration of calcium carbonate in the suspended solids of 5 mg l-1 has been reported in Lake Constance in the spring (Kuchler-Krischun and Kleiner, 1990).

The two main effects of the precipitation of carbonates on the transport of P in temperate hard water lakes are (a) the direct incorporation of P in the precipitated calcite, effectively removing dissolved inorganic P (and possible some dissolved organic P) from the water column, and (b) removal of particulate P bound in algae aggregated with calcite crystals. The aggregation of algal cells with the carbonate leads to increased settling rates and flux of P to the sediment.

The co-precipitation of P and calcite in lakes has now been widely documented (Kleiner, 1988; Jager and Rohrs, 1990; Otsuki and Wetzel, 1972; Murphy et al., 1983) and studied in laboratory conditions (Kleiner, 1988; House and Donaldson, 1986; House, 1990; Ishikawa and Ichikuni, 1981; House et al., 1986; Giannimaras and Koutsoukos, 1987; Hinedi et al., 1992). The results indicate that the loss of inorganic P may be predicted from the concomitant loss of calcium from the water. The prediction requires chemical data on the concentration of SRP, total dissolved calcium and the pH of the water (House, 1990). The amount of phosphate co-precipitated is slightly variable depending on the presence of other surface active compounds dissolved in the water (House et al., 1986). In general, solutes which inhibit precipitate formation are also likely to compete with phosphate ions for the surface and lead to lower surface densities of P and less co-precipitation. The loss of P from lakes by this mechanism is substantial with values between 25 and 45% of the total P removed from the epilimnion. It has also been suggested that in eutrophic hard-water lakes, a release of P from the sediment into the lake does not occur, so that the co-precipitated P is effectively trapped in the mineral substrate (Jager and Rohrs, 1990). High resolution NMR studies (Hinedi et al., 1992) have shown that at low P concentrations typically found under natural conditions, i.e.
< 0.79 µmol adsorbed per g of CaCO3, that the phosphate adsorbed to the calcite (and incorporated in the mineral) is most likely unprotonated and is not in the form of hydroxyapatite or amorphous calcium phosphates.

It is also known that, at high concentrations, adsorbed P effectively inhibits the surface nucleation of calcite and thus calcite precipitation. The "self-cleaning" mechanism may fail when the P loading increases to an extent that the precipitation reaction is inhibited. However, once growth occurs, co-precipitation of P continues (House, 1987; Kleiner, 1988; Giannimaras and Koutsoukos, 1987; Grases and March, 1990).

Another mechanism studied in detail is the formation of solid solutions of amorphous ferric phosphate in amorphous ferric hydroxide in soft waters. The solid phase is reported to have a solubility dependent on the pH and on the mol fraction of ferric phosphate associated with the hydroxide (x) (Fox 1991, 1993):

[8]

The reaction is probably the result of the interaction of phosphate with the surface hydroxyls similar to that invoked to explain the adsorption of P on minerals such as allophane, ferrihydite, goethite and iron oxides (Frossard et al., Ch. 7), but occurring during the formation of the iron hydroxide. Hence the composition of the solid solution may depend on the kinetics of the formation as well as the solution concentration of the matrix ions. In spite of this, it has been possible to formulate a solubility relationship describing the water composition in equilibrium with a particular solid solution. If the ratio of the solubility products of the iron phosphate and iron hydroxide is Ksp, then the solubility product for the solid is given by:

[9]

where [10]

and ai are the activities of the species i. The kinetics of the solid solution formation have not been studied in detail but a pseudo-equilibrium from under- and over- saturation appears to be feasible within a matter of days. Fox (1989) has examined the role of iron phosphates/hydroxides in determining the concentration of dissolved phosphate in laboratory experiments with waters from several rivers including the Delaware, Amazon and Negro. Studies on the Sepik River (Papua, New Guinea) and Hudson River (N.Y. State, USA) have supported the proposed mechanism.

The influences of these specific reactions in lake sediments have been less studied although the importance of redox conditions at the boundary layer in controlling the balance between ferrous, Fe(II), and ferric iron, Fe(III) and cycling of P have been known for a long time (Mortimer, 1941). Bacteria regulate the redox potential and thus indirectly affect abiotic reactions, e.g. when the potential falls below ca 230 mV the chemical reduction of Fe(III) occurs with the release of Fe(II) ions into the associated water. It is also known that anaerobic microbes mediate directly in the reduction of Fe(III) to Fe(II) and also influence the chemical reduction of Fe(III) by sulphides after microbiological reduction of sulphate.

Empirical approaches using laboratory microcosms to simulate processes in the field have yielded transfer rates, usually expressed in terms of the P flux across the sediment-water interface. When the flux is not diffusion controlled and there is sufficient mixing of the sediment and water, e.g. in suspended sediments, chemical/biological reactions determine the rates. In most river-bed and lake sediments, the release and uptake rates of P are likely to be controlled by mass transport in the sediment profile and at the water interface. The processes producing or consuming soluble P in the sediment will provide the concentration gradient or "driving force" for P movement to and from the water. Sediment processes include precipitation (some redox related), adsorption/desorption, microbial assimilation or mineralisation of organic P, bacterial/algal cell lysis and diffusive transport into biofilms, aggregates and mineral pores. These processes are influenced by many factors such as the interstitial water composition, temperature, water movement and sediment mineralogy, and are therefore highly specific to river or lake conditions.

For well-mixed conditions, e.g. suspended sediments or some surface river sediments, studies of sediment conditions imitating those expected in the field are possible. A parameter which has proved useful is the "equilibrium P concentration", EPC0, of the sediment. This is the concentration of SRP in a solution in contact with the sediment when the adsorption and desorption rates of P are equal, i.e. no net uptake or release of P occurs. Because of continually changing conditions, sediments are unlikely to be in equilibrium with their associated water. However, the difference between the EPC0 of a sediment and the SRP determined in the water is a indication of whether the sediment is releasing or adsorbing P. The procedure for the experimental determination of the EPC0 usually involves incubation of the sediment in a CaCl2 solution spiked with different concentrations of inorganic P, followed by separation of the sediment and determination of the SRP and EPC0. The EPC0 may be sensitive to the electrolyte (e.g. Ca2+) concentration (Klotz, 1988), and to other factors such as temperature and pH. Ideally a solution similar to the natural water should be employed, preferably the water originally in contact with the sediment. It is also essential that fresh rather than dried sediment is used because the adsorption/desorption properties of the sediment and the microbial activity are affected by drying. Attempts to sterilise sediments may cause interference of the preservatives with sorption sites, change the ionic strength or cause P release through cell lysis, and thus affect the kinetics of release and uptake. The exchange kinetics have fast and slow components (Fox, 1989; Chen et al., 1973), because rapid surface reactions co-occur with diffusion within the mineral matrix or in interstitial solutions within sediment aggregates, dissolution of phosphate containing minerals and biological degradation of organic P.

Results of EPC0 determinations have shown a reasonable correlation with the SRP concentration in the waters at the time of sampling (Meyer, 1979; Klotz, 1988; Klotz, 1991) with the SRP concentrations generally above the EPC0, indicating that the sediments are capable of P uptake. These results must be qualified, though, because the EPC0s were determined under standard conditions at 20°C which may have caused a systematic deviation between the SRP concentration and EPC0. EPC0s from the Maumee River, USA, have also been found to be positively correlated with the TP of the sediment (Green et al., 1978), showing the increased concentration of exchangeable P with increased total P of the sediment. No comparable measurements are available for suspended lake sediments.

Transfer of P to lake sediments occurs through deposition of particulate matter, such as organic detritus, riverine particulates and mineral precipitates formed in-situ such as calcite or Fe-hydroxides. Deposition usually exceeds the resuspension of sediments and soluble P release. Although, on an annual basis, the sediments are normally a net sink of P, the seasonal release of SRP is an important nutrient supply, particularly in shallow lakes (Cullen and Forsberg, 1988). This internal loading of the lake is an important factor during the recovery of lakes following a reduction in the external loading, e.g. through the introduction of tertiary sewage treatment or a reduction in detergent phosphate. Unfortunately there is no simple way to estimate the release rates of P from lake sediments. Most studies have measured the rates using intact sediment cores, and studies are difficult to compare because of differing experimental conditions, viz temperature, degree of mixing, redox potential and pH of the water, biological activity and mineralogy of the sediment. The most important controls are:

 

• the temperature of the sediment which influences the mass diffusion rates and microbial activity which in turn affects the redox potential in the sediment and Fe(II)/Fe(III) balance. An example are the measurements of Kelderman (1984) of the average sediment-water exchange flux of 12.9 mmol m-2d-1 at 5°C and 355 mmol m-2d-1 at 20°C measured for the Lake Grevelingen sediments;

• the bioturbation of the sediment by benthic organisms which, in some situations, overshadows effects of oxygen concentration or water mixing (Holdren and Armstrong, 1980);

• the redox potential of the sediment which influences the dissolution of Fe(III) minerals and concomitant release of P, and the mineralisation rates of organo-phosphates and release of SRP from cytoplasm (Montigny and Prairie, 1993) through the conversion of cell polyphosphates formed in oxic conditions to inorganic soluble phosphate in anaerobic environments;

• the mineralogy of the sediment which influences the chemical speciation of P. This has been demonstrated in numerous studies using a variety of sequential extraction schemes and sediment characterisation methods, e.g. (Hieltjes and Lyklema, 1980). In general calcareous sediments have a lower adsorption capacity than iron-rich ones or sediments dominated by clay minerals.

 

Measured release rates from intact sediment cores vary greatly from <1 µmol m-2 d-1 to 3000 µmol m-2 d-1 (Keizer et al., 1991; Wisniewski, 1991; Boers and van Hese, 1988; Holdren and Armstrong, 1980). The application of these results to predictions in the field are still uncertain and at best provide some guidance to the importance of internal loadings for particular lake conditions. Golterman et al. (1977) commented that the factors that "operate in different limnological environments cannot readily be predicted quantitatively without prior investigation of the particular water body. Thus when one is faced with a management or restoration problem, one can presume little on the basis of other work before one may usefully develop a model of P transfers." It seems that this situation has changed little over the last 17 years. Although the details of the individual processes controlling P dynamics in sediments are better understood, their application to complex environments is still limited.

 

Phosphorus cycling within aquatic ecosystems

There are two broad groups of processes involved in the P cycling within aquatic ecosystems: the "internal" cycle where P-transfer mechanisms are biologically mediated and the "external" cycle with physico-chemical complexation and release reactions (Golterman, 1975). Biologically mediated transfers occur rapidly, whereas the external processes are slow. Lean (1973) has proposed a quantitative, steady-state model of P exchanges in the epilimnion during the summer, which shows that the bulk of P is in the particulate fraction. Two other components are the colloidal and the low molecular weight P fractions. Transformations of the particulate and the colloidal fractions are slow. Only a small fraction is SRP which has an extremely short turnover time (e.g. Rigler, 1964). Flow rates between these components are greatly dependent on the season and to a lesser degree on the organisms present in the system.

Golterman (1984) combined both P and N cycles in a schematic model, which includes both the sediments and the pelagic zone of a water body (Figure 5), but excludes an anoxic hypolimnion where different reactions and transformations take place (Wetzel, 1983). Molecular N2 is fixed and incorporated in algal biomass (1). Inflowing water contains various P and N forms which are taken up by the phytoplankton (2), enter the dissolved pool (3), or sediments (4). Various forms of P and N can leave the system (5) and N2 can leave the anaerobic sediments through denitrification (6). There is a clear distinction between the pelagic zone and the sediments. In the open water, orthophosphate is taken up by phytoplankton (7) to be incorporated into the particulate or organically bound P fraction (8), and to be re-released as orthophosphate (9), by mineralisation of the organic material or as excreta. Phosphorus entering the sediments either as orthophosphate (10) or particulate organic P, undergoes transformations through redox reactions, mineralisation, adsorption, and desorption. Phosphorus is slowly released from the sediments (11) and enters the orthophosphate pool of the water. Both nitrate and ammonium are taken up by aquatic plants (12), and are incorporated into various organic fractions (13), either particulate or dissolved. Depending on the oxygenation of the water, nitrification and denitrification transformations can take place, to convert the organic-N into several oxidation states; NH4, NO2, NO3 and N2(14). Nitrate and ammonia then become available for uptake by aquatic plants. Ammonia is constantly released (15) to the water from the sediments, and consequently the C/N ratio of the detritus becomes too high for further mineralisation. When this happens, a considerable quantity of the N entering a water body can be fixed in the sediments. The above cycles can have turnover times of 10-20 times per annum depending on the trophic status of the water body and on climatic conditions (Golterman, 1984).

 

Figure 5. Simplified diagram of the P- and N-cycling in a lake (modified from Golterman, 1984).

A factor not considered in the model (Figure 5) is the suspended inorganic particulate material, which is common in many of the worlds’ freshwaters. These suspended materials have two major impacts on aquatic ecosystems: (1) they attenuate light entering aquatic systems and hence reduce phytoplankton productivity (Grobbelaar, 1985) and (2), being charged, they form adsorption and desorption surfaces. Sediments are important sinks and sources for P in aquatic habitats (Golterman, 1984). Grobbelaar (1983), Young and DePinto (1982), Viner (1988), and Engle and Sarnelle (1990) have shown that suspended sediments could supply phosphate to algae grown in culture, while the same was shown for bottom sediments by Golterman et al. (1969) and Grobler and Davies (1979). Grobbelaar (1983) demonstrated that N could also be adsorbed on suspended inorganic sediments and that algae could utilise this adsorbed fraction.

The quantities of P and N adsorbed on suspended sediments, as well as the mechanisms involved in release processes are still largely unresolved. Grobbelaar (1983) suggested that release might be due to extracellular products from the algae, which would release the adsorbed nutrients on the microzonal scale. The adsorbed fraction available to algae is extremely important, but this component is often ignored in the analyses of freshwaters. These bioavailable portions of P and N are important for understanding P- and N-cycling in lakes and in refining models aimed at water management strategies. Bioavailable P is defined as the sum of the immediately available P and the P that can be transformed into an available form by naturally occurring physical, chemical and biological processes (Boström et al., 1988b). This is also applicable to N. Whereas P adsorbs fairly strongly onto suspended inorganic particles, adsorbed N is readily available to algae (Grobbelaar, 1983). Grobbelaar (1992) has also shown that both the adsorbed P and N fractions on suspended inorganic particles exhibit a seasonality and that the quantities available to phytoplankton vary considerably (Figure 6).

 

Figure 6. Variations in the algal-available P and N on suspended sediments in the highly turbid impoundment, Wuras Dam.

Despite the wealth of information on P-cycling there are important gaps in our understanding (Taylor and Lean, 1991):

• It is still not possible to quantify phosphate at levels typical of P-limited waters, and SRP determination can greatly overestimate phosphate at low levels (comp. Herodek et al., Ch. 17).

• Bacteria dominate the uptake of phosphate in most lakes at low concentrations of SRP. This poses the question of how algae obtain P and why bacteria take up more inorganic nutrients than the primary producers?

• Zooplankton, although it regenerates P by concentrating it, does not have the expected impact on the P dynamics.

 

Phosphorus and eutrophication

Anthropogenic eutrophication has been identified a major factor affecting aquatic ecosystems since the turn of the century. Man has increased the supply of nutrients in precipitation, groundwater and runoff from land and consequently in flowing waters. There is consensus amongst limnologists that the term eutrophication is synonymous with increased growth rates of the aquatic biota and that this is due to perturbations of the system through man's activities. Both P and N have been linked to eutrophication (Table 1). The typical atomic ratio of phytoplankton biomass of 1P:7N:40C per 500 parts wet weight, indicates that P can generate 500 times and N 70 times its weight in biomass. This forms the basis of the numerous models predicting eutrophication, where perhaps the simplest are those showing a relationship between P and the chlorophyll content of a water body (equation 1; Vollenweider et al., 1974; Likens, 1972; Dillon and Rigler, 1974).

Therefore, the control of the P concentrations in aquatic systems has to be seen as the best means of controlling aquatic productivity and hence eutrophication. Nitrogen does not provide a suitable control, because both algae and bacteria can fix N2. At one stage it was thought that carbon was more important than P in the acceleration of eutrophication (Allen, 1972; Likens, 1972), but available inorganic carbon can limit phytoplankton productivity in eutrophic waters only under very specific conditions. This has been shown in very soft water lakes (very low alkalinity). Blue-green algae generally have lower Ks-values for inorganic carbon than most other algal groups, which explains their high productivity at low inorganic carbon concentrations. Diffusion of atmospheric CO2 is usually adequate to sustain the inorganic carbon requirements of the phytoplankton. The overwhelming experimental and applied evidence today shows the key role of P in the eutrophication of aquatic systems. Schindler (1974) demonstrated this in the experimental lakes experiments in which part of a lake was fertilised with P, N and C, whilst another part only received nitrogen and carbon. The basin which received the P very soon became eutrophied, while the basin that received only the nitrogen and carbon remained at prefertilisation conditions.

When P is added to a water body, a rapid increase in algal productivity occurs, but is not sustained (e.g. Mortimer and Hickling, 1954). It decreases rather rapidly and in a short time productivity is at levels comparable to those prior to the enrichment. Losses from the trophogenic region to the sediments are seen as the major reason for this (Figure 6, flow 10). Continuous inputs must be maintained to sustain the increased productivity. Such a continuous supply is generally referred to as the nutrient loading of a system. It is best described in terms of a mass balance of the nutrients (Vollenweider et al., 1974); for P it is:

DP/Dt = I - O - (S - R) [10]

where t = time, I = external nutrient load, O = nutrient loss by outflow, S = Nutrient loss to the sediment and R = nutrient regeneration from the sediment.

The relation between loading and concentration is given for orthophosphate in the water (Golterman, 1991):

PO4 - Pt = [11]

where L= loading (g m-2 y-1), z = mean depth (m), r= wash-out coefficient (t-1) and s = sedimentation coefficient (t-1).

Numerous models have incorporated nutrient loading as the variable to predict phytoplankton biomass or productivity. A reduction in nutrient loading is considered to be the only means to control eutrophication and this strategy has been followed in several countries. The threshold model, developed by the National Eutrophication Survey (Smith and Shapiro, 1981), predicts that the chlorophyll concentration in lakes would not decline unless the P was first reduced to a threshold concentration. They observed that in the 16 lakes investigated, a decline in chlorophyll followed a reduction in total P. However, the magnitude of the response differed between lakes and they found that some of the lakes responded in unique ways, which could not be predicted using the eutrophication model.

Tilzer et al. (1991) found in the mesotrophic Constance lake, that the SRP concentration has decreased about 50% over the past decade to a concentration of 1.6 mmol m-3. In spite of this, chlorophyll concentrations, water transparency and annual primary productivity, have not shown a downward trend. They concluded that the biomass accumulation is to a greater extent controlled by losses, mainly grazing by zooplankton and sedimentation, than primary resources such as P. More generally, eutrophication models may not predict responses adequately because:

• Ecosystems are controlled by a hierarchy of biotic and abiotic factors, and these controlling factors are a function of particular system properties.

• The response times to external perturbations as well as shifts in the internal control by feedback loops are highly variable between different components of a system.

• Because of random fluctuations, long-term observations are needed to determine the dominant controlling factors.

Schindler (1977) pointed out that controlling the P supply to a waterbody would cause an inversion in the N:P ratio with a consequent shift in the algal population from objectionable blue-greens to forms that are less objectionable. He also suggested that management decisions on nutrient control must be based on field tests and simple bioassays. A universal 1 mg P l-1 standard for effluents in sensitive catchments was legislated in South Africa, but this has been criticised (Pretorius, 1983), and Grobler (1985) has shown that as a result of hydrological variables in arid areas, the fate of P cannot be simulated in South African reservoirs. Production rates in turbid waters are governed both by nutrients and the underwater light regimes. Nutrients are only important when a favourable underwater light regime prevails and are of secondary importance when light is limiting (Grobbelaar, 1985, 1990). Models which are based on nutrient loading are not applicable to highly turbid systems. In addition to light effects, turbid waters have a greater resilience to enrichment and eutrophication than clear waters because of adsorption of nutrients to suspended sediment.

The consensus is that limiting nutrients do influence the overall productivity of freshwaters, but that each individual system must be evaluated for site-specific characteristics. Control of eutrophication would only be effective when a combination of rehabilitation measures are employed, including biomanipulation (Gophen, 1990).

 

---

Back to table of contents

Last updated: 02.07.2001