CHAPTER 2

PHYSICAL AND CHEMICAL PROCESSES

HYDROLOGY

FLOW

Variability of Flow Regimes

Flow regimes have their origin mainly in run-off from precipitation over the river basin. This may be in the form of rainfall in warm temperate and tropical areas, in which case the response of the low order streams to changes in precipitation is related directly to seasonality of pluviosity. In cold temperate or high altitude regims the precipitation is usually in the form of snow in which case release of water into the stream channel is more closely related to seasonal patterns of snow-melt. Retention and release of ground water, important factors in maintaining flow in the river channel during the dry season, contribute less to spates and tend to smooth out flows, reducing the amplitude but increasing the duration of a flood. The proportion of rain water that appears as run-off or as ground water depends on the type of terrain and its vegetation cover. Local difference in these factors can cause very different flood patterns even in two adjacent streams. Because of this, and because of the rapidity with which run-off from local rain storms can affect low order streams, graphs of water level or flow in the smaller water courses are apt to appear “spiky” with numerous rapid changes in flood condition. As streams amalgamate into larger and higher order rivers their individual regimes merge so the flow is smoothed (Fig. 2.1) and reflects the average of the precipitations over a progressively larger area. The fate of the flow thereafter depends on the form of the river basin and here reservoir rivers have internal features which reduce the variability of the flow, distributing it more evenly throughout the year and suppressing pronounced flood peaks. Such features as extensive swamps, forests, lakes or even floodplains themselves, may store water for later release. Alternatively the cumulative discharge of many tributaries arising in different latitudes and having correspondingly diverse flood peaks may also act in this way. The lower reaches of rivers, therefore, may tend to take on “reservoir” characteristics, a good example being perhaps the Zaire river below Kisangani. Even here there are two pronounced flood peaks (Marlier, 1973). The existence of swamp, forest and lakes in the river basin do not by themselves ensure the complete smoothing of flow and many flood rivers have all these features.

Velocity of flow

Although Leopold et al. (1964) maintain that the rate of flow remains more or less constant as one progresses downstream in rivers in the United States, it is the more general observation that flow rate decreases with decreasing slope. Thus lower order rivers tend to have faster flows consistant with the greater slope of upland streams, whereas the higher order streams have slow flows in the meandering lowland course. Some higher order rivers however, such as the Magdalena of Colombia or the Sepik of Papua New Guinea, do have considerable flow velocities in their potamon reaches with main channel rates exceeding 2–3 m/sec at low water. By constrast flow may cease completely in low order streams or even main channels during the dry season, and in many modified rivers abstraction of water for irrigation has reached such a level that there is rarely any net outflow to the sea. Obviously it is difficult to generalize as to the speed of current in various streams as it depends on many physical and climatic variables, but the range of velocities quoted by Zhadin and Gerd (1961) for Middle Asian rivers of the Soviet Union serves to illustrate the general slowing of the flow as the river proceeds downstream. Here flow velocity were 5–6 m/sec in the upper reaches, 3–3.5 m/sec in the foothills and 2–2.5 m/sec in the flatland reaches. Furthermore, velocities change over relatively small stretches of stream especially in the rhithron. For example, the rate of flow in Caucasian streams at low water was 1.25 m/sec over the riffles and only 0.6 m/sec through the succeeding pools.

Figure 2.1   Flood regime of rivers with different basin areas within the Chari-Logone river system showing the increasing smoothing of the flood curve with increasing size of basin

It is very important for living aquatic organisms within the river channel that the velocity of the water does not increase proportionally with discharge. Leopold and Maddock (1953) and Drury (1969) both present graphs which show that the mean velocity at any point on a river only doubles for a tenfold increase in disharge. Furthermore, since these figures apply to peak velocities at mid-channel the increases in velocity at the bottom or sides of the channel would be even less.

Same idea of velocity of flow in any reach can be obtained from the size of particles in the stream bed. Fig. 2.7 sets out the mean current velocity required to just move particles of various sizes along a stream, thus the existing bed deposits should indicate that flows are just below the equivalent rate listed. This serves only as a rough guide to the mean velocity which might be expected, as results from various authors do not agree all that closely. Furthermore, resistance to movement is not entirely a function of particle size for hard packed clay is far less easily entrained than is sand, even though its particle size is far smaller.

ORIGINS OF FLOODING

Flooding originates from three sources: (i) overspill from the river channel; (ii) local rainfall; and (iii) tides.

As flow increases a point is reached where the channel is no longer able to drain the volume of water passing down the river. Further rises in discharge above this point (bankfull level) result in overspill onto the floodplain. Here, because of the flatness of the terrain, increases in volume are achieved by lateral expansion rather than by increases in depth, and the water spreads slowly and diffusely outwards, hampered in its progress by the floodplain vegetation. Such flow is termed “creeping flow” and is depositional (i.e., silt comes out of suspension onto the plain), as opposed to the erosive flow where a strong directional current is confined within the main channel or along the terrace walls.

Figure 2.2   Flood regimes of typical river systems for various continents

Plots of the variation of water level, or flow of a flood against time (flood curves) for large rivers have the appearance of waves (Fig. 2.2), which usually are unimodal, although downstream of the confluence of two systems having differing flood regimes, polymodal floods may occur. The flood wave has amplitude (the total difference in level between maximum and minimum discharge) and duration, (the time taken to pass from one low water stage to the next). It moves down valley at a speed which may depend on many factors, including the slope and degree of enclosure of the river valley, the type of vegetation and the intensity of the flood. In the Niger river for instance (Fig. 2.3A), the flood crest takes over 100 days to move from Koulikoro to Malanville. As this is over 1 160 km, the average rate of travel is about 17 km/day. Similar speeds of transmission are found in the Senegal, where the 620 km from Galougo to Dagana take over a month (17 km/day) (Fig. 2.3B) and the Chari, where the 745 km between Mossala and N'Djamena are traversed in 42 days (18 km/day) (Fig. 2.3C). The flood moves down-valley at about 13 km/day in the Brahmaputra, 29 km/day in the middle Mekong and on the lower Mekong flood-plain 13 km/day. On the Kafue plain the speed of transmission is 11 km/day and on the Central Delta of the Niger 13 km/day. These figures are averaged over considerable distances and speed differ within various reaches of the same river. In general, the more expansive the plain, the less its slope and the greater its vegetation cover, the slower will be the velocity of the flood crest. The progression of a flood wave across a flood-plain is now especially easy to illustrate using satellite imagery. For example, four phases in the passage of the flood over the Yaeres floodplain are shown in Fig. 2.4 (after Quensiere, 1981).

Figure 2.3  Time taken for the flood wave to traverse the distance between selected points in three African river systems. The time taken for the flood wave to cover the distance between fixed points is indicated by the vertical lines showing the time of maximum flood and the line transposed from the preceding point upstream as shown in the graph; figures refer to the distance between one point and the next

In long rivers, where the timing of the rains is fairly uniform along their courses, the arrival of the main flood crest in downstream reaches may be delayed until the dry season. Such a delay is typical of the Niger river, where two floods occur at Malanville (Benin) (Fig. 2.3), one in August-November caused by rainfall and drainage of local rivers and a second in December-March due to the arrival of the flood from the headwaters of the river.

Figure 2.4  Successive stages in the flooding of the Yaeres by overspill from the Logone river (after Quensiere, 1981). ---- : Limits of total area flooded

As the flood wave passes along the floodplain there is a tendency for it to decrease in amplitude and increase in duration, provided no large amounts of water are added to it by other tributaries. This process is best seen on a river such as the Niger where the channel runs for a considerable distance through arid and semi-arid landscapes.

Precipitation on the floodplain itself, or on the basins of the lower order streams immediately surrounding it, saturates the soil and causes local flooding which often precedes the main flood by some considerable period of time. Early work by Svensson (1933) showed that in the Gambia river certain depressions are flooded only by local precipitation and are connected to the river and other low lying swamps by a single channel or creek. Such swamps may be unaffected by variations in river height, but fish may migrate to them by ascending the connecting channel. Early flooding of floodplain depressions by rainwater has been noted from several systems. Elsewhere in Africa, Carey (1971) described flooding of this type in the Kafue. The inundation of the Upper Chari (Bahr Salamat and Bahr Azoum) appears to be almost entirely by local rainfall (Durand, per.comm.), and in the Nile Sudd the swamps are filled one to two months before overspill from the river occurs. Extensive areas of the Venezuelan Ilanos (Matthes, pers.comm.) and the Gran Pantanal or Paraguayan Chaco of the Paraguay river flood in this manner, (Carter and Beadle, 1930). A similar type of flooding has been recorded from the Irrawaddy river in Burma by Ohya, (1966), where submersion by rainwater precedes the main floods by about one month. Flooding by local rainfall differs from overspill in that the net flow of the water is from the floodplain depression to the river rather than the other way round. Flooding with rainwater prepares the floodplain by saturating the soil and filling flood-plain depressions and replenishing the ground water so that relatively little of the subsequent overspill flood is absorbed by the ground.

Flooding caused by tidal action obviously affects only those parts of floodplains adjacent to the coast. However, secondary effects are produced by the backing up of freshwater floods by high tidal levels. Volker (1966) divided river estuaries into four reaches (Fig. 2.5); Zone A where vertical tides produce a reversal of the direction of flow and the penetration of saline water into the river channel, or onto the floodplain; Zone B where the river water remains fresh but tidal effects and current reversals still occur; Zone C where the water levels are still affected by the tides giving rise to differences in current velocity, although the flow always proceeds downstream and Zone D where water level and flow are affected by upstream discharge only; this zone merges into the main river upstream.

Figure 2.5  Zones of interaction between marine and freshwater in estuaries and coastal deltaic floodplains (after Volker, 1966)

WATER BALANCE ON FLOODPLAINS

The flood curve at any point in the river, results from discharge transmitted from upstream, input from local tributaries and precipitation in the immediate area. This can be represented diagramatically as in Fig. 2.6 where the x axis is time. The y axis may represent any of the indicators of water quantity, the most common of which are water level, discharge rate or storage volume. Although there is a continuous fluctuation in water level the flood can be broken into two parts. In the first, the water is confined in the river channel and alterations in discharge produce changes in depth and velocity. Changes in volume are caused mainly by increase in depth so long as the flow remains confined in the channel are caused mainly by increases in depth. This phase is delimited by the bankfull stage beyond which the water overflows the channel limits and floods the plain. Thereafter changes in inflow produce relatively little change in outflow, the excess water being temporarily stored on the floodplain where the volume increases mainly from the extension of the water area. The bankfull level is usually difficult to define precisely, as it varies from place to place in the river and also from year to year. The detailed hydraulics of floodplains are extremely complex and beyond the scope of this discussion. They have been the subject of many studies and attempts at mathematical modelling such as those elaborated by Weiss and Midgley (1975 and 1976).

Figure 2.6  Diagram of (A) a generalized annual flood regime, and (B) probability of occurrence of maximum and minimum water stages

The intensity of flooding varies from year to year in response to fluctuations in rainfall. Because of this variability, a distribution of the frequency with which any flood height occurs can be established over a number of years. Such distributions (Fig. 2.6) may be normal but are more often skewed. The number of years which must elapse on average between a flood of a particular height and the next flood of the same height is known as the return period. The 100 year flood, for example, is that level of flooding which can be expected once in 100 years. The probability of occurrence of any flood is the reciprocal of its return period; the 100 year flood having a probability of occurrence of 0.01 in any year.

Not only does the intensity of flooding vary from year to year, but there is increasing evidence that the variability is of a cyclic nature which is traceable to long-term global climatic patterns (Winstanly, 1975; Bryson, 1974). According to such cycles, years of exceptionally high flooding or of drought tend to be grouped together so there is a good chance of two or more years of intense flooding to appear in sequence or for one year of low flood intensity of be followed immediately by another poor year. When the latter happens drought conditions can ensue such as were shown in the Sahelian rivers between 1970–74 and again in 1980–85. The grouping of good and poor floods in this manner can have severe consequences on the ecology of the fish as we shall see in the section on the dynamics of floodplain communities. The variation in flood level is a normal phenomenon and a knowledge of the range as well as the mean value is essential to an understanding of the ecology of these systems. Differences in the amplitude of the flood curves as measured by maximum and minimum discharge rate are shown in Table 2.1. These are very variable and especially high values tending to infinity are obtained in those rivers where flow virtually stops during the dry season. An example of the effects of such variations on the actual areas flooded is that of the Senegal River as shown in Table 2.2.

Table 2.3 shows the mean rainfall over certain floodplains. In the tropics and sub-tropics, rainfall makes only a moderate contribution to the water balance of the plain and is largely exceeded by the evapotranspiration rates shown in Table 2.4. This excess of evaporation over precipitation causes the net outflow from some floodplains to be very much lower than the inflow. Thus, of the 7.11 × 10¹⁰ m³ of water entering the Central Delta of the Niger through the Niger and Bani rivers in years of normal flow, only 3.82 × 10¹⁰ m³ emerges at Dire, a net loss of 46 percent. Similarly, 2.7 × 10¹⁰ m³ of water enter the Sudd at Bor and only 1.4 × 10¹⁰ m³ leave it at Lake No, a loss of 48 percent of the water. In a few systems, the Okavango swamps for instance, evaporation almost balances inflow over the year so that there is no appreciable outflow from the system.

The high evaporation rates also mean that, to be permanent, pools or lagoons must either be filled annually to a depth greater than the loss by evapotranspiration (between 1.2 and 2.2 m from the data in Table 2.3), or must receive inflow from ground water.

CHEMISTRY

SEDIMENT LOAD AND TURBIDITY

There is a clear relationship between the velocity of the current in a river and the amount of matter carried suspended in its waters. The greater the current, the larger the load and particle size (see Fig. 2.7) that can be carried by the stream. This has many repercussions both for the morphology of the river and in the biology of many of the organisms living therein. Morphologically, because any river will tend to transport the maximum amount of material in suspension for any given current, sediment will be picked up or deposited as the current and the existing load vary. Thus slowing of flow results in aggradation, the deposition of part of the load leading to the development of many of the morphological features of the river, including levees, point bars and floodplains. Acceeration of the flow leads to the river picking up sediment causing erosion or degradation. During the flood cycle these two processes tend to alternate. Thus during the rising flood rivers are heavily charged with silt, but as the water spills over the flood-plain the general slackening of flow leads to a deposition of suspended matter producing decanted waters of some clarity. At falling flood acceleration in water velocity causes localized scour but as flow slackens progressively through the low water period the standing waters of the floodplain, and frequently of the main channel itself, deposit most of their load.

The amount of suspended matter in the water affects the penetration of light into the water, for example Bonetto (1980) found a relationship S = a e bh for the Parana river, where S is the light penetration as measured by Secchi disk, h is the velocity of the current as measured by the height of the river and a and b are constants.

Sediment yields of representative rivers of the world are shown in Table 2.5. The suspended matter carried by a river relative to its discharge is indicative of the stability of the system. Rivers which carry a large amount of sediment relative to their discharge, for example the Indus, tend to be unstable with shifting beds and rapidly evolving successions of floodplain feature, whereas rivers with low sediment yields relative to their discharge tend to change much more slowly.

Figure 2.7  The speeds of flow required to move mineral particles of various sizes (line plotted from data in Nielson, 1950); terminology of particle sizes from Gregory and Walling (1973)

NUTRIENT SPIRALLING

The idea of nutrient spiralling posits that a nutrient atom cycles through three generalized components of the stream ecosystem (Newbold et al., 1981). Each component is transported downstream with current so the path of the atom describes an imaginary spiral. The three components are: (a) the water in which the nutrient atom is dissolved; (b) the particulate phase in which the atom is sorbed onto particulate matter and is thus located in the bottom although downstream transport continues through saltaton, suspension, erosion, etc.; (c) the consumer phase in which the atom is incorporated into living matter. During this phase it may pass directly through a series of organisms before its release once more into the water phase. The total distance travelled downstream between one release into the water and the next, is termed the spiralling length (S). This length may be taken as a measure of the efficiency with which nutrients are used within any particular system, the shorter the spiralling length the more efficient the use in that any nutrient atom is utilized more frequently for any given length of stream.

During the consumer phase, most of the nutrients are held in micro-organisms, but invertebrates grazing in these and on higher plants can influence the spiralling length through accelerated release of nutrients. Collector and shredder organisms tend to shorten spiralling length whereas grazers may increase it.

CARBON AND ORGANIC MATERIAL

Rivers transport large, but usually not accurately quantified, amounts of dissolved and particulate organic material from the terrestrial component of the drainge basin to the oceans (See Fig. 2.8). Direct inputs usually consist of coarse particulate matter (C.P.O.M.) in the form of leaves, fruit, flowers, pollen, branches, etc. whereas indirect inputs consist of leachates and other dissolved material including inorganic carbonates from rocks and soils. Biological processes downstream tend to degrade the particulate organic matter from CPOM (particle size full size trees to 1 mm) thrugh fine particulate organic matter (FPOM - particle size 50 mm - 1 mm) to ultrafine particulate organic matter (UPOM - particle size 0.5 - 50 mm). Dissolved compounds increase in concentration down-stream and a wide range of amino acids, fatty acids, sugars, hydrocarbons and pigments have been recorded by Visser (1970) and Balogun (1970), either free in the water or absorbed onto sediment particles. At the same time living organism contribute to the CO₂ concentration through respiration.

Particulate carbon and compounds absorbed onto sediment particles follow similar processes to sediment and other nutrients in that they may be stored within the plain through deposition, may be recycled by biological agents and may be reincorporated into the flow towards the oceans through erosion, (Fig. 2.9). Storage rates may be quite high. For example, in the floodplains of small rivers of the forested zones of North America storage can amount to between 48 and 96 g C/m/yr (Sedell et al., 1974) and in mature floodplain swamps can be as high as the 113 g C/m/yr recorded from the Atchafalaya river (Gagliano and Van Beek, 1975).

Figure 2.8  A general model of the biogeochemistry of carbon in rivers (from Richey, 1981)

IONIC COMPOSITION

The chemical composition of river water depends on a wide variety of physical, chemical and biological features, but three basic mechanisms control surface-water chemistry: precipitation, the nature of the bedrock and the evaporation-crystallization process (Gibbs, 1970) (Fig. 2.10). In most of the large tropical rivers the ionic composition of the water derives primarily from the rain and the rock or sediments over which the river flows. In rivers flowing through hot arid regions, the rate of evaporation and crystalization of the concentrated salts exerts the main influence and flood areas located in such zones eventually transform into salt marsh habitats. Secondary influences on the ionic composition are exerted by macrophytes and phytoplankton, and to an increasing extent by the various industrial, agricultural and domestic human activities.

Figure 2.9   Organic carbon flow between a floodplain wetland and its bordering river (Mulholland, 1981)

Figure 2.10   Diagram of the basic mechanism controlling surface-water chemistry (Adapted from Gibbs, 1970)

Total dissolved solids (T.D.S.) may be taken as one convenient measure of the total ionic concentration in water. Naturally, the concentration and relative abundance of ions in river water is highly variable, although there is a tendency for larger rivers to average out the variability of the tributaries. Fig. 2.11, taken from Holland (1978), shows clearly that, while there is considerable spread in T.D.S. for any particular runoff the increase in concentration with decreasing runoff is unmistakeable.

Figure 2.11   Total dissolved solids in some major rivers as a function of the mean annual runoff (after Holland, 1978)

The mean composition of the river waters of the various continents of the world are shown in Table 2.6. Holland (1978) discusses in more detail the concentration and origin of most of the major ions present in these rivers.

Detailed lists of ionic content of individual tropical rivers are outside the scope of this book. Compilations of such data based on the work of numerous authors exist for Africa (Welcomme, 1972) and Latin America (Ardizzone and Ziesler, 1977). Observations from Asia have not been compiled systematically, but Jhingran (1975) cited data from Indian rivers and Johnson (1967) gave details of the chemistry of some Malaysian rivers. Studies on the significance of ionic transport on a world scale have been made by Meybeck (1976 and 1979) and Martin and Meybeck (1976 and 1979).

In those regions where the rocks are impoverished and soils leached, the main supply of nutrients comes from precipitation, such rivers being grouped in the lower right hand area of Fig. 2.10. In the Amazon basin for instance, the chemical composition of the blackwater streams of the Hylaea, or of the larger blackwater rivers, can resemble that of the rainwater. The frequent similarity of the composition of rain and river waters indicates that such ions as are leached from the soil may be balanced by those carried by the rain. Similarly the chemistry of the headwaters of the Paraná and Paraguay rivers are dominated by atmosperic precipitation (Bonetto, 1975) and in the same general area the conductivity of rainwater falling over the Mato Grosso, Brazil, is often higher than that of the streams (Green, 1970). In Africa, Visser (1974) noted that the composition of river and lake waters within a radius of 100 miles of Kampala indicated that rain contributed on average all of the sodium, ammonium and nitrate, over 75 percent of the chloride and over 70 percent of the potassium. In the Gombak river, Malaysia, Bishop (1973) considered precipitation as a very important factor in determining ionic concentrations.

The rivers of the three main tropical rainforest regions arise on very poor, leached, podsolitic soils. Sioli (1968) described such waters in the Amazon basin as more or less transparent, devoid of significant amounts of inorganic particles, but with very poor light penetration because of the brown colour imparted by dissolved humic substances. PH is usually very low and dissolved nutrient concentration poor. More detailed analyses of the composition of blackwaters of L. Tupe of the Rio Negro (Rai and Hill, 1981) show such waters to be biased to sodium and sulphate dominance. Similar waters have been reported from the headwaters of the Orinoco (Edwards and Thorne, 1970) and have also been described from both Africa (Matthes, 1964) and Malaysia (Johnson, 1968). Descriptions and analyses from the three continents are shown in Table 2.7.

^a    These figures are often based on only a few observations and do not therefore necessarilyreflect the full range of variations of the parameters measured

Blackwaters from the three continental areas are similar in general composition, except that sulphate is even more common in the Malaysian waters than in those of South America and Africa. In South America it is virtually absent in many blackwaters and in Africa is present only in those waters flowing from volcanic regions. Johnson (1968) described an aberrent type of blackwater from Malaysia which, by flowing over limestone, acquires higher total ionic concentrations of between 150–700 mg/l, accounting for the relatively high maximum values in Table 2.7. The “Gelam” type waters also had very low PH values of 3.5–4.9 showing considerable excess sulphate. A third series of blackwaters have become enriched by organic pollution and their PH was raised to 6 and above. This series, may indicate how blackwaters on other continents will react to pollution from urban sources as human settlement of these areas becomes more intense.

In addition to the blackwaters, the Amazon and Orinoco are fed by clear water rivers. They originate in the case of the Amazon from the massifs of central Brazil and in the Orinoco from the Guianian shield. These waters are as poor in nutrients as the black-waters and have similarly low PH values of between 4.0–6.6. They do not, however, have the dark staining by humic matter. A third category of Amazonian rivers, the white water rivers, more closely resemble the general pattern of other tropical rivers as described in the rest of this section and are the main source of nutrients in the basin where, by the deposition of their heavy silt loads, they maintain the productivity of the várzea flood-plains.

The lowland equatorial environments of the Amazon and the Zaire give these two rivers many common features. However, there are considerable differences between the two systems which arise from the altitude and relief of the basins as well as from the geological formation and the vegetation (Marlier, 1973). The forest is less extensive in the Zaire and the main channel of the river is divided into quiet and rapid reaches. Flooding is also less severe than in the Amazon.

Conductivity is another measure of the total amount of ions present in a body of water and is a useful approximation to chemical richness. As a measure, however, it does not give an indication of the actual ionic composition of the water and may, therefore, fail to convey information on limiting factors such as the lack of essential nutrients.

River waters are less variable than those of lakes and the conductivities of most major whitewater rivers tend to resemble one another, with a few exceptions. In fact variations between tributaries within any one river system are apt to be greater than those shown between different main river channels. This is because the lower order streams are more sensitive to local geological and vegetational patterns, whereas higher order channels average the conditions from a range of lower order ones (Table 2.8).

Within any one system conductivity changes throughout the year. The broad, ionic composition of the water is largely determined by the three processes mentioned in the introductory passage to this section. The concentration of dissolved substances which include organic compounds not measured by conductivity is influenced by five factors: (i) dilution effects, whereby flood or rainwater with weak ionic concentration reduces conductivity; (ii) solution effects, the “shore factor” of Braun (1952), whereby salts locked on previously dry land by decaying vegetation, animal dung, ash from burnt vegetation etc., enter solution as the flood waters extend over larger areas; (iii) wash-out effects whereby nutrient rich ground water is displaced into the river channel by increased infiltration; (iv) concentration by evaporation and (v) absorption by living components of the system. Normally, these effects combined tend to produce higher conductivities in the dry season than in the wet, both in lagoons and in the river channels, giving an inverse relationship between water depth and conductivity. The same trends have been recorded from the Oshun river (Egborge, 1971), the Ogun river (Adebisi, 1981), the Kafue river (Tait, 1967; Carey, 1971; and University of Idaho et al., 1971), the Senegal (Reizer, 1974) and the Zambezi (FAO/UN, 1969) in Africa. However, wash out of nutrient rich ground waters may cause local rises in conductivity in small streams during the rains. Several of these authors have also noted a secondary maximum of conductivity (due to solution effects) as water invades the floodplain but this is of short duration.

^a    These figures are often based on only a few observations and do not thereforenecessarily reflect the full range of the parameters measured

^b    Blackwater river

^c    Welcomme, 1972 compiled data received in response to a circular FAO questionnaire

The situation for Latin American rivers is less clear. Bonetto et al., 1978, show that, as may be expected, conductivity in the Laguna La Brava rose during the low water period. Also, when floods invade isolated floodplain lakes, they dilute the dissolved solids that have been concentrated during the dry season, (Bonetto, Dioni and Pignalberi, 1969). Schmidt (1972) noted consistently higher ionic concentrations during the dry season in the Amazon river as compared to the rains, and Junk (1973) remarked on a similar fluctuation between high conductivities in the dry season and lower conductivities in the flood in some floodplain lakes. However, both he and Schmidt (1972a) have also described lakes in which the conductivity is minimal during the dry season and maximal during the first phase of river inflow. The explanation advanced for this was that initial flooding by nutrient -rich whitewaters, and solution of salts from inundated lands, gives high ionic concentration during periods of increasing level. During the non-flood season water is diluted by drainage of nutrient-poor blackwater from the forest streams and by rain-water.

In the littoral zone of the floodplain, represented by the vanguard of the advancing flood water, nutrient-rich soil is freshly inundated and the solution of the salts produces a local increase in conductivity. This is undoubtedly of great significance in the productivity patterns of floodplains during the earlier part of the flood phase, when such shallow littoral zones must represent a considerable proportion of the total area of the aquatic system.

HYDROGEN ION CONCENTRATION (pH)

Differences in water chemistry including pH, lead to variations in the type of aquatic organisms which inhabit various waters. For instance distinctive communities of fish are particularly found associated with highly acidic waters; these often have only low resistance to more elevated pH's.

Forest rivers, with their characteristics blackwaters and abundant humic materials, are slightly to very acid with pH's ranging from 4 to neutrality. Savanna rivers are usually neutral or slightly alkaline (Table 2.8). Because there is decaying vegetation on the bottom of most standing waters of the floodplain, slight gradients often exist with higher pH values at the surface (Bonetto, 1975; Carey, 1971; Schmidt, 1973). Gradients in pH are also common at the waters edge where the newly flooded soil may produce either a local drop in pH, as in some African swamps, or a brusque rise in others. The sudden rise in the pH of floodwaters on the Central Delta of the Niger, to as high as 8, has been attributed by Blanc et al. (1955) to the effects of dung entering solution. Experimental evidence confirmed that cow droppings raise the pH of water rapidly. The effects of this are, however, temporary and more acid conditions are quickly restored. Laterite, which is a common soil in tropical regions, and the peaty types of soil found on the surface of floodplains are both capable of reducing pH (Mizuno and Mori, 1970). There is a tendency for swamps to become more acid as the dry season progresses; Carey (1971) and Holden and Green (1960) particularly having noted that pH is lower in swamps than in the rivre at this time. Drops in pH have also been noted by D'Aubenton (1963) under the floating vegetation mat of the flooded forest of the Tonle Sap. By contrast conditions in some open water lagoons can become increasingly alkaline as calcium salts are concentrated by evaporation (Holden and Green, 1960). The lower pH values in swamps usually cause a general drop in pH throughout the system when the acid waters are flushed out by rain or flood water with poor buffering capacity early in the flood season.

Diurnal fluctuations in pH are primarily associated with the respiratory and photosynthetic activities of the phytoplankton. A decrease in pH in the euphotic zone during the night, with a minimum early in the morning, is a common feature of Amazonian várzea lakes (Schmidt, 1973). Similar trends can reasonably be expected to occur throughout the day in the majority of floodplain lakes.

TEMPERATURE

Latitude, altitude degree of insolation, substrate composition, turbidity, ground or rainwater inflows, wind, and vegetation cover, can all influence the temperature of water in rivers and floodplain lakes. Generally, surface water temperatures follow the ambient air temperature fairly closely although under dry hot conditions this is more likely to be correlated with air temperature minima due to the cooling effect of evaporation. In the normal annual cycle in the tropics dry season water temperatures are higher than those of the wet season. In the temperate zone winter temperatures are lower than those in the summer and rivers may freeze for a considerable part of the year at higher latitudes and altitudes. At relatively low latitudes, or at high altitudes in the tropics, temperatures may still fall sufficiently to give conditions which are cold enough to cause growth checks and even fish mortalities. In the Paraná river, for example, a combination of sub-zero temperatures and low water caused massive mortalities in 1962 (Vidal, 1964).

Water in the main channel of a river rarely stratifies, as good mixing is maintained by the turbulence associated with river flow. On the other hand stratified conditions are common in even very shallow floodplain water bodies. Hastings (1972), for instance, recorded diurnal stratification in lagoons of the South Elephant marsh (Fig. 2.12) which were only 1.6 m deep. Similarly, Sanchez (1961) and Junk (1973) found 1–2 m deep lagoons from the Peruvian and Brazilian Amazonian várzea to be stratified at least temporarily, and Ducharme (1975) (Fig. 2.12), Mikkola and Arias (1976), Arias (1975) and Zarate and Cubides (1977) have all recorded stratified conditions from various cienagas of the Magdalena system ranging from between 1.5–6 m depth. In the dry season the Amazonian Lago do Castanho is only about 1.5 m deep, but clear stratification develops on windless days (Schmidt, 1973). In flooded condition the lake is up to 11 m deep and stratification is more persistent, although the temperature profiles are highly variable from one day to the next (Fig. 2.13) due to local changes in conditions. The main day-to-day fluctuations occur in waters of less that 3.5 m depth, indicating that only water down to this depth is normally affected by daytime heating. In the Paraná river, which is subject to lower temperatures especially in winter and during the night, inverse stratification has been noted by Bonetto, Dioni and Pignalberi (1969).

Figure 1.12   Diurnal cycle of temperature and dissolved oxygen in the Cienaga de Ayapel (Magdalena River) 24 October 1974 (After Ducharme, 1975)

Figure 1.13   Daytime temperature fluctuations and changes in the stratification of Lago do Castanho within short-time intervals during May 1969 (high-water). (After Schmidt, 1973)

It may be concluded that floodplain lakes not covered by vegetation normally stratify thermally during at least part of the year. In shallower lakes stratification due to surface heating reaches a maximum between 14.00 and 18.00 hrs, but tends to disappear at night or under the influence of strong or persistent winds. The difference in density per degree at high temperatures is greater than that at low temperature, for instance the drop in density between 29 and 30°C is about the same as that between 4 and 10°C. For this reason at high temperature density differences are quite large for even slight differences in temperature and the resistance to mixing is correspondingly high. In deep lagoons stratification may last for several weeks or even months.

Water shaded by flooded or fringing forest tend to be cooler and more thermally uniform than that exposed to direct sunlight, although thermal stratification has been described both from the flooded forest fringing the Tonle Sap (D'Aubenton, 1963) and from Igapó forests of the lower Rio Negro (Schmidt, 1976 and Geisler, 1969). Floating vegetation on the other hand gives rise to more varied temperature conditions. Because of the restricted evaporation and high absorption of sunlight in calm areas very high temperatures can be attained within masses of floating vegetation. Bonetto, Dioni and Pignalberi (1969) found exceptional temperatures of up to 40°C in the surface waters of an oxbow lake covered with Salvinia (Fig. 2.14). Similar temperatures have been found on the surface of Lago Calado in stands of Paspalum repens by Junk (1973). Temperatures fall off more rapidly than they do in open water due to shading by the vegetation. Floating mats of vegetation also encourage the maintenance of stratified conditions by damping wind and wave action.

Figure 2.14   An extreme case of temperature accumulation in the waters under a bed of Salvinia at midday. Solid line = open water; dotted line = water under vegetation. The arrow marks the air temperature. (After Bonetto et al., 1969)

An important but little investigated habitat is the shallow openwater littoral at the fringe of the flood zone. At the time of rising water this represents the first contact between the advancing water and the nutrient-rich dry soil. Because of its shallowness, daytime water temperatures may be very high and gradients are set up between the very edge of the zone and the cooler water offshore (Fig. 2.15).

DISSOLVED OXYGEN

The distribution of dissolved oxygen within the aquatic system is one of the main factors influencing the distribution of fish. In the rhithron waters are usually well oxygenated due to the turbulent flow across the riffles. However, in the tropics these zones are apt to become separated into a series of pools during the dry season when flow is minimal. At this time the pools may become deoxygenated or even anoxic due to the decomposition of organic material, the oxygen demand of the fish and to increased temperatures. In the Potamon the situation is more complex. As floodwaters invade the floodplain of the Parana river there is an initial rise in dissolved oxygen concentrations in open water and under floating vegetation. Similar effects have been noted from many other systems, particularly the Sokoto (Holden and Green, 1960) and the Kafue (University of Idaho et al., 1971) in Africa, the Paraguayan Chaco (Carter and Beadle, 1930) and the Magdalena (Arias, 1975; Zarate and Cubides, 1977) in South America. During the earlier part of the floods, flood or rainwater may flush deoxygenated water out of depression lakes and swamps remaining from the preceding dry season. There is, therefore, a tendency for dissolved oxygen levels to be low during the earlier period when water is rising, although they remain fairly high throughout the major part of the floods. Higher dissolved oxygen levels during the flood season are ascribed mostly to the aerating action of wind and the mixing brought about by turbulence. In some deeper lakes, however, stratified conditions with low oxygen tensions near the bottom may persist through out the year. Vertical oxygen gradients have also been observed in flooded Igapó forests in the Rio Negro basin by Geisler (1969) and Schmidt (1976). In deoxygenated water, H₂S concentrations may build up through the decomposition of bottom mud and detritus.

Figure 2.15   Changes in the physical and chemical conditions with distance from the waters edge in a floodplain lagoon: (x--x--x) at 0800 hrs •--•--•) at 1200 hrs. (After Welcomme, 1970)

In the dry season dissolved oxygen conditions are linked to a number of factors including the size of the water body, biological oxygen demand (BOD) of organic detritus or pollutants, degree of thermal stratification, vegetation cover, phytoplankton development and wind action. In general there is a tendency for even comparatively shallow bodies of water to stratify with higher dissolved oxygen concentrations near the surface. Wind action is generally insufficent to ensure complete mixing in deeper lakes where strong thermal stratification also exists. In shallower water bodies which lack vegetation cover, wind action may ensure mixing through the whole water column (Schmidt, 1973). However, resistence to mixing is comparatively high for the reason given in the preceding section. Ducharme (1975) considered that, whereas wind is the major oxygenator in large lakes, dissolved oxygen concentrations in smaller pools are largely controlled by the photosynthetic activity of the phytoplankton. This means that smaller water bodies show greater diurnal changes in dissolved oxygen than do larger ones. Phytosynthetic activity raises the oxygen content of the water during the day, particularly near the surface, but dissolved oxygen is withdrawn from water by plants and the biochemical oxygen demand of silt and suspended organic solids and levels drop during the night (Fig. 2.16).

Mass fish mortalities have been noted from Brazil (Brinkmann and Santos, 1973) and the Kafue flats (Tait, 1967). In both cases the mortalities were attributed to sudden reductions in temperature cooling the surface water of floodplain lakes which, coupled with strong winds, caused a rapid breakdown of stratification, abrupt deoxygenation and contamination with H₂S to lethal levels. Mortalities on the Apure river floodplain have also been attributed to deoxygenated conditions (Matthes, pers.comm.). These arise when depressions containing unburnt grasses are inundated with rainwater. Because of the persistence of thermal stratification the upper layers of the water (epilimnion) retain oxygen and the fish survive there. However, the oxygen rapidly becomes depleted in the lower layers and H₂S accumulates. Later the turbulence associated with the entry of river flood waters mixes the layers rapidly killing the fish. Other massive fish kills caused by the abrupt purging of oxygen depleted waters from swamps have been recorded by Bryan and Sabins (1979) for the Atchafalaya river, a tributary of the Mississippi. Such mortalities appear widespread in tropical floodplains as they have been reported from many systems. Similar mortalities have been reported from cold water floodplains where high B.O.D. originating from pollution reduce dissolved oxygen levels below ice cover in winter to levels below those tolerated by fish.

Floating mats of vegetation are an extremely common feature of tropical rivers and floodplain lakes. The effects of these are similar whatever species of plant forms the mat. Fig. 2.16 for instance shows the effect of stands of Paspalum repens on oxygen concentrations in Lago Calado of the Amazonian várzea, but similar graphs have been drawn for the water column below Eichhornia crassipes (Arias, 1975 and Zarate and Cubides, 1977), Salvinia herzogii (Bonetto, Dioni and Pignalberi, 1969), Leersia hexandra (Junk, 1973), Cyperus papyrus (Carter, 1953) and, Vossia (Tait, 1967). The general effect of plant cover is to reduce dissolved oxygen concentrations, frequently to zero (Beadle, 1974) with a coincident development and release of H₂S. Because of the large amount of decaying organic matter trapped in the root masses of floating plants, the deoxygenation often reaches a maximum in this zone and sometimes rises slightly with increasing depth thereafter. The degree of deoxygenation below the floating mat depends on water movements, for in strong currents the oxygen gradient beneath the mat may be eliminated. It is, therefore, to be suspected that conditions under vegetation mats are less extreme during the floods, when the constant movement of water may maintain at least some level of dissolved oxygen. The density of the plant cover also influences the extent to which water becomes deoxygenated; the more dense the cover, the less oxygen below it. Plant cover contributes to low dissolved oxygen conditions in two ways. Firstly, a considerable biochemical oxygen demand is created by both trapping organic matter and by the decay of their own vegetative parts, thus removing oxygen from the water. Secondly, re-aeration of the water is prevented by reducing wind and wave action on the surface of the water, and by producing shaded conditions unsuitable for phytoplankton growth.

Figure 2.16   Diurnal changes in dissolved oxygen content and temperature in open waters and under floating vegetation (Paspalum repens) in Lago Calado, 8 September 1968. (After Junk, 1973)

Oxygen gradients also seem to exist in the shallow littoral zone (Fig. 2.15). Wave and turbulence at the waters edge may here assure a restricted belt of oxygenated water, especially at night, even though the water a little further offshore becomes completely deoxygenated. This effect, if widespread, may be of considerable importance in the ecology of the fish, as many of the juvenile forms are to be found in this zone.