Principal Investigators: M. Thoms and R. Norris
Research Officers: M. Parsons and G. Ransom
Cooperative Research Centre for Freshwater Ecology
Monitoring River Health Initiative Technical Report Number 23
Environment Australia, 2002
ISSN 1447 1280
ISBN 0 642 54889 7
Assessment of stream condition using physical and chemical parameters will provide information on the physical and chemical health of streams, as well as providing information that is complementary to the AusRivAS biological assessments of stream condition. Thus, the development of a standardised physical and chemical assessment protocol will need to merge aspects of physical assessment with aspects of biological assessment. The following sections expand on the work included in the review of physical stream assessment methods (Parsons et al., 2000; and see Section 2.3) and describe the process used to derive the philosophy and content of the physical assessment protocol. Section 4.2.1 examines the criteria required for the physical and chemical assessment protocol. Section 4.2.2 then evaluates the performance of four existing methods of stream condition against these criteria. Section 4.2.3 makes recommendations about the inclusion of existing river assessment methods into the protocol and Section 4.2.4 outlines the major components that are to be included in the protocol, in light of the evaluation of existing methods. The methods of physical stream assessment included in the evaluation (River Habitat Audit Procedure, Index of Stream Condition, Habitat Predictive Modelling and River Styles) correspond with the methods presented at the Habitat Assessment Workshop. Each method is also examined in detail in the review document of Parsons et al. (2000).
Ideally, a standardised physical and chemical assessment protocol will:
In addition, compatibility of the protocol with AusRivAS will require:
The representation of these criteria within the five stream assessment methods examined at the habitat assessment workshop is summarised in Table 4.3. The relationship between each stream assessment method and each desired protocol criteria is discussed in Section 4.2.2.
Habitat Predictive Modelling (Davies et al., 2000) is clearly the most advanced method in this category. Habitat Predictive Modelling has the ability to predict the occurrence of local stream features from larger scale catchment characteristics. River Styles (Brierley et al., 1996) is the only other method with predictive capability. However, the predictions made in River Styles are based upon expert geomorphological interpretation of the behaviour of river systems and this knowledge is not easily transferable to a standardised physical and chemical assessment protocol. The Index of Stream Condition (Ladson and White, 1999) and the River Habitat Audit Procedure (Anderson, 1993) were not designed to have predictive ability.
| Criteria required for the physical and chemical assessment protocol | Existing physical assessment methods | |||
|---|---|---|---|---|
| River Habitat Audit Procedure | Index of Stream Condition | River Styles | Habitat Predictive Modelling | |
| Ability to predict the physical features that should occur in disturbed rivers and streams | N | N | P1 | Y |
| Ability to assess stream condition relative to a desirable reference state | Y | Y | Y | Y |
| Use of a 'rapid' data collection philosophy | Y | Y | N | Y |
| Use of physical and chemical variables that do not require a high level of expertise to measure and interpret | Y | Y | P2 | Y |
| Use of variables that represent the fluvial processes that influence physical stream condition | Y | Y | Y | P3 |
| Outputs that are easily interpreted by a range of users | Y | Y | N | Y |
| Applicability to all stream types within Australia | P4 | P4 | P4 | P4 |
| Incorporation of a scale of focus that matches the scale of biological collection within AusRivAS | Y | Y | P5 | Y |
| Collection of physical parameters that are relevant to macroinvertebrates | P | P | P | Y |
| Outputs of physical condition that are comparable to AusRivAS outputs of biological condition | N | N | N | Y |
All of the stream assessment methods have the capacity to assess stream condition against a reference state. However, each method differs in its delineation of that reference state. The River Habitat Audit Procedure (Anderson, 1993) uses a rating system where the condition of a site is expressed as a percentage of the original pristine values, functions or utilities that are retained. However, the designation of pristine values for each data component is somewhat subjective. For example, a riparian width of 50m is designated as pristine. Local reference sites can be used to re-scale the condition ratings to reflect inherent local or regional features, although this re-scaling process is still dependent on the original designation of pristine. The Index of Stream Condition (Ladson and White, 1999) also uses a rating system to assess stream condition. Reference values or ranges are defined for each indicator and allocated rating scores. For each site, the rating scores for each indicator are summed to produce an overall index. The rating scores reflect coarse differences between excellent, moderate and poor stream condition, however, it may not provide the resolution to reflect a continuum of stream condition with many states.
Delineation of the reference condition in River Styles (Brierley et al., 1996) is descriptive only. It is based on expert geomorphological knowledge of a river system in relation to its historical and catchment context. Reliance on expert geomorphological knowledge to determine reference condition is not conducive to a standardised physical and chemical assessment protocol. However, the placement of a river into its historical and catchment context may be an aspect of reference condition that is important when examining physical stream condition, and which is not addressed in detail by other methods.
Habitat Predictive Modelling (Davies et al., 2000) uses the same regional reference approach as AusRivAS. Reference sites are chosen to represent minimally disturbed conditions, on the basis of criteria such as land-use, absence of point source pollution, riparian vegetation, geomorphology and general catchment condition. These reference sites form the basis of the predictive model. The advantage of the regional reference site approach is that assessment of condition at a test site is derived from comparison with a group of reference sites with similar environmental conditions (Norris, 1994; Reynoldson et al., 1997). However, the Habitat Predictive Modelling study used the AusRivAS reference sites, which were originally chosen on the basis of characteristics related to biota. Habitat Predictive Modelling may be improved by selecting reference sites on the basis of characteristics that are related to physical stream condition.
The River Habitat Audit Procedure (Anderson, 1993) and Habitat Predictive Modelling (Davies et al., 2000) are designed specifically to use rapid data collection methods. The River Habitat Audit Procedure collects a comprehensive set of physical measurements and Habitat Predictive Modelling measures the physical parameters chosen for use in AusRivAS. For both methods, the amount of time spent at a sampling site is approximately one hour, although there is also an office data collection component. The Index of Stream Condition (Ladson and White, 1999) is also a rapid method and field information can be collected in approximately two hours. However, the collection of hydrology and biological data adds a substantial office component to this time. River Styles (Brierley et al., 1996) was not designed to fit rapid a sampling philosophy. However, many of the physical and geomorphological variables collected in River Styles have potential for adoption into a standardised physical and chemical assessment protocol (see Section 4.2.2.4).
The River Habitat Audit Procedure (Anderson, 1993) and the Index of Stream Condition (Ladson and White, 1999) are designed for use by a range of operators. As such, the physical and chemical variables are collected using standard and rapid techniques. For example, the River Habitat Audit Procedure includes variables such as bank condition and riparian vegetation characteristics, and the Index of Stream Condition includes sub-indices such as bed and bank stability and longitudinal connectivity of riparian vegetation. In the Index of Stream Condition, consistency of interpretation and measurement is achieved through the provision of reference photographs and in the River Habitat Audit Procedure, consistency is achieved through a two-day training program. Likewise, Habitat Predictive Modelling (Davies et al., 2000) was developed using the AusRivAS data, which included physical and chemical variables that could be easily measured and interpreted by a range of users.
River Styles (Brierley et al., 1996) aims to assess the geomorphological character of a river and thus, it includes specialised geomorphological measurements. Different sets of variables are measured at the catchment, reach and geomorphic unit scales, however, many of these variables correspond to those used in other methods. For example catchment area, reach vegetation cover, stream order and landuse are measured using standard techniques. The more specialised variables measured in River Styles include catchment elongation ratio, meander wavelength, planform geometry, width to depth ratio, stream power, bankfull discharge and sediment grain size distribution, among others. With appropriate operator training, most of these variables could be collected in the field by agency staff, using rapid data collection techniques. However, it is unclear whether these variables require specialised field measurement equipment.
The River Habitat Audit Procedure (Anderson, 1993) measures variables that essentially 'summarise' fluvial processes. Features such as channel shape, sediment particle size, aggradation and degradation of the bed, sediment compaction and instream bars are visually assessed. Thus, many of the River Habitat Audit Procedure variables represent important fluvial processes and utilise rapid collection methods. River Styles (Brierley et al., 1996) also measures fluvial process variables, but in a more specialised manner. As mentioned in the preceding section, the more specialised variables measured in River Styles include catchment elongation ratio, meander wavelength, planform geometry, width to depth ratio, stream power, bankfull discharge and sediment grain size distribution, among others. The direct process based nature of these variables suggests that they may be important inclusions in the physical and chemical assessment protocol.
The Index of Stream Condition (Ladson and White, 1999) is designed to be a holistic assessment of stream condition, on a scale that is commensurate with management concerns. The sub-indices used in the method represent the broad influences on stream condition and indicators within these sub-indices measure specialised aspects of condition. The hydrology, physical form and streamside zone sub-indices are the most relevant to fluvial processes. As with the River Habitat Audit Procedure these sub-indices are 'summary' rather than 'direct' measures of fluvial processes. Conversely, Habitat Predictive Modelling (Davies et al., 2000) was developed using the physical and chemical variables collected in AusRivAS, and thus, these variables predominantly reflect fluvial factors that are related directly to macroinvertebrates. However, Habitat Predictive Modelling contains scope to incorporate different types of variables and it is potentially possible to develop a model that contains fluvial process variables (e.g. River Styles), fluvial 'summary' variables (e.g. River Habitat Audit Procedure and Index of Stream Condition) and biologically relevant variables (e.g. AusRivAS).
The outputs of the River Habitat Audit Procedure (Anderson, 1993) can be simplified into a skeleton map or pie chart that shows the condition of river sections within a catchment. Likewise, the sub-indices of the Index of Stream Condition (Ladson and White, 1999) are added together to form an overall index score that represents a continuum of excellent to poor stream condition. Habitat Predictive Modelling (Davies et al., 2000) produces AusRivAS-style observed:expected ratios, however, the designation of these ratios into bands representing different levels of impairment has yet to be investigated for physical features. River Styles makes a detailed assessment of the geomorphological character and behaviour of a river within its catchment and historical context and thus, outputs are not summarised into an index or score. Maps of channel character under pre and post-disturbance conditions can be constructed for different rivers, although in comparison to the outputs of the other methods, these maps provide a more indirect, expert interpretation based assessment of stream condition.
Of the four stream assessment methods considered in this document, none are presently being applied Australia wide. Each method has potential applicability to a range of stream types, however, modifications to the variables collected and the delineation of reference condition may be required to reflect inherent local or regional variability in stream types.
The focus of the AusRivAS study design is on the reach scale. At each site, the length of the sampling reach is defined as 10 times the mean stream width. Macroinvertebrates are collected from different habitats (riffle, main channel, edge, macrophyte or sandbeds) within this reach. It is assumed that a macroinvertebrate sample collected from a 10m transect in each habitat will be representative of the fauna present in that habitat along the entire reach. In addition to macroinvertebrates, physical, chemical and habitat features are measured within the whole reach (e.g. riparian vegetation, substratum, water chemistry, HABSCORE) or within individual habitats (e.g. water velocity, substratum, FPOM, CPOM). A set of large-scale map measurements is also collected (e.g. landuse), along with a set of measurements that indicate geographic position of the sampling site (e.g. altitude, latitude, longitude, distance from source). Thus, the main scale of focus of AusRivAS is the reach scale, although macroinvertebrates are collected on a smaller scale within this reach, and physical, chemical and biological variables are collected across different scales. In addition, the statistics used to derive the AusRivAS predictive models are based on differences in the spread of riffle samples across the landscape.
The sampling focus of the River Habitat Audit Procedure (Anderson, 1993), Index of Stream Condition (Ladson and White, 1999) and Habitat Predictive Modelling (Davies et al., 2000) is also predominantly at the reach scale. The River Habitat Audit Procedure divides streams into homogeneous stream sections then samples representative reaches within each section. The Index of Stream Condition divides streams into reaches and then samples at three measuring sites within each reach. Habitat Predictive Modelling is based on the AusRivAS data and uses large-scale catchment characteristics to predict reach or individual habitat scale physical features. However, each method uses a different definition of a reach. Reaches within the River Habitat Audit Procedure are equivalent to two meander wavelengths or pool/riffle/run sequences. Reaches within the Index of Stream Condition are usually 10-30km long, and measuring sites within these reaches are 430m long. Habitat Predictive Modelling reaches are the same as AusRivAS reaches and are equivalent to 10 times the mean stream width. Despite differences in the definitions of a reach, the scale of measurement for these physical assessment methods is generally at a local-scale, in comparison to some larger scale context. This is commensurate with the scale of focus of AusRivAS.
River Styles (Brierley et al., 1996) has a multi-scale focus that is designed to encompass hierarchical controls on the behaviour of river systems. Catchment scale boundary conditions dictate the range of behaviour within sub-catchments, planform attributes describe river behaviour at the reach scale and structural attributes describe river character at a geomorphic unit scale. The multi-scale approach used in River Styles facilitates examination of geomorphological character and behaviour, and interpretation of stream condition in different sized 'pieces' of river systems. Although not commensurate with the scale of focus of AusRivAS, the multi-scale approach of River Styles may be a useful addition to the standardised physical and chemical assessment protocol because it encompasses the geomorphological processes that influence channel features at different hierarchical levels.
AusRivAS uses macroinvertebrates to assess stream condition. As such, the physical, chemical and habitat variables that are collected in AusRivAS are specifically related to the distribution and ecology of the fauna. While these variables may reflect the environmental requirements of macroinvertebrates, it is unclear whether they cover the full spectrum of factors that may indicate the condition of the stream from a physical perspective.
The relevance of the AusRivAS physical and chemical variables to physical stream assessment was examined at the habitat assessment workshop (see Appendix 1). Generally, some of the current AusRivAS variables are able to indicate aspects of physical stream condition. These include factors such as riparian vegetation, water chemistry and inorganic substratum. However, there are also other variables not collected in AusRivAS that are important for assessing physical stream condition. These include factors such as hydrology, channel complexity, channel morphology, valley characteristics and floodplain characteristics, among others. These factors are also important biologically, because they indirectly influence ecological processes or the provision of instream habitat. Thus, a standardised physical and chemical assessment module should include variables other than those collected in AusRivAS.
A comprehensive spectrum of variables that indicate physical stream condition are encompassed by the River Habitat Audit Procedure (Anderson, 1993), the Index of Stream Condition (Ladson and White, 1999), Habitat Predictive Modelling (Davies et al., 2000) and River Styles (Brierley et al., 1996). However, no one method contains all the variables relevant to physical stream assessment. Habitat Predictive Modelling is based primarily on the physical and chemical features important to macroinvertebrates, but also includes a set of larger scale predictor variables that are related to catchment scale morphology and processes. The Index of Stream Condition collects a range of physical and hydrological indicators that are related to physical stream condition and the River Habitat Audit Procedure collects a comprehensive set of physical stream features. River Styles measures variables related to geomorphological structure and process. Thus, the variables included in a standardised physical and chemical assessment protocol could be drawn from the different methods, to represent both biologically relevant and geomorphologically relevant factors.The main output of the AusRivAS predictive models is an observed:expected taxa ratio. The deviation between the number of taxa expected to occur and the number of taxa that were actually observed at a test site is a measure of stream condition. Other outputs from the AusRivAS predictive models include the probability of occurrence of individual taxa at a test site, the impairment band of the observed:expected ratio and an observed:expected ratio for the SIGNAL index. Habitat Predictive Modelling (Davies et al., 2000) is the only stream assessment method that matches AusRivAS outputs, because it can produce an observed:expected ratio that compares the local habitat features expected to occur at a site against the local habitat features that were actually observed at a site. Despite its demonstrated ability to predict habitat features, Habitat Predictive Modelling adapts a biological monitoring technique to the assessment of physical condition and thus, there are several analytical limitations associated with this method. In particular, Habitat Predictive Modelling currently predicts categories of habitat features, rather than continuous data. Investigations into new analytical techniques that can predict continuous data are currently under way. In addition, the relationship between actual instream condition and the observed:expected ratio needs to be investigated for physical data.