CHAPTER 5

RIVER FISHES AND THE RIVERINE SYSTEM

NUMBERS OF SPECIES IN RIVER SYSTEMS

DIFFERENCES IN NUMBERS OF SPECIES BETWEEN SYSTEMS

The considerable differences in the numbers of species inhabiting the various river systems are largely attributable to the size of the river as represented by its basin area or some correlate of it such as length of main channel or stream order. A relationship of the form N = fA^b (where N = Number of species and A = basin area in km₂) can be fitted to the dispersion in Fig. 5.1. When all points are included a regression, N = 0.297 A^0.477 is obtained. In fact there are differences between different geographical areas which become apparent when individual blocks of data are analysed separately as follows:

Figure 5.1   Number of species of fish present in different river systems plotted according to their basin areas: (•) South America; (o) Africa; (▭) Asia; (x) Europe; () North America

North flowing Siberian and European Soviet Rivers:

N = 2.76A^0.19 (N = 6, r = 0.91)

South flowing European rivers:

N = 0.6A^0.14 (n = 11, r = 0.72)

Africa:

N = 0.449A^0.434 (n = 25; r = 0.91)

South America:

N = 0.169A^0.552 (n = 11; r = 0.95)

Daget and Economidis (1975) have independently fitted similar log-log regressions to date from sets of small rivers in Portugal and Greece obtaining:

Greece:

N = 2.319A^0.24 (n = 12; r = 0.94)

Portugal:

N = 1.786A^0.19 (n = 12; r = 0.92)

which agree well with the regression established for Europe. Similar scatter plots for Australian rivers also conform to these general principles with the distribution for nine North flowing tropical rivers being located above the world mean regression line and five of the six points for temperate rivers being located below the line. (Bishop and Forbes, in press). These relationships enable us to conclude that while species diversity increases with basin area at all latitudes, it does so faster as one approaches the tropics - as indicated by the larger exponent at lower latitudes. Because a log-log relationship does not completely describe the distribution of points, the intercept, which Daget uses as a relative index of faunal richness, is probably not a reliable predictor for basins of extremely small area. Many reasons have been advanced for the increased diversity at low latitudes and these are discussed more fully by Lowe-McConnell (1975). As to the influence of size, the number of ecological niches is probably greater in larger river systems than in small ones. Meandering creates a regular series of habitats in the main course of the river and some floodplain lakes are often isolated from the rest of the system for long periods of time. Furthermore, similar habitats in the sub-systems are often separated by considerable distances of inimicable biotope, leading to the formation of distinct groups of species adapted to similar conditions along the length of the river. Roberts (1973) has described the influence of such factors on the faunas of the two largest rivers of the world, the Amazon and the Zaire. However, such differences do not occur solely in the larger systems for they have also been noted in small rivers and streams of sixth order or lower by Kuehne (1962); Barrel, Davis and Dorris (1967), Whiteside and McNatt (1972), and Platts (1979) from the United States, and Bishop (1973) from the Gombak river in Malaysia. Gorman and Karr (1978) also found that fish community diversity is correlated with stream habitat diversity in selected Indiana and Panama streams, and Horwitz (1978) demonstrated similar relationships in various types of rivers in Illinois, Missouri, Ohio and Wyoming. Furthermore, diversity of habitat increases with increasing stream order.

The aquatic components of river basins resemble island fauna in their relative isolation one from another. The mathematical relationship between the size of islands and the number and diversity of species inhabiting them has been examined by McArthur and Wilson (1967), who concluded that the equilibrium between the extinction of species and the colonization with new species differs according to the size of system; larger systems favour higher diversity perhaps because of the type of biogeographical factors mentioned above.

Although total numbers in large systems tend to be high, groups of species are often located in very different portions of the system, thus only a certain percentage concerns the floodplain or its fishery at any one point. Some species also are limited to the turbulent waters of rapids and headwater tributaries and are rarely, if ever, found in the slower flowing floodplain reaches. Nevertheless individual fisheries in larger rivers do have many more species available to them than those of smaller streams, and catches consisting of over fifty species in one type of gear are not uncommon in the biggest systems. Figure 5.2 illustrates the diversity in form of representative species of various genera of tropical river systems.

Figure 5.2  Representates of genera of fish from tropical river systems

RELATIVE ABUNDANCE OF SPECIES WITHIN ONE SYSTEM

In his review of data covering a number of areas and taxa, Preston (1962 and 1962a) found that within a particular taxon and region, the relative abundance of species followed a lognormal distribution (i.e., the logarithms of the species' abundance were normally distributed). These distributions were of course truncated at very high and very low levels of abundance, but over the observed range distribution of this type described most of the sets of data well. In theory there are an infinite number of lognormal distributions that could describe the relative abundance of a given number of species, but Preston found a very close empirical relationship between the number of species in an assemblage and the precise form (i.e., the variance) of the observed lognormal distribution of relative abundance. He proposed that distributions obeying this relationship be called “canonical distribution”. Daget (1966) concluded from two samples from the flooded plain of the Benue river, that the distribution of numerical abundance and rarity of fishes in the sample locality, did indeed conform to the canonical type. A similar conclusion was reached by Loubens (1970) on the basis of 15 samples from various parts of the Lake Chad-Chari river complex. Here samples were distributed lognormally, both when analysed by number and by weight, although in the latter case the correlations were less exact. With smaller sample areas a simpler exponential relationship between ranked species (R) and number (N) of the form N = ab^R was equally applicable. The extension of this type of analysis to other series of data from other water bodies throughout the world demonstrates the wider application of these principles, both to areas sampled by complete fishing and to samples obtained with individual types of gear. Preston (1969) expanded the idea of relative abundance of taxa in one place at one time to suggest that commonness and rarity are also distributed lognormally in space and in time.

These conclusions are of interest when considering the community dynamics of floodplain ecosystems insofar as they affect the dominance of species especially favoured by the fishery. They also imply that the fish catch will tend to be dominated by only a few species even in communities of high specific diversity and abundance. For this reason changes in population structure under exploitation deserve further investigation. Particularly interesting is Preston's contention that the distribution of commonness in any sample of a population is a truncated lognormal, having the same modal height and logarithmic distribution as the “universe” from which it is drawn. The abundance structure of fish isolated within floodplain depressions and lagoons may thus be linked by similar relationships to the abundance structure of the fish community in the river as a whole.

SUB-POPULATIONS

There is very little information on the fine structure of species distribution of fishes inhabiting rivers. From the behaviour of freshwater fish elsewhere it might be expected that any one species is far from homogeneous and the sub-populations or stocks might be set up within the system. In the Volga river the existence of such populations was indicated by the continued segregation of a species (Abramis brama) into a number of local stocks after impoundment to form a reservoir of the riverine type. The individual stocks (see Fig. 5.3) maintained distinct, isolated reproductive areas and areas for feeding and over wintering. The feeding sites were separated from the breeding sites by considerable distances (Poddubnyi in Mordukhai-Boltovskoi (1979). In rivers such as the Niger, Zaire or Zambesi where floodplain reaches alternate with stretches of rapids, conditions seem particularly suitable for the type of genetic isolation which leads to the setting up of separate breeding populations.

In the Mekong, Sao-Leang and Dom Saveun (1955) noted that populations north of Khone falls migrate downstream of them, whereas fish from the southern reaches often move upstream of the falls. Blackfish species, particularly those whose longitudinal migrations are minimal, seem likely candidates for the evolution of sub-populations along the length of the river. The more mobile whitefish, however, have much more opportunity for dispersal and mixing of genetic strains, but even here there are some indications that sub-populations are formed. There are hints that some species have a homing instinct. Bonetto found tagged individuals of Prochilodus platensis in the same floodplain pool in successive years in the Parana River, as did Holden (1963) with individuals of Lates niloticus and Hydrocynus lineatus in the Sokoto. This, of course, may mean simply that the fish have not moved at all in the intervening flood period, but it does imply a certain degree of territoriality in the species concerned. Furthermore, Godoy (1959 and 1975) removed tagged individuals of Prochilodus scrofa from one site on the Mogi Guassu to place several hundred kilometres away and even located on different branches of the same river system. Some fish returned rapidly to the site from which they were captured originally, indicating the type of homing ability which is often associated with distinct stocks of migratory species.

Figure 5.3  Feeding and breeding locations of different stocks of bream (Abramis brama) in a river and riverine reservoir (Volga system). (After Poddubnyi, 1979)

The existence of separate sub-populations of species in river systems has also been proposed on other grounds. According to Durand and Loubens (1969), Alestes baremoze has two populations, one of which is migratory within the Chari and Logone rivers, the other resident in Lake Chad. A similar separation has been attributed to Basilichthys bonariensis on ecological evidence. Here there is an estuarine population in the Rio de la Plata, and a riverine population in the Parana. The riverine population can also be distinguished by its faster and more consistant growth rate (Cabrera, 1962). From the results of their tagging experiments, Bonetto and Pignalberi (1964) suspected the existence of an upstream and downstream population of Prochilodus platensis. Prochilodus platensis also has two body forms, ‘longilineas’ and ‘brevilineas’ which may depend on the early nutritional history of the individual fish (Vidal, 1967). Studies on a more restricted area of the La Plata estuary have also shown this species to have sub-populations with different growth characteristics (Cabrera and Candia, 1964), whereas Parapimelodus valenciennesi, a blackfish species from the same region, only has a single homogeneous population (Cabrera et al., 1973; Candia et al., 1973). Bayley (pers.comm.) has also collected meristic evidence of the existence of several distinct sub-populations of Prochilodus insignis in the Amazon. Separate sub-populations of migratory fish have also been identified with differing life-history strategies adapted to the range of physical and climatic conditions in the streams in which the fish spawn. The anadromous Alosa sapidissima studied by Legget and Carscadden (1978), for example, shows a gradation in fecundity and migratory distance in sub-populations associated with East Coast American rivers. These authors' arguments can probably be extended to other species and may apply to other systems which are sufficiently geographically or spatially diverse as to have a range of climatic conditions over their length.

There are indications that many riverine species exist in two behavioural phases which may correspond to sub-populations. One phase is migratory showing characteristic separation of breeding and feeding grounds. The other is static and is more inclined to form territories. Such behaviour has been described for Hilsa ilisha (Pillay and Rosa, 1963) in India and for Gobio gobio and Rutilus rutilus in Europe (Stott, 1967). The existence of such alternative forms would explain the ease with which some species adapt to changes in flow within rivers and readily adopt more sedentary habits after flood control or impoundment.

There appears to be good circumstantial evidence that geographic sub-populations exist in rivers and there is support for this from other inland waters such as the Great Lakes of N. America (Loftus, 1976). However, there are only few cases where such subpopulations have been positively identified. Whether or not a population of fish is disassociated into separate stocks at a sub-population level is of great significance to the management and conservation of fisheries for that species. Where separate subpopulations exist local depletion of the fish fauna by overfishing or by other environmental pressures is less likely to be compensated for by recolonization from larger populations

SIZE OF SPECIES

The species of fish inhabiting floodplain rivers cover a wide range of size as is illustrated by the maximum recorded lengths of the individual species from communities inhabiting three typical rivers in Fig. 5.4 in which the histograms represent the number of species classed according to their total lengths. Sizes normally span about 3 orders of magnitude (i.e., from about 1.5 cm to 1 500 cm). There tends to be a high proportion of fish of very small adult size (less than 10 cm) in rivers both in feeder streams or rocky headwaters and in the potamon reaches of the river. Small size is advantageous both in the riffles of the rhithron and in most floodplains, as pygmy species can mature more rapidly (often within one year) and can seek refuge in the root masses of vegetation in the interstices of pebble masses and other small crevices. Equally they can colonize the surface area of the water and more readily exploit the neuston or allochthonous food sources found there. Most river systems have a few species of truly gigantic size. In Latin America the characteristic giant species are Arapaima gigas, Pseudoplatystoma spp. and Brachyplatystoma spp., in Africa, Lates niloticus and in the Mekong Pangasianodon gigas, although these are by no means the only fish attaining lengths of 1.5 m or more.

Figure 5.4  Histograms showing the proportion by number of species of different maximum lengths in three typical tropical river systems

DISTRIBUTION OF SPECIES IN RIVER SYSTEMS

DISTRIBUTION IN SPACE AND ZONATION

Zonation

Many attempts have been made at establishing general principles regulating the zonal distribution of living organisms in rivers. Illies and Botosaneanu (1963) summarized the schemes proposed by a number of authors and themselves suggested a separation of the river into creon, rhithron and potamon zones based on faunistic criterea: A specific rule describing the distribution of Northern European species was advanced by Huet (1949) on the basis of the slope and width of any particular reach of the river (Fig. 5.5). More recently Vannote et al. (1980) suggested the river continuum concept which posits an orderly downstream progression of organisms as discussed in Chapter 3. These attempts mostly originated in the North Temperate zone and pre-suppose an idealized river form with a continuously decreasing gradient from source to mouth. As such they are useful within areas where the original climatic or morphological conditions prevail, but are transposed to the tropics only with difficulty. The temperate pattern of zonation has been demonstrated by Bishop (1973) for the Gombak river (Malaysia) and in other Malaysia rivers the cyprinids, at least, show a clear succession from Tor tambroides and Acrossocheilis hexagonolepis in upper rapids reaches, Leptobarbus hoeveni, Puntius bulu and P. daruphani in intermediate streams and P. jullieni only in the lower potamonic reaches (Tan, 1980). Watson and Balon (1984a) also found highly structured communities to exist in the Baram river of North Borneo. Here fish partitioned first into layers in the water column, surface, pelagic, benthic and substratum niche types. A tendency was noted for there to be an increase in the number of species inhabiting the bottom zone with increasing distance downstream. Furthermore a related reduction in niche size with the increasing numbers of species was indicated by a decrease in mean life span and adult size of the individual species.

Figure 5.5  Graphical representation of the relationship between slope of river bed and channel width and fish community in European rivers. (After Huet, 1949)

In Africa the Illies-Botosaneanu classification has only been applied successfully in some South African rivers (Harrison, 1965) or in the Luanza, a high altitude tributary of the Zaire river system (Malaisse, 1976) where there is a succession of rhithronic and potamonic reaches similar to those in Europe. In most parts of the continent zonation is much more obscure as is shown by examples from West Africa (Ogun river, Sydenham (1978); Bandama river, Merona (1981); Ebo Stream, Ghana, Lelek (1968)) or from Central Africa (Kalomo river, Balon and Coche (1974) and the Luongo river, Balon and Stewart (1983)).

Provided there are no geographical discontinuities such as large waterfalls, the increase in diversity along many rivers tends to be by addition of species to those present rather than by the replacement of species. This is particularly true of the tropics although Horwitz (1978) remarks on this aspect of the increase in diversity of fish faunas in streams of the mid-Western United States. In Africa and parts of Asia most large river systems have such discontinuities so that there are sufficient important exceptions for this not to constitute a general principle. For example, the Zambezi River is divided into three faunistic zones, the middle of which between the Cahora Bassa rapids and the Victoria Falls is the poorest, or in the Nile, the Kabalega Falls between Lake Kyoga and Lake Mobutu, marks a discontinuity between the generalized nilotic fishes downstream and a distinct riverine and lacustrine fauna upstream. Instead it is possible from the work that has been done, to separate the causes of species distribution in water courses under two main headings: geographical factors and geomorphological factors.

Geographical factors influence the taxonomic differences that may be encountered within a river basin. For instance, the isolation of sub-populations of species in small order streams can result in specific divergence over a long time period. In this way it is possible to observe two similar species occupying identical or similar niches in two separate streams of the same basin. This isolation is to a certain extent reinforced by the behaviour of the species themselves, which are often small, sedentary and have a short life cycle, which may contribute to a rapid rate of speciation. This tendency for species to diverge in the lower order streams of a river basin could account for much of the increasing diversity of species shown by larger river basins in Fig. 5.1 and for the relatively large proportion of species of small size in river fish communities. A second mechanism influencing distribution is river capture when species may be exchanged between systems. Mahon (1984) discusses some of these mechanisms in connection with the differences between fish faunas of similar river systems in Europe and N. America. In North America the more normal divergence has occurred with occupation of head water streams by numerous small cyprinical and percid species. In Europe the rivers are still occupied by larger cyprinid species which inhabit the potamon and only migrate to the rhithron to spawn. The headwater streams there tend to be used mainly by the young cyprinids. This difference is attributed mainly to the affect of glaciation.

Whereas geographical considerations may influence the distribution of fish species between river systems, the distribution of the various types of species within any one system is more likely to be controlled by the geomorphology of any particular river reach. Thus, it is useful to distinguish two major fish communities:

(i)   communities of rhithron-like or rapids zones; and

(ii)   communities of the potamon,

certain amount of interchange between these two communities may occur, particularly by certain elements of the potamon fauna which enter rapids to breed. Rhithronic fishes tend to be relatively static within their preferred zone, although juveniles of migratory potamonic species or even marine anadromous forms such as the salmon form temporary elements of their communities. Potamonic fishes can, however, be divided into two fairly distant groups on the basis of their behaviour.

(a) The first group of fishes avoid severe conditions on the floodplain by migration to the main river channel and frequently by more extensive movements in the river beyond the floodplain area. Members of this group are recognized a “Piracema” or “Subienda” species in Latin America, in the Mekong system are termed “white” fishes, and in other river systems may be classified as rheophilic. Species of Cyprinidae in Asia, Africa and Europe and Characoidei in Africa and South America are conspicuously members of this group, sometimes undertaking very spectacular migrations. Some siluroids and mormyrids are also migratory in behaviour. A few species are confined to the river channel at all times and never penetrate the plain.

(b) The second group of fish consists of species which have considerable resistance to deoxygenated conditions and which are termed “black” fish in Southeast Asia or limnophylic elsewhere. Their movements are therefore more limited than those of the “white” fish. They frequently remain in the standing waters of floodplain during the dry period and if they move to the river they remain within the vegetated fringes or in the pools of the river bed as it dries. Most siluroids belong to this category together with ophiocephalids (channids) anabantids, osteoglossids, polypterids and lung fishes.

Although it is very rough, the distinction between blackfish and whitefish is very useful as a first level ecological classification of species and will be retained in this book for this level of discussion. The response of the two groups to varying conditions on the floodplain are very different and have important implications for the management of the ecosystem and the fish stock.

Because many tropical rivers pass through several successive alternations between calm and rapids reaches, these two faunas likewise alternate along the length of the stream, which gives a type of zonation based purely on the flow or bottom characteristics peculiar to the river in which they are found as, for instance, in the Luanza River (Balon and Stewart, 1983). Species of several families have become adapted to life in rapids and as Poll (1959; 1959a) points out for the Zaire River, the inhabitants of downstream rapids are generally representatives of families which are normally found in the potamon, but which have occupied the rapids, rather than representatives of those families, such as the Kneriidae or Amphiliidae which are native to the rapids of the headwater streams. Asian and South American rivers tend to be of a more traditional form with steep mountainous headwater streams giving way to flat lower courses which are sometimes of extreme length. In such rivers the discontinuity between the rheophilic fishes of the headwater streams and the communities of the lentic potomon downstream is apt to be particularly abrupt.

Conditions become increasingly estuarine toward the mouth of the river where saline waters penetrate many kilometres upstream, particularly in lowland rivers of slight slope. A pronounced zonation of species occurs according to their salinity tolerance, although the zones move up and down river as the interface between saline and freshwater varies with the state of the tide and the flood regime of the river. Three groups of fish are to be found in this transitional zone.

 (i)  Freshwater stenohaline species which enter the zone during the flood and retreat upstream at low water according to the penetration of the saline tongue.

(ii)  Marine stenohaline species which follow the influx of marine waters into the river for feeding.

(iii)  Euryhaline species which move little but which adapt to the changing salinities of the water. These may be of freshwater origin (e.g., members of Cichlidae, Cyprinodontidae and some Siluroidae) or of marine origin (e.g., members of Clupeidae, Atherinidae, Mugilidae, Lutjanidae, Sciaenidae, Ariidae, Pomadasydae, Gerridae, Carangidae, Centropomidae, Eleotridae and Gobiidae).

Several species migrate between the river system and the sea either for breeding or feeding. Anadromous forms, where the breeding cycle is completed in freshwater, include several estuarine species of marine origin, the Eleotrids Batanga lebrotonis, for example, which undertake limited migrations upstream, as well as coastal marine species such as the clupeids which sometimes migrate over long distances in the river. In the temperate flood rivers of Southern Europe the sturgeons (acipenseridae) and salmonids are the main anadromous fishes. Truly Catadrcmous species are somewhat rarer in large tropical rivers, although eels are present in some systems, noteably Anguilla nebulosa in the Zambezi and its tributaries. Many normally marine species enter the lower reaches of rivers to feed during the dry season and return to the sea during the rains.

Habitats of River Systems and the Accompanying Floodplains

A number of areas can be considered distinct habitats from differences in morphology, chemical and physical conditions, presence or absence of vegetation, structured objects such as rocks, wood or other cover and abundance and type of food. Table 5.1 based on such ecological characteristics, lists the major habitats found in river systems.

The main channel

Fish in the rhithronic reaches are distributed either among the riffles or in the deeper pools. Although the riffles present a reasonably homogeneous environment they are structurally complex and require specific adaptations either to resist the current and the turbulent conditions found on the surface or to live in the crevices and holes in the rocks. Large, swift swimming, streamlined species sometimes penetrate those turbulent waters but either pass through in an upstream migration or drop back to the shelter of the pools. Riffles are also favoured environments for deposition of eggs of some species and for the subsequent development of their larvae and fry. The adaptations to swift current also fit fish to exploit the available food sources in the riffles. Swift swimming forms pick up allochthonous material falling on the surface of the water or drift organisms. Fishes with adaptations to cling to the surface of the stones often have sucker like mouths with which to feed on epilithic or epiphytic organisms whereas the diminutive or elongated species that inhabit the interstices in the rocks are particularly well situated to prey upon the numerous insect larvae crustacea that inhabit the bottom.

Pool habitats are much more varied and fish tend to segregate both by depth and by distance from the shore. Three communities can be readily distinguished within the water column. A pelagic community which tends to consist of small species, silvery in colour with upward facing mouths, a mid-water community of larger silvery fishes streamlined with terminal mouths and a bottom living community of drab coloured fishes with dorsally humped profiles and ventrally positioned mouths. These various forms are well illustrated by the communities of cyprinids in a small Borneo stream as described by Inger and Chin (1962) Fig. 5.6. The slacks near the bank are often heavily vegetated with floating and emergent aquatic and everchanging terrestrial plants. Here bottoms are usually composed of foodrich detritus from leaf fall and from silt deposition in the quieter waters.

The slack areas at the fringes of pools, the limited floodplains of the rapids reaches and perrenial streams of the lowland zone have similar faunas many of the elements of which recur within the floodplains of the potamon. Because there are a variety of microhabitats in the slacks small size is a great advantage. Spatial niche selection seems very highly developed and all levels of the watercolumn are occupied as in the example of the Borneo streams above. In a tropical African forest stream several genera of cyprinodonts occupy the water surface, particularly Epiplatys and Aphyosemion although some Characins such as Brycinus macrolepidotus are also found there. The mid-waters support clouds of dwarf Barbus and Aphyosemion species which are more often than not associated with the shade of floating lily leaves. The rich detritus of the bottom attracts several species of small mormyridae, Neolebias, Barbus and Labeo together with various small or dwarf ciclids such as Pelvicachromis, Thysia ansorgii, Hemichronis bimaculatus. Similar elongated fishes are found in the rocks also occur in the floating vegetation at the fringes of the pools as the sinuous habit is equally adapted to such conditions. Mastacembelus and Calamoichthys particularly are conspicuous in such environments as they also frequent similar vegetation in the potamon.

Semi-permanent channels, which are more characteristic of low order streams, frequently become broken-up into a series of rocky riffles, and pools which may become deoxygenated or totally anoxic as the season advances. Very great densities of fish may be found in the pools, which essentially represent the sump into which most of the population drains. Lowe-McConnell (1964), for instance, found 870 fish belonging to 36 species in a pool of 19 m³. This she attributed to a three dimensional use of space within the pool, together with an alternation of activity between two very different nocturnal and diurnal fish faunas. The species composition of main river pools is rarely the same as that of floodplain pools, and within the pools segregation appears to occur by bottom type. Lowe-McDonnell recorded 44 species of fish in the pools of the Rupununi river, of which 37 (84%) were found over only one type of substrate.

During the floods the main river channel is filled with rapidly flowing water, but at its margin the floating vegetation fringe merges with the flooded banks of the plain itself. Large masses of vegetation become detached and are swept downstream as floating islands. As a consequence the channel is usually comparatively sparsely populated at this time, although some species, for instance Petrocephalus bane or Hydrocynus forskahlii, never move out of the river on to the plain (Daget, 1954). In Latin American rivers, the upstream migration, “Piracema” or “Subienda”, of the major characins during low water and the beginning of the flood also ensures that some species are present in the channel during at least the earlier phases of rising water. The fringing vegetation and floating sudd islands shelter juvenile fish or small vegetation dwelling species, which may become distributed throughout the system in this way.

Figure 5.6  Segregation by depth of cyprinid species in a Borneo stream. (From data in Inger and Chin, 1962)

In the Potamon most species of fish move to the main channel during the dry season and settle in different habitats along its length. Some dry season habitats can lie a considerable distance away from the main floodplain, and may even be located in entirely different aquatic systems such as the sea or major freashwater lakes.

In the permanent channels during the dry season, fish also separate by depth, type of bottom and vegetation cover. Many smaller species inhabit the root masses of the floating vegetation at the edge of the river, finding there both abundant food and shelter. Several species have specific adaptations to this habitat. The “upside-down” swimming position of some Synodontis species enables them to browse on the root fauna, and the serpentine shape of Mastacembelus or Calamoichthys enables them to weave among the entangled stems of the plants.

Deeper areas of the main channel attract the larger fish species. Pangasius sutchi, for example, migrates to the deepest reaches of the Mekong river between Sambor and Stung Tren during the dry season (Sao-Leang and Dom Saveun, 1955). The larger individuals of Lates niloticus are reputed to frequent the deeper parts of the African rivers within its range. Similarly Carey (1967a) noted that individuals of several species were much larger in the Kafue river than in the adjacent lagoons.

Several species are pelagic within the main channel, and also in the larger permanent water bodies. In African rivers, small groups of Brycinus macrolepidotus are to be seen cruising at the surface under overhanging vegetation, and small clupeids characins or cyprinids occur in the surface waters of all three continents. The surface film presents a specialized habitat occupied by small cyprinodonts which are found in all quiet stretches of river and floodplain alike.

Three other special habitats have been noted by Lowe-McConnell (1964, 1967) and Mago-Leccia (1970). Some fishes burrow into sand bottoms. Gymnorhamphichthys hypostomus of the Rupununi and Venezuelan savannah rivers has an elongated snout which facilitates respiration while buried. Potamotrygon hystrix and Xenogoniatus have also been recorded from the bottom of Venezuelan streams. A somewhat similar habit has been noted by Daget (1954) and Cromeria nilotica which burrows in sandy bottoms of the Niger river when alarmed.

Leaf litter on the bottom of forested streams and pools also yield a rich harvest of small species some of which do not occur elsewhere in the system. Lowe-McConnell listed Agmus lyriformis and Farlowella sp. to which Mago-Leccia added Aequidens, Apistogramma, Orinocodoras, Corydoras, Agamyxis, Homodiaetus and Hyphessobrycon. The crevices and hollows of the mass of decaying branches which also accumulate in such creeks, are inhabited by several small types of fish. Lowe-McConnell reported having collected 17 species and over 200 individuals in two hours from split logs and branches. In the Rupununi these were mostly catfishes, such as Platydoras, Trachycorystes, Pseudopimelodus, Hoplosternum or Ancistrus. In the Apure river some gymnotids also have been recorded from the crevice dwelling habitat including Electrophorus electricus, Sternopygus macrurus and Apteronotus albifrons, although catfishes such as Panaque nigrolineatus also occur there. Both Lowe-McConnell and Mago-Leccia maintained that crevice dwelling is associated with nocturnal habits. Specialized habitats of this type have only been studied in Latin America, but it seems probable that they also exist in both African and Asian inland waters.

The blind arms and backwaters of the main channel remain in communication with the main river at one end, but are otherwise lenthic, having many of the characteristics of floodplain lagoons. They are usually especially rich localities, as silt accumulates in them giving rise to plankton blooms and increased primary production. They are sheltered and accumulate tree trunks, branches, rocks and other types of cover providing material for shelter so many fishes enter them as a refuge from the current of the main stream. As such backwaters are of particular importance as major concentrators of ichthyomass, and in rivers such as the Danube and the Mississippi are the main surviving features of the flood system. In the Mississippi particularly they are identified by Schramm et al. (1974) as providing essential habitats for feeding and reproduction in at least 57% of the fish species recorded from the reach between St. Louis and Cairo. Guillery (1979) also identified a movement of 15 mainstream species on to the floodplain coincident with flooding of which some adults and juveniles remained trapped in sloughs and blind areas during low waters.

When inundated the plain has a rich mosaic of habitats although there is little information on the distribution of fishes amongst them. A basic difference exists between forested and savannah plains, although originally many plains that are now exposed supported greater tree cover. This means that much of the present day fauna on the now denuded plains has become modified in recent times as human activities have worked upon the environment. Flooded rain forests themselves appear to be varied habitats.

Observations on the fish faunas of flooded forests are limited, but indications from Grand Lac of the Mekong (Bardach, 1959), the Zaire river (Matthes, 1964) and the Amazon (Roberts, 1973) attest to the variety of species in such fish communities. Goulding (1980) particularly has investigated the fish fauna inhibiting the Amazonian Igapo showing the importance of the fruit and seed bearing trees to a variety of fishes, including Colossoma macropomum, C. bidens, Brycon spp., Myleus spp., and Serrasalmus spp. Occupancy of the flooded forest by these species is closely allied to the distribution of these favoured fruit bearing trees.

On floodplains with more restricted gallery forest or bush scrub, the submerged branches and roots provide a feeding substratum and concealment for many species. Mago Leccia (1970) lists several fishes from the Orinoco river which occur among flooded scrub. These include nocturnal armoured catfishes such as Hypostomus and Pterygoplichthys and the gymnotids, Sternopygus macrurus, Rhamphichthys rostratus, Adontosternarchus sachsi and Eigenmannia. Diurnal forms such as the cichlids Astronotus ocellatus and a multitude of characids including Triportheus, Cynopotamus, Astyanax, Moenkhausia and Thoracocharax were also found. The flooded scrub habitat is common on most tropical plains, and as observed above, probably more nearly approximates to the original condition of most of them. Thus, fish communities of this type are presumably widespread although they have not been specifically described from other systems.

The floating meadows, which now seasonally cover many of the world's savanna floodplains, appear at first sight to contain little variation. Closer examination reveals a fine texture allied to contour. There are deeper places around lakes and depressions which are often free of vegetation or have a flora more typical of permanently wet areas, and shallower places over the levees and terraces. Local differences in vegetation and bottom type are associated with these features and it may be assumed that fish segregate accordingly. There is no doubt that certain areas do have particular attraction for characteristic species. The most important factors controlling such distribution are probably dissolved oxygen concentration, depth, substrate and vegetation cover.

The littoral zone of the plain, that area of interface between land and water, is usually colonized by young fish. In Africa these are almost entirely cichlids of the general Tilapia, and Sarotherodon or Oreochromis and cyprinodonts, which have specific tolerance of the elevated temperature found there.

On some of the most extensive plains, where sheet flooding is common, there is an intermediate zone where the areas flooded by rainwater meet those inundated by the rising river level. This zone marks the boundary between two types of water of different productivity and may limit the distribution of fish on the plain. Blanc, Daget and D'Aubenton (1955) distinguished two zones on the Niger river floodplain, a) a zone corresponding to the major bed of the river which was rich in fish and b) a peripheral flood area which was less densely colonized. A similar distribution pattern has been reported from AraucaApure flood system by Matthes (pers.comm.), from the flooded forests of the Mekong by Le Van Dang (1970) and from the Sudd of the Nile River (Mefit Babtie, 1983). Some of the patchiness in fish distribution on other large floodplains may be attributable to such differences in water quality.

In the dry season most of the plain is drained leaving only the network of depression pools, lagoons and swamps, some of which dry out and some of which persist until the next flood. Most fish leave the plain at this time, but a certain section of the community remains in these standing waters. Of the fish remaining, a proportion is composed of species which would normally retreat to the river channels, but which have become isolated in various depressions. The majority of these die through desiccation of the water body in which they find themselves, through deoxygenated conditions, or through exposure to excessive temperatures. Sane find their way to deeper lakes where they may survive.

An assortment of fish stay on the plain. Daget (1954) and Blanc, Daget and D'Aubenton (1955) listed Marcusenius senegalensis, Pollimyrus isidori, Petrocephalus bovei (mormyrids), Gymnarchus niloticus, Heterotis niloticus, Ctenopoma, Parachanna, Polypterus, Synodontis, Clarias, Hepsetus, Auchenoglanis and Heterobranchus as comprising the characteristic fauna of the floodplain pools of the Niger river. These genera recur on floodplains throughout their range in Africa, which in some cases is very widespread. In the Mekong, the blackfish assemblage described by Sao-Leang and Dom Saveun (1955) contains a similar group of fishes, many of which belong to the same families and closely resemble the African species. Major examples such as Ophicephalus striatus, O. micropeltes, Anabas testudineus and Clarias batrachus remain in depression pools throughout the dry season to spread over the plain during the flood. Species recorded from the lagoons and pools of the Apure river, Venezuela, by Mago-Leccia (1970) included Hoplias malabaricus, Serrasalmus notatus, Callichthys callichthys, Hoplosternum littorale, Pseudoplatystoma fasciatum, Sorubim lima. Pimelodus, Leporinus and Pimelodella. Similar assemblages, often involving the same species or genera, have been described from the Rupununi by Lowe-McConnell (1964), from the Parana river by Bonetto et al. (1969) and from the Magdalena by Kapetsky et al. (1976).

There are considerable differences in the species composition of fish population from floodplain pools of the same system, as typified by the data given in Fig. 5.7. Attempts have been made to correlate these with the number of variables. Welcomme (1975) traced the increased specific diversity of populations with size of pool in the Oueme system (Table 5.2). That size of pool can influence species composition has also been found by Lowe-McConnell (1964) who found that larger species inhabited the larger bodies of water. Holden (1963) found the same effect on the Sokoto river and noted that individuals of the same species tended to be larger in the bigger pools.

Figure 5.7  Percentage representation of major fish groups in eight floodplain depression lakes of the middle Parana system. (After Bonetto et al., 1969a)

^a   up to 500 m²
^b   500–5 000 m²
^c   over 5 000 m²

The influence of size of pool may depend on the heightened dissolved oxygen concentration arising from lessened vegetation cover and improved aeration by wind. Such relatively favourable conditions allow many more species to survive, including seme which would normally return to the river channel in the dry season. Greater area al so leads to a greater diversity of habitat, with more varied bottom types, submersed, floating and emergent vegetation, and open water in the place of the densely packed stands of aquatic plants found in the smaller pools. On the basis of samples from a selection of floodplain pools from the Sokoto river, Holden (1963) concluded that several species were distributed according to substrate. Although no species was confined to any one bottom type, certain species, such as Alestes dentex, were more common over sand. Oreochromis galilaeus was found more frequently over mud, and Tilapia zillii over intermediate bottoms. There is little explanation for the variability in species composition found by Bonetto et al. (1969) (Fig. 5.7) in the pools of the Parana river, either in terms of vegetation cover or size of water body, although Cordiviola, de Yuan and Pignalberi (1981) attribute the variations in these and other lagoons to a number of factors. Firstly, geographical location in the river valley determines overall species composition with Triportheus paranensis, Moenkhausia dichroura, Acestrorhynchus falcatus and to a certain extent Prochilodus platensis being dominant to the North (Corrientes area) and P. platensis alone being dominant to the South (Santa Fe area). Elsewhere Bonetto (1975) has stressed the abundance of mud eating fish in lagoons whose bottom is of fine particulate organic material. Secondly specific content of the catch may be generally related to a complex of topographical and biological factors related to the biology of the species. Thirdly, while some species are of constant occurrence in the ponds over at least five years the presence of other species appears much more haphazard. Depth and exposure may also influence the distribution of species composition of the shallow sheltered bay habitat of Magdalena river lagoons, where Potamotrygon and Pseudoplatystoma dominate, as opposed to the open water where Hemiancistrus, Triportheus and Prochilodus are more cannon. Other species such as Plagioscion are distributed more or less indifferently.

How much the distribution of species among the dry season habitats is a matter of chance, with only those specially adapted species surviving adverse conditions, and how much is a matter of deliberate selection of habitats is not clear. It seems likely that those blackfish species, which always remain on the plain, seek out, or maybe never leave, the vicinity of the depressions where they pass their lives and the year-to-year recurrence of such species in any one lagoon would support this. With migratory species, on the other hand, there would appear to be a considerable element of chance and the precise composition of the fish fauna of the larger lagoons is probably determined by the haphazard trapping of such individuals. The shoaling habit of many of these fishes would lead to them being either isolated in large quantities, or not present at all, giving very skewed distributions of abundance.

There is some evidence from the Rupununi swamps that species do select their dry season habitats very closely. In Lowe-McConnell's list of 129 species, 79 percent were found in only one habitat (excluding channels draining the floodplain which would be likely to have a high proportion of transient species). More generally, however, distribution patterns do not appear to be so exclusive and preferences are shown by greater concentrations in one locality rather than another. Such selection may be by species, although different life stages or age groups of the same species may also segregate in this way. Introspecific resource partitioning of this type has been shown for fish from headwater streams in U.S. rivers such as the pearl dace, Semotilus margarita where in summer the 0 year class occupies shallow pools and deep channels and age 2+ fish occupy deep channels and pools. This segregation is also reflected in differences in feeding pattern which may reduce competition between age classes in this omnivorous species (Tallman and Gee, 1982). This type of segregation is especially common in those species which utilize the rhithron either as a full time habitat or periodically for breeding. There the use of riffles for depositing spawn, for the development of the fry and the growth of juveniles is widespread, whereas the pools serve to house the larger individuals of the resident species. Oreochromis species also have different habitats for feeding of juveniles (nurseries in the littoral zone) and adults (the bottom in deeper water). They also segregate for breeding, with the mature males being found over distinct nesting sites on the bottom, non breeding adults, both male and female, staying in mid-water adjacent to the breeding sites, and brooding females carrying eggs in their mouths being located in sheltered brooding sites near the nurseries. Not all fishes show such complex distribution patterns, but in many, some phase of the life cycle (usually, but not always the immature stage) is passed in a habitat other than that frequented by the other age groups of the species.

DISTRIBUTION IN TIME AND MIGRATION

Types of Migration and Movements

Migrations of Adult Fish

Because the best breeding habitat rarely coincides with the best feeding habitat most species have two distinct centres of concentration and fish have to travel, sometimes over long distances, between the two. Two components of such movements have been recognized by Daget (1960) for tropical African species. His categories of longitudinal and lateral migration are of general application to flood rivers everywhere. Longitudinal migrations are those which take place within the main river channel, and lateral migrations are those whereby fish leave the main channel and distribute themselves over the floodplain.

Four phases of fish movement have been identified separately by Blache et al., (1964) and Williams (1971) from the Chari and Kafue rivers respectively. Combining their two classifications, six main phases in the distribution of fish emerge:

(a)   longitudinal migrations within main channel: these are usually upstream, but not always so;

(b)   lateral migrations on to the floodplain;

(c)   local movements on the floodplain and distribution among flood season habitats;

(d)   lateral migration from the floodplain towards the main channel;

(e)   longitudinal migrations within the main channel: these are usually downstream but not always so;

(f)   local movements within the dry season habitat: this may be the river, adjacent lake or in sane cases the sea.

Although these are broadly applicable to the majority of individuals of most fish species inhabiting flood river systems, some species are confined to one habitat only. It is also not certain that all individuals of the mobile species do in fact undertake migrations every year. Within the above pattern three distinct groups of freshwater species can be distinguished.

(i) The “blackfish” species, whose migrations between dry and wet season habitats are restricted and which, at the most, undertake lateral migrations to the fringes of the main channel. These species are more normally confined to the plain spreading over it during the floods from the residual pools and lagoons which are the dry season habitat;

(ii) Those species which undertake moderate movements within the river, but which spawn on the floodplain. The major migrations are for dispersal to dry season habitats such as those of the African Brycinus leuciscus described by Daget (1952), or Crossocheilus reba (Tongsanga and Kessunchai, 1966), and other Mekong species (Bardach, 1959). Migrations to the favoured breeding places on the floodplain may be either upstream or downstream and rarely involve the formation of large shoals. Migrations of this type appear to be mainly for the avoidance of unfavourable conditions on the plain;

(iii) The “whitefish” species which undertake an upstream migration during the dry season or early in the wet season. Such migrations are usually linked both to reproduction and to the need to escape the adverse conditions of the downstream river channels and lakes in which water levels and dissolved oxygen concentrations may become dangerously lowered for sensitive species.

The migrations of freshwater fish in the tropics are best known from the South American continent where a wide variety of migrations occur. A number of medium sized systems seem to have migratory fish populations showing similar characteristics with a single seasonal movement to and from a down river feeding zone and an up river feeding zone. Such patterns have been reported from the Rupununi river with Boulangerella cuvieri, Hydrolicus scomberoides and Myleus pacu (Lowe-McConnell, 1964), the Pilcomayo river with Prochilodus platensis (Bayley, 1973), the Sao Francisco river with Prochilodus sp. and Duopalatinus emarginatus (Paiva and Bastos, 1982) and are presumably repeated throughout most moderate size rivers of the continent which carry characin and siluroid populations. The most complete studies of this pattern of migration are from the Mogi Guassu in which a succession of fish species move up-river at rising water (Godoy, 1975) (Fig. 5.8). A similar but more complicated sequence of movements is shown by several species from the River Magdalena. Here there are two seasons of activity, the first the major “Subienda” involves a migration from the extensive deltaic downstream floodplain to the more shallow upstream reaches in February and March. Prochilodus reticulatus is the main species involved, although it may be accompanied by others including Brycon moorei, Pimelodus clarias and Pseudoplatystoma fasciatum. This is followed by a downstream movement in April to June and a second, minor, upstream migration the “mitaca” in July to September with a final downstream movement in October-December. Prochilodus reticulatus is also involved in similar migration upstream from the floodplains of the Catatumbo river at its point of inflow into L. Maracaibo.

Figure 5.8  migration of fish within the Mogi-Pardo-Grande river system. (After Godoy, 1975)

In larger rivers the migration patterns tend to become more complex as migrations within the tributary rivers at the main river channel are mixed with movements between these various elements of the system. Thus different but equally complicated patterns have been reported from the Parana river system and from the Amazon with some observations from the Orinoco indicating an intermediate level of complexity.

Two main species Prochilodus platensis and Salminus maxillosus has been studied in the Parana and Uruguay rivers by Bonetto and Pignalberi (1964), Bonetto et al. (1971) and Bonetto et al. (1981) although several other migratory fishes were also investigated by them. The studies which combined observation and tagging experiments show the movements of fish in the Parana and its tributaries to be elaborations of the simple upriver and downriver translocations between feeding and breeding sites. Prochilodus shows fairly straight forward movements in the peripheral rivers such as the Pilcomayo but in the main stream individual fish appear to move upstream or downstream more or less indiscriminately. This seeming confusion arises from the size of the system, the thermal regime in the South which is severe enough to influence behaviour, the assumed existence of separate upriver and downriver populations and to the presence of several major tributaries. A further example, Salminus maxillosus, moves downstream in the Parana river in October whereas the same species is ascending the Bermejo river at the same time. In an attempt to resolve these complex patterns Bonetto et al. (1981) proposes the diagram in Fig. 5.9 to represent current thinking on fish movements in this river. In at least one species from the Parana, Luciopimelodus pati, migration occurs in late summer on falling water when fish travel upstream for 400–600 km to arrive at the spawning sites just before the next rise in water level. The curious feature of this migration lies in the behaviour of the male fish which tend to remain upriver to an increasing degree as they grow older so that females between 40–70 cm in length come to dominate the downstream populations.

Figure 5.9  Summary diagram of principal movements of fish within the Parana river. (Bonetto et al. 1981)

Migrations of fish in the various components of the Orinoco system are still very imperfectly understood, although observations have been made on Semaprochilodus laticeps in the lower reaches of the river from Caicara to the delta (Novoa, 1982). Behaviourally this species resembles P. platensis in that eggs and young fish drift downstream from an upriver spawning site, move laterally into lagoons where they grow and subsequently emerge to migrate up the main channel to spawn some two to four years later. This pattern of movement produces the type of upstream increase in mean adult size observed by Goulding (1981) for the Madeira and for male Luciopimelodus pati in the Parana and would suggest that in the Orinoco at least one discrete stock of S. laticeps is involved.

In the Amazon several types of migrations appear to be undertaken by the same species. Many of the larger characin genera, Colossoma, Brycon, Mylossoma, Triportheus, Leporinus, Schizodon, Rhytiodus, Prochilodus, Semaprochilodus, Anodus and Curimatus show a pattern of migration originally described as “Piracema” by Ihering (1930) and Geisler et al. (1973). Recent investigations by Goulding (1980) in the Madeira river, Goulding and Carvelho (1982) in the Amazon and Ribeiro (1983 and a) in the Rio Negro have shed more light on the nature of these movements which can be broken down into three components (Fig. 5.10).

Figure 5.10   Diagram of assumed migration patterns of prochilodontid fishes in the Amazon and Madeira rivers

1. The cycle starts with a migration downstream in a tributary river on the rising flood which culminates in breeding at the mouth of the tributary. Eggs and young fish then drift downstream in the main channel until of sufficient size to move into marginal vegetation of the floodplains or other suitable nursery sites.
2. Adult fish then reascend the same tributary and disperse into the flooded forest to feed.
3. As water levels fall vast shoals of fish move down the tributary, up the main channel and into the next tributary upstream. It is this phase of the migratory cycle which is referred to as the true “Piracema”.

This type of migration results in a lack of juvenile fish in the tributaries and an accumulation of larger, older fish as one moves towards the headwaters and is probably adaptive to the very poor trophic conditions for juvenile fish in the flooded forests fringing the black-water tributaries.

In Africa characins and some siluroids are conspicuous among the migratory species, but the cyprinids also show this type of behaviour. Potamodramous migrations, whereby adult fish leave lakes and ascend rivers during the floods to spawn in the upstream swamps, have been described by Whitehead (1959) for 18 migratory species in the Lake Victoria/Nzoia river systems, and De Kimpe (1964) for some mormyrids, characins and cyprinids of the L. Mweru/Luapula river system. Carmouze et al. (1983) distinguish three groups of migrants in the Lake Chad basin. True migrant species, Alestes baremoze, Brachysynodontis batensoda, Distichodus rostratus, Marcusenius cyprinoides, Petrocephalus bane and possibly Hemisynodontis membranaceus, Labeo senegalensis and Hydrocynus brevis made large-scale longitudinal movements for breeding. Such movements were usually between the lake and the Yaeres floodplain, but a further species A. dentex appeared to reproduce in the upper reaches of the Chari and Logone rivers and hence made even more extensive migrations. Mixed migrants made similar large scale longitudinal movements whose motives were less certain as both adult and juvenile fish were involved. These species included Schilbe uranoscopus, Synodontis schall, Hyperopisus bebe, Mormyrus rume and Eutropius niloticus. Finally a well defined class of lateral migrants was present whose movements between the river channel and the floodplain were mostly associated with feeding.

A similar group of species is involved in an annual migration up the Senegal river from the lower reaches around Dagana to as far up river as Dioulde-Diabe where further movement is asserted by a shallow rocky sill across the river bed. The fish then move laterally across the plain during the floods and return to the river in falling water. The downstream movement was found by Reizer (1974) to take the form of a drift produced by the failure of fish to combat the current in the main channel. As larger fish are generally more capable of combatting the current there is a tendency for them to be swept less far downstream and a noticable gradation in mean size results in smaller fisher nearer the mouth of the river.

Long distance spawning migration has also been recorded in S.E. Asia where Probarbus jullieni migrate up the Mekong river to spawning sites having shallow water, moderate current and sandy bottoms. Sritingsook and Yoovetwatana (1976). Day (1958) notes that there are several components of the fish communities of Indian rivers which move long distances to the spawning sites in the hill tributaries of the major rivers. Species of the genus Tor are particularly famous for their ascent of the Himalayan streams. Several cyprinid species migrate during the early part of the spring flood in the Tigris/Euphrates system (FAO/UN, 1954). Notable amongst these are Barbus xanthopterus, B. grypus and Aspius vorax whose adults spawn on upstream gravel beds, and whose young later drift downstream to occupy the floodlands in the lower reaches of the river complex. Other cyprinid species remain downstream to spawn in the river channel or in the swamps, e.g., Barbus sharpeyi or B. luteus.

Not all migrations are for breeding, however, as Thynnichthys vaillantii appear to move mainly to avoid adverse conditions in their downstream habitat during the dry season. This species migrates out of the flood lakes of the Mahakan system of Borneo during falling water and moves progressively upstream as water levels and dissolved oxygen concentrations fall downstream. The movement is accompanied by gonadial ripening but spawning is not reported to occur until the fish have returned to the lakes during the rising flood (Saanin, 1953). The abandoning of downstream lacustrine habitats due to unfavourable conditions has also been recorded among the Mesopotamic fishes which also move out of the marshes and floodplains of the Tigris/Euphrates at low water.

In European rivers many species are normally long distance migrants to the extent that Mahon (1984) comments on the prevalence of this behaviour in European rivers as contrasted to the relatively low percentage of migrant species in equivalent North American waters. Most of these species still undertake seasonal migrations although the highly modified state of most waters has limited these severely. Many records exist of migrant species which have disappeared following the damming of rivers but Zambriborsch and Nguen Tan Chin (1973) cite examples of continuing semi anadronous migrations by Aspius aspius, Blicca bjoerkna and Abramis brama in the Kiliya arm of the Danube, and Belyy (1972) attributed similar behaviour to Lucioperca lucioperca from the Dneiper.

In estuarine deltas where sea water penetrates upstream in the dry season, freshwater species also undergo local movements to avoid the saline conditions. Reizer (1974) noted that the more sensitive species move as a wave for up to 100 km in front of the saline tongue. In the Senegal river these are Hyperopisus bebe, Mormyrus rume, Mormyrops deliciosus, Marcusenius senegalensis, Alestes dentex, Citharinus citharus, Labeo senegalensis and Schilbe mystus. Similar migrations undoubtedly occur in all other rivers having a long zone of interaction between sea and fresh waters.

Marine and brackish water species also show a variety of migration patterns. Entry into the lower reaches of the river with the saline waters during the dry season may be regarded as an extension of the estuarine environment. However, penetrations into the freshwaters further upstream are also a common feature which has been described from many systems. Fish of marine origin regularly move many hundreds of kilometres up such rivers as the Niger in Africa (Reed et al., 1967), the Mekong in Asia (Shiraishi, 1970) or the Magdalena in South America (Dahl, 1971). These penetrations of.the freshwater habitat may be for feeding or for breeding. Some notable anadromous migrations also occur, for instance that of Hilsa ilisha, which moves up many Indian and Southeast Asian rivers to spawn (Pillay and Rosa, 1963), the Golden perch, Macquaria ambigua, which migrates up the Murray river in Australia (Butcher, 1967). The salmonids of Europe of N. America or the many migratory species of the Volga river which moved into the headwaters of the river each year prior to the construction of the dams there (Poddubnyi in Mordukhai-Boltovskoi (1979). Other less spectacular breeding migrations are those of the eleotrids, such as Batanga lebretonis or Dormitator latifrons, which enter rivers to spawn on the flooded vegetation fringes of the lower reaches.

Movement of Juveniles

Originally it was suggested that only the longitudinal component of migration was active, being directed by physiological stimuli, and that lateral migration was more of a passive affair with fish being swept by the rising floods on to the plain. It has since become apparent from the orderly succession of the migrations, and from the way in which fish often swim against the current to gain access to the plain, that this is not generally the case. In fact it would appear from the available but rather circumstantial evidence that most movements of healthy adult fish are directed rather than passive. Passive drifts do occur, nevertheless, where the eggs and larval forms of several species profit by the current for transport from the spawning site. Observations from the Danube and the Tigris/Euphrates systems and from various rivers in the United States (e.g., Muth and Schmulbach, 1984) have shown the earlier phases of return movements of younger stages to be by passive drift with the current. Similarly Godoy (1959) noted that the eggs of Prochilodus scrofa are semi-pelagic and are carried downstream by the current from the spawning sites which are in mid-channel. The eggs, however, develop rapidly and the fry soon take refuge in flooded oxbows only to be washed out later by falling water when they swim downstream to the major feeding grounds. In the Pilcomayo river, lateral floodplains are absent in the upper courses and Bayley (1973) supposed that the eggs and fry of P. platensis are swept directly downstream. Similar drifts are suggested by Novoa (1982) for Semaprochilodus laticeps in the Orinoco, several species of characins in the Amazon (Goulding, 1980) and Hilsa ilisha in the Indus (Islam and Talbot, 1968). Reynolds (1983), also draws attention to the differences in migration patterns in the Murray-Darling river as a correlate of reproductive strategy. Here only those fish with buoyant eggs such as the golden and silver perches undertake extensive upstream movements in order to allow the eggs space for their downstream drift. Some light is shed on the nature of the passive larval drift through work in the Missouri river (Harrow and Schlesinger (1980) where a rich and varied ichthyoplankton was composed of larvae of Aplidonatus grunniens, catastomids, cyprinids, centrarchids, sturgeon and paddlefish. The larvae disappear from the drift abruptly at about 8–12 mm in length either by catastrophic mortality or because they leave the drift to occupy more protected and food rich habitats. The latter view is supported by the rapid build-up of post-larval and juvenile fish in the backwaters of the river coincident with this disappearance and would also indicate purposeful movements beyond a certain size. Similar conclusions were reached by Nezdoliy (1984) for Lucioperca lucioperca, Perca fluviatilis and Abramis brama fry in the Ili river when the initiation of active migration was marked by schooling behaviour.

The mechanisms for “passive drift” movements of juvenile fish may be complex as Pavlov et al. (1977) found in the Volga and Kuban rivers where juveniles of several cyprinid, clupeid and gobiid species move downstream from the spawning grounds to the feeding grounds by drift. This occurs at night in clear water rivers probably due to the loss of visual fix by the young fish, thus there is a marked diurnal migration pattern. In the Kuban, a turbid river, the drift continued around the clock. Vertical distribution of the young fish differ considerably during the 24 hour period and horizontal distribution patterns appear to be species specific with some species located in the middle of the channel and others along the banks. Vertical distribution is also apparently highly biased in some species such as Catastomus commersoni which tend to occupy the surface of the water rather than the bottom, particularly as individuals increase in size (Clifford, 1972). Such distributions are regarded by Clifford and by Pavlov et al. as an active process controlled by definite behavioural reactions. Therefore the downstream movement, although passive in the sense of propulsion, may involve some active participation thereby differing from the distribution and transport pattern of inert matter or plankton within the stream.

The larvae of some species which lay their eggs in floodplain lagoons congregate on the bottom and show the peculiar behaviour of frequently ascending to the surface and then sinking back down again. Daget (1957) noted this behaviour in Heterotis niloticus and interpreted it as a respiratory adaptation. However, similar behaviour in Labeo niloticus (Fryer and Whitehead, 1959), and Lucioperca lucioperca (Belyy, 1972) and several cyprinid species from the Lli River (Nezdoliy, 1984) is thought to be to catch currents for transport to the river and eventually downstream. In fact Belyy found that fry were moved considerable distances in this manner and it could be that the same behavioural pattern serves the two functions equally.

The downstream movement of juvenile fish within the main channel is an important phase of the total migratory pattern of long distance migrants. It ensures the return of stocks to areas vacated by the adults, and in many cases, where the feeding grounds differ from the breeding grounds, enables the young fish to reach sources of food. However, the passive drift type of behaviour contrasts with that of fry spawned in the mountain headstreams of Indian rivers, which according to Day (1958) take their time migrating downstream and are often halted for one or more dry seasons in the residual pools of the main river channel. This type of behaviour would seem better adapted to continual survival in torrential streams where the loss of control inherent in the transport by drift may rapidly carry the fry outside the range of the preferred habitat.

Distance and Speed of Movement

The main long distance migrations of tropical freshwater fish are those undertaken by the South American characin species. Tagging experiments, particularly those undertaken on the Parana system, have yielded considerable information on the extent of such movements. Bonetto and Pignalberi (1964) tagged 40 000 fish of which 70% were Prochilodus platensis. A second and third series of taggings (Bonetto et al., 1971, 1981) gave more details of migration distances and routes for this and other species. In this series of experiments the maximum distance travelled downstream by P. platensis was 650 km at 3.3 km/day; although the mean migration speed was 7 km/day. In 1962 the maximum distance covered downstream was 500 km at a mean speed of 5.8 km/day. A considerable proportion of the fish did not migrate but stayed near the point of release. Salminus maxillosus tagged in the same series of experiments travelled further, 1000 km in 60 days (16.7 km/day). This and other results are summarized in Table 5.3.

In the Mogi Guassu and Rio Grande sub-system of the Parana, 17 species of characin migrate every year over a total distance of 1228 km between the rhithronic upriverspawning sites and the potamonic feeding sites downstream. Recorded speeds upriver ranged from 5–8 km/day for Prochilodus scrofa, 2.5–10.0 km/day for S. maxillosus and 3 km/day for Leporinus copelandii. After spawning fish moved downstream at between 3–5 km/day.

Bayley (1973) also recorded migrations of P. platensis in the Pilcomayo tributary of the Parana. Here he assumes the fish to travel at least 450 km from the downstream floodplains to the upriver spawning site. Migrations of the same type were described by INDERENA (1973) from the Magdalena river, Colombia, where the “Subienda” migration of Brycon moorei, Pimelodus clarias, Prochilodus reticulatus and Pseudoplatystoma fasciatum, among other species, from the cienagas of the floodplain to the breeding grounds in small Andean tributary streams is about 500 km.

Tagging experiments in the Sao Francisco river conducted by Paiva and Bastos (1982) showed that distances travelled by different species ranged from 3 km (Hoplias malabaricus) to 530 km (Duopalatinus emarginatus). Daily migration rates ranged from 0.03 km/day in Leporinus to 13.9 km/day in Duopalatinus and one specimen of Prochilodus travelled 250 km downstream in 30 days (6.6 km/day).

In Africa dry season migrations are associated mainly with dispersal of fish in the river system. In the Niger river, for instance, Daget (1952) studies Brycinus leuciscus which travelled at 1–1.5 km/hr up to a maximum of about 9 km/day. Total distances traversed were as great as 400 km before the construction of the Markala dam. Also in the Niger fishermen from the middle reaches between Ayourou and Niamey in Niger maintain that many species move upstream at the beginning of the rising flood. Fish are full of eggs at this time and it is suspected that the main destination is the floodplain of the Central Delta some 400–650 km distant. The floodplains in the middle reaches of the Niger are very poorly developed and probably do not provide an adequate area for breeding in those species requiring this type of habitat. Many migratory species including Alestes baremoze and A. dentex are indicated by Blache, Miton and Stauch (1962) and Carmouze et al. (1983) as undertaking potomodromous movements of up to 650 km between Lake Chad and the upstream swamps. Reizer's (1974) observations in the Senegal river also indicate maximum distances of 400 km for the main migratory species which here include Alestes, Brycinus, Citharinus, Distichodus, Labeo, Lates and Clarias. Other recorded African breeding migrations are of shorter, but still impressive lengths. The ascent of species from Lake Victoria up to Nzoia river in Kenya during the flood were classed by Whitehead (1959) as long duration (80 km or more), Barbus altianalis; medium duration (15–25 km), Labeo victorianus and Schilbe mystus and short duration (up to 8 km), Brycinus nurse. Another Labeo, L. altivelis migrated up to 150 km up the Luapula river from Lake Mweru at the beginning of the floods. Williams (1971) tagged several species of fish as they left the Kafue river plains and found that most undertook movements of up to 60 km upstream and 120 km downstream.

Among Asiatic species the Indian major carps appear to move only locally and mainly laterally on to the floodplains within the Ganges and other river systems (Jhingran, 1968, for Catla catla, and Khan and Jhingran, 1975, for Labeo rohita). Other species such as the Mahseers (Tor spp) undertake extensive migrations within the Gangetic system running up the major tributaries to the foothills of the Himalayas to spawn. In the Mekong several species have been recorded as covering large distances. Shiraishi (1970) quoted migration distances of up to 1 000 km between the delta and Vientiane during the wet season. There appears to be little evidence to support this, however, and more reasonable estimates were given by Bardach (1959) for the movement of Pangasius sutchi from the Grand Lac to the Khone falls, a distance of some 3–4 00 km. Pantulu (1970) did mention that Pangasianodon gigas is suspected of spectacularly long migrations, although other species in the Mekong, including P. pangasius, P. sanitwongsei, Cirrhinus auratus, Probarbus jullieni and Thynnichthys thynnoides only undertake medium to long-range migrations. Another species of Thynnichthys, T. vaillanti has been described as migratory up to 450 km upstream of its wet season habitats at low water (Saanin, 1953). Long distance migrations also occur in Ctenopharyngodon idella and Hypophthalmichthys molitrix in the Amur River which cover up to 1 200 km from downstream feeding grounds to upstream breeding sites (Krykhtin and Gorbach, 1981).

The Golden Perch (Macquaria ambigua) of the Murray-Darling river system of Australian are recorded by Butcher, (1967 and Reynolds (1983) as undertaking migration of 800 to 1 000 kms from the lower reaches of these rivers to the headwaters to spawn. The average speed of the fish was 3 km/day and a maximum speed of 15.7 km/day. A fourth species from this system, the Silver Perch, Bidyanus bidyanus was also found to move up to 520 km upstream.

Many anadromous fish travel considerable distances within freshwater for instance Huso huso and Acipenser gueldenstaeti are both recorded to have moved up to 3000 km up the Volga river from the Caspian sea. Such journeys within the rivers may represent only a small part of the total distance travelled as species such as Salmo salar may have a preliminary route of several thousand kilometres in the sea.

Timing of Migration

The timing of the initiation of longitudinal migration varies considerably according to various groups of species.

Movements of fish in the Mekong river, recorded from the commercial fishery at Khone falls (Chanthepha, 1972) showed two separate groups. The first, consisting of cyprinids, passed upstream in November to February and probably represented dispersal migrations after leaving the floodplain. The second group, mainly of siluroids, passed up through the falls from mid April to July and were possibly pre-spawning migrations. Alternations between dispersal and spawning migrations in the same species have also been recorded in Latin America. For instance, Prochilodus mariae in the Orinoco shows two phases of movement, one reproductive at low water (February-May) and one for dispersal from September to December just after peak flood. In this the species is accompanied by Brycon sp. Curimatus sp. and some predatory siluroids such as Pseudopimelodus and Sorubimichthys. Other species adapt different timings in the same river. Thus Colossoma brachypomum migrates at low water, while Semaprochilous laticeps and S. kneri emerge from the lagoons and migrate up the main channel to breed early in the rising flood May to July. Similar plurimodality of migration is demonstrated by the Prochilodus and Semaprochilodus of the Amazon. In most other systems in Latin America the Prochilodontidae and certain siluroids complete their major upstream movements at low waters. Thus in the Parana, Mogi Guassu and Magdalena rivers the fish time their movements so as to arrive at the upstream spawning sites as the flood begins to rise. Goulding (1981) also describes the migration of many species of catfish up the Teotonio cataracts of the Madeira river at low water, and here only one species, Goslinia platynema, was observed to migrate at high water.

The pattern also appears in the Lake Chad basin where Alestes baremoze move up the Chari and Logone rivers at low water, arriving at the entrance to the Yaeres floodplains at the beginning of the flood (Stauch, pers.comm.), in the Senegal river where the longitudinal migrations of Alestes during the dry season bring them to the channels opening on to the floodplain as the floods rise and in the movements of Brycinus leuciscus and B. nurse within the Central Delta of the Niger, although in these species the migration is for dispersal rather than spawning (Daget, 1952).

Other families of migratory fish, particularly the mormyrids in Africa, some siluroids and most cyprinids in Africa and Asia and the Golden perch in Australia normally initiate their riverine spawning migrations as the floods appear. In some cases, for example, Labeo victorianus and other small cyprinids in Lake Victoria, riverine migration may be preceded by a preparatory movement to the river mouth in the lake itself (Cadwalladr, 1965). However, the arrival of the first freshets of the floods in the river seems to trigger longitudinal migratory behaviour in the majority of species. This migratory urge appears to be transfomed into a general impulse for lateral movements as the bankfull stage is reached and floods spill on to the floodplain. The observations of several workers that many fish migrate actively against the current up channels where water is flowing out of the plain, rather than to enter passively on inflowing currents, points to the physiological orientation of such movements.

Migration on to the floodplain seems to develop as an ordered sequence of species. Sao-Leang and Don Saveun (1955) described the movements of fish in the Mekong as a series of waves with siluroids entering early in the sequence and whitefish later. In the Niger (FAO/UN, 1970), swamp tolerant species which often stay in the larger lagoons over the dry season, tended to enter first from the river. Earlier entries included species of Clarias, Distichodus, Citharinus and Labeo, whilst Brycinus, Tilapia, mormyrids, Schilbe and Synodontis entered second. In the Kafue flats, Williams (1971) has observed a different sequence that, to a certain extent, contradicts that observed above. Hare, Clarias, Schilbe, Barbus and Tilapia migrated on to the plain while the water was still quite low and Serranochromis and Haplochromis only moved on to the plain later. The early movement of Tilapia and the clariids is confirmed by the University of Idaho et al. (1971) for the same area. Because of the lack of information, which is difficult to collect, the situation with regard to movement during rising water is far from clear, although the concept of phased migration of species on to the plain is probably sufficiently well-established.

During normal flooding there appears to be adequate space on the floodplains to accomodate both the resident black fishes and the immigratory individuals of the more mobile species. Benech and Quensiere (1983), however, concluded that, during periods of drought, the sedentary stocks can restrain the annual colonization of the floodplain by riverine fishes. In the Yaeres, the disproportions between resident and immigrant species increased in those years when floods were less intense and was minimal during the period when virtually no wetting of the floodplain occurred. The effect may be due in part to the prior occupation of such living space as is available on the plain, and in part to the reduction in possible ingress channels with lessened flooding.

Movements off the plain are more easily studied and these too show that fish leave in a distinct sequence. This appears to be roughly the reverse of the sequence with which fish enter the plain. In the Mekong, studies by Blache and Goosens (1954) and by Sao-Leang and Dom Saveun (1955) from trap catches of over 100 species showed that movement at the level of Quatre Bras occurs between October and February. Here the sequence of migration was apparently conditioned primarily by the size of the fish, the larger fish and larger species leaving first. Similar observations on size sequence in migration have been made by Saanin (1953) in the Mahakan system where the larger individuals of Thynnichthys vaillantii left the floodplain lakes and migrated upriver before the smaller fish, and in the Mogi Guassu where smaller species always precede the larger ones in the upstream movement.

Lunar phase also influenced the timing of the migration of whitefish in the Mekong which takes place only during the second quarter to full moon each month. Thus there were monthly waves of migration in which a few of the largest fish moved in October and November. In December to January mixed groups of large fish including Cirrhinus auratus, Osteocheilus hasselti, Pangasius sutchi, Pangasius larnaudi, Belodontichthys and Cyclocheilichthys were caught. In February smaller species such as Thynnichthys thynnoides, Cirrhinus jullieni and Botia modesta dominated in the catch.

Studies by Durand (1970 and 1971) on the succession of species passing through the El Beid river which drains the Yaeres, also pointed to there being a definite sequence of species which correspond to different water masses. The first group, which moved at high water and had a very clear peak in November and December, consisted of Marcusenius cyprinoides, Hyperopisus bebe, Alestes dentex and Labeo senegalensis, preceded by a group of accompanying species, Alestes baremoze, Polypterus bichir, Hydrocynus brevis and Lates niloticus and followed by Distichodus rostratus, Oreochromis aureus, Pollimyrus isidori and Distichodus brevipinnis. A second well-defined group consisting of Oreochromis galilaeus, Brienomyrus niger, Barbus and Clarias species was very abundant in January and Ichthyoborus besse, Synodontis auritus, S. schall and Schilbe mystus appeared most strongly in February. The first group corresponded to fish migrating at the end of the Logone floods, the second and third groups move with the water draining off the flooded plain. More detailed descriptions of this phenomenon by Benech and Quensiere (1982,1983 and 1984) confirmed the various species groups and their order of migration and linked these to flood conditions on the plain. A strong diurnal pattern was also demonstrated with diurnal, nocturnal and crepuscular migrants. Lunar phase influenced the nocturnal migrants but this was considered of secondary importance to the hydrological phase. Certain species may change their lunacity depending on the flood period. Thus Hyperopisus bebe, which is purely nocturnal on the floodplain during the rising flood, is active both day and night at the beginning of the flood and as the water is draining from the plain, i.e., at times of migration.

This example of the migrations from the Yaeres serves to illustrate the probable complexity of migration patterns elsewhere and in some ways similar to the phased appearance of young fish on shoreline nursery habitats in a North American stream studied by Floyd et al. (1984). Here an ordered succession of species, from cottids through percids, catostoids, centrarchids to ictalurids, occupied the habitats. The phasing of arrival on the shoreline was linked to the time of breeding of the species in a somewhat extended season. Nevertheless, sufficient numbers of species were present in the habitat at the same time as to call for their partitioning of the resource. Residence time on the shoreline ranged from three to sixteen weeks depending on species and represented a way station between the spawning habitat on the shoreline or the riffle systems and the adult habitat.

That adult fish leave the floodplain before the young-of-the-year is borne out by many authors. For example, Motwani (FAO/UN, 1970) remarked that older age groups leave the swamps of the Niger and Benue river first. The juveniles remain on the floodplain until the later stages of its emptying. The University of Idaho et al. (1971) and Williams (1971) noted the same phenomenon in the Kafue river. In the Yaeres, the breeding fish often do not return through the El Beid river, preferring to re-enter the lake via the main channels of the Chari and Logone rivers. Thus Durand (1970) was able to record that young fish make up 95 percent of the El Beid catch by number and weight. This phenomenon is mainly caused by the peculiar flow patterns of this system (Fig. 5.11) which only allow the adult fish to fight against the currents entering the plain through the levees and thus to regain the main channel of the river. The young fish are not strong enough to do this and are more strongly influenced by the current which directs them down the El Beid as it drains the plain. More extreme examples of delayed migration are shown by the juveniles of Prochilodus platensis, which remain in the floodplain pools of the Parana river during an entire dry season before entering the river after the next flood (Bonetto, 1975). Prochilodus species in the Apure river show similar behaviour (Matthes, pers.comm.). That return migrations by juveniles are more complex than just a passive movement under the influence of the movement of water as it leaves the plain is indicated by the different behaviour of two very similar species of small cyprinid in a stream system flowing into Lake Victoria (Welcomme, 1969). Here the adults of both species left the upstream swamps towards the end of the flood. The juveniles of one species, B. kerstenii remained only a short time in the swamps, moving quickly to the drainage channels and the river and migrating shortly after to the lake. B. apleurogramma on the other hand tended to stay in the swamps until they were almost dry, moving to the drainage channels and later to the river where it remains throughout the dry season. These movements occurred at quite characteristic sizes. By contrast a third sequence of movement is shown by a small cichlid which inhabited the same system and did not enter the lake.

External factors which might stimulate spawning migrations have been discussed for many years, somewhat inconclusively. In view of the great differences in timing both of the initiation of movement and of spawning among the many species inhabiting tropical river systems, it is likely that there are a whole range of fine tuning stimuli which influence the various species in different ways. Because the end of the flood tends to be less predictable than the beginning, and the behaviour of the fish at this time is not apparently subject to a strong physiological impulse such as reproduction, it seems reasonable to suppose that simple mechanical and chemical factors would be sufficient to initiate movement out of the floodplain habitats. It is important to the survival of the species that such signals should anticipate the onset of conditions that would be lethal to the majority of the population. Nevertheless, it does seem that stimuli for return migrations are often less than totally effective in this respect, as huge quantities of fish are stranded and die every year.

The generally noted tendency for both the larger species and the larger individuals of each species to leave the floodplain earlier than the smaller fish is probably indicative that depth is one of the major factors controlling this. The fact that the large fish often fail to move out of deeper floodplain pools would also tend to support this supposition. Dissolved oxygen concentration and temperature also seem to be of major importance in determining the distribution of fish within the system, and changes in these factors are liable to provoke fish into leaving regions where conditions are less than optimal.

Figure 5.11   Migration patterns within the Yaeres-Lake Chad system. (Adapted from Durand, 1970)

Light plays a very strong role in regulating the time of migration. Several families move mainly at night. In Asia these include siluroids and ophicephalids, in Africa siluroids, ophicephalids and mormyrids, and in South America siluroids and gymnotiids. Other fishes including cichlids, cyprinids and some characins, tend to move by day, yet others, such as the small mormyrids which move up the Nzoia River fron Lake Victoria to spawn concentrate their movements at dawn and dusk (Okedi, 1969). There is, in this manner a possibility for round-the-clock utilization of both the migratory pathways and the slack water refuges and resting places. That diurnal sequences of migration may be quite complex is evidenced by Lowe-McConnell (1964) observations on the Rupununi system. Here, the periods for peak migration were more closely defined as follows: dawn and early morning - Cichla ocellaris, Osteoglossum; daytime - Serrasalmus nattereri, Metynnis sp., Geophagus jurupari, Cichla ocellaris, Cichlosoma severum, Leporinus friderici; late afternoon - Brycon falcatus, Metynnis sp.; evening (just after dark) - Prochilodus insignis, Schizodon fasciatum; night - Pimelodus and other catfishes.

The strong influence of lunar phase on the timing of the migration of whitefish in the Mekong has already been mentioned. A similar phenomenon has been described for Brycinus leuciscus in the Central Delta of the Niger by Daget (1952). This characin species forms extensive shoals during moonlit nights for their upstream dispersal migrations. When there is no moon the shoals dissociate. Migration is also initiated as soon after the draining of the floodplain as is consistent with lunar phase, and because of the topography and drawdown regime of the central delta, four different migratory groups are formed according to the successive coincidence of full moon and low water as the flood recedes downriver. Other characin species, Brycinus nurse, Alestes dentex and A. baremoze also migrate upstream but are not ordered by lunar phase. Upstream movements of B. nurse in fact always precede movements of B. leuciscus from the same portion of floodplain.

Potamodramous migrations would appear to have several advantages for the fish species undertaking them. Many species may migrate primarily to avoid unfavourable conditions in the lower reaches of the river, but the majority of migrations seem directed at reaching localities suitable for reproduction or feeding. By placing the young fish nearer the headwaters, or where appropriate in the main stem of the river, their journey downstream can coincide with the flood wave over several hundred kilometres. Juvenile fish can thereby arrive at floodplain reaches suitable for feeding grounds, which may differ from the rocky, sandy and turbulent areas chosen for breeding, and thus ensure maximum exposure to zooplankton in the floodplain pools. Furthermore, in forested nutrient poor rivers, such as the Amazonian tributaries, the fringing flood forests are so lacking in nutrients that the young need to be transported downstream to find suitable feeding and nursery habitats. This opinion is also advanced by Krykhtin and Gorbach (1981) for Ctenopharynogodon idella and Hypophthalmichthys molitrix from the Amur river. The eggs of these species originate from spawnings on upstream grounds more than 1200 km from the mouth of the river, which distance rules out the possibility of eggs or larvae being swept out to sea. Spawning at the earliest phase of the rising flood also ensures that the eggs and larvae drift downstream and that the juvenile fish are thus positioned opposite suitable floodplain foraging areas at the time of changeover from zooplankton to mixed feeding some four and a half days after hatching. At the current velocity of 4.7 km/hr purtaining in the Amur even this short time lapse may mean a drift over 500 km between hatching and seeking refuge in the side arms of the floodplain.

A similar strategy, that of placing the young on the floodplain as early as possible also is followed by these upstream migrants who time their arrival at the channels opening on to the floodplain to coincide with the first flow of water on to the plain. In addition to the benefits for feeding noted above, upstream migration by adult fish prior to spawning must have a role in counteracting downstream drift of larvae and fry in those species that are obligate main channel spawners. Although not analysed in detail by any studies so far it wouldappear that migration distances upstream as described for the Pilcomayo (Bayley, 1973), the Parana (Bonetto, et al. 1971, 1981) or the Amur (Krykhtin and Gorbach (1981) are of the same order as the distance drifted downstream by larvae and fry given the current velocities of the rivers and development times of the fishes concerned. Obviously during drift type movement there must be considerable dispersal of the population which may ensure mixing of stocks over such a long river as the Parana. In other systems, such as the Orinoco, it is, however, clear that separate sub-populations are maintained in different reaches of the river.

There has been much discussion of the role of this potomodromous habit in African species, Jackson (1961) maintaining that it is principally a device to protect the young from predation, whereas Fryer (1965) considered it mainly as a mechanism to secure dispersal over the whole river course. As in so many such arguments both participants are probably partly correct and there is no doubt that the use of upstream swamps and spawning habitat does have the double advantage of presenting the young fish with a rich habitat in which to start life, while at the same time giving considerable shelter from the predation to which they would be exposed in the adult lacustrine habitat.

ADAPTATIONS TO EXTREME ENVIRONMENT CONDITIONS

Many of the habitats within river ecosystems have extreme physical or chemical conditions which call for special adaptations on the part of the fish inhabiting them. Many of the adaptations are behavioural, involving migrations or local movements whereby the adverse conditions are avoided. A certain section of the fish fauna, however, has specific anatomical or physiological adaptations which permit the species concerned to survive low dissolved oxygen concentrations or even complete deoxygenation, high temperature, desiccation, poor light conditions or strong current.

LOW DISSOLVED OXYGEN CONCENTRATIONS

One of the main factors determining the distribution of fish within the river and floodplain system is the availability of dissolved oxygen. As has been shown above low dissolved oxygen concentrations or completely anoxic conditions are common on floodplains at certain times of the year. Fish have become adapted to these conditions in a variety of ways but some species which inhabit the lentic waters of the floodplain are strangely sensitive to low dissolved oxygen levels. Serrasalmus nattereri and S. rhombeus for instance show the first symptoms of asphyxia when the oxygen falls below only 20 percent saturation. Such species are especially sensitive to sudden drops in dissolved oxygen and are among the first to suffer catastrophic mortalities. There are two sources from which fish may obtain supplementary oxygen in poorly aerated waters. These are:

(a)   the air above the water, and
(b)   the thin, but well oxygenated surface layer which is often only a few millimetres deep.

Many of the blackfish species inhabiting swamps have modifications which allow them to benefit from one or another of these sources. For example, Carter and Beadle (1931) found that eight out of twenty species occurring in the Paraguayan Chaco had anatomical respiratory modifications for air breathing, whereas the rest used the surface layer as a source of oxygenated water.

Adaptations for Air Breathing

The development of specialized organs which enable fish to breath air has occurred independently in many taxa and in all zoogeographic regions. Respiratory modifications have been centred around three main anatomical systems, the mouth and digestive tract, the gills and branchial chamber, and the lung or swim bladder. These are discussed in detail by Carter (in Brown, 1957), Norman (1975) and various contributions in Hughes (1976) particularly that of Dehadrai and Tripathi.

Several species have become so dependent on air that they die if prevented from reaching the surface. Thus the lung fishes are obliged to breathe at frequent intervals, and the paiche (Arapaima gigas) needs air every 10–15 minutes when adult and more frequently when young (Sanchez Romero, 1961).

Modification of the digestive tract for air breathing is made possible by the stopping of feeding during the dry season (Lowe-McConnell, 1967), at just those times when deoxygenation is most extreme. Most parts of the alimentary canal have been modified in one family or another. The mouth cavity and pharynx are highly papillated and well supplied with blood in Electrophorus electricus which surfaces to gulp air. Air bubbles are passed backwards to lodge inside the heavily vascularized stomach of Ancistrus, Corydoras, Plecostomus spp. and Hypostomus, the intestine of Hoplosternum or Callichthys, or the rectum of Misgurnus fossilis.

The branchial or pharyngeal cavity has become modified by simple vascularization in Hypopomus and in particular in the synbranchid eels such as Monopterus and Symbranchus marmoratus. Mastacembelus spp. secrete a protective slime over the unmodified gills which permits a limited amount of aerial respiration. Three different families have developed diverticula of the branchial cavity (Fig. 5.12). These are least developed in the Channidae whose supra-branchial chambers are simply lined with a richly vascularized epithelium. The Anabantidae have labyrinth organs elaborated from the first gill arch. In the clariidae the II and IV gill arches have become modified into arborescent organs in many genera, and in Heteropneustes (Saccobranchus) fossilis the branchial chamber is extended backwards along the body.

Figure 5.12   Adaptations for air-breathing in (A) Channidae, Channa; (B) Anabantidae, Ctenopoma; (C) Clariidae, Clarias; (D) Heteropneustidae, Heteropneustes

Only the lung fishes (Dipnoi) and bichirs (Polypteridae) have true lungs, but several physostomous families have modified swim bladders which act in an almost identical manner. These included Osteoglossidae (Arapaima gigas), lepisosteidae (Lepisosteus sp.); Gymnarchus and several species of Mormyridae, Erythrinidae (Erythrinus and Hoploerythrinus), Notopteridae, and Umbridae (Umbra). Young forms of Lepidosirenidae, Polypteridae, Osteoglossidae and Gymnarchidae have external gills which are resorbed during development when the lungs or swim bladder take over the main respiratory function.

The ability to breathe atmospheric oxygen enables fish to colonize waters which would be otherwise uninhabitable thereby reducing interspecific competition. Furthermore as comparatively few ichthyophage predators have developed these mechanisms fish colonizing such deoxygenated waters are relatively free of predation (Junk, Soares and Carvalho, 1983). It has the additional bonus that it permits overland movement. Migration of fish over dry land, or at least over damp, swampy grounds have been recorded in several species, particularly various clariids in Africa and Asia, but also in the erythrinids and perhaps the callichthyids and loricariids of South America (Kramer et al., 1978). Such mobility permits fish to escape desiccating water bodies and also to colonize pools isolated from the main water mass.

Dehadrai and Tripathi (in Hughes, 1976) mention the energy cost of air breathing. Young Ophicephalus punctatus kept in 40 cm of water, surfaced 1 879 times per day at a cost of 161 cal/day. In 2.5 cm of water the same species surfaces 482 times per day at a cost of 92 cal/day. Fish living in deeper water consume about 1.5 times as much food as those from the shallower waters presumably to compensate for the additional energy required. Another disadvantage of the air breathing habit is the increased vulnerability to predation of the fishes by both aquatic and airborne predators during their journey to and from the surface and at the surface itself (Kramer et al., 1982). Kramer and Graham (1976) suggested that, in many species from many families this is to a certain extent mitigated by the social behaviour of synchronous breathing by a number of individuals which space their breath randomly. This type of behaviour has been observed in the laboratory in Ancistrus chagresi (Loricariidae), Hoplosternum thoracicum (Callichthyidae), piabucina festae (Lebiasinidae) and Trichogaster leeri (Anabantidae), and in the field in Piabucina and Hoplosternum. Other species from other families (Polypteridae, Lepisosteidae, Notopteridae and Megalopidae appear to behave in a similar manner from descriptions in the literative, indicating that the adaptation may be widespread.

Adaptations for Using the Surface Layer

In species lacking specific physical or physiological adaptations the practice of utilizing the better oxygenated surface layers of the water is common. For instance Kramer (1981) recorded 93% of non-airbreathing fishes in Panama as breathing at the surface layers and later Kramer and McClure (1982) extended these observations to the Amazon where 31 species belonging to 14 families made increasing use of the surface layer as oxygen concentrations dropped. Thus aquatic surface respiration was considered to be a specific adaptation to hypixia in fishes inhabiting shallow tropical waters.

Anatomical modifications which enable certain small species of fish to use the oxygenated surface film of the water particularly efficiently have been described by Lewis (1970). These adaptations include the small dorsally-oriented mouth and dorso-ventrally flattened head found in most cyprinodonts. Lewis carried out experiments to show that in deoxygenated conditions Fundulus, Poecilia and Gambusia all adopted a characteristic posture at the water surface where they can survive indefinitely, provided a critical population density is not exceeded. Kramer and McClure indicate similar morphology and behaviour in species of cichlids (Pterophyllum) and osteoglossid (Osteoglossum). Some larger fishes too have adopted this strategy and in addition to their posture at the surface Colossoma macropomum and Brycon melanopterus develop dermal enlargements of the lower jaw during hypoxic conditions which facilitate the influx of surface water into the buccal cavity. These enlargements are lost in oxygen rich water (Braum and Junk, 1982).

Physiological Adaptations

Physiological mechanisms in some species allow the fish to withstand low dissolved oxygen concentrations, although not necessarily to survive under complete anoxia. For instance Blazka (1958) found that Carassius carassius can tolerate anoxia for at least two months at low temperature. At higher temperatures tolerance time is lessened. Much of this is due to the composition of the blood which in some species, noteably those inhabiting oxygen-poor waters, is relatively unaffected by changes in 0₂ uptake produced by increases in CO₂ tension (Lagler et al., 1977). There is in fact a great diversity of response by fish to respiratory stress as described by Fry (in Brown, 1957). Work by Powers et al., (1979) on Amazonian fishes has particularly drawn attention to differences in the affinities of the blood of different species of fish for oxygen. Their work was summarized by a series of curves (see Fig. 5.13) which show fish from running waters where there is much dissolved oxygen to have blood of lower affinity than fish from still waters with low dissolved oxygen tensions. Fish from the margins of rapids occupy an intermediate position. By extension fish occupying rapids which dry up during the dry season may change the O₂ affinity of their blood to correspond to prevailing conditions.

ADAPTATIONS TO RESIST HIGH TEMPERATURE

Temperatures at the fringes of the flooded plain, and in shallow water bodies, may rise as high as 40°C. Certain species of fish, particularly the juveniles, tend to prefer such areas, using them as nurseries. Here they benefit from the higher temperature and greater availability of food to grow faster, and also as a partial refuge from predation (Welcomme, 1964). There is experimental evidence that species inhabiting these warmer water areas have a much greater physiological resitance to the effects of high temperatures than most other species. For example, Fig. 5.14 illustrates the difference between the survival of Tilapia zillii and Haplochromis spp. under experimental high temperatures. T. zillii survive indefinitely at 38°C, a temperature at which half of the sample of Haplochromis would die within two minutes.

There are indications that such thermal tolerance may be cyclic, at least in some species. Johnson (1976) has shown that there is a daily rhythm of thermal tolerance in Gambusia affinis which rises some 3 degrees between morning and mid-day and falls away again towards evening.

Figure 5.13   A comparison of O₂-Hb affinity for water breathing fish from lotic habitats in the Amazon system. Thick solid lines enclose the range of blood equilibrium curves for species from slow flowing zones; Thin solid lines those for species from rapidly flowing water; Dashed lines those from rapid waters but which inhabit streams or river margines. (From Powers et al., 1979)

Figure 5.14   Survival of Haplochromis species and Tilapia zillii in high temperatures when acclimatized to 27°C

ADAPTATIONS TO RESIST DESICCATION

Very few species are adapted to survive desiccation and annual losses of fish trapped in temporary water bodies is enormous. Some species however survive the dry period by cocooning. The African lung fishes Protopterus annectensand P. aethiopicus burrow into the bed of a drying pool and secrete a cocoon of hard slime in which they rest, coiled, so the mouth is upwards and connected to an air passage. Fish have been recorded as having survived over a year of aestivation. Some murrels (Ophicephalidae) are reputed to survive short periods of drought in a similar manner as do the synbranchid eels of the Amazon (Kramer et al., 1978). Dehadrai and Tripathi (in Hughes, 1976) also report that Clarias and Heteropneustes take refuge in soft mud in drying pools, and Donnelly (1978) summarizes reports from Africa of Oreochromis mossambicus and Clarias gariepinus survival in wet sand. Bruton (1979), in his review of the literature on survival of habitat desiccation by air breathing clariids, supports the idea that Clarias species can survive for some time in burrows in damp mud or wet sand. This ability, however, does not seem to extend to survival under totally dry substrates as in the case of Protopterus to which the construction of a coccoon is unique.

Several species of cyprinodont in Africa and South America can maintain permanent populations in temporary aquatic habitats. These are annual fishes which complete their life cycles in as little time as a few weeks and which have several adaptations including drought resistant eggs, embryonic diapause, rapid hatching responses to rainfall all of which adapt them to withstand prolonged desiccation (Simpson, 1979).

The lowering of the water level in their environment, and possibly the associated physical and chemical changes in water quality, appear to act as the stimulus for reproduction. The eggs are shed on the bottom where they settle or are pushed into the mud by the parent fish which die shortly afterwards. The eggs and growing embryos may stop their development for variable periods and such arrests (diapauses) may occur at three stages, described by Wourms (1972) as Diapause I (dispersed cell phase), Diapause II (long somite embryo) and Diapause III (pre-hatching). In some species the arrest is facultative at Diapause I and II, but obligate at Diapause III. The different combinations of Diapause can generate 8 different distributions of total development time as shown in Fig. 5.15. In this way a single egg population can give rise to several sub-populations which allow for the repeated loss of individual eggs under conditions which can start development but do not allow maturation and successful reproduction. This “multiplier” effect guarantees that some portion of the egg stock will survive to reproduce.

Figure 5.15   “Multiplier” effect of various combinations of diapause on egg stock of annual fish. (After Wourms, 1972)

ADAPTATIONS TO POOR LIGHT

Levine et al. (1980) and Levine and MacNichol (1982) have drawn attention to the necessity for fishes to have eyes which are fully adapted to the environments in which they live and to their own particular position in that environment. The quality and quantity of light within river systems varies considerably according to habitat. Thus in some mountain streams or clear water rivers the limpidity of the water permits great light penetration and only a gradual modification of the spectrum. However, more commonly, the highly turbid conditions of rivers and their floodplains leads to distinctly murky conditions. Equally, the poor penetration and high spectral distortion of blackwaters requires specialized visual adaptation. Differences in visual structures have been found among various species from the Amazon which are directed towards life in the diverse habitats (Menezes et al., 1981). Equally, specializations are found which correspond to the fishes accustomed position in the watercolumn or to its preferred period of activity (Levine and MacNichol, 1982) which may be termed “visual riches”. Some behavioural quirks may be explained in this manner although at present there is insufficient evidence for or against such hypotheses. For instance, the migration of prochilodontid characins from the turbid floodplains and white water rivers to the mouths of blackwater rivers for breeding might be explained by differential visual requirements.

Many species either supplement, or completely supplant, vision with other organs, particularly in those fishes which live at or near the bottom of the river. Thus barbels are found in most cyprinids and siluroids, some of which are eyeless (e.g., the cetopsid catfishes). An alternative system is the elaboration of electric organs to send impulses which presumably serve not only as communication but also to define the environment. The gymnotids of South America and the mormyrids of Africa are noted examples of this adaptation which in the mormyrids is coupled with greatly reduced eyes.

ADAPTATIONS TO RESIST STRONG CURRENT

Species of fish inhabiting the rocky riffles of the rhithron or the rapids reaches of larger rivers are highly adapted to the turbulent conditions. There are three main groups of such fishes; (i) Those which cling to the surface of vegetation and rocks, (ii) Those which take refuge from the current in the crevices and holes between rocks and (iii) those which can swim sufficently fast as to resist the current.

The first group is particularly well represented by members of the family Amphiliidae in Africa whose several genera are all adapted in various ways to life in strong currents. They are all elongated, streamlined and with the slightly humped form that results in the fish being forced on to the bottom by the flow. In addition the various genera possess a variety of suckers or enlarged fins with which they cling to the substrate (Fig. 5.16a). For example, Amphilius spp. have sucker like mouths and stiff pectoral spines. Doumea spp. have enlarged rigid pectoral spines and Phractura clauseni has been observed to cling to the edges of leaves with suffered maxillary borkals. Similar structures are found in the Mochokidae where Chiloglanis spp. have elaborate pectoral spines and in the Bagridae with some Auchenoglanis spp. Sucker like mouths are also found among the Cyprinids in both Garra and Labeo species. Numerous other families and genera have species with similar modifications throughout the world. Among which the Gastromyzontidae, Cyprinidae, or Sisoriidae in Asia and the Astroblepidae in South America.

Two particular adaptations fit fish for life in the interstices of rocks and anchored vegetation, sinuous, serpentine form and small size (Fig. 5.16b). Both are found in rapids fauna where a number of genera from various families, for instance the clariidae, with Gymnallabes and Clariallabes, the mormyridae with Mormyrops, the mastacembelidae with Mastacembelus and the cichlids with Gobiocichla in Africa. The cyprinids and the Gobiidae with a wide variety of species in Asia, Europe and N. America, the Trichomycteridae in Latin America and the Galaxiidae in Australia have all developed elongated or pygmy species.

Figure 5.16   Adaptions to swift current in some African rheophilic species.
(A) illustrates mouth suckers, stiffened barbels and stout pectoral spines;
(B) illustrates slim or serpentine form