The General Integrative Watershed Model


The General Integrative Watershed Model (GIW):

A GIS based, Landscape Ecology Approach to Large River and Watershed Environmental Management


Roman G. Kyshakevych


Abstract: Rivers of many scales have acted as conduits of transportation and communication, sources of agricultural irrigation, hydropower, potable water, and sinks of anthropogenic waste. As such, humans have always tried to manage rivers, and in the process, have often erred to extent of mismanagement. This has often led to compromising the ecological integrity of rivers, especially of large rivers. Most lotic models are "channel-centric" and there has been no generally accepted viable models that could serve as a framework for management of an entire watershed system.

The newly constructed General Integrative Watershed model (GIW), herein, introduced and implemented, however, is 1) holistic, 2) applicable to both anthropogenically impacted and pristine watersheds, 3) cosmopolitan because it is applicable in any lithologic, topographic, biotic, and climatic regime, 4) a mechanistic, discrete patch system permitting cross-watershed comparisons, 5) modular, treating tributaries as units, and 6) the model is simple and easily implemented.The GIW was implemented using the French Creek watershed in northwestern Pennsylvania. The GIW predicted differential influences of various geologic bedrock, glacial deposits, and land use attributes on French Creek water quality. The model was also used to discern relationships between mussel associations and terrestrial watershed attributes.




    Human cultures and societies are intimately associated with rivers. This is as real today as it has been historically, when whole civilizations developed around the watersheds of the Tigris and Euphrates, the Nile, the Indus, the Ganges, the Dnieper, the Amazon, the Magdalena, the Mississippi, the Ohio, etc. Rivers and streams of many scales have acted as conduits of transportation and communication; sources of agricultural irrigations, hydropower, sources of potable water, and sinks of anthropogenic wastes. As such, humans have always tried to manage rivers, and in the process have often erred to the extent of mismanagement. A river system represents an invaluable resource, a sustainable “natural capital” that is essential for long-term human habitation.

Mismanagement or lack of management has, invariably, compromised the ecological integrity of rivers. Rivers have been viewed as "interdependent combinations of aquatic and terrestrial landscapes" (Johnson et al., 1995: 134). Because rivers are intimately tied to watershed-wide physical, chemical, and biological processes, it is widely recognized that any attempt to partially restore this "ecological integrity" involves complex watershed-wide management techniques (Gore and Shields, 1995).

However, there is a lack of viable hypotheses or models that could serve as frameworks to manage, and help restore and maintain lotic systems diversity and functions: "...the current hypotheses for lotic systems are essentially descriptive and do not provide mechanisms that can be used to direct management efforts...Better methods and tools are needed not only to guide management, but also to predict a river's physical and biological characteristics along its length" (Johnson et al., 1995: 138).

The newly constructed General Integrative Watershed model (GIW) model is designed to fulfill the need for a viable, mechanistic model that will act as a framework for the study and prediction of physical and biological characteristics and their relationships, and guide management and restoration of rivers from a holistic, watershed-wide perspective.

The model is designed to help in the management of a watershed in a cost effective way. Cost being always a primary concern, the GIW channels scarce resources to mitigate and manage watershed in a more direct “big bang for the buck” manner.

From the GIW perspective, rivers are thought of as sustainable physical entities: part of a hydrological cycle running on solar power. It is generally well established that there exists a integral relationship between water and life, therefore, an underlying tenet of the GIW is that on the continents there exists an equally integral direct correspondence between the ecological integrity of a river system and the ecological integrity of the biotic component (which includes the anthropogenic part) that lives on the watershed of said river system. Therefore, in the short, medium, and long term, it is essential to use the newest technology available to manage organically, and holistically the mitigation, maintenance and improvement of a river to assure quality of life.

Ultimately, the management of a large river system is an open-ended commitment. Key to the model is the information database on which it relies. Data would include chemical, geological, biological, anthropogenic land use information about the main stream, its tributaries, and the entire watershed system. Theoretically, the resolution of the data about a complex system such as a watershed is infinite. Nonetheless, the GIW model is based on the principle that the first step in the management and in the mitigation of a river system is to gather information and establish a viable information system: a database.

Furthermore, the GIW model is based on well founded and accepted fluvial geomorphic principles that have been researched and studied for more then a century. The mitigation of a river system is a complex process. It involves not only the application of standardized and proven practices to control effluent quality but also political and social aspects that affect land use practices in culturally impacted watersheds.    



                                                  FLUVIAL GEOMORPHOLOGY

Fluvial systems are complex and incorporate many interactive features. Any successful integrated model of such a system must take into account and reflect salient features impacting the system. The GIW model is theoretically based on the principles of fluvial geomorphology.

Fluvial systems serve many functions. Geomorphically, they are important agents of physical erosion and are responsible for major alteration of landscapes. Physical weathering and erosion are largely caused by moving water.  This surface water driven erosion is the greatest factor responsible for landscape geomorphology as expressed in the classic model by Davis (1899). Davis' model describes geomorphic evolution throughout time, as rugged tectonically induced landscapes are reduced to relatively flat erosional surfaces called peneplains or pedeplains (Schumm, 1977). Ecologically, fluvial systems provide habitats for continental biota.

Rivers are part of the hydrologic cycle. The hydrologic cycle is driven by the potential energies of solar heat and gravity converted into kinetic energy manifested by precipitation, and the flow of surface and subsurface waters. There are two interrelated components of fluvial systems: the terrestrial watershed drainage basin area which provides water, sediment, and organic material, and the stream channel with its own physical properties (Schlosser, 1991).

 The geographical area of production, from which water drains, is termed a watershed, drainage basin, or catchment area.

Fluvial systems can be more fully understood in the context of the watershed concept. Watersheds are characterized by fluvial patterns dictated by many variables such as tectonism, underlying geology, and climate. Attribute features of watersheds include basin topography, geology, and vegetational characteristics. Characteristics associated with channels include sinuosities, valley types, floodplains, levee banks, types of bars, substrate, pools, riffles, water depth and velocity. Fluvial systems (watersheds and channels), are categorized by stream order, a somewhat arbitrary method of classifying streams, and by watershed size, channel measurement, and discharge (Strahler, 1964).

Fluvial waters chemically reflect the composition of runoff surfaces and sub-surfaces through which groundwater flows. Water is an important component of chemical weathering. During precipitation, depending on duration, intensity, soil type, rock type, vegetation, slope and other factors, some water may infiltrate the substrate surface and undergo chemical reactions.  Most chemical reactions at the earth's surface include water. Such reactions include: oxidation reactions where iron bearing minerals react with the oxygen in water; hydrolysis reactions involving decomposition in the presence of water; hydration, the addition of a water molecule to the molecular structure of a mineral; and carbonation, the reaction of minerals with carbon dioxide dissolved in water (Bloom, 1978). The fluvial characteristics mentioned above and the aquatic habitats therein can be best understood only as products of the watershed area as a whole.

            There are two main sources of water in a stream: ground water flow and surficial flow (Horton, 1945). Precipitation either flows overland into streams, or it infiltrates the soils and rocks of the basin, and accumulates as ground water, supplying streams with base flow. The amount of water which enters the subsurface depends on infiltration rates (Horton, 1945). If the amount of precipitation exceeds the infiltration rate, water flows over the surface in sheets and rills, and ultimately accumulates in first-order streams.

Rates of infiltration and overland flow are all dependent on the surface and subsurface compositions of basins, which may include, for example, impermeable clay-rich soils, pastures, forested areas, and exposed rocks. However, the amount of water yielded by a watershed basin is influenced by several variables in addition to landscape "texture", or composition. Drainage density and ratio, area, and drainage relief (steepness of slopes) are important watershed characteristics (Strahler, 1957). Schumm (1967) considers several other variables to be important, such as time, geology, climate, vegetation, denudation, relief, hydrology (runoff and sediment yield per unit area), and drainage network morphology. Gregory and Walling (1973) categorize the myriad of watershed variables into three main groups: 1) topographic characteristics which include watershed area, stream orders, stream network density, basin and channel length, basin and channel shapes, and basin and channel relief; 2) geological characteristics; and 3) vegetational characteristics.

           Area is the most ubiquitous variable of watersheds (Gregory and Walling, 1973). Anderson (1957) characterized watershed basin area as the "devil's own variable" because all other variables are, in some way, correlated to area. Classifying watershed basins according to area has led to stream network ordering schemes (Gregory and Walling, 1973).

Horton (1945) devised a method of stream ranking which designated all unbranched headwater "finger-tip" tributaries as first-order streams which when joined with other first-order streams are designated as second-order streams. Similarly, second-order streams when joined with other second-order streams are designated as third-order streams, and the process is repeated in a recursive fashion for the entire watershed stream network. Ultimately, the highest order stream can be traced back to the one headwater first-order stream which deviated least from the direction of the highest order mainstream (Figure 1).

Stream order has a logarithmic relation with watershed basin area: i.e. stream order increases as the logarithm of the area. For example, if a watershed of 10,000 square miles develops a stream of order "n" then a watershed of 100,000 square miles would be drained by an "n+1" order stream (Horton, 1945). The relationship between stream order and basin size is a product of a deterministic balance of two opposing tendencies: 1) minimum overland flow distances, and 2) maximum channel size for maximum discharge rates (Woldenberg, 1969).

If frictional losses due to overland flow were negligible, then a large watershed basin would be drained by one large channel. On the other hand, if larger stream size was a negligible factor in maximizing discharge, then there would be many small streams and many orders. "Competition between small inefficient streams which are accessible, and large efficient streams which are inaccessible (to points within the basin) finally leads to an equilibrium hierarchy of a given number of orders and a given number of basins per order" (Wodenberg, 1969: 103).  

Horton's (1945) method of ordering streams was thought to be too subjective and complicated. Nonetheless, Leopold and Miller (1956) found that Horton's stream ordering methodology is a useful tool for a quantification of several fluvial characteristcs.





Figure 1. The Horton and Strahler methods of stream and segment ordering (from Gregory and Walling, 1973: 43).



Stream order is correlated to stream length, drainage area, to number of streams of a given order, and to channel slope (Figure 2).

Strahler (1952) proposed a scheme like Horton's, where the combination of first-order streams produced second-order streams, combining second-order streams produced third-order streams, and so on. However, in Strahler's system the highest order was given to a segment rather than to the whole stream as in the Horton ordering scheme(Figure 2).

            The limitation of the Strahler method of stream ordering is that stream segment order (n) can only be altered by equal order tributaries. Smaller size tributaries (n-1,n-2,n-3, etc.) do not alter the main stream segment (n). This shortcoming does not accommodate the increasing size of the stream segment due to the contributions from the smaller order tributaries. Currently, however, Strahler's stream ordering method is widely used by scientists and managers as a method of quantifying and standardizing fluvial characteristics. Two problems arise from the use of these stream ordering schemes: 1) how are perennial, intermittent, and ephemeral streams to be included in stream ordering; and 2) which map scale should be universally used in order to standardize the determination of stream order? (Hughes and Omernik, 1983).

Furthermore, although stream order is useful as a means of comparing relative stream and watershed size within a homogeneous physiogeographic and climatic area, the method is also erroneously used to compare characteristics such as stream size, watershed size, and discharge in disparate, widely separated watershed basins under different geographic and climatic regimes (Hughes and Omernik, 1983). Therefore watershed area and mean annual discharge per unit area (i.e., unit discharge in cubic meters per second






Figure 2 Relation of number of streams and drainage area to stream order (from Leopold and Miller, 1956: 203-206).



per square kilometer) are more accurate and useful parameters for determining stream size, watershed size, and discharge (Hughes and Omernik, 1983).

            The most salient topographic characteristics of watersheds are drainage density, relief and slope, shape and pattern, and area (Gregory and Walling, 1973). Geology, climate, soils, vegetation coverages, and anthropogenic land use also greatly affect watershed dynamics.

Drainage density, defined as the average length of streams within a basin per unit area, is a useful index that quantitatively describes drainage development. The development of drainage patterns reflects the different influences of geology, vegetation, soils, land-use, and other attributes (Horton, 1945).

            Horton (1945) defined stream frequency as the number of streams per unit area. Stream frequency is stream-order dependent whereas stream density is independent of stream order (Gregory and Wallis, 1973). Morisawa (1959) in an analysis of Mill Creek, Ohio, showed that stream density did not vary greatly within the Mill Creek watershed, however the number of streams differed greatly according to order. The frequency of first-order streams within the basin was by far the greatest.



Stream Order            Number of Streams       Ave.Basin Area        Stream Density

       1                                           104                          6.97(105 ft2)                 5.45 (mi/mi2)

2                                            22                         33.73    "                      7.02    "

3                                              5                       161.97    "                      6.06    "

4                                              1                       747.14    "                      5.66    "


(Morisawa, 1959; from Bloom, 1978: 203)


        The frequency of first-order streams within a watershed, therefore, is of primal importance, "few first-order tributaries and negligible overland flow enters a trunk stream directly, and these are the original source of most of the sediment load" (Bloom, 1978: 218). Most of the erosional work within a watershed basin takes place at the first-order level, where "during erosion of a drainage basin the zone of maximum erosion migrates toward the head of the basin" (Schumm, 1977: 69). All streams of higher orders are, in essence, trunk streams which act to transfer sediment and discharge waters produced and accumulated in the first-order drainage basins which make up (from the above table) the great majority of the watershed area.

            Drainage density increases with relief. Relief and slope affect the movement of water through a basin. The formation of the watershed itself is a product of water movement. The resulting drainage network, recursively, affects the manner in which water and sediment are discharged. Mean sediment yield (Figure 3) plotted on semilog paper against mean relief/length ratio, shows how rapidly denudation increases with the increase of relief (Schumm, 1977).

            Watershed shape and drainage pattern are highly interrelated factors. When comparing two watershed basins with the same channel size and roughness, but different long profile slopes, the hydrographs would show higher discharges, shorter discharge increase times and shorter lag times for the basin with the higher relief ratio (Figure 4). As seen previously the higher energy provided by steeper relief yielded greater amounts of sediment (Figure 3). Watershed shape also can affect the hydrograph, based on work done by DeWiest (1965), as depicted in Figure 5. However, watershed basin shape is drainage pattern dependent, as seen in Figure 6, following Strahler (1965), the efficient watershed (E) has a hydrograph with a slower rise but higher peak, whereas, in the more restricted watershed (D) the hydrograph has a conversely quicker rise but lower peak.

Watershed area is the single most significant characteristic because it is in some way always related to other watershed characteristics and parameters. In a theoretical geographic area, where all other topographic characteristics, geology, climate and vegetation are homogeneous, and precipitation inputs are uniform, discharge intensity would be dependent on area only (Morisawa, 1962),


                                                             Q = ¦ (A)


Such a theoretically uniform geographic area is rarely found in actual fluvial settings; nonetheless, the above discharge-to-area relationship has been used for many years to predict flooding events. As seen in Figure 7(A), based on several hydrographic areas of the United States, mean annual flooding [measured as mean total discharge (m3/s] per watershed) increases with watershed area. However, this simple discharge-to-area equation can be problematic. For example, as depicted in Figure 7(B), if mean annual flooding is measured per unit area (m3/s/km2) within the watershed then flooding decreases with unit area. This inverse relationship is mainly due to the fact that discharge rates, and sediment yields are greater in smaller (Figure 8), steeper (Figure 9), watershed basins.

Watershed basin drainage densities, relief, slope, shape, drainage pattern, and area all reflect the lithologic structural control exerted by the geologic composition of the watershed. Rivers that drain areas of different rock types and soils manifest different morphologies, due to differences in lithologic porosity, permeability, and erodibility




Figure 3

Figure 3 Relationship of mean annual sediment yield to relief/length ratio (from Schumm, 1977:21)



figure 4


Figure 4 The influence of basin relief upon stream hydrograph form is contrasted for two basins with different long profiles (from Gregory and walling, 1973: 269).



figure 4


Figure 5. The significance of drainage basin shape (A,B,C) on hydrograph form (from Gregory and Walling, 1973 : 269).

Figure 6



Figure 6. The effect of drainage networks (D,E) with contrasted restrictions of hydrograph forms. In contrast to (E) in the more restricted watershed (D) the hydrograph shows a quicker rise but lower peak (from Gregory and Walling, 1973: 269).




Figure 7


Figure 7 The relations of mean flood and watershed area for eight areas of the U.S. B) Mean annual flood per unit area plotted against watershed area in watershed of the U.K. and Wales (from Gregory and Walling, 1973:266).


Figure 8

Figure 8 Sediment delivery ratio as a function of drainage are (from Schumm, 1977:72).


Figure 9

Figure 9. Relation between valley slope and drainage area (from Schumm, 1973:338).




(crystalline vs. sedimentary rocks) (Schumm, 1977). Porosity and permeability of rocks allow for storage of water, thus affecting watershed discharge. Conversely, impermeable aquicludes may facilitate runoff discharges. Rock types affect the rate of weathering, sediment yields, and chemical solutes supplied to the channel.

Climate, soil and vegetation are highly interrelated factors of watershed dynamics. Amount of precipitation is a function of climate which, in turn, determines the quantity of water which is available as input to the watershed system. As shown in Figure 10, climate (P-E index = Mean monthly precipitation * Intensity of precipitation) is a key factor in determining watershed flow regimes such as drainage density. Water is essential to chemical and physical weathering of parent material, which in turn, contributes to the formation of different soils.  Climate is also important in influencing the vegetational cover of the watershed.   

Pedogenesis can be attributed to the combined influences of climatic, biotic, and topographic factors. Soils can be formed in situ from parent rock or deposited as allochtonous material by glaciers. The fundamental significance of soil in watershed dynamics is that it acts to either contain or discharge water (Gregory and Wallis, 1973). Precipitation will either runoff the surface of soils or it will infiltrate the soil profile. Infiltration rate is the maximum rate at which water can enter the soil. The proportion of runoff discharge or infiltration will depend on several factors such as type of soil, topography, precipitation intensity, and land use. The manner in which water is stored and transmitted through soils influences the production of sediment and water-borne solutes which, in turn, ultimately influences the hydrologic and morphologic make-up of the stream channel.

Vegetation in the watershed basin is important because it influences 1) water input by the processes of interception and evaporation; 2) water storage within the soil and plant mass; and 3) output of water and sediment into the stream channel. Vegetation cover intercepts the amount of water that can reach the ground by precipitation. The reduction of interception caused by the removal of vegetation (clear-cutting or burning) results in increases of runoff and loose soil detachment. As seen from Figures 11 and 12, the time-to-peak-flow is shortened by the loss of vegetational canopy cover density during a small storm, but there is no change in time-to-peak-flow in the same watershed during a large storm. Water and sediment yields also increase with decrease in canopy cover density, although differences due to variable precipitation intensities are quite evident.

Infiltration is increased by vegetation which resists runoff. Maximum infiltration is found in fully established forests, and minimum infiltration and maximum runoff occur in vegetationally denuded areas. Evapo-transpiration is the water which is returned to the atmosphere following uptake by plants and evaporation from leaf surfaces. Vegetation affects the amount and movement of water and sediment through the watershed system (Gregory and Wallis, 1973).

Land use is understood to be an alteration of the composition of the watershed landscape due to human activity  (Trimble, 1997) (Figure 13). These anthropogenic processes change the vegetational, pedogenic, and morphologic composition of the watershed. Therefore, they influence  water runoff, infiltration, evapotranspiration, and erosional dynamics of the basin. These changes, in turn, affect the amount and composition of water and sediment delivered to streams.





 Figure 10. Indicates the relationship between drainage density and climate (the P-E index =   Mean monthly precipitation        Intensity of precipitation) (from Gregory and Walling, 1973:270).



Figure 11


Figure 11. Effect of canopy cover density on watershed water hydrograph (from Ruh-ming, 1979 : 9-61).



figure 12


Figure 12.   Effect of canopy cover density on watershed sediment hydrograph (from Ruh-ming, 1979:9-62).



figure 13

Figure 13. Stream hydrographs of an area before and after urbanization. Urbanization results in a decrease in lag time and an increase in the peak discharge of the stream (from Ludman and Coch, 1982:222).



THE RIVER CHANNEL                     

Rivers can be classified according to several criteria. Tectonic-geologic controls separate most rivers into two groups depending on their ability to adjust river shape and gradient. Bedrock-controlled rivers adjust channel banks and beds to lithologic conditions produced by tectonic processes.  Alluvial rivers, however, adjust channel banks and beds to material transported and deposited by the rivers themselves (Schumm, 1977).

River channel morphology depends largely on two variables which are the two major products of watershed dynamics already discussed: discharge and sediment. Discharge dictates stream channel size, capacity and competence. Sediment is sorted according to size by channel hydrological dynamics during the transportation process.

            Transported sediment, termed "sediment load", is one variable by which channels are very often classified. The term "sediment load" not only defines the type of sediment (clay, silt, sand, cobble, boulder), but it also implies the mode of transportation (bed-load or suspended-load), which is discharge dependent. Alluvial channels can primarily be classified according to sediment load: bed-load, mixed-load, suspended-load (Schumm, 1977). The channel types, although they exist along a continuum, exhibit a range of straight, meandering, and braided channel forms. Straight channels are bed-load channels, meandering channels are suspended-load channels, and braided channels are bed-load rivers with islands of deposited sediment (Schumm, 1977).        

                                                           Channel Forms

River channel systems can be classified into a continuous spectrum of river types where the two end members are meandering rivers and braided rivers (Boggs, 1987).

Meandering rivers, unlike braided rivers, tend to be confined to a single channel, though anastomosing rivers such as the Amazon, are a type of multiple channel meandering rivers. Meandering rivers tend to have a low gradient (the Amazon, for example, drops only 65 meters over 3000 km), low velocity, and carry light, fine grained sediment. Braided rivers, on the other hand, are high energy systems.

            Rust (1978), however, classified rivers, in a more rigorous way, according to the degree of channel meandering (sinuosity) and channel braiding (multiplicity). Sinuosity is the ratio of the thalweg (depth at the deepest part of the channel) to the valley length. A ratio of 1.5 is the boundary between low and high sinuosities. The braiding value of "1" is a single channel river, more then "1" constitutes multiple channels. Therefore, the Rust classification of rivers is subdivided into 4 types of rivers:

1)      meandering where sinuosity is > 1.5 and braiding is < 1   

2)      braided where sinuosity is < 1.5 and braiding is > 1

3) straight where sinuosity is < 1.5 and braiding is < 1

4) anastomosing where sinuosity is > 1.5 braiding is > 1

The most common river types in nature are braided and meandering (Rust, 1978).         Friend (1983) classifies rivers either by channel patterns or sediment loads. Channel patterns are described as braided, meandering, or straight. Sediment loads are designated as suspended and dissolved loads, mixed bed and suspended loads, and, lastly, heavy and coarse bed-loads.

Schumm (1977), defined two parameters which control river geometries: (i) load characteristics such as suspended, mixed, or channel bed loads, and (ii) stability which is dependent on sediment size, sediment load velocity or stream power. Suspended load rivers are characterized by high sinuosity. Mixed load (channel bed and suspended loads) rivers tend to be meandering. Bed load rivers tend to have a braided character. Rivers which are sinuous single channel or anastomosing are confined by the deposition of mud and silty waters carrying a high proportion of fine suspended loads.

In an ideal river model, however, a river evolves longitudinally, from a straight type in the upper reaches of the watershed, to a meandering river in the lower portions of the river. In such a river, the bar structures change accordingly down-stream, from mid-channel compound bars, to bank-attached compound bars, to point bars at the lower reaches (Brierley, 1991). For example, in a study done by  Davies et al. (1978), on the river systems that drain the active Fuego volcano in Guatemala, it was evident that channel patterns, morphologies and flow characteristics change in response to rapid down stream gradient change. Fluvial sediments in transport show downstream changes in texture and composition which are closely related to fluvial mechanics. Decreasing slope results in major downstream modification in grain size and sorting. Sediment composition is also modified. There is an exponential downstream decrease in abundance of volcanic rock fragments. Mechanics of grain transport such as impact and saltation influence particle fragmentation (Davies et al., 1978).

            Meandering rivers can be classified into five categories: muddy fine grained, sand beds with mud, sand beds without mud, gravelly sand beds, and coarse gravel, and no sand (Jackson, 1978). Sediment deposits in meandering rivers accumulate in several different depositional settings: channels, point bars, levees, flood basin, oxbow lakes, and chutes (see Figure 14).

            Braided rivers are in the distal parts of alluvial fans, glacial outwash plains, and mountainous reaches of watersheds. These environments are often sediment rich, vegetation poor and with high water discharge. Depositional structures most evident are longitudinal, transverse, and lateral bars (Figure 15).

            The principles of fluvial geomorphology discussed above provide the theoretical basis of the GIW fluvial model.  Several concepts are important to the model. The model is built around the stream ordering scheme of Strahler (1952) where a second-order stream is a product of the confluence of two first-order streams, a third-order stream






Figure 14. The morphological elements of a meandering river system (from Boggs, 1987:355).



figure 15


Figure 15. The morphological structures in braided rivers (from Boggs, 1987:350).



results from the confluence of two second-order streams, and so on. First-order streams are considered accumulators of non-point discharge from sources such as sheet-flow and rills. As Morisawa (1959) demonstrated with Mill Creek a watershed surface is mostly drained by non-point discharge into first-order streams. The discharge in streams of second order and higher is mostly the result of point discharges of tributaries. In other words, all streams of higher orders are, in essence, trunk streams which act to transfer sediment and discharge waters produced and accumulated in the first-order drainage basins which make up the great majority of the watershed. Watershed area is the single most significant characteristic in determining discharge of any given watershed.



                        Mussels are great biomonitors. They will be used to demonstrate the usefulness of the GIW model to find relationships between in-stream habitat quality, water quality, and land use on the watershed. Bivalves, one of the five classes of the phylum Mollusca, are invariably aquatic and possess two calcareous shells (Figure 16). These animals lack a head and the accompanying cephalic sensory organs, radula and jaws. The mouth and the anus are located at opposite ends of the body. Bivalves have a foot which is usually used for infaunal burrowing (Cox, 1969). Except for the freshwater species, most bivalve species are benthic shallow marine, and live infaunally, epifaunally, attached to some substrate, or are free-swimming.

            Bivalves adapted to and colonized freshwater environments as early as the Carboniferous (Cox, 1969). Freshwater mussels have colonized freshwater habitats since at least the beginning of the Mesozoic. The order Unionoida, a subset of the subclass Paleoheterodonta, is divided into two superfamilies: Unionoidea and Etherioidea (Bogan, 1993). The other major subclass of freshwater bivalves is the Heterodonta which are subdivided into the orders Veneroida and the Myoida. The orders, superfamilies and families of these two subclasses are as follows (Bogan, 1993: 600):


               Subclass: Paleoheterodonta

                   Order: Unionoida

                      Superfamily: Unionoidae

                                  Family:      Unionidae



                       Superfamily: Etherioidea

                                  Family:      Mutelidae




               Subclass: Heterodonta

                   Order: Veneroida

                                 Family:      Corbiculidae






                   Order: Myoida

                                  Family:      Corbulidae




            Freshwater environments are ephemeral in comparison to marine environments which are much more stable over geologic time. It is important to realize, however, that freshwater environments, although ephemeral, are not younger, as a type of environment, than marine or terrestrial environments (Gray, 1988).        

Freshwater bivalves evolved from marine ancestors (see Appendix). Freshwater environments presented certain physical problems which bivalves had to overcome. Not only are lotic (see glossary) environments ephemeral but they are also, but they are also dominated by current. Adult freshwater mussels are mostly infaunal siphon bearing filter feeders; generally, they are sedentary (they can and do move locally, sometimes seasonally, and under conditions of drought and water-lowering).








Figure 16. Anatomy of a living bivalve (from Stearn and Carroll, 1989:117).



Except for the Corbiculidae and the Sphaeriidae, Heterodont reproduction strategy involves a ciliated free swimming larval type called the veliger. The Corbiculidae and the Sphaeriidae females release already formed juvenile bivalves directly into the water (Bogan, 1993). Dispersal and colonization of river systems by the Paleoheterodonta, however, is carried out parasitically, by a variety of larval types (glochidia, haustoria, and lasidia), utilizing varying parasitic strategies on host fish.

Unionoideans, after hatching from fertilized eggs, pass through a larval stage called a glochidium. The glochidia develop in a marsupial-like pouch between the adult's gills. The glochidia feed on the egg yolk and at a certain time, when the hatching female mussel comes in contact with a proper host fish, the glochidia are expelled from the pouch and attach themselves to the fish by sticky threads and/or sharp hooks on the miniature valves. The glochidium remain on the host fish until they grow into a miniature complete clam, at which time they drop to the bottom of the stream where they continue to grow (Cox, 1969). The female Lampsilis ovata, for example, attracts the host fish by mimicking, with modified mantle flaps, a small prey fish (Figure 17) and this serves as "bait" for a predatory fish which, in turn, serves as host for the larva (McMahon, 1991).

The Mutelidae haustoria larva develop on a tube attached with small hooks to the side of a fish. After metamorphosis takes place, the juvenile bivalve is released.  The Mycetopodidae lasidia larva strategy, unlike the haustoria tube-like attachment, involves a cyst-like attachment to the body of the host fish. The parasitic larval-stage adaptation, distinguishes these freshwater mussel groups from all other bivalves. The evolutionary advantage is obvious. Because of this adaptation freshwater bivalves have been able to disperse up streams and between streams when lotic system dynamics change through





Figure 17a. Highly modified mantle edge of a female Lampsilis ventricosa which exhibits a fish-like lure complete with eye spot and tail (from Kaat, 1982:198).


Figure 17b. Highly modified mantle edge of a female Mucket (Actinonaias ligamentina) which exhibits a fish-like lure complete with tail (Foto: Roman G. Kyshakevych).


Figure 17c. Highly modified mantle edge of a female Mucket (Actinonaias ligamentina) which exhibits a fish-like lure complete with tail (Foto: Roman G. Kyshakevych).


Figure 17d. Highly modified mantle edge of a female Mucket (Actinonaias ligamentina) which exhibits a fish-like lure complete with tail (Foto: Roman G. Kyshakevych).


geomorphic processes such as stream "piracy" (headwater capture) and coastal flooding (Sepkoski, 1974).

According to Bogan (1993), however, unionoid faunas are found in freshwater environments throughout the world: Australia (Hyriidae: 17 species), India (Unionidae: 52 species; Margaritiferidae: 1 species), China (Margaritiferidae: 1 species; Unionidae 37 species); Southeast Asia (Margaritiferidae: 1 species; Unionidae: 32 species), Europe (Margaritiferidae: 2 species; Unionidae: 8 species), Africa (Margaritiferidae: 1 species; Unionidae: 26 species), South America (a dozen Hyriidae species), North America (Margaritiferidae: 5 species; Unionidae: 292 species).

There has been considerable speculation regarding the mechanisms responsible for this global distribution of Unionoideans. For example, Hyriidae are only found in South America and Australia. Why? The discussion centers around  two possible theories that explain Unionoidean cosmopolitanism. One theory maintains that Unionoidean  populations were widespread throughout Mesozoic freshwater environments. With changing environments, due to geomorphic processes, these faunas became isolated and localized. The opposing theory proposes that Unionoideans simply migrated from one freshwater system to another as the tectonically driven, continental geography changed and regions devoid of species were simply uncolonized (Taylor, 1988).

Many factors threaten freshwater bivalve populations throughout the world: pollution, modification of channels, flood control. Bogan (1993) argues that since bivalves are generally long lived, deleterious environmental stresses are not immediately apparent. In many species, as discussed above, the "weak-link" in the bivalve life cycle is the larval-stage reliance on the obligate fish host. When the host fish  populations are exacerbated, the bivalve populations become "functionally extinct" and with time, "totally extinct" when the last remaining individuals die.



Freshwater mussels have been widely used as monitors of anthropogenic stress to lotic and lentic (see glossary) systems. They readily sequester and accumulate organic and inorganic contaminants in both soft tissue and shell, and, as has been the case with marine bivalve species, are excellent indicators of environmental pollution. In addition, their extended life spans and sedentary modes of life enhance their use as biomonitors.

The distribution and diversity of freshwater mussels in lotic habitats reflect the intensity of environmental perturbation because this group appears particularly sensitive to changes in water chemistry and sedimentation. In this study, freshwater mussel diversity and distribution along French Creek were used to help demonstrate implementation of the GIW model.

Since the turn of the century, shell growth increments from a variety of molluscan species have been analyzed by ecologists, archaeologists, and paleontologists (Rhoads and Pannella, 1970; Kyshakevych, 1994)) to establish "after-the-fact" records of many types of environmental perturbations: tidal (Richardson et al., 1980), diurnal/nocturnal and seasonal periodicities, floods, temperature changes, past El Niño/La Niña events (Rollins et al., 1987), anthropogenic pollution (Fritz and Lutz, 1986; Hutchinson et al., 1993; Salánki et al., 1989; Solé et al., 1995; Muncaster, et al., 1990), shell growth increments have been also widely used in archaeological research (Koike, 1973).         

External events, such as pollution, thermal shock, and drought disrupt growth and alter the structure, morphology and the internal chemistry of the shell (Fritz and Lutz, 1986). These alterations affect bivalve physiology by changing metabolic composition within the extrapallial fluid, which in turn, induces the structural, morphological and chemical changes within the shell. Environmental perturbations alter internal shell growth patterns by modifying (1) the microgrowth increments bounded by layers of higher concentrations of organic material, and (2) the arrangement of microstructures. These shell modifications are manifested as dark bands in the shell structure (Rhoads and Lutz, 1980).

Mussel populations have also been used to study the effects of environmental changes over time. Spatial and temporal distribution, abundance, growth, and biomass of freshwater mussel populations have been used as biomonitoring indices of perturbations in lentic (Nalepa and Gauvin, 1988; Bailey and Green, 1989; Huebner et al., 1989), and lotic systems (Watters, 1992; Kat, 1982; Sepkoski and Rex, 1974; White, 1977; Starnes and Bogan, 1988; Vannote and Minshall; 1982).

In general, bivalves make ideal biomonitors because they often have long life spans, and as discussed earlier, are not highly motile, spending their natural lives in geographically restricted areas. Freshwater mussel life spans have a wide range. Bauer (1992), reported that in freshwater pearl mussel (Margarifera margarifera) populations, life spans varied from 30 to 132 years. Neves and Moyer (1988) reported Virginia riverine species with longevities that ranged between 22 and 56 years. Harmon and Joy studied longevity in West Virginian paper pondshell mussels (Anodonta imbecillis) and determined the average age to be approximately 10 years. Longevity is important as a biomonitoring characteristic because mussel physiologies and communities reflect environmental perturbations over long periods of time.

Mussels also live sedentary life styles. Mussels spend their life span in a geographically small area where they drop-off, from the host fishes, in their post larval condition and anchor themselves to the substrate. The foot is mostly used as a digging mechanism, and it is not very efficient for locomotive purposes. Large, old mussels such as the mucket (Actinonaias ligamentina), for example, are often found buried among heavy, large cobbles (Kyshakevych, field observations), leading to the speculation, that it is possible that they have be been interred in the same location throughout their ontogeny. Their sedentary stability and longevity are, therefore, important long term, geographically restricted biomonitoring characteristics.




            In French Creek, Unionoideans have been present since at least the last Pleistocene glaciation. According to Johnson (1980), Unionoideans are found to the north of the maximum southernmost advance of the Pleistocene glaciers. The French Creek watershed drains a heavily glaciated area. Johnson maintains that these mussels migrated up the Ohio from some southern refugia and are not autochthonous. There are 27 species of mussels in the French Creek watershed (Bogan (personal communication), August 1996). The list of French Creek mussels is as follows (from Acker, 1996):


Family- Ambleminae

          1) Clubshell (Pleurobema clava)

          2) Longsolid (Fusconaia subrotunda)

          3) Rabbitsfoot (Quadrula cylindrica)

          4) Round Pig (Pleurobema coccinem)

          5) Spike (Elliptio dilitata)

          6) Three Ridge (Amblema plicata)


          Family- Anodontinae

          7) Creek Heelsplitter (Lasmigona compressa)

          8) Cylindrical Papershell (Anodontoides ferruscianus)

          9) Elktoe (Alasmidonta marginata)

         10) Fluted Shell (Lasmigona costata)

         11) Giant Floater (Pyganodon or Anodonta grandis)

         12) Kidney Shell (Ptychobranchus fasiolaris)

         13) Paper Pondshell (Utterbackia or Anodonta imbecillis)

         14) Squawfoot (Strophitus undulatus)

         15) White Heelsplitter (Lasmigona complanata)


         Family- Lampsilinae

         16) Black Sandshell (Ligumia recta)

         17) Eastern Pondmussel (Ligumia nasuta)

         18) Fatmucket (Lampsilis siliquoidea)

         19) Mucket (Actinonaias ligamentina)

         20) Northen Riffleshell (Epioblasma torulosa rangiana)

         21) Plain Pocketbook (Lampsilis cardium)

         22) Pocketbook (Lampsilis ovata)

         23) Rainbow Mussel (Villosa iris)

         24) Rayed Bean Mussel (Villosa fabalis)

         25) Snuffbox (Epioblasma triquetra)

         26) Wavy-rayed Lampmussel (Lampsilis fasciola)


According to the Western PA Conservancy, French Creek acts as a refugia for mussels because it contains 56% of all the surviving mussel species found in the state. Both the Clubshell (Plerobema clava) and the Northen Riffleshell (Epioblasma torulosa rangiana) mussels are listed as critically imperiled under the U.S. Endangered Species Act.

1) Key to the success of Unionoidea in freshwater lotic environments is the evolution of the obligate parasitic larval stage on host fish. This allows the sedentary freshwater mussel to disperse and colonize far-flung freshwater environments.

2) Bivalves are ideal long term, geographically localized biomonitors.





        Over the past few years, the use of geographical information systems (GIS) grew proportionally with a widening range of applications in government, industry, and academics dealing with management, planning, and research in engineering, business, health, natural resources and environment. GIS is a vital tool for the succesful implemention of environmental management and research.

Environmental and ecological applications are many and varied. For instance, GIS was used in the assessment of ecosystem "health" in the agricultural Rio Grande Valley to discover which features in the intensively impacted landscape were important to maintain native flora and fauna diversity (Whitford et al., 1996). In 1995 the U.S. Forest Service conducted the Southern Appalachian Assessment (SAA) to help resource managers plan and manage 37 million acres covering seven states of the Southern Appalachian region. The region is complex in land use patterns (agriculture, forestry, mining, urban, recreational), and natural ecosystems which support a wide variety of wildlife species (Fehringer et al., 1997). GIS technology figure prominently in this study, as well.

What is a GIS and why is it proven to be so useful in resource management and conservation? According to Environmental Systems Research Institute (ESRI), a GIS is:

"An organized collection of computer hardware, software, geographic data, and personnel designed to efficiently capture, store, update, manipulate, analyze, and display all forms of geographically referenced information" (ESRI, 1990: 1-2).

The National Science Foundation's (NSF) National Center for Geographic Information and Analysis (NCGIA) defines a GIS as "...a computerized data base management system for capture, storage, retrieval, analysis, and display of spatial (locationally defined) data" (Huxbold, 1991: 29).

Geographic information systems "contain map information stored in digital form in a data base" (Huxbold, 1991: 25). A geographic information system, in other words, is a digital amalgamation of geographic maps and data base management system (DBMS). Traditional cartographic maps representing a certain geographical area provided a limited amount of information. By including a data base management system, digitized graphic map images provide much greater information about geographic spatial features and attributes. Although statistical, spreadsheet, and drafting software applications can be used to handle spatial data, only GIS can carry-out spatial analysis by processing queries and operations on georeferenced data (Figure 18).

                                                                   Spatial Analysis

"Analyzing data involves the determination of patterns of data associated with locations and the manipulation of location related data to derive new information from existing data" (Huxbold, 1991: 57).

Spatial analysis involves defining the dynamic relationships among topological data structures (points, lines, polygons) in terms of distance, direction, and connectedness (contiguity). Just as the data-base analysis of attributes involves data entry, data retrieval, sorting, etc. Therefore, it is spatial analysis in conjunction with data-base manipulation what makes GIS a unique tool because it can answer two basic questions: "1) Where is...(an object); and 2) What is at...(a certain location)?" (Huxbold, 1991: 58). A map can provide information about the first question only if the object sought is plotted on the map. However, a GIS can find the object and from its attribute cartographic data locate the object on a map. The second question is more difficult to answer from a map because it depends on the attribute or set of attributes plotted on the map at a particular location. A GIS system, on the other hand, can have a data-base with information about many different attributes for a single geographic location (Huxbold, 1991). For example, spatial analysis of a single geographic coordinate may have information about a certain power line tower (point entity), along a road (line entity), built on a certain geological formation (polygon entity). An answer to he query: "How many mussels beds are there in French Creek?" does not require georeferenced data. However, a question such as "How many mussel beds, and where are they in French Creek between Venango and Meadville?" requires georeferenced data (latitude and longitude) about the mussel beds. These queries can only be answered by GIS (ESRI, 1990).

"The reference to spatial and analysis most often applied to GIS implies not only an ability to map information and refer to features that can be located, but also to identify relationships among mapped features and process their geometric characteristics for analyzing data in a spatial context" (Huxbold, 1991: 29).

In essence, a GIS system is composed of a data-base management system (DBMS), and the capabilities to perform spatial analysis. A database management system is software that allows for the use and modification of data in a database.






Figure 18. Depicts how the real world consists of many geographies which can be represented as a number of related data layers (from Environmental Systems Research Institute, 1990:1-2).






An integrative fluvial environmental management model which focuses upon the watershed system is best approached by the hierarchical, holistic multidisciplinary philosophy embodied by "landscape ecology". To appreciate the intrinsic value of this, herein proposed, general integrative lotic model it is essential to provide the necessary background of existing lotic ecological models.

According to Minshall (1985, 1988), holistic approaches to the study of stream ecology began with the work of Howard Odum in the 1950's. Over the past few years several ecological concepts have combined biotic and physical attributes (both natural and man-made) into holistic models describing the ecology of lotic systems. Two main historical concepts are the river-continuum concept (RCC) (Vannote et al., 1980) and the flood-pulse concept (Junk et al., 1989). During the past few years, the holistic approach of "landscape ecology" pioneered by German geographer Carl Troll (1950) has made an impact on integrative approaches to ecology. American proponents of landscape ecology, Forman and Grodon (1986: 595), defined landscape as "the study of the structure, function and change in a heterogeneous land area composed of interacting ecosystems". A landscape is described as a "mosaic where the mix of local ecosystems or land uses is repeated in similar form over a kilometers-wide area" (Forman, 1995: 13). These landscape mosaics, which can be of different scales, are, in turn, composed of homogenous areas (mosaics) called patches. Patches and patch dynamics are central concepts of landscape ecology.





                                                  RIVERS AS CONTINUA

The river-continuum concept (RCC) (Vannote et al., 1980), based on geomorphological principals (Leopold et al., 1964), states that in lotic systems, there exists a physical gradient that changes longitudinally along the length of the stream from the headwaters, through the middle reaches, to the confluence or delta. For example, substrate sizes change longitudinally from coarse in the headwaters to fine in the low reaches. Using the stream classification of Strahler (1957), headwater streams are characterized as 1-3 ordered streams, medium size streams were 4-6, and large rivers are greater then 6th-order streams. The derivation of the concept was based on the notion that biological communities adjust and become established in accordance to the physical longitudinal changes and adjustments of stream width, depth, velocity, sediment load, and other stream channel morphological and hydraulic parameters.

According to the RCC, the headwaters streams are influenced by riparian vegetation much more than are the lower reaches. In the headwaters, therefore, there is more allochthonous detritus material contributed to the stream and more shading of the stream which reduces autotrophic production. Autotrophic organisms, such as green plants, are more prevalent in more "open" and light intense mid-reaches. This relation is reversed in the lower reaches where stream size increases and riparian contribution of allochthonous organic material decreases and within-channel autotrophic primary production increases. This change is measured by the change of the gross primary production over respiration ratio (P/R).

The RCC further postulates that this biological gradient is evidenced by the "morphological-behavioral adaptions of running water invertebrates that reflect shifts in types and locations of food resources with stream size" (Vannote et al., 1980: 132). These morphological differences of invertebrates are functionally categorized as: shredders, collectors, scrapers and predators. Shredders feed on coarse particulate organic matter (CPOM, >1mm). Collectors filter from the water column or from the substrate fine particulate organic matter (FPOM, micron-1mm) and ultra-fine particulate organic matter (UPOM, 0.5-50 microns). Scrapers specialize in grazing on algal-rich surfaces. Predators are mostly invertivors and piscivors.

Figure 19 indicates that shredders and collectors are found mostly at the headwaters where CPOM, FPOM, UPOM have a predominantly riparian provenance. Therefore, the RCC postulates the importance of riparian contribution in the headwaters. "Headwater streams represent the maximum interface with the landscape and therefore are predominantly accumulators, processors, and transporters of materials from the terrestrial system" (Vannote et al., 1980: 133). With the increase of stream size, the nutrient particles decrease and the number of collectors increases. With stream size increase there is also an increase in  autotrophic production and number of scrapers.

In the low reaches of the river, according to the RCC, mostly collectors are found. Predators are ubiquitous through out the system. Most predators in the headwaters are invertivors, and downstream, both invertivors and piscivors. Mid-size rivers have the broadest range of abiotic parameters and, therefore, the greatest diversity (Johnson et al., 1995).



river continua



Figure 19. A depiction of the relationship between stream size and the progressive shift in structural and functional attributes of lotic communities (from Vannote et al., 1980:132)





"Major bioenergetic influences along the stream continuum are local inputs (allochthonous litter and light) and transport from upstream reaches and tributaries. As a consequence of physical and biological processes, the particle size of organic material in transport should become progressively smaller down the continuum...and the stream community response reflect progressively more efficient processing of smaller particles" (Vannote et al., 1980: 133).


Lastly, in every part of the stream continuum there is organic material processed, stored and released. In the RCC, released material is defined as "leakage". The released material is transported downstream where other communities are structured to capitalize on this "windfall". Over evolutionary time, according to the RCC, the lotic community evolves to reduce this leakage of organic matter. In this time scale, there was a spacial shift of communities along the continuum. This shift has two components: 1) a down stream vector that involves aquatic insects, and 2) an upstream vector which includes molluscs and crustaceans. Over evolutionary time, insects, believed to be of terrestrial provenance, most likely made the transition into aquatic environments via fluvial headwaters where there was a pronounced interface between land and water. The bivalves and the crayfish, conversely, shifted upstream from marine environments. This is the reason why, according to RCC, maximum species diversity is found along the midreaches of streams (Vannote et al., 1980:135).

The RCC has two additional corollaries: the resource-spiraling concept and the serial discontinuity concept. The resource-spiraling concept (Elwood et al., 1983, Mulholland et al., 1995, 1985, Newbold et al. 1981, 1982, 1983), maintains that resources do not just flow downstream uninterrupted but are absorbed by organisms and released as detritus and waste, and then re-absorbed and re-released further downstream creating a virtual spiraling effect (Johnson et al., 1995). Nutrients are transported downstream by recycling. Upstream nutrient cycling affects nutrient concentration which, in turn, will have downstream effects on communities which rely on those nutrients. Therefore, as the RCC deals with the longitudinal transport of material downstream as a combined effect of physical and biotic factors, the resource-spiralling concept deals with the downstream transport of nutrient in a conceptualized helical process (Newbold et al., 1982). "The term spiralling refers to the interdependent processes of cycling and downstream transport of nutrients in a stream ecosystem" (Newbold et al., 1983: 1249).



                                                       THE FLOOD PULSE CONCEPT


The river continuum concept (RCC) deals with the longitudinal physical relationships between the headwaters and mouth of a stream, and how biotic communities have adapted to those geomorphic parameters. The flood pulse concept (FPC), however, hypothesizes lateral relationships between the stream channel and the floodplain, and their effects on stream biota.

According to Junk et al. (1989), the RCC has two limitations: (1) formulation of the RCC was based on temperate streams, which usually are strongly anthropogenically impacted, and then extrapolated to all rivers; and (2) the RCC was restricted to aquatic environments that are permanent and lotic. The flood pulse concept (FPC), however, explains the relationship between the physical habitat and biota based on evidence gathered on unmodified, pristine temperate, subtropical, and tropical large river-floodplain systems (Bayley, 1995).

The flood pulse concept (FPC) proposes (Figure 20) that the "overwhelming bulk" of riverine animal biomass is derived from autochthonous floodplain production of organic matter and not organic matter derived from downstream transport, as postulated by the RCC. The concept also holds that, under the same hydrological regime, the longitudinal position of the floodplain along the drainage network is of little importance. Junk et al. (1989) go one step further: "We postulate that if no organic material except living animals were exchanged between floodplain and channel, no qualitative and, at most, limited quantitative changes would occur in the flood plain" (Junk et al., 1989: 112).      The concept deals only with predictable, seasonal, time-dependent flood pulses. Under these circumstances floodplain biota develop various adaptive life cycle and production strategies designed to take advantage of the raised hydrological input of the "moving littoral" (Bayley, 1995). Floodplain, according to the concept is defined as "areas that are periodically inundated by the lateral overflow of rivers or lakes, and/or by the direct precipitation or groundwater; the resulting physicochemical environment causes the biota to respond by morphological, anatomical, physiological, phenological, and/or ethological adaptions, and produce characteristic community structures" (Junk et al., 1989: 112).

Thus, "Life cycles of biota utilizing floodplain habitats are related to the flood pulse in terms of its annual timing, duration, and the rate of rise and fall" (Junk et al., 1989: 116). The floodplain is termed the "aquatic/terrestrial transition zone" (ATTZ). The term "river floodplain system", within the concept, includes the channel and the ATTZ. The flood pulse, according to Bayley (1995), is not to be thought of as a disturbance





Figure 20. Depiction of the flood pulse concept (from Bayley, 1995:154).



whereas man-made flood control structures such as levees, and dams, are considered disturbances.

The FPC emphasizes, therefore, the existence of two phases in the floodplain cycle: the flooding aquatic phase, and the inter-flooding terrestrial phase. During the flooding phase, in inundated areas, lentic environments develop which are dominated by main channel flood waters. Conversely, during base flow, the floodplain reverts back to a terrestrial habitat dominated by local conditions, such as rainwater, spring water, and not by the main channel. Biota adjust accordingly, although there are species that are not easily categorized by terms such as "aquatic" or "terrestrial" because they are equally productive during both phases (Bayley, 1995).

Although, according to the FPC, the floodplain clearly receives organic and inorganic nutrients from the channel, which  reflects the chemical composition of the drainage basin, most of the nutrients found in the floodplain, however, are due to autochthonous processing and production and are quite different from the main channel. For example, Junk et al. (1989) point out that gaseous inorganic compounds such as CO2, O2, H2S, CH4, and N2 are produced and consumed in the floodplain differently than in the main channel. They cite examples of the Amazon River floodplain where flood waters quickly become thermally stratified. Decomposing organic material on the bottom of these high temperature inundated floodplains consume more oxygen and release greater amounts of CO2 than main channel waters. Initially during the flooding phase (Figure 4.4) primary production is greater then decomposition; however, as water levels stop rising the rate of decomposition rises and this decreases the concentration of dissolved oxygen (DO). When drawdown takes place, nutrient runoff and nutrient concentration increases phytoplankton production. The amounts of organic compounds such as carbon, although transported to the floodplain from the channel, are negligible in comparison to autochthonous, in situ floodplain production (Junk et al., 1989; Bayley, 1995).

Dissolved solids such as halite, due to high rates of evaporation, have a higher concentration in the floodplain than in the main channel. In addition, levels of dissolved solid nutrients are never limiting factors in the main channel, whereas, in the floodplain nutrients such as phosphorous and nitrogen can limit productivity. In-suspension inorganic particulate matter, however, is less important in the channel then in the floodplain. Suspended sediment increases turbidity, hindering photosynthetic production in the channel. However, in the floodplain, due to lower hydraulic energies, suspended particulates settle out on the floodplain, and depending on the nature of the sediment, possibly increase fertility.


                                                      LANDSCAPE ECOLOGY


"An ecology of the landscape is nascent and timely. The field will be developed by basic biologists and ecologists, as well as geographers, foresters, planners and landscape designers, social scientists, wildlife biologists, agriculturalists, and others. Important principles will be drawn from each field, but a central body of landscape ecology theory is likely to emerge concurrently. Such theory may suggest novel solutions to critical environmental problems" (Forman, 1983: 535).

Landscape ecology centers on three main characteristics of the environment:

1) the spatial interrelationships among elements (ecosystems, ecotopes) of the landscape; 2) the functional interactions such as the flow of energy, nutrients, species among these elements; and 3) the naturally and anthropogenically induced changes in spatial and functional relationships over time (Forman, 1983; Schlosser, 1991).

In landscape ecology, landscapes are vertically composed of overlying (topological), heterogenous land attributes such as geology, soils, forests, and landuse. Each attribute, in turn, is composed of horizontal (chorological), heterogenous elements. In the forests attribute, for example, chorological elements would be a tree gap and stand. According to Zonneveld (1990), the science of landscape ecology is the study of the relationship between topological units (attributes) and chrological units (elements). 

A basic concept of landscape ecology is that it is a hierarchically structured, holistic discipline. Both hierarchical structure and holism merit a clarification. According to Zonneveld (1990), holism is a philosophy formulated by J.C. Smuts in the mid 1920's, which states "that reality consists of wholes in a hierarchical structure in the sequence: atoms, molecules, minerals, organisms, human society, the world as a total ecosystem, the galaxy, and the cosmos" (Zonneveld, 1990: 9).

From a scientific point of view, holism allows the studying of whole systems without knowing all the details of the internal functions. These systems can be classified as "black boxes" where only the inputs and outputs are known and the details of how the input is processed within are invisible. This a common phenomenon in everyday life. For example, in the use of an application software such as a word processor on a computer a known input goes through the keyboard and a known output is manifested on the CRT screen, but all the hard/software processes within the computer are invisible to the user. Therefore, the "black box" approach in holism simplifies the understanding of very complex structures and processes by reducing details (Zonneveld, 1990).

From an aerial photograph, land seems to be a spatial arrangement of heterogenous mosaic-like patches and corridors with boundaries, on a background matrix. These heterogenous horizontal patterns of patches and corridors are caused by 1) different substrates, soil types, land forms such as hills, wetlands, rivers (corridors) which cause biotic differences in vegetation, 2) natural disturbances such as fires, winds, pests, and 3) human disturbances such as farming, lumber harvests, clear cutting, urban development, roads and canals (corridors), and other landuses. The mosaic pattern of patches and corridors on the background matrix, according to landscape ecology, changes over time. Therefore, Forman (1995) identifies three types of spatial elements: patch, corridor, background matrix.

A patch is defined as a "wide, relatively homogenous area that differs from the surroundings" (Forman (1995: 39). A patch is a horizontal, bounded, landscape element of a given size, shape, and compositional attribute. These patches are dynamic because the parameters change over time. Due to an anthropogenic disturbance, for example, part of a wooded patch may become an agricultural patch, changing the size, shape, and attribute value of the patch composition within a landscape. These patch dynamics have ecological implications with changes in fauna and flora diversity, substrate composition due to erosion, and hydrologic processes such as increases in runoff (Andersson and Sivertun, 1991).The second type of building block of landscapes is a corridor. Contiguous and far-flung patches within a landscape are connected by corridors. Non-anthropogenic corridors include streams, and animal trails. Roads, canals, pipe lines, and bridges are examples of anthropogenic corridors. According to Forman (1995, 1990), corridors may perform several functions. A corridor may act as conduit along which objects (trains), animals, nutrients (in a stream), move between patches. Also a corridor may act as barrier which inhibits the movement of objects between patches, or it may act as a filter which differentially controls movement by type of object or by rate. Corridors may also serve as habitats, such as hedgerows, roadside ditches, telephone poles for rodents, fishes, frogs, birds.

Stream channels and the adjacent riparian, wetland and floodplain zones are considered corridors that transport organic (nutrients) and inorganic (sediment) matter, and act as habitats for a great variety of fauna and flora. These fluvial corridors are, and historically have been, important elements of human society providing irrigation, drinking water, generating power, transportation, fisheries, recreation, and sinks of societal wastes.

Stream corridors are themselves extremely patchy. The channel can constitute a mosaic of well-defined patches which reflect gradients, overall watershed hydrology and anthropogenic land-use patterns, and the underlying morphology, leading to forming pools and riffles of different depths, flow velocities, and substrates. The mosaic of channel patches, in turn, is manifested by patchiness of the diversity and abundance of life. Matrix is defined as "the background ecosystem or land use type in a mosaic characterized by extensive cover, high connectivity, and/or major control over dynamics" (Forman, 1995: 39). The predominant land area cover is used to determine the matrix type. The landscape can be mainly composed of homogenous or mixed heterogeneous patch covers of forest and/or grassland ecosystems, farmlands and/or urban areas, which although not contiguous, nonetheless, are highly connected throughout the landscape. The background matrix affects the overall dynamics of the landscape by controlling the flow of materials (water, nutrients, sediment), and the movement of species.  

Integral to landscape ecology is hierarchy theory which deals with organized complexity. "Hierarchically organized systems can be divided, or decomposed, into discrete functional components operating at different scales" (Urban et al., 1987: 121) An example of spatial hierarchy is shown in Figure 21 (from Forman, 1995: 12). The planet is subdivided into continents and oceans, continents are further subdivided into regions, regions into landscapes, landscapes into local ecosystems and each of these hierarchical levels has its own scale. Landscape is defined as a kilometers-wide scaled area composed of local ecosystems. (Forman, 1995).

In landscape ecology, the hierarchy paradigm permits, first, definition of functionality of elements (patch-corridor-matrix) in a system and, secondly, explanation of how these elements are related to each other. According to Urban et al. (1987), natural landscapes are often not easily decomposed because spatial elements are not easily discernable; however, in many cases, complex natural systems can be decomposed to a set of hierarchical elements.

These elements, within the vertical hierarchical structure, are organized according to scale of function. Measuring scale as a function of space and time, low-level elements within the hierarchical structure are small and frequent (high frequency, fast), and high





Figure 21. Spatial hierchy on land (from Forman, 1995 : 12).





level elements are large and infrequent (low frequency, slow). Higher levels cannot only be larger and slower but also can control and/or contain lower levels. Landscapes are such high-level elements. Landscapes are spatially nested with each higher element containing lower level elements. Figure 22 demonstrates such a hierarchy of a forested landscape. Each lower element is spatially smaller than the elements above and nested within the higher elements: stand elements, for example, are smaller (1-10 hectares) and are part of watershed elements (100-1000 hectares). Hierarchy of scales allows elements to be isolated and at the same time recognizes that other scales are relevant also.

Among these elements there is also a behavioral interaction hierarchy. According to Urban et al. (1987), using the above forested landscape as an example, stands, watersheds and landscapes interact to generate ever higher level behaviors, and all behaviors at each level can only be understood in context of the other levels of behavior. Higher level elements define the context for lower level behaviors: watershed morphology determines size of tree stands (Urban et al., 1987).

Flows of energy, nutrients, biota (animal migrations) link the hierarchy of spatial elements vertically and horizontally. Therefore, to understand the dynamics of any one element within the hierarchical structure, according to Forman (1995), it is necessary to know at least three hierarchical linkages: 1) the encompassing element/s above; 2) the other elements which compose the same level; and 3) the component elements at the next lower level. These three linkages control the pattern and processes within each element (Forman, 1995).


                                FOUR LEVELS OF A FOREST HIERARCHY

Level                                         Boundary Definition                        Scale

LANSCAPE                              PHYSIGRAPHIC PROVINCES                   1000s ha

WATERSHED                            LOCAL DRAINAGE BASIN;                     100s-1000s ha

                                           TOPOGRAPHIC DIVIDES


STAND                                   TOPOGRAPHIC POSITIONS;                  1s-10s ha

                                            DISTURBANCES PATCHES


GAP                                       LARGE TREE’S INFLUENCE                    0.01-0.1 ha




Figure 22. Four levels of hierarchy and scale (from Urban et al., 1987: 122).



                                  THE GENERAL INTEGRATIVE WATERSHED MODEL                               


     Modeling has been a very useful technique for simulating and understanding natural environments, and implementing and evaluating the effect of environmental management policies. River geomorphologists such as R.E Horton (1945), A.N. Strahler (1957), and M.E. Morisawa (1962), as discussed in chapter 1, developed quantitative analyses of fluvial systems which were then applied in physical and mathematical models by hydraulic engineers.

The construction of a model involves three stages: conceptualization, implementation, application and results (Martin, 1968). Conceptualization of a model is a process that is subdivided into several tasks:

(a) The system to be simulated must be defined in precise terms. The system is broken down into modular functional subdivisions which, in turn, depend on the complexity of the system.

(b) The theoretical basis of the model is formed from educated, informed hypotheses and assumptions about the system.

(c) Based on the hypotheses and assumptions, the model rationale is established, reflecting the "real world" system being simulated. According to Martin (1968), several questions about the "real world" system must be answered during ecological model development. What environmental factors affect the system? What are the interactions between humans and the system? What are the functions of the system? Are the functions deterministic or stochastic? In a deterministic model there are no variations to the outputs due to chance. Without probabilistic elements, the same inputs produce the same outputs. In a nondeterministic, stochastic model, conversely, due to probabilistic elements and variables, the same input might not produce the same output.

(d) All system parameters and variables are defined.

(e) A conceptual model is described in abstract terms and concepts which explain the process of how a "real world" system is replicated by the model (Martin, 1968).

Model implementation involves transposing the model from a conceptual to a robust working model. The first step is to construct a logical flow diagram of the model. The flow diagram reflects the modularization of the system into component functions. Each module simulates a system function, as in the solution of complex problems, where the problem is broken down into several simple solvable component problems, and then reassembled as a solution of the complex problem.

Modern high-level (where each statement corresponds to several "low level" machine language instructions) computer languages reflect this philosophy with highly modular languages where subroutines are considered as recursive independent modules. Modularization also allows for refinement and flexibility of the model. Individual modules are refined and altered to reflect the real world system conditions and functions more precisely.

 In this implementation phase of the model, all mathematical equations and variables are expressed explicitly. According to Martin (1968), at this point the model is checked for validation by asking: Is the model conceptually valid? If the model is deemed valid then the logical flow diagram of the conceptual model is translated into a computer program flow diagram, and then coded. Finally, the model is applied by going into the field to collect data, and the results are analyzed.






Natural environment geometries may be so complicated that many problems can be solved with the aid of models. Physical models are scaled down replications of natural systems. Physical models are used to study, at low cost and minimum risk, the interactions in natural river environments, and between man-made structures and river environments. Examples of three dimensional mobile physical models are the steel-plexiglass constructed flume used to simulate valley development, and the 9x15 m container called the Rainfall-Erosion Facility (REF) at Colorado State used to study drainage network and alluvial fan formation (Schumm, 1977). These models are referred to as "mobile" because they are constructed of materials that can be changed and molded by model water flows. In the case of the steel-plexiglass flume used by Shepherd and Schumm (Schumm, 1977), the "mobile" material that simulated the stream channel bed rock was a mixture of fine sand and kaolinite applied wet and allowed to dry so that when subjected to flow it simulated channel erosion, deposition, lateral channel migration, etc.

However, scale models do not faithfully simulate all fluvial processes, so parameter distortions are used in models to heighten a response of particular interest. Distorted physical models weaken the geometrical analogies to a natural system but they are still useful (Simons, 1979).


                                       DIGITAL MATHEMATICAL MODELS


     The physical model can be referred to as an analog computer model. Since the advent of digital computers, virtual digital modeling has also been widely use. Digital computer modeling, however, requires a numerical basis. The philosophy of mathematical modeling is based on the notion that all natural phenomena can be described by an appropriate equation. Hydraulic engineers developed many mathematical models that dealt with aspects of fluvial water and sediment movement through the use of equations that described various aspects of river dynamics. However, due to the difficulties in the application of these mathematical equations, engineers and fluvial geomorphologist still resort to physical models (Simons, 1977).

Mathematical models can be classified into three types: statistical regression models, "black box" models, and physical process models. Some models are hybrids of two or all of the preceding types of models (Simons and Li, 1979). Regression models are not adequate because they require great amounts of data to describe certain relationships among parameters. In addition, according to Simons and Li (1979), regression models cannot adequately predict time/space processes.

So called "black box" models tend to be simplistic because they reduce all the complex internal system parameters and processes strictly into input/output data. An example of this type of modeling, is one where complex water discharge and hydraulic processes within a watershed are reduced to a formula such as "Q = CAI" (where Q is discharge, C is a runoff coefficient that is also a reduced parameter representing complex hydraulic processes such as infiltration, evaporation, sediment detachment, and I is the precipitation input variable). The input into the "black box" is the rainfall, the output is the watershed discharge. Such a model is useful because it is simple but it does not accurately, or in detail, represent the system.

However, accuracy is also a function of scale. In the modeling of a particular river reach, hydrologic and, ecologic parameters, and details of mechanisms at work within the system are important. On a large landscape/watershed scale system, detailed accuracy of entities or mechanisms on a small area of the landscape might be irrelevant when considering inputs and outputs that affect or are affected by the large scale system as a whole. These entities and mechanisms can be represented by a limited number of properties called "diagnostic characteristics" which can be used to formulate useful empirical generalizations. In these large scale systems, the "black box" approach is plausible and effective (Zonneveld, 1990).

Physical process mathematical models do not reduce small scale complex systems into subsystem input/output "black boxes", but rather decompose complex systems into modular components using simpler subsystem mathematical models. Using the above example, a complex system such as the hydrology of a watershed would be segregated into component localized phenomena such as soil infiltration, sediment detachment by rain drops, and runoff over differential substrates. Each of these geographically restricted hydrologic phenomena would be treated in a modular fashion. Each module would be refined, calibrated and customized, by the use of site-specific variables, to the physical environment being simulated. These models tend to be more flexible and replicate more accurately a physical process (Simons and Li, 1979).

            However, rarely can river channels be described as simple one-dimensional entities. Much more complicated mathematical two-dimensional models describe in more detail sediment and water flows along two axis, longitudinal and transverse. Two dimensional models describe flow phenomena along irregular stretches of river where channels are irregular in size and shape, sediment density is not uniform, inconsistent flow velocities produce eddies and back flow, and channel bed resistance is irregular. "It is apparent that the model complexities and computational efficiencies depend on the number of governing equations being solved in the dominative direction(s). A relatively simple model that can provide adequate information is always desired" (Chen, 1979: 10-70).    

            These mathematical models were utilized to deal with problems in river construction projects such as bridges, dams, canals, irrigation systems, and flood control. The U.S. Army Corps of Engineers, in the early 1960's, established the Hydrologic Engineering Center (HEC) which developed six computer-based river models for different aspects of watershed and channel dynamics. The HEC-1 model, for example, calculated the precipitation runoff from watersheds. The HEC-4 model calculated and generated data on flow volumes of any stream length. The HEC-6 model calculated water surface and stream bed profiles, water velocity, water depth, energy slope, sediment load, and gradation of sediment load (Thomas, 1979).



                                         CONCEPTIONALIZATION OF THE

                                  GENERAL INTEGRATIVE WATERSHED MODEL  


Over the years the realization, from a resource and environmental management perspective, has been that river environments are products of much larger and complex watershed physical processes (Johnson et al., 1995; Sparks, 1995).

Because streams and rivers are products of entire watersheds, the basis of research and management of river and stream resources and environments has to be approached from a broad, all-inclusive watershed context. Most models, either physical or mathematical, constructed by geomorphologists and hydraulic engineers are too specific and narrow in scope. Usually these models are designed to measure one or a set of parameters that are site-specific, in conjunction with some construction project. These models are also one dimensional in that they describe only abiotic, physical phenomena, such as flows, discharge, sediment loads, aggradation, erosion, deposition, gradients, sinuosities, facies, etc.

The conceptual models of stream/river ecologists, on the other hand, have been criticized for being only applicable to pristine environments. As is the case with the mathematical models of hydraulic engineers, these conceptual models also tend to be "channel-centric" in the sense that biota are seen only in the context of the immediate channel environment. These models also have been criticized for being only descriptive, and not quantitative (Johnson et al., 1995). Some of these descriptive concepts also have been deemed latitude-specific. The RCC model (see chapter 4: 89), for example, is more applicable to mid-latitude temperate streams and rivers, whereas, the "flood-pulse" concept (see chapter 4) is more applicable to the tropical riverine systems of the lower latitudes.

A new watershed wide model is need that is properly integrative and includes a) all the principles, measurements, and quantitative methodologies of fluvial geomorphologists and hydraulic engineers; b) the concepts and hypotheses of stream/river ecologists; and c) with the growing popularity of landscape ecology as a powerful comprehensive concept, the model should include the geographical dynamics of natural and anthropogenically-impacted fluvial systems within the context of landscape ecology.

The proposed model should not only be integrative interdisciplinarily, but it should also be general and flexible so as to be universally applicable to different fluvial environments, different latitudes, different climatic and geologic regimes, catchment area sizes, and anthropological impacts. The model should be of modular construction so that different components of the system could be refined and augmented according to research or managerial requirements.

The proposed GENERALIZED INTEGRATIVE WATERSHED MODEL (GIW) is such a model. The model is based on five assumptions:

1) watershed systems are composed of tributary/watershed subsystems. The majority of these component tributary/watersheds are 1st order systems. Based on this hypothesis, it is assumed that most of the discharge into channels of ³ 2nd  order rivers is due to discharge from tributaries, therefore, rivers of ³ 2nd  order are defined as "trunk" river systems;

2) discharges into channels of main trunk rivers form a heterogenous mosaic of patches which are geographically delineated by tributaries and point source discharges;

3) the magnitude of tributary and point discharges into corresponding river trunk patch discharges, denoted by the discharge proportion ratio (PQ), is proportional (in the case of tributary discharges) to tributary/watershed system geomorphic variables (such as watershed area, catchment relief, channel gradient, climate). In the case of point discharges, the amount of discharge is directly proportional to source variables (such as quantity of municipal, industrial, and agricultural waste) affecting the magnitude of discharges;

4) the composition of tributary discharges, within the river trunk patch discharges, is proportionally affected by the % of total area within the tributary/watershed system covered by an attribute such as geology, soils, and land use. Denoted as the attribute discharge (AQ), the tributary discharge is assumed to be a product of interactions between precipitation (infiltration, runoff, evapotranspiration, etc.) and attribute coverages. In the case of point discharges, the composition of these discharges is affected by such variables as type and concentration of polluting agents;

5) and lastly, it is assumed that the heterogenous composition of these main trunk patches, induced by tributary/watershed system, and point discharges, in turn, differentially affect the biotic (flora and fauna) and abiotic (chemical, nutrient, sediment) makeup within the trunk river patches.

Indeed, the first assumption, as it has been discussed earlier, is based on the hypothesis that water and sediment flows in rivers have their provenance in the physical processes associated with first order streams. For example, according to Morisawa (1962), for example, in the drainage network of Mill Creek, Ohio, there are 104 first order streams with an average basin area of 6.97 x 105 ft2 per stream for a total of 725 x 105 ft2, whereas, the single 4th order stream, Mill Creek, which reflects the area of the entire watershed drains an area of only 747.14 x 105 ft2.

This demonstrates that the great majority (96%, in the above example) of the catchment precipitation, the ensuing evapotranspiration, infiltration, runoff and other physical hydrological processes within a watershed occur at the first order level. Likewise, direct precipitation into a channel is a negligible part of the total trunk discharge. Ground water does contribute, depending on parameters such as precipitation rates, vegetation cover, and soil infiltration rates, to the total discharge of a trunk stream/river. Most of the sediment flow has its source in first order surficial streams (Bloom, 1978; Strahler, 1965). It could be argued, therefore, that any stream above a first order stream can be considered a trunk stream.

Thus, any watershed area drained by a trunk stream/river of order (n), where (n ³ 2), is subdivided by watershed areas drained by streams/rivers of order (< n), where any stream "X" of order (n ³ 2) is considered, hypothetically, to be a trunk stream whose discharge Qx is composed of the sum of discharges Qy and Qz from streams "Y" and "Z" of orders (< n), such that Qz + Qy = Qx (Yen, 1979).


                                                         Qz              Qy





Water and sediment flow within the trunk stream is considered to be exclusively a product of tributary inputs. Downhill surficial riparian contribution and direct "into-the-channel" precipitation are considered minimal. Groundwater contribution to discharge, if measurable and mappable, can be included in the GIW model. However, sediment contributions from ground water sources are considered negligible.

 A tributary's discharge (Q), into a trunk river can be expressed as a function of the geomorphology of the tributary's watershed, such that


                                          Q = ∫(A, L, 1/F,1/S,1/Rc,1/Rh)


where A = drainage area, SL = total stream length, F = frequency of stream channels, S = channel gradient, Rc = basin shape, Rh = basin relief. Since cumulative channel gradient and relief ratio measure the same parameter one of them can be eliminated. Therefore, discharge of a tributary is a function of drainage/watershed area, stream length, frequency of stream channels, basin shape, and basin relief (Morisawa, 1962).

Furthermore, "black box" reduction on this relationship could be done in the case of small watersheds where climatic conditions, drainage patterns, and topography are relatively uniform, then stream length and frequency, and basin shape and relief could also be viewed as constant and thus eliminated from the equation. Under these conditions discharge could be considered as function of watershed area.

The second assumption of the model postulates trunk river channel discharges are subdivided into a system of patches which are delineated, and quantitatively and qualitatively impacted, by tributary and by point and non-point discharges. According to Foreman (1995), the spatial analytical methodology of landscape ecology (patch dynamics) can be applied to river channels and the adjoining flood plain. These are highly heterogeneous patchy systems. A river is viewed as a mosaic of distinct patches defined by water depths, substrate differences, riffles, pools, sand bars, and other channel morphological differences. Pringle et al. (1988) applied patch dynamics to lotic systems, classifying several types of patches: 1) patches of nutrient and their effects on periphyton communities and nutrient spiralling; 2) riparian leaf litter patches; 3) beaver-induced patches; and 4) flood plain patches.

The GIW model, however, defines "hydro-patches". These are heterogeneous patches within the channel flow which differ according to the composition, in solution and suspension, of discharge waters. As seen in Figure 23, each discharge input into the trunk river channel and flood plain delineates a patch which corresponds to a discharge source. Size and shape of patch is highly correlated with many hydraulic parameters such as: magnitude of tributary discharge, main channel flow parameters, mixing dynamics, and position and size of the subsequent tributary discharge.

In the GIW model there are two types of patches: "large" patches (LP), and constituent "small" patches (SP). The LP boundaries in a (n) order river are delineated by (n-1) order tributaries. The LP's are subdivided into SP's delineated contiguously by tributaries of £ (n-2) order, point and non-point sources (see Figure 23). The upper boundary of the LP begins at the confluence of the (n-1) tributary and the lower boundary ends directly before the confluence of the subsequent (n-1) tributary. The discharge within the LP changes with the addition of each SP discharge. 

Each tributary's watershed is considered as a modular subdivision of the main trunk river watershed, and the biotic and abiotic "impact" upon the main trunk river patches, depends on (a) the tributary's proportion of discharge, and (b) the tributary's relative longitudinal position along the trunk stream.

 The third assumption maintains that the quantitative impact of a tributary/watershed system on a trunk river patch, denoted as a parameterless "proportion of discharge" ratio:

                                                     PQ = QT  / QR + QT


where (PQ) = the proportion ratio of the total discharge in a trunk river patch contributed by a patch tributary; (QT) = tributary's total discharge; (QR) = total main trunk river discharge at the upper boundary (at the tributary) of the patch. From Figure 25, for example, based on several parameters such as area, etc., tributary/watershed system "B" discharge = 10 cfs., therefore, (QT) = 10, and the total upstream discharge (QR) = 15 cfs., thus,

                                                        PQ  = 10 / (15 + 10)



the proportion of the total discharge in patch "B" contributed by tributary "B" is (PQ) = 0.4. In other words, 40% of the total discharge in patch "B" has its provenance in tributary/watershed system "B".

Tributary "C's" proportional contribution to patch "C" discharge is 67% (PQ = .67),


                                                     PQ =  50  /  (25 + 50)


where (QT) = 50, (QR) = 75. Similarly, tributary "D's" discharge proportion ratio in patch "D" is (PQ) = .25 where (QT) = 25 and (QR) = 100. The greater the proportional discharge ratio (PQ), the greater is the tributary's discharge contribution to its corresponding patch.

As a result, the longitudinal upstream-downstream position of a tributary's discharge into the main trunk affects its impact on the trunk patches. The farther downstream a tributary discharges into the main trunk river, the less is its proportion of discharge to the corresponding patches. All other parameters being equal, (Figure 25) although tributary "B" has a smaller overall watershed discharge (10 cfs.), it has a greater discharge proportion (PQ = 0.40) in patch "B", than does tributary "C" which has a larger overall discharge (25 cfs.) but a smaller discharge proportion ratio (PQ = .25) in patch "C". In other words, the impact of tributary discharge on its corresponding patch is relative to its upstream-downstream position along the trunk river.

As seen from the example in Figure 24, where the discharges from tributaries/watersheds "A", "B", "C", "D" are the same, the proportional discharge ratios in the corresponding patches differ according to the position of tributary/watershed system discharge along the trunk river. This is due to the cumulative effect of upstream discharges on downstream patches.

Point source discharges such as municipal and industrial wastes, and non-point source discharges such as sheet flow agricultural runoffs are also assumed to produce a system of trunk river channel patches. The relative magnitude of tributary, point and non-point source discharge impacts within a watershed is important for cost- and time-effective evaluation, planning and management of biota, water quality, and reparation projects.

The fourth assumption is a "black box" proposition which maintains that abiotic factors (chemistry, nutrients, sediments) in each river patch can be considered proportional products of all the hydrologic complexities (evaporation, infiltration, surficial runoff) involved in the interaction between precipitation and the corresponding tributary/watershed systems coverage attributes (geologic regimes, soils, and anthropologic land use). This interaction between precipitation and coverage attributes is denoted as the % of attribute affected discharge (AQ)(per attribute), such that,

                                                               AQ =    A       PQ

where (A) = % of the total area of a tributary/watershed system covered by some attribute, the tributary/watershed discharge proportion ratio is (PQ), and (AQ) equals the percent of discharge in the patch affected by precipitation and attribute interactions within the tributary/watershed system.

For example, referring to Figure 25, if tributary/watershed system "B" is 50% covered by forest, and, from the previous example, the discharge proportion ratio (PQ) for patch "B" is 0.4, then (AQ) = 20% (50% x 0.4). In other words, 20% of the total discharge in patch "B" is affected by the forest coverages in tributary/watershed system "B". Likewise, if tributary/watershed system "D" is 100% covered by forest, and, as we have seen, (PQ) for patch "D" is .25, then (AQ) = 25% (100% x .25). Thus, 25% of the total discharge in patch "D" is affected by the forest coverages of tributary/watershed system "D".

The GIW model is flexible because each of these "black box" modules can be made more transparent. Each tributary/watershed system module can be modeled much more rigorously and precisely. Mathematical and statistical models can be constructed that separate hydraulic phenomena into distinct components such as infiltration and runoff through different soils, infiltration and runoff through different kinds of bed rock coverages, raindrop impact on different soil types, evaporation and different vegetation types, etc. The only factors that limit the transparence of any module are time and money.

The fifth assumption is based on the well-established hypothesis that different chemical and nutrient compositions, and sedimentary loads, in freshwater fluvial systems, affect flora and fauna differentially (Pringle et al, 1988; Ward and Johnson, 1989; Schlosser, 1991).




Figure 23. Patch system.




even tribs

Figure 24. Equal area tributary system.





uneven tributaries


Figure 25. Unequal area tributary system.






A useful analogy of system and subsystem modules is that of an automobile. An automobile is a system composed of a variety of "subsystems", such as the cooling system, the electrical system, steering system, the hydraulic system, suspension system, etc. Each of these subsystems can be considered as an independent module which contributes, with a certain input/output, to the overall operational integrity of the "automobile" system. This is a deterministic system which for a certain input (gas) will produce a certain output (distance/time). There are stochastic, non-deterministic elements, which can make the system act chaotically, such as a road pot holes, a leaking hydraulic line, a grounded electrical system, a broken rod, axle, etc. Any disturbance to a subsystem module can, negatively or positively, influence the behavior of the system as a whole.

The GIW watershed system model, resembles the automobile model in that it is a system which is deterministic and composed of modular subsystems. However, there is an aspect of the automobile-watershed analogy that is not parallel. The watershed system modeled by the GIW is recursive. Natural watershed systems are composed of similarly scaled-down watershed subsystems. Automobiles are not composed of small scaled-down automobiles. Indeed, this modular aspect models accurately "real world" watersheds.

The GIW model also proposes that each component watershed subsystem module has a corresponding river which is a constituent tributary of the watershed system trunk river. The trunk river system, in turn, is also recursive because it is replicated in each subsystem watershed tributary, i.e. each tributary discharge is composed of smaller tributary discharges (with the exception of 1st order rivers which discharge is a product of rill and sheet flow runoff). This recursive and modular aspect of rivers is upheld by geomorphic evidence (Morisawa, 1962). These concepts of the GENERAL INTEGRATED WATERSHED model watershed/river systems accurately represent "real world" natural systems.

As it was established by geomorphologists and hydraulic engineers, trunk river discharge is overwhelmingly a product of tributary discharges, and therefore the trunk river reflects the downstream, cumulative effect of tributary discharges. Each tributary and its discharge is a product of the fluvial dynamics (precipitation, substrate geology, substrate soils, land use: forests, agriculture, urban) of its correspondent watershed. Each tributary discharge, therefore, impacts the trunk river proportionally to the tributary's watershed composition and fluvial dynamics.

Applying patch dynamic principles, channel discharge is viewed as a system of contiguous patches of flow waters composed of different materials, elements, nutrients in solution and suspension. The point of discharge into the main trunk channel marks the upper boundary of any patch. Patch sizes and shapes are dynamic and fluctuate according to the magnitude of seasonal discharge, discharge composition, channel flow dynamics and parameters. These heterogeneous channel patch discharges, in turn,  differentially impact the biotic composition of the channel. However, the GIW model is deemed to be deterministic because the same input (precipitation) in a given watershed, will produce the same output (discharge).

The GIW model, therefore, is a deterministic recursive modular model, where trunk river watersheds are composed of independent (component tributary/watersheds subsystems) modules which produce variable-dependent proportional discharges, in magnitude and composition, into the trunk river. These modules are subject to their own local climatic, geologic, biotic, and anthropogenic land use regimes. Each of the modules can also be broken down, in a similar fashion, into a lower order level of component modules.

Based on the watershed hydrologic hypotheses and assumptions discussed above, the proposed GENERAL INTEGRATIVE WATERSHED MODEL  reflects the "real world" trunk river/watershed system and, therefore, it is a valid empirical model.



                                          IMPLEMENTATION OF THE

                             GENERAL INTEGRATIVE WATERSHED MODEL


Figure 26 shows the flow diagram for the GIW model. As shown, the model is composed of four main modules: the point/nonpoint module, n-order main trunk river module, the tributary module, and the patch module.

In the n-order trunk river module, head water watershed parameters for calculating discharge (area, precipitation, and other runoff coefficients), attribute data, biotic data (vertebrates, invertebrates, plants, etc.), and abiotic data (nutrients, water chemistry) is inputted into a main trunk system component where discharge (QR) is calculated. The output is discharge (QR), attribute area A, and biota population data.

The tributary module input consists of tributary watershed parameters used for calculating total tributary discharge (QT), plus attribute data, biotic and abiotic data. The tributary module output consists of tributary discharge (QT), total area per attribute (A) within tributary watershed, and tributary abiotic and biotic data.

The point source module is composed of inputs of type (organic and inorganic) of pollutants, concentration of pollutants, and amount of discharge per pollutant, attribute data (agricultural, industrial, urban waste discharges) and outputs are total point source discharge (QT), and attribute data.

The patch module is the main program module of the GIW model (shown in detail in Figure 27). As discussed before, large trunk river patches (LP) in a (n) order river are delineated by (n-1) tributaries. These large patches are, in turn, subdivided into smaller patches delineated by tributaries of £ (n-2) order, and point sources. The algorithm steps for the patch module are:

1) a (n-1) order tributary confluence sets the upper large patch (LP) boundary;

2) QT the discharge from the tributary as a proportion of the total discharge for the entire watershed is calculated (with data from GIS) according to adopted standard as discussed previously;

3) QR the cumulative discharge in the channel patch as proportion of the entire watershed (the sum of all tributary watersheds);

4) PQ = QT/(QR+QT) the tributary proportion ratio is calculated in a subroutine, this value represents the total discharge in a channel patch from tributary watershed provenance;

5) AA = the ratio of the tributary watershed covered by a given attribute (as derived using GIS) to the total area of the tributary watershed;

6) AQ = AA x PQ is calculated which represents the proportion of the channel patch discharge derived from tributary watershed attribute coverage runoff;

7) TA = (AA/100) x QT the tributary watershed attribute coverage as proportion of the entire watershed is calculated in a subroutine;

8) the cumulative value of QR is updated: QR = QR + QT;

9) AC = AC + TA is the proportion of cumulative discharge in the channel patch which was derived from a given attribute coverage runoff;

10) QA = (AC/QR) is the total proportion of discharge from an attribute runoff within the channel patch discharge;

11) all subsequent £ (n-2) order tributary, point, or non-point confluences or effluents set small patch (SP) boundaries;

12) recursively, QR = QR + QT, for the £ (n-2) order tributary and point source discharges are added to the trunk river discharges;

13) the PQ values are calculated for all subsequent intra-patch (£ (n-2) order tributaries and point sources) discharges and added to the (n-1) PQ value;

14) the AQ values are calculated for all subsequent channel intra-patch (£ (n-2) order tributaries and point sources) discharges and added to the (n-1) AQ value;

15) if the subsequent tributary is = (n) then it delineates the lower boundary of the large patch (LP), and a new upper boundary is set for the contiguous large patch system (LP);

16) if the subsequent tributary is > (n) then the trunk river has either become a tributary to a larger river system or it has discharged into a basin; output is produced, program stops.




patche model


Figure 26. Schematic of the General Integrative Watershed Model (GIW).




Figure 27


Figure 27. General Integrative Watershed Model (GIW) patch module.



The GIW model output includes:

1) a list ranking main trunk river (LP and SP) patches according to proportion of discharge ratio (PQ) values;

2) a list ranking main trunk river (LP and SP) patches according to the % of attribute affected discharge (AQ), per attribute (geology, soils, land uses);

3) a list ranking (LP and SP) patches according to abiotic parameters (water chemistry, in-suspension load, in-solution load, nutrients).

All pertinent data is georeferenced, and orders of tributaries are predetermined. The difference in composition (AQ), magnitude (QR), and proportion ratio (PQ) of the discharges between the upper and lower boundaries of any patch is such that,


AQ at lower patch boundary = AQ at upper patch boundary + AQ from tributary/point

QR at lower patch boundary = QR at upper patch boundary + QT from tributary/point

PQ at lower patch boundary = PQ at upper patch boundary + PQ from tributary/point


The GIW patch (LP and SP) output values are always the values at the lower patch boundary.



There exists a need for a large river model which would serve as a conceptual framework for environmental management. The are several a priori requirements which the model should meet:

1) The river model should be holistic in scale. Rivers should be seen as products of watershed-wide dynamics. It should, potentially, be able to integrate all the variables of a complex watershed system. 

2) Most of the large river systems in the world are culturally impacted. The model should be equally applicable to pristine rivers as to anthropogenically impacted systems.

3) The model should be general enough to be applied to rivers in any lithological, biological, and climatic regime.

4) The model should be mechanistic and standardized in its application. This allows for cross lotic systems comparisons.

5) The model should be of modular construction allowing for coarse- and high-resolution analysis of watershed biotic and abiotic dynamics.

6) The model should be readily adoptable for use by lotic environmental managers because of simplicity of application and ease of use with pre-existing commercial data-base, GIS software and hardware tools.


GIW CONCLUSION...(click here)