Professional Documents
Culture Documents
PROJECT 4:
STREAM ECOLOGY
Managing and harvesting urban stormwater for stream health
LITERATURE REVIEW
SEPTEMBER 2010
Executive Summary
Why are waterway ecosystems important?
Humans have always depended on the products and services of healthy river and stream
ecosystems, not only for water and waste disposal, but for the important food and fibre sources
provided by their plants and animals. Rivers and streams play other important roles, particularly the
retention and transformation of materials and pollutants which are generated in their catchments
(Alexander, et al., 2000). They can help to protect downstream estuaries and bays from excessive
nutrients and are thus vital in protecting important coastal fisheries.
Humans also value waterways for their intrinsic and biodiversity values. Freshwaters, despite only
making up 0.009% of the world‘s water, contain at least 40% of the world‘s fish species and a quarter
of all vertebrate species (Millennium Ecosystem Assessment, 2005). Unfortunately, the rate of
extinction in freshwater systems is greater than in any terrestrial ecosystem (Dudgeon, et al., 2006).
The degradation of waterways through urbanisation has resulted in the loss of many riverine species,
and threatened human wellbeing and livelihoods through phenomena such as algal blooms and
toxicant accumulation. The natural form of urban waterways has typically been altered, either directly,
or though changes in hydrology, which acts as a ‗master variable‘ driving patterns and processes in
waterways (Poff, et al., 1997).
So why should we protect and attempt to rehabilitate and restore our urban waterways? The intrinsic
and human values described above provide, in themselves, a compelling rationale. Healthy streams,
rivers, estuaries and coastal waters are a fundamental prerequisite to achieving sustainable
management of our cities; our streams must be maintained in a healthy state for not only this
generation, but for future generations (United Nations General Assembly, 1987).
However, there are also some simple economic arguments for taking action to prevent stream
degradation. Preventing decline through appropriate stormwater management will be considerably
less expensive than the cost of restoring already degraded waterways (Rutherfurd, et al., 2000).
Fortunately, the integration of stormwater harvesting and other carefully-designed stormwater
management provides a great opportunity to protect and restore urban waterways.
Hydrology
Urbanisation profoundly changes the hydrologic cycle, increasing the volume, frequency and flow rate
of storm flows. Evapotranspiration is also greatly reduced, decreasing from around 80% of mean
annual rainfall in the pre-developed situation to around 15% for impervious areas. The increases in
runoff result in increased frequency and magnitude of disturbance to channel substrates and to
aquatic ecosystems. Whilst the impacts of urbanisation on stormflows are generally well recognised,
the impacts on low-flow hydrology are less well understood, and yet have critical impacts on both
the ecological and social value of waterways. The loss of baseflow due to decreased infiltration
caused by impervious areas, results in once perennial streams becoming ephemeral, and has
consequent impacts on water quality, particularly during summer periods.
Many studies have investigated potentially useful indicators for use in assessing the catchment-scale
impacts of urbanisation. Such indicators need to have ecological relevance and be able to be
measured and modelled. They should be sensitive to the urbanisation gradient, rather than external
A number of studies have shown that pollutant concentrations are well predicted by impervious areas,
but importantly, Hatt et al‘s (2004) study showed that the effective imperviousness of a catchment
gave the best prediction of a wide range of pollutants, indicating the direct role which stormwater
discharge has on receiving waterway water quality. A new approach to the design of urban drainage
systems is thus required if we wish to avoid water quality degradation due to urbanisation.
The selection of water quality indicators must take into account the nature of the waterway. For large
systems, loads targets provide a suitable measure, whilst concentration targets (such as those
provided by ANZECC or local SEPPs) provide the most ecologically-relevant target for smaller flowing
waterways, which have limited buffering capacity.
Geomorphology
The natural geomorphic function of urban waterways is severely impacted by the increased volume of
water generated by the urbanised catchment. The increased flow removes much of the mobile
sediment load and preferentially erodes the channel. With conventional urban hydrology there is
limited potential to replace (and retain) mobile sediments and hence the natural functioning, namely
erosion and deposition in a form of quasi-equilibrium. This includes ubiquitous features of a ‗natural‘
channel such as benches and bars. Unless we reduce those flows responsible for mobile sediment
transport returning much of this functioning is not feasible, necessitating the protection of artificially
‗static‘ channels with hard or soft engineering works.
In addition, the increased channel capacity of urban streams confounds the hydrologic alterations
induced by the urban catchment. Reducing channel capacity and re-engaging the floodplain will
reduce the accelerated channel degradation resulting from both the incised channel and altered
hydrologic regime. Within the urban environment space is the greatest constraint where ‗internal
floodplains‘ may be required to alleviate impacts on the channel while preventing flooding into the
development boundary.
Ecology
Given the impacts of urbanisation on hydrology, water quality and geomorphology, it is n surprise that
there are major consequent impacts on the ecological values of streams. While the underlying
mechanisms are complex, they can be classified into catchment-level impacts (largely hydrology and
water quality) and local impacts (largely changes in organic matter supply). The water quality and
hydrologic impacts described above cause increases in the frequency of both hydraulic and water
quality disturbance, change the structure of in-stream habitat, and directly disturb biota.
The majority of ecological impacts are therefore a result of catchment scale processes, whilst local-
scale impacts (e.g. riparian vegetation) may play a role in some circumstances. However, catchments
with even a small coverage of connected imperviousness are likely to have sufficiently altered
hydrology and reduced water quality that even extensive riparian restoration works may show minimal
improvement in stream ecological condition.
Ecological responses to catchment urbanisation can be measured using a range of indicators. For
example, primary producers (e.g. algae) will increase as a result of increased nutrient concentrations
(Hatt, et al., 2004; Newall & Walsh, 2005; Resh, et al., 1988; Sonneman, et al., 2001). Ecosystem
Perhaps the most commonly used indicators have been macroinvertebrates, given their sensitivity to
water quality changes and the presence of well-established protocols for comparing impacted and
reference populations. Urban streams typically, show an increased abundance of ‗pollution tolerant
taxa‘, with reductions in the more ‗pollution sensitive taxa‘ including Ephemeroptera, Plecoptera and
Trichoptera (Paul & Meyer, 2001; Suren & McMurtrie, 2005; Walsh, 2004). Urban streams are also
typified by organisms with rapid life histories, such as oligochaetes and chironomids. These taxa are
highly tolerant to hydrologic disturbance (both in terms of high flows and drying) because they spend
a relatively short time as an aquatic larva. Fish have also been used as indicators, with urbanised
streams typically having much less diverse fish populations (Arthington, et al., 1983).
Importantly, because a wide range of ecological indicators are highly sensitive to the impacts of
stormwater runoff, significant response by their indicators is seen at even very low levels of
urbanisation. This makes these indicators very useful in assessing the degree to which urbanisation
is impacting on the waterway ecosystem, but, conversely, means that rehabilitation efforts (e.g.
through stormwater management and harvesting) will not show an ecological response until the
urban-runoff impact has been reduced to very low levels. Whilst stormwater managers may
understandably wish to have a more linear (and thus progressive indicator), no studies to date have
been able to identify any. Given the role of indicators in identifying the ecological condition of
waterways, the search to find a more progressive indicator, while perhaps helpful in terms of
garnering political and social support for interventions, will probably provide little ecological insight.
The sharply non-linear response of currently indicators accurately identifies the situation; streams are
ecologically degraded wherever significant stormwater inputs are permitted to occur.
As has been described above, the reverse is true for urban stormwater; the frequency and volume of
stormwater runoff into waterways is far greater than in the pre-developed state. Urban stormwater
runoff, delivered through conventional drainage systems is a complex environmental flow problem
that can, in large part, be solved through harvesting of stormwater before it reaches aquatic
ecosystems. This is an unusual ―win-win‖ situation, in that retention of that water for human uses will
have a positive environmental outcome in protecting the waterway. However, it is critical that urban
stormwater harvesting systems harvest water before it enters the waterway, rather than harvesting
water from the waterways, leaving flow and water quality disturbance upstream unaddressed.
Stormwater harvesting can – and should – thus be applied to help restore flow regimes and water
quality. To do so requires careful integration of harvesting (which has the potential to reduce flow
volume and peak discharge rates) with carefully chosen techniques to restore the low-flows which
have been lost due to urbanisation. A wide range of WSUD techniques, such as vegetated infiltration
and biofiltration systems, can be used to restore both the evapotranspiration and infiltration fluxes.
Interim recommendations
We propose the following initial set of ecologically-based principles for the design of integrated
stormwater management, incorporating both stormwater harvesting and measures focussed at
restoring baseflows and water quality:
1. Minimize uncontrolled storm flows to the pre-developed runoff frequency.
2. Restore baseflows, by delivering filtered flows to the stream through treatment measures that
ensure flow rates do not exceed pre-urban subsurface flow rates. Appropriate maximum flow
rates can be estimated from baseflow separation analysis in reference streams, or by
assessment of infiltration capacities of native soils in the catchment.
Figure S1. Annual volume of runoff from 1 ha of impervious surface, partitioned into two parts: the volume that needs
to be passed through filtration systems to restore lost subsurface flows (grey polygon), and the volume that needs to
be retained in the catchment and not delivered to the stream (through evapotranspirational loss or through use and
export from the catchment through the wastewater stream (adapted from Zhang et al. (2001).
Hydrological effects, scales and integration; how should stormwater harvesting be integrated with
other stormwater management measures to restore all components of the flow regime? At what
scales should these systems be located? To what extent do flow regimes needs to be restored in
order to see an ecological response? What is the ecologically appropriate flow regime in the context
of a channel enlarged by urbanisation?
Impacts on water quality regimes; what will be the consequences of harvesting at different scales
on the water quality regime in the receiving waters, and what ecological consequences will that have?
Will different waterway types have varying tolerance to harvesting-induced water quality changes? Is
catchment-scale water quality treatment enough, or is floodplain re-engagement necessary to restore
pre-development nutrient retention of urban streams, in order to protect downstream waterways?
Geomorphic processes and consequences; what is the effect of urbanisation on sediment budgets
and what effect will stormwater harvesting have on these budgets. Do we need to restore the
sediment regime? How does the reduction in mobile sediments impact on geomorphic functions of
the channels? What level of dynamism is acceptable or feasible for an urban stream, and how can
we design our stormwater system to facilitate this ‗desirable‘ level of dynamism. Will channel
intervention be necessary, or will restoring pre-development flows be enough to allow channels to
Ecological responses and thresholds; What mitigating factors might explain variations in the level
of ecosystem response to urban impacts? How will stream ecosystems respond to stormwater
harvesting? How should we monitor and report these responses in ways that are ecologically
meaningful?
To use urban stormwater management and harvesting to maintain the stream in its intact
natural state
Driving factor indicators: maintain natural flow volume and variability and pollutant concentrations
and loads; natural levels of erosion and sediment deposition, supply and retention of coarse
particulate organic matter (CPOM), natural frequency of floodplain engagement, adequate riparian
vegetation buffer width and condition
Ecological indicators: achieve presence of sensitive species (e.g. high EPT, riparian plants), natural
levels of primary and secondary productivity and natural levels of nutrient and sediment retention
(with respect to reference conditions).
Driving factor indicators: reduce flow volume, magnitude, duration and frequency of stormflows,
restore baseflows and reduce pollutant concentrations and loads; retain appropriate levels of mobile
sediment, reduce erosion and restore near-natural sediment supply levels, restore supply and
retention of CPOM and as much as possible the natural frequency of floodplain engagement, maintain
an adequate riparian vegetation buffer.
Ecological indicators: achieve presence of sensitive species (e.g. high EPT, riparian plants),
reduced levels of primary and secondary productivity and increased retention of nutrient and sediment
(with respect to reference conditions).
We hypothesise that the ‗initial principles‘ guiding the design and operation of stormwater harvesting
systems, integrated with other WSUD measures aimed at restoring pre-development water quality
and hydrology, will be capable of delivering this vision. The Cities as Catchments Project, and in
particular the opportunity to monitor real systems, gives us an ideal framework in which to test this
hypothesis.
References ............................................................................................................................................ 47
Table of Tables
Table 1. Hydrologic indicators suitable for assessment of urbanisation impacts, as suggested by
DeGasperi et al. (2009). ........................................................................................................................ 10
Table 2. Hydrologic indicators used by Fletcher et al. (2007) for evaluation of stormwater harvesting
impacts. ................................................................................................................................................. 13
Table 3. Typical urban stormwater pollutants and sources (after: Duncan, 1995; Lawrence & Breen,
2006). .................................................................................................................................................... 15
Table 4. Summary of typical water quality values for runoff from urban, agricultural and forested
catchments, urban streams, and secondary treated sewage. .............................................................. 16
Table 5. Example of pollutant load reduction targets (Source: Wong, 2006). ..................................... 17
Table 6. Summary of ecological responses to urbanisation currently reported in the scientific
literature. ............................................................................................................................................... 31
The animals and plants of rivers and streams historically provided important food and fibre sources for
human settlements, and this remains true in many parts of the world, such as the great basins of the
Mekong and the Amazon.
Riverine ecosystems are also important for their retention, transformation and export of materials that
are generated by catchment physical and biological processes. The transport of sediments in rivers is
a central determinant of landforms (Knighton, 1998) and the productivity of estuaries and coastal
waters is strongly determined by the input of nutrients from rivers, particularly during floods
(Loneragan & Bunn, 1999). Thus coastal fisheries are dependent on the health of their receiving
rivers and streams.
However, the health of coastal waters and the productivity of coastal fisheries depend on an
appropriate balance of nutrient delivery. Eutrophication resulting from excess nutrient export from
degraded rivers is a threat to many coastal waters globally, particularly those downstream of large
cities (e.g. Harris, et al., 1996; Palmer, 2004). Small streams and their riparian zones are critical parts
of the landscape for the retention of nutrients to prevent excessive export, particularly nitrogen
(Alexander, et al., 2000), primarily because a large proportion of the water they carry flows through
channel and riparian sediments that are hotspots for critical nutrient transformation and removal
processes such as denitrification (McClain, et al., 2003). In addition to protecting downstream waters,
this important capacity of healthy streams to retain and transform contaminants also serves to purify
the water in which riverine plants and animals live (and which humans extract).
Yet, river ecosystems, so central to human well-being, carry only 0.0002% of the world‘s water, and
cover less than 0.1% of the world‘s surface (Millennium Ecosystem Assessment, 2005). Given their
relatively small coverage of the planet, the diversity of life in rivers is astonishing. Freshwaters (which
includes lakes as well as rivers, making up 0.009% of the world‘s water) are home to at least 40% of
the world‘s fish species, and a quarter of all vertebrate species. The taxonomy of freshwater biota is
woefully understudied compared to terrestrial biota, yet they still account for 6% of all described
species (Dudgeon, et al., 2006). While the extraordinary biodiversity of rivers and lakes is not as
widely appreciated as that of, say tropical rainforests and coral reefs, the threats and rates of
extinction are much greater in freshwaters than in any of the most affected terrestrial ecosystems
(Dudgeon, et al., 2006).
In part, the increased vulnerability of freshwater biodiversity and freshwater ecosystems in general, is
a result of their position in the landscape. At the bottom of valleys, they are the natural destination of
the products of land use change. Even in the absence of land use change, the biodiversity of rivers
and the ecological processes that support it (and human civilization) depend strongly on physical
processes originating in the catchment, with processes in the riparian zone and channel having the
most immediate influence.
This narrow, utilitarian approach to rivers has resulted in a grand diminishment of the benefits that
rivers provide the world as healthy ecosystems, such as their remarkable biological diversity, and the
retention and treatment of nutrients and other contaminants. Poor management of the world's rivers
Historically, the development of cities has followed a pattern of initial dependence on freshwater from
the local river, increasing use of the river for disposal of wastes, compromising its use as a water
source, followed by the implementation of engineering solutions to source more distant water for the
city. With increasing affluence, modern cities have introduced environmental regulation to limit
wastewater disposal into rivers, but the use of rivers for the disposal of urban stormwater remains the
norm (Walsh, 2000).
Improved environmental regulation has resulted in improvement in the aesthetic appearance and
reduced health risks associated with primary contact of urban rivers. However, the continued,
unregulated discharge of urban stormwater to rivers has largely limited any improvement in urban
rivers to aesthetics and reduced human health risk (Walsh, 2000).
Urban dwellers value the parklands and open space afforded by urban streams (Lackey, 2001;
ResearchWise, 2004). Increasingly, cities are turning to their rivers as attractive open space, but the
rivers in most cases are diminished in their ecological structure in function by catchment processes,
and local engineering of their banks. The protection or restoration of biodiversity or ecological function
in rivers with substantial catchment urbanisation, to a condition close to that of streams with less
intense human land use, has yet to be achieved.
The renewed and growing enthusiasm of the world‘s urbanites for their rivers and streams is
encouraging. However, the perception of today‘s generations of improved urban rivers arises from a
starting point of sick polluted rivers of 20-30 years ago. A lowering of expectations of the environment
over generations is a common occurrence because the rate of environmental change exceeds
generation times (Pauly, 1995). We must therefore be careful to set objectives for environmental
improvement based on what is achievable rather than what is perceived by the community as
acceptable.
The physical setting of a river strongly influences how the flow regime translates into hydraulic
conditions experienced by, and available to, the riverine biota (Poff, et al., 2010). The translation of the
flow regime into small-scale hydraulic conditions has important implications for ecological processes
(Newall & Walsh, 2005). Indeed, Stratzner and Higler (1985) state that ‗physical characteristics of flow
(“stream hydraulics”) are the most important environmental factor governing the zonation of stream
benthos on a world-wide scale‘ (pg. 127).
The abundance and diversity of aquatic biota is closely associated with geomorphic aspects of
channel morphology and functioning (Bledsloe, 2002). Physical features such as riffles, pools and
bars, and channel characteristics such as width, depth and sinuosity drive the translation of flow
regime into hydraulic conditions. For example, greater variation in bed elevation will provide fish and
invertebrates with longer pool persistence during drying periods and refuge during higher ‗disturbance‘
flows. Local characteristics such as channel geometry, floodplain height and streambed composition
are determinants of the impact of events, such as whether a given flow will create a bed-moving
disturbance or an overbank flow (Poff et al. 2010).
Natural channels are inherently dynamic in their physical form. Their characteristics vary over time and
space with changes in environmental controls and channel morphology being a function of the flow
and sediment regime delivered to the channel (Gilvear, 1999; Knighton, 1998). Not only do channel
features respond to these inputs, and translate flow into erosive or transportational components, but
the channel itself is a deformable boundary. The altered boundary then impacts differently on
sediment or velocity.
The health, biodiversity and human value of urban waterways is also driven by water quality, which
has both short-term and long-term effects. Water quality is strongly linked with hydrological changes;
increases to runoff coefficients will result in increased mobilisation and transport of contaminants from
the catchment into the receiving waters. Aquatic organisms are typically sensitive to elevated levels of
toxicants such as heavy metals, as well as the accumulation of excessive nutrients, which can cause
eutrophication and the development of (often toxic) algal blooms (Novotny & Olem, 1994; Novotny &
Witte, 1997). Over the longer-term, pollutant accumulation in sediments can result in continuing
degradation of the ecosystem, even after the pollutant source has been addressed.
However, there are also strong economic arguments for the protection of the existing values of urban
waterways. Considerably less investment is required to protect or prevent decline than is required to
restore already degraded waterways (Rutherfurd, et al., 2000).
In the case where we are uncertain as to the value of ecological health in urban streams, based on
current knowledge and values, the precautionary principle should be adopted. This is analogous to
environmental flows for rural streams where the dramatically increased understanding in the last
decade has brought considerable investment and attention to the issue. In many cases the actions are
too late for effective action, particularly actions which are affordable.
In the case of urban streams, we suggest that there is strong evidence that stream and river
ecosystems that support biota and ecological functions similar to streams with little human impact is
possible; that the (generally low) expectations and perceptions of urban dwellers of their streams can
be exceeded. Such a future is dependent on new approaches to stormwater management that
address the important drivers of degradation, as identified above.
Ideally the only difference between these two visions is time. The ultimate hope would be to return all
urban streams to a condition near to their pre-developed state, but clearly this will only be possible
over very long timeframes. The second vision thus sets out the goals for the condition of currently
degraded streams in the foreseeable future, without precluding future rehabilitation to a close to intact
natural state.
To use urban stormwater management and harvesting to maintain the stream in its intact
natural state
Driving factor indicators: maintain natural flow volume and variability and pollutant concentrations
and loads; natural levels of erosion and sediment deposition, supply and retention of coarse
particulate organic matter (CPOM), natural frequency of floodplain engagement, adequate riparian
vegetation buffer width and condition
Ecological indicators: achieve presence of sensitive species (e.g. high EPT, riparian plants), natural
levels of primary and secondary productivity and natural levels of nutrient and sediment retention (with
respect to reference conditions).
To use urban stormwater management and harvesting to re-establish ecological processes and
patterns characteristic of the stream in its natural state
Driving factor indicators: reduce flow volume, magnitude, duration and frequency of stormflows,
restore baseflows and reduce pollutant concentrations and loads; retain appropriate levels of mobile
sediment, reduce erosion and restore near-natural sediment supply levels, restore supply and
retention of CPOM and as much as possible the natural frequency of floodplain engagement, maintain
an adequate riparian vegetation buffer.
Ecological indicators: achieve presence of sensitive species (e.g. high EPT, riparian plants),
reduced levels of primary and secondary productivity and increased retention of nutrient and sediment
(with respect to reference conditions).
In a natural catchment, the vast majority (typically 80-95%) of precipitation which falls on a catchment
will be evapotranspired back to the atmosphere (Figure 1, Figure 2), with only a relatively small
percentage (typically 5-20%) resulting in streamflow (Argue, 2009; Zhang, et al., 1999). The
streamflow coefficient (proportion of catchment rainfall which ends up as streamflow) will depend on
factors such as slope, soil and vegetation type and of course climate. Evapotranspiration will occur
both in the soil and in the vegetation canopy, as a result of interception, where precipitation is
Figure 1. Relationship between annual rainfall and evapotranspiration for a range of vegetation types (Source: Zhang,
et al., 1999).
Figure 2. Typical water balance in natural and 50% impervious urbanised catchment (Base diagram adapted from
FISRWG, 1998).
In an undisturbed catchment, the frequency with which surface runoff will occur is limited by the
‗sponge effect‘ of the catchment soils (as well as interception by vegetation). For example, Hill et al
(1998; 1996) suggest that initial loss values (the amount of rainfall required before runoff is produced)
range from around 10 to 50 mm, with a typical value of around 25 mm. Not surprisingly then, natural
catchments produce direct surface runoff to streams very rarely. This has both hydrological and water
quality implications, since in a natural catchment runoff events capable of mobilising and transporting
sediments and pollutants in waterways happen very rarely (Walsh, et al., 2005a; Walsh, et al., 2009).
The effects of urbanisation on the urban water balance are well documented, with the key
observations being major reductions in evapotranspiration along with major increases in both the
frequency and volume of runoff (Fletcher, et al., 2007; Walsh, et al., 2009). Typically a reduction in
groundwater level is also observed although this may be obscured by loss of evapotranspiration (due
However, the impacts are far more complex than a simple increase in flow rates, with the following
changes also commonly observed:
typically decreased dry weather flows (often referred to as baseflows). The magnitude of this
change will depend on the balance between loss of infiltration and loss of evapotranspiration
within the catchment;
more frequent surface runoff inputs (with consequences for pollution inputs) as a result of the
decreased initial loss (from around 25 mm in a natural catchment to around 1 mm in the
urbanised catchment);
decreased lag-time (time of concentration) and recession time;
increased frequency and duration of dry spells (cease-to-flow periods); and
Perennial streams becoming ephemeral (due to the loss of baseflows).
These changes are conceptually represented in Figure 3. Storm events, whilst becoming more
frequent, decrease in duration, with shorter lag-time and recession (Burns, et al., 2005). Urban
catchments are thus typically referred to as ‗flashy‘. Urban catchments feature a reduced capacity to
attenuate runoff because storm runoff is quickly routed to receiving waters, indicated by relatively
smaller lag times compared to non-urban catchments (Leopold, 1991).
Figure 3. Schematic illustration of the pertinent impacts of urbanisation on hydrology at the catchment scale
(Marsalek, et al., 2007).
In addition to its impacts on stormflows, urbanisation impacts the low flow aspects of the hydrologic
regime. DeGasperi et al. (2009) showed that in the Puget Lowland (Washington, USA) urbanisation
has likely increased the frequency of days in which streamflow is less than 50% of the annual average
streamflow. Konrad (2000) reached a similar conclusion, proposing that urbanisation increases the
frequency of streamflows below the annual average streamflow. DeGasperi et al. (2009) suggest that
the duration of these low flows is decreased following urbanisation.
There does appear to be more consistency regarding the impact of urbanisation on winter baseflow.
Konrad and Booth (2005) showed that with increasing urbanisation, winter baseflow decreased,
implying that winter baseflow is less influenced by catchment physiographic characteristics than is
summer baseflow.
Figure 4 provides a schematic summary of the impacts of urbanisation on flow regimes, using real
data from two catchments – Olinda Creek (with very low levels of urbanisation) and Brushy Creek
(with moderate levels). The impacts of even this relatively low level of imperviousness on peak flows
are obvious, as is the reduced storm recession time and the reduction in both summer and winter
baseflow.
Generally, a hydrologic indicator reveals some information relating to a component of the hydrologic
regime. The various components of the hydrologic regime include: volume, frequency, duration, rate
of change, flashiness, magnitude and timing.
Richter et al. (1996) and subsequent studies (Richter, et al., 1998; Richter, et al., 1997) describe 32
different hydrologic indicators. DeGasperi et al. (2009) adopted these hydrologic indicators to
evaluate the hydrologic impacts of urbanisation on streams in the Puget Lowland (Washington, USA).
They found that only some were significantly correlated to both benthic index of biological integrity
scores and measures of urbanisation (Table 1), making them potentially sensitive metrics of the
impacts of urbanisation on receiving water health.
Component
of Hydrologic
Definition Significance
hydrologic indicator
regime
A low flow pulse is defined as a daily flow The number of low pulse counts
observed which is below or equal to a set increased for a rapidly urbanizating
threshold (estimated based on pre- catchment. Moreover, a negative
developed data). The threshold is set at 50 correlation between benthic index of
Low pulse
percent of the long-term daily average flow. biological integrity scores and low pulse
count This indicator is the number of times in each count was observed. Finally, the number
calendar year that distinct low flow pulses of low pulse counts for a forested
occurred. The units of this indicator are catchments was around three times as
count. less compared to urbanised catchments.
The choice of indicators requires careful consideration of the catchment characteristics. In particular,
catchment size will have an important influence, since, for example, larger catchments are likely to
have less variable flows. In other words, care should be given to the use of indicators which are
related to variables besides those relating to urbanisation (e.g. catchment area, annual rainfall etc).
Similarly, a small upland ephemeral (i.e. with streamflow during only part of the year) stream will have
very different ecological dependency on flow than a lowland perennial (permanently flowing) stream.
Baker et al. (2004) developed the Richards-Baker Flashiness Index (Equation 1). This hydrologic
indicator, measures the degree to which streamflow varies throughout a period of interest. For
example, the R-B Index for pristine streams is typically close to zero since streamflow tends to
gradually rise and fall throughout any given year. In contrast, the R-B Index for urban streams is
usually close to one because the capacity of urban catchments to attenuate surface runoff is low.
Indeed, Bressler et al. (2009) showed that the R-B Index is significantly correlated with measures of
Equation 1
Where qi is the daily streamflow, qi-1 is the daily streamflow on the previous day and n is the number of
days throughout the period of interest.
Konrad (2000) proposed the hydrologic indicator T Qmean which describes the fraction of time during a
water year that the average daily flow rate is greater than the annual average daily flow rate of that
year. The value of TQmean is negatively correlated with measures of urbanisation. For example, Konrad
(2000) showed that for streams in the Puget Lowland (Washington, USA), the difference between the
average TQmean for urban streams and that for non-urban streams, was statistically significant.
Furthermore, Booth et al. (2004) showed that benthic index of biological integrity (B-IBI) scores were
positively correlated with TQmean. For example, streams with T Qmean values of around 0.35 or greater,
tended to feature a good diversity of macroinvertebrate species.
While hydrologic indicators such as the R-B Index and TQmean describe daily fluctuations in streamflow,
other indicators relate more to floods. For example, Konrad (2000) proposes the hydrologic indicator
CVAMF, which measures the coefficient of variation of annual flood size, and the T xyr indicator, which
measures the frequency of time when streamflow is greater than the flood with a magnitude of 1 in X
times per year. This indicator was found by Konrad (2000) to be well correlated with aquatic
ecosystem health (as measured by the B-IBI score).
Kennen et al. (2010) explored the hydro-ecological relationships of 67 streams in north-eastern USA,
using 171 hydrologic indicators to characterize the hydrologic regime of their study streams. Similar to
Booth et al, they found that only a small number of these critically influenced ecological condition.
These most pertinent indicators included:
The mean of all April flow values over the entire record (this criticality of seasonal flows may
relating to the timing of important biological events);
Low flood pulse count. Computed as the average number of flow events with flows below a
th
threshold equal to the 25 percentile value for the entire flow record (number of events/year);
High-flow frequency. Computed as the average number of flow events with flows above a
threshold equal to 75 percent exceedence value for the entire flow record (number of
events/year); and
The mean of the minima of all April flow values over the entire record.
In a review of hydrologic indicators, Olden and Poff (2003) suggest that a small number of indicators
can robustly characterize a streamflow regime, proposing indicators including pulse count, mean
annual runoff, changes in flow and flow minima over given periods. Kennen et al. (2008) explored a
range of ecologically relevant hydrologic indicators at 856 monitoring sites in New Jersey, United
States. They suggested that the indicator NSTORM (average number of storms per year which
produce quickflow) was significantly correlated to macroinvertebrate assemblage structure, due to its
impact on habitat and channel form. Steuer et al‘s (2010) study of 83 hydrologic condition metrics took
a similar approach, relating the metrics to algal, invertebrate and fish communities across the USA.
Whilst a number of metrics were found to be relevant, the frequency of high-flow events
(events/month, where an event is defined as when flow exceeds 9x median flow) was found to best
explain ecological condition. Not surprisingly, Steuer et al also found that the hydrological indicators
were themselves explained by catchment urbanisation. They note, however, that their study of the
relationship to catchment land use was limited by not having access to connected imperviousness
data rather than total imperviousness.
Whilst catchment-scale indicators are necessary to understand the relationship between hydrology
and receiving water condition, the impacts of urbanisation on catchment hydrology will be managed
across the catchment, at a range of scales. For stormwater managers – and for the designers and
operators of stormwater harvesting systems, there is thus a need to understand how these catchment-
scale hydrologic indicators are impacted by changes to stormwater flows at the site-scale (be it
allotment, streetscape or precinct).
Wong et al. (2000) continued this line of enquiry, comparing the consequences of maintaining natural
drainage channels with that of lining or piping some or all drainage channels within the catchment.
They concluded that 80-90% of the increase in peak flows is explained by the nature of the drainage
connection, rather than simply the proportion of the catchment which is impervious. However, whilst
effective imperviousness (EI) might be a useful landscape indicator, it remains an oversimplification,
as it categorises impervious areas in a binary manner, as either connected or unconnected. A given
impervious area may contribute runoff in some storms and not others. So, runoff after a small storm
from a particular non-connected surface may not wet its surrounding pervious surfaces sufficiently to
initiate overland flow to the nearest waterway or to result in discharge from the nearest sealed,
connected drain (Walsh, et al., 2009). However, from a larger storm, runoff may occur either from
rainfall intensity greater than the soil infiltration or from saturation of the soil, and the consequent
runoff may be sufficient to form a direct hydraulic connection between the unconnected surface and
the waterway.
In response to the need to develop a more sophisticated approach, several studies (Ladson, et al.,
2005; Ladson, et al., 2006; Walsh, et al., 2005a) identified the frequency of runoff as a useful indicator
of stormwater hydrologic impacts, reasoning that in a natural catchment runoff frequency is limited by
interception, infiltration and large ‗initial losses‘ (Hill, et al., 1998; Hill, et al., 1996). For example,
Walsh et al [, 2005 #62} used empirical evidence to propose that the frequency of stormwater
discharge explained variation in ecological condition of 16 catchments. In 2009, Walsh et al refined
the runoff frequency metric, proposing retention capacity, which measures the capacity of a given
stormwater management measure to reduce the runoff frequency back to the pre-urban frequency.
Whilst such frequency-based indicators are easy to understand, measure and model, they measures
only one aspect of the hydrological disturbance caused by imperviousness, ignoring changes in
volume or changes in low flow hydrology.
In 2007, Fletcher et al. proposed a wide range of indicators, in a modelling study of the potential
impacts of stormwater harvesting on catchment hydrology. They undertook modelling at a sub-
Table 2. Hydrologic indicators used by Fletcher et al. (2007) for evaluation of stormwater harvesting impacts.
Analysis
Category Indicator (and abbreviated name) Unit
timestep
Total runoff daily ML/yr
Runoff
Frequency of surface runoff daily times/yr
Duration (total time of low flows) daily days/yr
Average length of low-flow spells daily days in a row (average/yr)
Number of low-flow events daily events/yr
Flow Spells
Duration (total time of high flows) daily days/yr
Average length of high-flow spells daily days in a row (average/yr)
Number of high-flow events daily events/yr
3
Q1month hourly m /sec
3
Q3month hourly m /sec
3
Peak Flow Q1year hourly m /sec
3
Q1.5year hourly m /sec
3
Q5year hourly m /sec
Flow Duration
Integral of the flow duration curve hourly Integral of curve
Curve
In the most recent developments in this area (Fletcher, et al., in press; Walsh, et al., 2010), a more
integrated set of site-scale indicators (all measured relative to the pre-urban state) has been proposed,
including:
1. The frequency of runoff discharged directly to waterway (i.e. the frequency of high-flow
disturbance);
2. The volume of subsurface (filtered flow) discharged from the site (i.e. the volume of baseflow
contribution); and
3. The mean annual flow volume from the site.
Such an integrated suite of indicators forms potentially a useful basis for developing stormwater
management objectives aimed at restoring the post-development hydrologic regime as close as
possible to the pre-development level.
2.1.4 Summary
Urbanisation profoundly changes the hydrologic cycle, increasing the volume, frequency and flow rate
of storm flows, whilst commonly reducing baseflows through reductions in infiltration.
Evapotranspiration is also greatly reduced, decreasing from around 80% of mean annual rainfall in the
pre-developed situation to around 15% for impervious areas. The increases in runoff result in
increased frequency and magnitude of disturbance to channel substrates and to aquatic ecosystems.
Many studies have investigated potentially useful indicators for use in assessing the catchment-scale
impacts of urbanisation. Such indicators need to have ecological relevance (i.e. be shown to be
related to ecological indicators) and be able to be measured and modelled. They should be sensitive
to the urbanisation gradient, rather than external factors (e.g. catchment size). The indicators found to
be most useful relate to frequency and duration of high flow events, flashiness, mean annual runoff
Issues related to these pollutants include impacts on ecosystems due to toxicants in the water column
and sediments, impacts on ecosystems due to nuisance plant growth, asphyxiation of respiring
organisms due to depletion of oxygen, modified primary production as a result of light attenuation by
particles, and smothering of benthic organisms by sedimentation (ANZECC/ARMCANZ, 2000). The
principal sources of common stormwater pollutants are summarised in Table 3.
Pollutant
demanding
Source
Synthetic
nutrients
organics
oxygen
metals
solids
oils
atmospheric deposition
plants & plant debris
soil erosion
cleared land
fertilizers
human waste
animal waste
vehicle fuels & fluids
fuel combustion
vehicle wear
industrial and household chemicals
industrial processes
paints and preservatives
pesticides
stormwater facilities
The mobilisation and transport processes that influence urban stormwater pollution are complex and
highly variable, both in space and time (Duncan, 1995). However, the primary processes include:
wet and dry atmospheric deposition;
interception on vegetation and anthropogenic above-ground structures;
buildup of contaminants on impervious surfaces;
washoff from both pervious and impervious surfaces into formed channels and pipes; and
transport along channels and pipes (Duncan, 1995).
The contribution of these processes to stormwater pollution varies according to the pollutant type. For
example, dry and wet atmospheric deposition typically supplies as much nitrogen as is washed off in
urban runoff (Duncan, 1995). In contrast, dissolution and washoff from roof materials is a major
source of copper (Cu) and zinc (Zn) (Duncan, 1995). It is also important to note that buildup of
pollutants on urban surfaces is typically very high compared to washoff loads from any one event,
suggesting that in many cases the buildup is not a limiting factor in determining washoff loads
(Duncan, 1995).
Table 4. Summary of typical water quality values for runoff from urban, agricultural and forested catchments, urban
streams, and secondary treated sewage.
Typical urban
Variable Urban Agricultural Forest Secondary
a b b a stream water
(mg/L) runoff runoff runoff sewage a
quality
TSS 250 (13-1620) 186 79 25 2.5-23
TP 0.6 (0.1-3) 0.54 0.072 8 0.02-1.2
TN 3.5 (0.5-13) 3.9 0.83 35 0.39-4.9
+
NH4 0.7 (0.1-2.5) 20 0.002-0.16
NOx 1.5 (0.4-5) 10 0.34-3.2
BOD 15 (7-40) 15 1.0-4.0
Cd 0.002-0.05 0.025 0.002 <0.0005
Cr 0.02 0.01 -
Cu 0.4 0.038 0.03 0.001-0.017
Pb 0.01-2.0 0.025 0.02 <0.002-0.024
Zn 0.01-5.0 0.2 0.1 0.009-0.14
a
(Lawrence & Breen, 2006); only for cities with separate stormwater and sanitary sewer systems.
b
(Duncan, 2006a; Duncan, 1999)
Such targets are widely applied in Queensland, Victoria, and New South Wales. Loads targets can be
readily reported against using currently available modelling tools such as MUSIC (Gold Coast City
Council, 2006) and have been considered as useful given their integrative nature.
However, whilst loads-based treatment targets are suitable for lentic waterways such as lakes,
estuaries and bays which have a high buffering capacity due to their large volumes, they are not
appropriate for smaller flowing (lotic) waterways (Taylor, et al., 2005; Walsh, et al., 2004). Indeed lotic
systems such as small streams are far more sensitive to temporary spikes in stormwater pollution
(Roesner, 1999; Walsh, et al., 2005a), thus concentration-based treatment targets would be more
appropriate for these receiving waters.
In Australia many concentration targets exist, such as the national ANZECC guidelines for freshwater
and marine waters (ANZECC/ARMCANZ, 2000). Similar concentration targets exist for most
Australian States and Territories. They are generally derived from an assessment of receiving water
requirements and are typically risk-based, in that they specify maximum permissible probabilities of a
given threshold concentration (e.g. median concentration of TSS not to exceed 20 mg/L). Such
guidelines use trigger levels, below which there is a low risk of harm to the ecosystem. When trigger
levels are exceeded, further investigation to understand ecosystem response is required.
It is possible that concentration targets may not all be achievable using current stormwater treatment
technologies (Hatt, et al., 2009; Walsh, et al., 2010). In this case, a compromise ―best practical
treatment‖ target may need to be considered as an interim measure only.
It is thus critical that appropriate indicators and targets be chosen for the particular receiving water of
interest, with concentration targets used for lotic receiving waters and loads targets used for large,
well-buffered systems. This also means that where stormwater is to be infiltrated, it should be
required to meet water quality objectives to ensure that groundwater is not polluted. As a starting
point, the same targets as applied to surface receiving waters could be used.
While imperviousness is an important source of pollutants, the main driver of pollutant transport in
urban catchments is the drainage system, which efficiently delivers runoff and pollutants to receiving
waters.
2.2.4 Summary
Like hydrology, changes to water quality are a ubiquitous response to urbanisation. A wide range of
pollutants, including toxicants, nutrients oxygen-demanding substances, and of course sediments, are
likely to occur in elevated concentrations and loads in catchments subjected to urbanisation. The
increased concentrations and loads in waterways results from both increased generation (through land
use activities) and mobilisation/transport processes (due to the creation of hydraulically efficient
drainage networks).
A number of studies have shown that pollutant concentrations are well predicted by impervious areas,
but importantly, Hatt et al‘s (2004) study showed that the effective imperviousness of a catchment
gave the best prediction of a wide range of pollutants, indicating the direct role which stormwater
discharge has on receiving waterway water quality. A new approach to the design of urban drainage
systems is thus required if we wish to avoid water quality degradation due to urbanisation.
The selection of water quality indicators must take into account the nature of the waterway. For large
systems, loads targets provide a suitable measure, whilst concentration targets (such as those
2.3 Geomorphology
Alterations to hydrology of streams in urban catchments, and of the catchment landscape, are
accompanied by altered sediment dynamics (Wolman, 1967). Sediment dynamics, which incorporate
the spatial and temporal presence and variation in constituents of the bed and bank materials, is an
ecologically important aspect of stream health. It is well established that reduced sediment supply to a
stream increases the energy available to degrade the channel for a given flow, or alternatively stated,
stream energy increases as sediment supply decreases (Schumm, 1977).
Sediment budgets in the urban environment are relatively poorly understood, with the majority of
research in this field focused on forested and rural catchments (Nelson & Booth, 2002). Sediment yield
from an urban catchment is generally considered to decrease, although these conclusions refer to
coarse-grained sediments (bed load) more commonly than fine-grained suspended sediment
(Bledsloe, 2002; Gurnell, et al., 2007). Walsh et al. (2005b) highlighted the inconsistency across
studies in suspended sediment yield response following urbanisation.
The prevailing model of sediment dynamics related to land use is that of Wolman (1967), who
conceptualised the changes in sediment delivery to streams relative to catchment land-use change
(Figure 5). He found catchment-cover change from forest to crops resulted in significant increases in
sediment load to receiving streams. The most dramatic change in sediment dynamics resulting from
land-use change results from construction impacts of a developing catchment (3 to 5 times higher; see
Keller, 1962), although this is likely to overestimate the contemporary contribution of suspended
sediment to the stream system during construction due to improvements in construction practice. Most
pertinently Wolman (1967) suggested that sediment yields post-construction phase can be lower than
for the intact catchment, though, the findings for the established urban phase were the least
compelling, based predominantly on images of drain sedimentation.
Figure 5. Conceptualised change in sediment delivery to streams relative to catchment land-use (after Wolman 1967).
Note the uncertainty regarding the urban phase (circled).
Gurnell et al. (2007) suggested that reduction in sediment supply in urban environments is a result of
erosion-resistant sealing of catchment surfaces. This theory, when comparing urban to the former
agricultural land use, is intuitively convincing considering the reduced areal extent of exposed soils
and reduced potential for sediment liberation, in comparison to the increased frequency and volumes
of runoff.
In the Issaquah Creek catchment, Washington, Nelson and Booth (2002) found development to almost
double sediment production even though relatively little sediment was liberated directly from the urban
areas. The increase was primarily attributed to sediment production resulting from discharge-induced
channel erosion (20% of the total sediment budget). However, the approach used to determine pre-
and post-urbanisation channel capacity, namely an empirically derived relationship between channel
capacity and the 2 year ARI, is simplistic (as acknowledged by the author). Other approaches are
discussed in this section.
The increased sediment transport capacity resulting from conventional approaches to managing
stormwater (and its subsequent increased runoff volume) has been well demonstrated (Bledsloe,
2002; Grove & Ladson, 2006; Pomeroy, et al., 2008).
Of the three phases of sediment movement in the urban environment; supply (source), transport (and
erosion) and deposition, the latter is the most poorly understood. This, no doubt, has much to do with
the absence of sediment deposition to inspire these studies and the prevalence, and immediately
relevant concerns, surrounding erosion. Issues result from both decreased sediment loads (e.g.
accelerated clearwater erosion) and increased sediment loads (e.g. smothering of substrates, Wood &
Armitage, 1997). In addition, better understanding the dynamics of sediment deposition in urban
stream is particularly pressing particularly when compared with rural streams, because fine-grained
sediments play an important role in the storage and transport of contaminants.
There are differing findings on the effects of urbanisation on sediment calibre but in general a
progression towards finer sediments (‗sediment fining‘) is observed. Booth and Jackson (1997)
suggested that this fining with urbanisation is explained by the dominance of overland flow driving
increased sediment transport. This is supported by Gurnell et al. (2007), whereas Pizzuto et al. (2000)
found that gravel-bed urban streams, when compared with rural streams, were lacking the finer
particles ranging in size from sands to pebbles which they suggested had been selectively removed,
resulting in coarsening. This may suggest a bimodal distribution following the adjustment of streams to
urbanisation: increased clays/silts, decreased sands and gravels, and a dominance of coarser material
(cobbles, boulders) where present. Interestingly, overall increases in sediment size have been
observed due to intervention activities (Grable & Harden, 2006), which is discussed further in the
following section.
Irrespective of sediment size it appears the mobilisation effectiveness of stormwater flows from urban
catchments cannot be discounted. Booth and Henshaw (2001) suggested that no unconsolidated
sediments in urban stream channels are immune from disturbance given adequate upstream
urbanisation.
In theory, as a stream adjusts over time to the hydrologic and sedimentologic inputs it should be
capable of reaching a relative stable phase without intervention. In reality, many stream responses are
non linear and, irrespective of the stability of the inputs with time, the channel remains dynamic
(Rhoads, et al., 2008). Stream states have been referred to as ‗dynamic equilibrium‘; moving between
phases of erosion and deposition (Hack, 1960); or ‗quasi-equilibrium‘: whereby channels undergo
adjustment to the range of discharges and sediment loads experienced and continue to change
A ‗metastable‘ state has been used to define riverine ecosystems (Thorp, et al., 2006) and this may be
an appropriate way to consider geomorphic functioning of urban streams. Metastability is a long lived
unstable or transient state. In essence, the stream can exist in a number of ‗stable‘ states and hover
between those states for long periods of time depending on the conditions, changing readily between
a more stable or less stable condition. While semantics can often become tiresome there is a need to
define, at least conceptually, the desires for dynamism when considering river management in a
physically constrained environment.
In an urban setting attainment of any form of ‗equilibrium‘ or ‗stable‘ state is most challenging for two
reasons: a lack of adjustment room, a.k.a. riparian ‗buffer‘; and the magnitude of channel adjustment
required to cater for the flow volume and hydrologic regime delivered to the channels through
stormwater systems. Whether urban stream channels will ever achieve a quasi-equilibrium appears to
still be an open question (Grable & Harden, 2006).
Notwithstanding, Booth and Henshaw (2001) found that the age of the upstream development was
closely related to the rate of channel change. Streams are most stable downstream of more
established developments. They suggested the ‗reason for this influence was enigmatic‘, yet, it seems
quite plausible that (as they further suggest): a) the longer the period of adjustment the closer to a
state of quasi-equilibrium, b) erodible sediments are likely to have been removed and low-erosivity
sediments remain, and c) undergo a stabilising influence from cementation of suspended sediments
and regeneration of vegetation.
Booth and Henshaw (2001) suggested there is a potential for research bias through focusing on
degrading channels. Attention is most commonly, and not surprisingly, paid to particularly dramatic
channel changes during urbanisation of a catchment. Geomorphologists are inherently interested in
measuring change.
Channel capacity has been found to increase by a factor of at least 2: Pizzuto et al. (2000) found an
increase of 2.3 times; Gregory et al. (2002) 2-2.5 times; and MacRae (1997) 4.2 times. These
increases in capacity are not uniformly distributed by bed and bank. Pizzuto et al. (2000) found
capacity increase solely resulted from widening, attributed, in part, to armouring of bed sediments (a
common geomorphic response to the removal of the finer-grained sediments).
Channel incision is a well known geomorphic response to either increased flow, decreased sediment
load, an oversteep channel gradient, or decreased calibre of sediment inputs (Lane, 1955). Incising
channels tend to degrade vertically, lowering the bed, prior to a phase of lateral degradation, eroding
the channel banks and widening the channel. This process can be cyclical. The classic representation
of channel incision is provided by Schumm (1984), as shown in Figure 6. Where flow and sediment
regimes are not conducive to recovery, the return of an inset channel (Type V) is unlikely.
Fluvial scour removes individual sediment particles or aggregates by water flow. This occurs when the
force applied to the bank by flowing water exceeds the resistance of the bank surface to these forces
(Abernethy and Rutherfurd, 1999). The removal of bank material is therefore closely related to near-
bank velocity conditions and in particular to the velocity gradient and turbulence close to the bank,
which determines the magnitude of hydraulic shear (Knighton, 1998). The process of scour and fill has
been stated as contributing to the adjustment of size and shape in natural channels (Leopold et al,
1964), though, with increased flow, decreased sediment supply and incised channels containing flow
the dominance of scour over fill is likely to be significant.
Mass failure or slumping occurs when large segments of the bank break off through the process of
erosion. This process is generally triggered when a critical stability condition is exceeded, either by
reduction of the internal strength of the bank, due to sub-areal preparation, or a change of river
geometry, commonly through fluvial scour (Abernethy and Rutherfurd, 1999). Therefore the
susceptibility of river banks to mass failure depends on their geometry, structure and material
properties (Knighton, 1998). The collapsed blocks produced by mass failures may break on impact
and be removed or they may remain intact to be eroded by hydraulic action, sometimes protecting the
lower bank from further erosion (Knighton, 1998). A number of factors influence erosion by mass
failure including: bank and material composition, climate, subsurface conditions, channel geometry
and bioperturbation. In particular, flow characteristics, such as rates of fall, play a considerable role in
the potential for mass failure.
Sub-aerial preparation (drying and desiccation) occurs when bank areas are exposed to air (above the
waterline). This includes piping, rain splash, rill erosion, stock trampling and desiccation (Abernethy
and Rutherfurd, 1999). Cycles of wetting and drying are especially important as they cause swelling
and shrinkage of the soil, leading to the development of fissures and tension cracks which promote
failure (Knighton, 1998). Desiccation causes extremely dry and cracking bank material which is highly
erodible (Abernethy and Rutherfurd, 1999).
The physical structure of the stream can affect how a flow translates to an important ecological event
such as a disturbance (Poff et al. 2010). Booth and Henshaw (2001) highlighted the impact of
morphologic simplification on biological and aesthetic values and demonstrated that ‗channel instability
does correlate with low quality habitat’ (p. 19).
The role of wood and other roughness elements in maintaining the diversity and stability of a channel
through the provision of roughness cannot be underestimated. The removal of wood from an otherwise
stable stream may be the tipping point for catastrophic channel change (Booth & Henshaw, 2001).
While channel stability is often the aim of the river engineer it generally competes with the desires of
the river manager or geomorphologist sympathetic with ecological functioning. The attainment of
quasi-equilibrium may not in fact be the ideal state for an urban stream where some semblance of
ecological functioning is desirable. Booth and Henshaw (2001) highlighted that the re-equilibration of
a channel following major incision does not necessarily coincide with the return of improved habitat
quality.
Planform change is often the result of interventions such as meander cutoffs or channel straightening
(Nelson, et al., 2006; Rhoads, et al., 2008). Historically, obvious benefits to this approach have been
highlighted including: increasing hydraulic efficiency (hydraulic conveyance), reducing erosion and
reducing maintenance (Wolman, 1967). This, however, often only transfers the problem, and channel
straightening inevitably leads to increased hydraulic gradient and an erosional response (Rhoads, et
al., 2008)
The concept of an ‗equilibrium‘ stream implies stability in elevation, gradient and channel form
(Wolman, 1967). This concept might be appealing from an engineering viewpoint but ignores many of
the aspirational goals for a healthy stream ecosystem. Modes of adjustment, including the presence of
erosion and substrate sediment dynamics, can be important elements of a healthy stream (Florsheim,
et al., 2008; Wolman, 1967). The ecological and biological environment, adjusting to the input
variables, is unlikely to be dynamic if stream morphology is moribund. Stabilised urban channels have
been found to exhibit demonstrably lower habitat values (Gurnell, et al., 2007).
The role of engineering protection in urban streams is often considerably more influential than
hydrologic or sedimentologic change. The geomorphic response of a stream to changed flow and
sediment inputs is often moot if protection prevents this adjustment. The vexing problem is that to
In understanding the relative role of urbanisation in shaping urban channels the modifying role of
preparatory land uses cannot be discounted. Channel incision is often ongoing, and related to urban
stormwater as a driver, but these adjustments may be minor in comparison to channel degradation
prior to urbanisation. Catchment clearing, and clearing of the riparian zone, encourages channel
incision through altered streamflows and reduced structural stability via vegetation.
Incision processes depend on land use and the hydrogeology of the channel in question. For poorly
drained riparian lands physical interventions such as channel dredging confines flow, decreases
infiltration and reduces flow attenuation to the trunk channel. Direct intervention in the channel with the
express intention of drainage ‗efficiency‘ leads to an enlarged and straightened stream. Increased
containment of flow leads to higher stream powers greatly exacerbating the already increased energy
from stormwater runoff. For well drained land, commonly associated with steeper terrain, stream
modification is most commonly related to flow detention through the construction of online storages.
This often leads to an alternating cut and fill channels.
The potential for abundance and diversity of bed and bank habitats in urban streams, and the
relationship of these features to mobile channels, supports efforts to focus on urban runoff to streams
(the cause) rather than reactively treating the stream (the symptom).
Theoretically, power functions relating the overall discharge in a river channel to river dimensions
(Leopold & Maddock, 1953) clearly suggest that for increased discharge (up to 10 times the volume)
there will be a corresponding increase in channel width and depth.
EarthTech (2006) highlighted the increase in energy expenditure (not excess energy) in an urban
stream, Little Stringybark Creek, based on standard urban stormwater design. Peak total energy
expenditure was found to be bimodal: greatest during low flows (0.1 m3/s) and medium frequency
events (1 in 4 yr ARI approx.) and generally 4 to 5 times greater than for natural conditions. WSUD
was found to return total energy expenditure to levels similar to a condition closer to that of a
forested/cleared catchment.
For ease of extrapolation urban stream degradation is most commonly related to a surrogate measure
of flow hydrology: Total Imperviousness Area (TIA) or more appropriately Effective Impervious Area
(EIA). Based on an empirical study nearby Seattle, Washington, Booth and Jackson (1997) found that
6 percent EIA for a watershed was a threshold beyond which the receiving channel was significantly
wider.
Moving beyond a static consideration of the impacts of effective impervious area Booth and Henshaw
(2001) analysed an 11 year data set to identify EIA relative to the rate of channel change. They found
that the rate of channel change is poorly correlated with EIA. Some sites with up to 40 percent EIA
The relative sensitivity of streams to urbanisation propagates the uncertainty surrounding response.
Not surprisingly, streams in western Washington‘s granular hillslope deposits considerably greater
change occurred compared with those in cohesive silt-clay deposits (Booth & Henshaw, 2001). In
addition to local conditions, the poor correlation between rate of channel change and EIA may
represent stabilisation of channels over time, or the inadequacy of EIA as a surrogate measure of the
impact of urbanisation on channel morphology.
Channel capacity and erosion of urban channels are most commonly related to the 1 in 2yr ARI event
(Bledsloe, 2002; MacRae, 1996; Nelson & Booth, 2002). This appears to be a relic of the early findings
of Wolman and Miller (1960). MacRae and Rowney (1992) found channel erosion, based on scour
potential, following urbanisation to be better associated with moderate flow events: sub-bankfull flows
with recurrence intervals between the 1 in 0.5 to 1 in 1.5 year ARI. The greatest sediment transport
potential was found by MacRae and Rowney (1992) to occur at moderate depths: 0.5 to 0.85 of
bankfull depth. The greatest increase in scour from pre to post-urbanisation corresponded to flow
events less than 0.7 bankfull depth. They concluded that urbanisation shifts channel forming
dominance from bankfull flows to more frequent smaller events.
Urban flow attenuation is commonly, particularly in the United States, threshold based. The ‗duration
standard‘ for detention basin design is aimed at maintaining the aggregate of post-development flows
at or below a threshold for sediment mobility (Booth & Jackson, 1997; Pomeroy, et al., 2008).
Common practice is to relate channel erosion to hydrologic thresholds for sediment mobility to
determine ‗geomorphically effective‘ or ‗geomorphically detrimental‘ flows, e.g. one half of the 2 year
pre-development flow: Booth and Jackson 1997, using excess energy expenditure or erosion potential
(EP) (Bledsloe, 2002; Booth, 1990; Elliot, et al., 2010; Grove & Ladson, 2006; Pomeroy, et al., 2008;
Tilleard & Blackham, 2010). The erosion potential index is continuously simulated to assess ‗work
done‘ on the channel above the critical shear stress (Bledsloe, 2002; Pomeroy, et al., 2008). Although,
critical shear stress values are rarely field verified (e.g. Grove & Ladson, 2006; Pomeroy, et al., 2008;
Tilleard & Blackham, 2010) often relying on the flume study results of Chow (1959). Furthermore, the
concept that stream power can be used to determine urban stream stability has been refuted by
Stacey and Rutherfurd (2007) based on a study of over a 1000 sites in Fairfax County, Virginia, USA.
They found that substrate type provided a statistically greater influence on stability , but could not
define stability (Stacey & Rutherfurd, 2007).
A ‗peak standard‘, or ‗peak shaving‘ approach, whereby peak flows pre and post are matched, has
been found to be considerably less effective in reducing elevated sediment transport rates because
The four key contributions which fluvial geomorphology can make in this regard are highlighted by
Gilvear (1999, p. 229):
1. to promote recognition of lateral, vertical, and downstream connectivity in the fluvial system
and the inter-relationships between river planform, profile, and cross-section;
2. to stress the importance of understanding fluvial history and chronology over a range of time
scales, and recognizing the significance of both palaeo and active landforms and deposits as
indicators of levels of landscape stability;
3. to highlight the sensitivity of geomorphic systems to environmental disturbances and change,
especially when close to geomorphic thresholds, and the dynamics of the natural systems;
and
4. to demonstrate the importance of landforms and processes in controlling and defining fluvial
biotopes and to thus promote ecologically acceptable engineering.
The challenge is to appropriately describe the geomorphic response of urban streams to altered
stormwater and sediment inputs, reducing uncertainty and quantifying risk. As identified by Rhoads et
al. (2008):
Uncertainty itself derives largely from a lack of information on how the streams might respond
to naturalization. Alleviation of this uncertainty involves the implementation of various
evaluation methods to generate information on potential or actual geomorphological
performance. (p. 225)
A dilemma facing the geomorphologist rehabilitating an urban stream is the appropriate level of
intervention. The geomorphic form and function of streams cannot be ‗recolonised‘ at the same rate
nor the same extent, as the biological condition can under ideal circumstances. Physical form can
either be reinstated through intervention, e.g. physically recreating pools, bars and riffles using
machinery, or, with appropriate hydrologic and sediment inputs e.g. allowing for self adjustment. While
the former is an immediate response, the period of response for the latter, even with appropriate
inputs, could be measured in decades rather than years (the latter comparable to biota recovery times
with favourable conditions). This raises the question should we intervene to expedite the recovery
process?
According to Rhoads et al. (1999) rehabilitation programs should focus on creation or naturalisation in
order to improve the health and value of a system. Newson (2002) suggested intervention was often
necessary: ‗assisted natural recovery‘.
The success of appropriate rehabilitation ‗will be difficult unless public expectations are tailored to an
appropriate understanding of stream dynamics‘ (Rhoads, et al., 2008 p. 227). The steps to stakeholder
confidence include highlighting the values of geomorphically functioning streams and improved
understanding of channel response in urban environments so that outcomes and associated risks can
be practically assessed.
There appears little need for monitoring protocols specifically tailored to urban streams as inputs
(sediment and water) and degrees of freedom (width, depth, planform etc.) are comparable to non-
urban streams. The most considerable differentiator is the scale of modification to inputs. In this regard
the rates of change should be greater. The constraints on degrees of freedom, such as rock
protection, have obvious implications for channel stability but, if suitably sized by the engineer, there is
little potential for channel change in these locations. In these locations ‗environmental‘ monitoring is a
fruitless exercise merely demonstrating the river engineer‘s ability to design for a rare event.
A number of approaches are available to the geomorphologist monitoring geomorphic change in urban
streams over timescales which are short term (<1 year), medium term (<= 5 years) and long term (> 5
years). For example, short to medium term change may be detected through: deposition mapping and
suspended sediment concentrations; and medium to long term changes may be detected through:
repeat cross sections and planform alignment mapping. Where cross sectional or alignment
comparisons are to be made the importance of permanent benchmarks and accurate survey
techniques cannot be over emphasised. Channel changes are often barely discernible from potential
operator or equipment errors.
Monitoring geomorphic change must be spatially targeted to those sites in which change is likely within
management timeframes. Timeframes for geomorphic adjustment, even within urban streams, are
particularly greater than that experienced for ecological attributes. Geomorphic change is also highly
dominated by thresholds (Booth & Jackson, 1997; Schumm, 1977) and as such a linear response is
rare.
2.3.4 Summary
In summary, while impervious cover and channel degradation are often correlated, altered stormwater
inflows are cited as the driver of physical form degradation in urban streams (e.g. Grable & Harden,
2006; Walsh, et al., 2005b). In order to maintain (or reduce) the channel capacity of urban streams,
and increase the potential to reach a quasi-equilibrium, reductions are required in the overall
stormwater volumes to the channel. Reductions are also required in the frequency and duration of
scouring flows which have been found to be more prevalent in urban stormwater regimes (MacRae &
Rowney, 1992). Returning some of the physical components of physical habitat and a healthy stream,
such as benches, bars, and spatially variable and temporally mobile substrates, is not feasible unless
the components of the stormwater flow regime are amenable. Most importantly, the focus of the river
manager should move beyond reactively patching the symptom (bank and bed erosion) rather than
pursuing a solution to the problem (an increase in geomorphically effective flows).
In addition to the physical sciences there are further social and institutional challenges faced by the
geomorphologist. If ecological health of urban streams is a priority then there are essentially three
steps in the process for geomorphic recovery of an urban stream: 1) - Convincing river managers,
engineers and applied geomorphologist, who have historically been focused on channel ‗stabilisation‘
in urban settings, that ‗naturally‘ functioning stream morphology leads to ecological health; 2) -
Determining the level of intervention, and the riparian land required to return geomorphic functioning to
the urban stream, or in the case of mildly impacted streams whether ‗assisted natural recovery‘
(Newson, 2002) will suffice; and 3) - Determining the feasibility of returning the flow and
sedimentologic regime required to facilitate geomorphic functioning within the constraints of the urban
environment. It is important to accept that ‗natural‘ geomorphic functioning may not be desired by the
community or managers in an urban setting.
2.4 Ecology
2.4.1 Overview
For the last decade, scientists have been studying the effects of urbanisation on aquatic ecosystems
(Paul & Meyer, 2001; Suren & McMurtrie, 2005; Taylor, et al., 2004; Wenger, et al., 2009). There is
now a well-described suite of impacts of urbanisation on stream and river ecosystems, and an
emerging deep understanding of the mechanistic bases for those impacts. Studies of the effects of
urbanisation have moved from being correlational studies between land use and ecological structure
(e.g. Sonneman, et al., 2001) or function (e.g. Imberger, et al., 2008), to more sophisticated spatial
analysis of ecological response to elements of urban land use in relation to flow paths, that has
identified urban stormwater runoff delivered through conventional drainage systems as the most
plausible driver of urban related degradation of stream ecosystems (Wenger, et al., 2009).
Increasingly, having determined some of the cause and effect relationships underlying urban impacts
on streams, attention is moving to how best mitigate those impacts. It is timely therefore to summarise
the described impacts of urbanisation on stream biota within an organised framework that identifies
how best to mitigate these effects. In keeping with this broader review, we will focus on stormwater
runoff as a primary mechanism of urban degradation.
Figure 7. A conceptual model of the impacts of urbanisation on stream ecosystems (after Walsh et al. 2005b). The
primary driver of stream degradation in conventionally drained urban catchments (A) is most likely urban stormwater
runoff that is delivered to streams through stormwater drains and pipes. These drainage systems efficiently transport
pollutants generated in urban catchments to the stream every time there is enough rain to elicit impervious runoff,
resulting in more frequent and larger storm flows with increased concentrations of nutrients and toxicants. A
secondary driver often associated with urbanisation is the loss of riparian forests, which compounds the effects of
stormwater runoff on sediment and organic matter dynamics and growth of autotrophs (algae and bacteria) in-stream.
These changes compound the direct influence of stormwater runoff on the animals inhabiting the stream. If
impervious surfaces are permitted to drain to pervious land (B), or to stormwater retention measures, the effects of
stormwater can be attenuated before reaching the stream. Where the effects of stormwater runoff are adequately
reduced, the local scale effects of riparian condition and function are likely to become a more dominant driver of in-
stream ecological structure and function. Upward and downward arrows denote increased and decreased effects,
respectively, and delta denotes a change of unspecified direction.
The catchment-scale effects of urban stormwater runoff are transmitted as the previously described
changes to storm flows, dry-weather flows and their quality. Stormwater runoff can also effect organic
matter supply, through leaf material entering streams via stormwater drains distributed throughout the
Effects of hydrologic change have complex consequences for stream ecosystems but can be broadly
grouped into the effects of disturbance (both increased ‗flashiness‘ of flows and increased drying),
impacts on channel form and sediment composition, and reductions in retentiveness of organic matter.
Direct effects of disturbance on aquatic biota have been well studied in non-urbanised settings (e.g.
Engelhardt & Kadlec, 2001; Stanley, et al., 2010; Townsend, et al., 1998). These impacts are as a
result of a number of factors including increased abrasion from particles entrained in the flow,
disturbance of the stream bed and direct wash-out of biota (both sloughing of algae and loss of
animals downstream). Drying of streams also impacts biota through dehydration and through a
requirement for recolonisation or emergence from resting stages when flow returns (e.g. Lake, 2003).
Complex interactions between stream channel form and drying determine the degree of impact of
drying; as deeper areas may form pools that act as refugia (Lake, 2003).
Altered hydrology may also alter in-stream habitat structure. Accumulations of wood and leaves
(snags) in streams are important sites of biodiversity and are often highly productive (Bilby & Likens,
1980). Accumulations of organic matter can also contribute to alterations in stream channel form
through the generation of eddies and backwater areas that affect sediment accumulation (Bilby, 1984).
Changes in the stream hydrologic regime flushes out accumulations of wood and leaf litter, resulting in
reductions in the standing crop of these materials and further simplification of urban stream channels
(Roy, et al., 2005).
Accumulation of toxins in sediments (derived from polluted stormwater runoff from the catchment) has
been posited as a possible driver of reductions in in-stream biodiversity, based on mesocosm
experiments using wetland sediments (e.g. Anson, et al., 2008). However, a recent study comparing
biota sampled from the streams along an urban gradient concluded, and biota colonizing sediments
from the same sites in controlled mesocosms, found no evidence that differences in sediment quality
explained the observed declines in biodiversity with increasing catchment-scale urban impact (O'Brien,
et al., 2010).
The majority of early studies on the effects of urbanisation were based on inference of catchment-
scale effects of urbanisation to correlational studies of urban density (usually measured as total
imperviousness). More recently studies have sought to infer the most important local-scale
mechanisms caused by urbanisation. Most recently considerable advances have been made in
seeking to separate the different aspects of catchment urbanisation that are the most plausible drivers
of stream ecosystem change. A recent assessment of the most important hydrologic flow paths for
transmitting urban impacts to streams, identified the stormwater drainage network as the most
plausible pathway by which stream biota are degraded (Walsh & Kunapo, 2009). The identification of
stormwater runoff directed through a piped drainage network has been identified as potentially the
primary driver of environmental harm to stream ecosystems (Walsh, et al., 2005b).
Local inputs of organic matter are important both in the provision of habitat and resources for
consumption (Bilby & Likens, 1980). High rates of local supply may maintain these functions even
when retention of material is relatively low. Where riparian vegetation along streams is intact it may
also function to filter surface and shallow sub-surface flows of water before they enter the stream
channel (Gregory, et al., 1991; Lowrance, 1998). However, the interception and filtering of water
quality to streams are likely to be limited in reaches in which stormwater runoff is directed directly
through pipes to the stream (Walsh, et al., 2007), or where stream incision has resulted in hydrologic
isolation of the stream channel from the floodplain (Groffman, et al., 2003).
Yet few studies have unambiguously demonstrated a strong effect of riparian forest on stream
ecosystem structure and function. Some studies have posited that riparian forest might ameliorate the
effects of catchment urbanisation, but these have been based on limited data (Morley & Karr, 2002) or
have not clearly accounted for spatial correlations between catchment urbanisation and riparian forest
cover (Moore & Palmer, 2005; Urban, et al., 2006). However, correlational studies that have explicitly
sought to separate the influences of catchment-scale and riparian effects have consistently found that
catchment-scale urban density has a stronger effect on biotic assemblages than riparian forest cover
(Roy, et al., 2005; Thompson & Parkinson, in press; Walsh, et al., 2007). Similarly, an experimental
manipulation of light climate in streams across an urban gradient found that the effect of light on algal
biomass accrual and assemblage composition was relatively small compared to differences in
catchment urban density (Catford, et al., 2007).
Increased biomass Water quality (C) + Light (L) Taylor et al. (2004)
Primary producers
Shift to eutrophic species Water quality (C) + Disturbance (C) Newall and Walsh (2005)
Algae
Increased biomass Water quality (C) + Light (L) Suren (2009)
Aquatic plants Increased invasive species Water quality (C) Suren (2009)
Ecosystem function Litter supply (L) + Water Quality (C) Imberger et al. (2008)
Litter decomposition Increased Abrasion (C) Meyer et al. (2005)
Nitrogen cycling Altered Riparian condition (L) Grimm et al. (2005)
Sediment toxicity and water quality (C) Serena and Pettigrove (2005)
Mammals Reduced occurrence of resident platypus
Impacts of multiple stressors (C) Danger and Walsh (2008)
A number of studies in recent years have investigated the effects of urbanisation on the processing of
organic matters. Decomposition of leaf litter and woody inputs into streams is dependent on a number
of environmental conditions which are altered by urbanisation. Studies on leaf breakdown rates have
suggested that that leaf litter breaks down significantly faster in urban streams, due to physical
fragmentation associated with increased stormwater inflows (Meyer, et al., 2005), or through
increased microbial breakdown resulting from high nutrient availability (Imberger, et al., 2008).
Changes in the type of litter provided can also alter function; reduced decomposition rates have been
found in urban streams where the predominant litter is from exotic species (Miller & Boulton, 2005).
However the rates of organic matter processing are strongly influenced by retention within the stream
channel. Hydrologic impacts can greatly reduce the availability of organic matter in urban streams due
to washout from simplified channels during high flows.
Broad scale changes in nitrogen cycling have been shown along urban gradients (Grimm, et al.,
2005). Nitrogen is supplied to urban systems in a variety of sources throughout the catchment,
including aerial deposition from exhaust fumes, delivered to streams through stormwater drainage
systems. Processing rates are generally high, although the disconnection of floodplains from streams
(Section 2.3) can result in drying of soils and a reduced capacity to process nitrogen (Goldman, et al.,
1995; Groffman, et al., 2003). The ways in which in-stream nutrient limitation is influenced by
urbanisation have been surprisingly little studied (but see Chessman, et al., 1992).
2.4.5.3 Macroinvertebrates
Macroinvertebrate communities have been widely studied in urban streams for over a decade. The
impacts of urbanisation on patterns of diversity (the number of taxa) and composition (the types of
taxa and their numerical abundance) are well understood, although the nature of the cause and effect
relationships between urbanisation as a stressor and changes in invertebrate communities are not
always consistent or clear. Studies from a range of urban streams have found that urban streams
have lower diversity than similar non-urbanised streams (e.g. Blakely & Harding, 2005; Paul & Meyer,
2001; Walsh, et al., 2001). Reduced diversity has been attributed to a number of mechanisms
associated with urban stormwater runoff, including direct toxic effects on some biota (Iannuzzi, et al.,
2004), reduced habitat heterogeneity in urbanised streams, particularly in substrate (Roy, et al.,
2003), reduced availability of resources (Moore & Palmer, 2005) or more generally by the complex
multiple stressors delivered by urban stormwater runoff (Walsh, et al., 2001).
The altered composition of stream macroinvertebrates in urban streams provides some clear
evidence of the likely drivers for reduced diversity in these systems. Organic pollution is a major
factor altering invertebrate community structure and diversity. Urban streams typically, show an
increased abundance of ‗pollution tolerant taxa‘, with reductions in the more ‗pollution sensitive taxa‘
Urban streams are also typified by organisms with rapid life histories, which are correlated with small
body size and with taxonomic groups such as oligochaetes and chironomids. These taxa are highly
tolerant to hydrologic disturbance (both in terms of high flows and drying) because they spend a
relatively short time as an aquatic larva. Walsh et al., (2001) investigated the effects of differing levels
of urbanisation around metropolitan Melbourne, Australia on macroinvertebrate communities. These
authors hypothesized that urban streams were impacted by increased levels of effective
imperviousness in the catchment, leading to low water quality and flashy hydrology, resulting in
communities with high abundances of pollutant tolerant taxa, an hypothesis that was supported by
subsequent studies that separated the effects of general urban density from stormwater drainage
connection (Walsh, 2004; Walsh & Kunapo, 2009).
A number of studies have posited the existence of thresholds of imperviousness before an effect on
macroinvertebrate assemblage composition is detected (e.g. Moore & Palmer, 2005; Morse, et al.,
2003). All such studies have used total imperviousness (TI) as their measure of urban density, and
have demonstrated a wide range of assemblage response at low levels of TI, which could be
explained by the wider variation in drainage connection at low levels of TI (Hatt, et al., 2004; Roy &
Shuster, 2009). Some studies, such as Wang et al. (2001), have claimed to use connected
imperviousness, but have in fact only adjusted total imperviousness by a fixed proportion, where
direct estimates of connected imperviousness (CI) have shown that CI and TI are not consistently
related, particularly in less densely developed catchments (Roy & Shuster, 2009). To date the only
studies that have directly estimated CI by spatial analysis have demonstrated a strong negative
decline in macroinvertebrate diversity (and other ecological indicators) with increasing CI, with no
evidence of a threshold before an effect is observed (Walsh, et al., 2005a; Walsh & Kunapo, 2009).
This suggests that the hydrologic and water quality disturbance of urban stormwater runoff has strong
negative impacts on stream ecosystems at very low levels of conventionally drained catchment
urbanisation. Disturbance is an important structuring force in stream communities, with major
influences on body form, life histories and community composition (Resh, et al., 1988; Townsend, et
al., 1997).
Changes in invertebrate communities may also be driven by changes in the nature and volume of
organic matter in urban streams. The absence of accumulations of leaves and wood in urban streams
reduces the availability of both habitat and food resources for groups such as shredders (aquatic
invertebrates which feed on leaves and wood), although most species of shredders also happen to be
species with low tolerances to physical disturbance and pollution.
Changes in invertebrate communities can also occur as a result of urban-induced changes from
native to exotic riparian trees (Miller & Boulton, 2005). There are also other, more complex
interactions between riparian vegetation and aquatic invertebrate communities. Riparian vegetation
can also enhance dispersal and reproductive success in the adult stages of aquatic
macroinvertebrates (Samways & Steytler, 1996), although urban infrastructure has been shown to be
a significant barrier to dispersal of flying aquatic insects (Blakely, et al., 2006). The broader food web
consequences of urbanisation are the product of interactions between multiple stressors and their
direct and indirect effects via all parts of the aquatic food web. In general these have been little
studied, although Faeth et al. (2005) found that compared to nearby control streams the importance of
predation by birds was greater in urban streams, and that resource fluctuations were lower, with
highly complex food web impacts on a large number of taxa.
Amphibians show similar patterns, although they are notably intolerant of many pollutants (particularly
heavy metals), and are generally more intolerant of urban water quality than fish (Barrett, et al., 2010).
Invasive fish are generally highly tolerant of disturbance, and can be favoured over native species in
urbanised streams. In temperate areas invasive species such as the eastern mosquitofish (Gambusia
holbrooki), and carp (Cyprinus carpio) are commonly present, and appear to outcompete native fish in
urban settings. An exception is brown trout, which are absent from the degraded urban streams of
Melbourne (Danger & Walsh, 2008). Danger and Walsh also noted a weak positive correlation
between the native Galaxias maculatus and CI. They posited that this species, which is found less
commonly in rural streams that contain trout, could use the sub-optimal conditions of metropolitan
streams as a refuge from competitive exclusion by trout.
Fortunately, a number of indicators are readily measured and have been identified as being highly
sensitive to impacts of urbanisation (making them effective indicators). For example,
macroinvertebrates are commonly used, including the widely applied Ephemeroptra, Plecoptera,
Trichoptera (EPT) index EPT (Roy, et al., 2003). In Australia both the AusRIVAS and SIGNAL indices
have been used, with SIGNAL considered to be more effective across urban gradients (Walsh, 2006).
As identified in Table 6, some fish can also be effective indicators, as can algal biomass and
composition.
There is a difficult paradox in the effectiveness of indicators for assessing the impacts of urbanisation
– and conversely – for predicting the impacts of stormwater management aimed at redressing these
impacts. Given their sensitivity to the impacts of urban stormwater runoff, all ecological indicators
currently in use show a very steep decline with even very low levels of stormwater runoff (as
measured by directly connected imperviousness). Whilst this makes the indicators useful in
assessing stormwater impacts, it means that the reverse trajectory (along the rehabilitation pathway)
will not see a substantial ecological response until impacts have been brought back to these very low
levels (Figure 8). Whilst stormwater managers may understandably wish to have a more linear (and
thus progressive indicator), no studies to date have been able to identify any. Given the role of
indicators in identifying the ecological condition of waterways, the search to find a more progressive
indicator, while perhaps helpful in terms of garnering political and social support for interventions, will
probably provide little ecological insight. The sharply non-linear response of currently indicators
accurately identifies the situation; streams are ecologically degraded wherever significant stormwater
inputs are permitted to occur.
2.4.6 Summary
This review shows the major impacts that urbanisation can have on the ecological values of streams.
While the underlying mechanisms are complex, they can be classified into catchment level impacts
(largely hydrology and water quality) and local impacts (largely changes in organic matter supply).
The majority of ecological impacts are as a result of catchment scale processes (Table 6). Our review
therefore clearly indicates a disjunct between the scale of the main impacts (predominantly
catchment-scale urban stormwater runoff) and the scale of most restoration activities (predominantly
local scale riparian replantings). Riparian restoration will result in some improvement in ecological
condition (Moore & Palmer, 2005; Thompson & Parkinson, in press) however the amount of
improvement will be contingent on the condition of the catchment (Hatt, et al., 2004; Roy, et al.,
2006). Catchments with even a small coverage of connected imperviousness are likely to have
sufficiently altered hydrology and reduced water quality that even extensive riparian restoration works
may show minimal improvement in stream ecological condition. One of the major mechanisms
underlying this is likely to be the interaction between organic matter supply (which is restored by
riparian restoration) and organic matter retention (a function of hydrology). Riparian restoration
activities are most likely to see major increases in stream condition where the overall catchment
condition is highest. Similarly, riparian restoration activities will be most effective if they also seek to
address both catchment scale and local scale impacts of urbanisation. Where riparian replantings can
be paired with in-catchment retention of stormwater to reduce total runoff volume, reduce the
frequency of uncontrolled stormwater flows, and restore the quality and flow regime of filtered sub-
surface flows, ecological improvement is likely to be greater.
Most environmental flow problems arise from water being extracted for human use: the challenge for
environmental flow researchers and practitioners is how to distribute the remainder for maximum
environmental benefit (Arthington, et al., 2010). The starting point in considerations of the effects of
water extraction from aquatic ecosystems is that there is a monotonic decline in the condition of
aquatic ecosystems with increasing extraction of water from them (Figure 9). This focus on extraction
of water from aquatic ecosystems, compounded with the tendency for water resource managers to
prefer centralized systems, leads to a tendency of urban water managers to first consider extraction
from urban rivers and drains when identifying urban stormwater harvesting projects (Knights &
McAuley, 2009; Newton & Ewert, 2009). Such a conception of stormwater harvesting has led to a
misconception that urban stormwater runoff has some environmental flow benefit (Victorian
Government: Department of Sustainability and Environment, 2006). As is clear from this review, in
fact the reverse is true. Urban stormwater runoff, delivered through conventional drainage systems is
a complex environmental flow problem that can, in large part, be solved through harvesting of
stormwater before it reaches aquatic ecosystems.
Figure 9. Conceptual graphs of ecological and human value of water. I. adapted from Gleick and Palaniappan (2010),
assumes that any extraction from aquatic ecosystems has some negative ecological impact, predicting a monotonic
decline of increasing gradient with greater extraction. The benefits accrued by the human population rise linearly with
the volume extracted. Beyond peak ecological water (P: Gleick & Palaniappan, 2010), any increase in human benefit is
outweighed by reduced ecological benefit. II. illustrates substantially different trends in ecological and human cost
and benefit with increasing retention and use of stormwater before it reaches aquatic ecosystems. No stormwater use
(A) results in ecological degradation of receiving waters. It also presents greater costs in microclimate control and
drainage than if stormwater was used sustainably. Using a volume of stormwater equivalent to the volume lost to ET
in pre-urban state (B), if coupled with infiltration systems to restore lost sub-surface flows, provides maximum
environmental benefit, by maximising performance of the infiltration systems. Using all available stormwater runoff (C)
has an environmental cost by reducing subsurface flow delivery to stream. However, in many urban settings this loss
can be compensated by increased infiltration in non-treed open spaces, or by leakage of water supply systems.
The general conclusion from these works, consistent with the review of hydrologic indicators in
Section 2.1.3 is that the primary hydrologic changes driving ecological degradation centre around the
increased frequency, magnitude (and worsened quality; see Section 2.2.3) of storm flow events, and
the reduced duration of low flows. Several of the studies aiming to identify the change in hydrologic
patterns driving ecological response not focussed on or even identified the increase in total volume
caused by urban stormwater runoff. While this change might not be a direct driver of ecological
response it casts the problem of stormwater runoff as a different class of environmental flow problem.
As described in Section 2.1.1, urban runoff volumes are increased due to the loss of
evapotranspiration which occurs with urbanisation. As the proportion of rainfall that becomes
streamflow in vegetated catchments, and the proportion of rainfall that runs off impervious surfaces
are well known, it is a simple task to estimate the volume of new, excess water that is generated by
impervious surfaces (Figure 10 and Figure 11). In most cities, at least 60-90% of impervious runoff is
water that would never have reached the stream in the pre-urban catchment.
While it is not necessarily a problem of direct ecological relevance, the excess volume of stormwater
aggravates the challenge of retaining and treating stormwater adequately to provide filtered flows that
could mimic lost subsurface-fed dry-weather flows. The literature reviewed in this document suggests
that there are four the critical elements of the hydrograph that are likely to be the primary drivers of
urban stream degradation, from which we propose objectives that effective stormwater management
should target.
1. Minimize uncontrolled storm flows. The increased frequency of hydraulic and pollutant
disturbance from stormwater drainage flows has been identified as a primary driver of
ecological degradation in streams (see Section 2.1.3). While a natural forested stream might
receive one or two substantial floods a year that are associated with increased hydraulic
disturbance and delivery of increased contaminant concentrations, streams receiving urban
runoff receive such flows every time it rains enough to generate impervious runoff. A primary
objective of stormwater management is thus to reduce the frequency of piped flows as close
to the pre-urban level as possible (Walsh et al., 2009). Such objectives have recently been
mandated for federal projects in the US (US Environmental Protection Agency, 2009),
expressed as a requirement to retain the 95th percentile rain event on site.
2. Infiltration flows must be delivered to the stream through treatment measures that ensure flow
rates do not exceed pre-urban subsurface flow rates. This objective aims to restore lost dry-
weather flows. Appropriate maximum flow rates can be estimated from baseflow separation
analysis in reference streams, or by assessment of infiltration capacities of native soils in the
catchment (or a combination of these two techniques).
Figure 10. Estimated annual runoff coefficients (C) from impervious surfaces (open triangles) from sites across the
Melbourne region as a function of mean annual rainfall (R). Impervious runoff was estimated from daily rainfall data at
each of 11 sites (3–45 years of data), assuming an initial loss of 1 mm/d. Regression line: C = 0.234 + 0.203*log10(R).
R2 = 0.90. Annual runoff coefficients from 12 streams with forested (closed circles), grassland (open circles) or mixed
forested and grassland catchments (grey circles) across the Melbourne region as a function of mean annual rainfall.
The lines surrounding these stream points are the relationship between streamflow derived by Zhang et al. (2001) for
grassland (dashed curve) and forested catchments (dotted curve) of the world.
It could be argued that the natural flow paradigm may not be entirely appropriate for urban streams,
because their physical form is generally different compared to their natural condition (Chin & Gregory,
2005). However, it is certain that the flow regime must be restored back towards the pre-development
regime, if ecologically successful stream restoration (sensu Palmer et al 2008) is to be possible.
12
10
0
Total Freq. time no. av. time no. av. Q1 Q3 Q1 year Q1.5 Q5 Integer TSS TN TP
Runoff Surface events Length events Length month month years years of flow
Runoff
Runoff Low flows High flows Peak flows Pollutant loads
Figure 12. Relative changes to flow and water quality indicators from natural conditions for a catchment with no
permanent baseflow (i.e. typical small upland catchment). The post-development ratio of the pre-developed level
(which is denoted by the dotted line) is shown for (a) urban development at 14% imperviousness, and (b) after
application of stormwater harvesting. The peak flow indicators for 1 and 3 month ARI have not been calculated,
because at an hourly timestep for a catchment with no baseflow, they are zero for the pre-development case.
12
10
0
Total Freq. time no. av. time no. av. Q1 Q3 Q1 year Q1.5 Q5 Integer TSS TN TP
Runoff Surface events Length events Length month month years years of flow
Runoff
Runoff Low flows High flows Peak flows Pollutant loads
Figure 13. Relative changes to flow and water quality indicators from natural conditions for a catchment with no
permanent baseflow (i.e. typical small upland catchment). The post-development ratio of the pre-developed level
(which is denoted by the dotted line) is shown for (a) urban development at 70% imperviousness, and (b) after
application of stormwater harvesting. The peak flow indicators for 1 and 3 month ARI have not been calculated,
because at an hourly timestep for a catchment with no baseflow, they are zero for the pre-development case.
Other studies have demonstrated the benefits of stormwater harvesting in terms of reducing flow
peaks downstream. For example, even a distributed network of rainwater tanks, specially designed to
provide permanent freeboard by having a ‗trickle-outlet‘ at a pre-determined point below the top of the
By its very nature, the harvesting of stormwater will affect downstream water quality. It will reduce
pollutant loads to receiving waterways, simply as a function of the water taken away (depending on
the use and disposal of that harvested water). For example, if urban stormwater has a typical
concentration of around 2.5 mg/L TN (Duncan, 2006; Taylor et al., 2005; Brombach et al., 2005), each
ML of water harvested will reduce the N load to the receiving water by 2.5 kg. For example, Hatt et
al.‘s (2006a) review found significant reductions in N loads to receiving water, and quantified the
equivalent cost of this reduction, using well-established costs for constructing stormwater treatment
wetlands. The influence of stormwater harvesting on pollutant concentrations, however, is more
complex. If only roofwater, with its lower concentrations of particulate pollutants (Duncan, 2006), was
harvested, the concentration at the receiving water may increase. In the case of harvesting of overall
stormwater runoff, this is unlikely to be the case. However, no known monitoring data yet exist on the
impact of stormwater harvesting on downstream pollutant concentrations.
In summary, the potential for stormwater harvesting to assist in mitigating impacts of urban runoff
seems to be justified by the (few) studies undertaken to date. However, clear consensus on the
performance and effectiveness has not yet developed, meaning that guidance on system design to
optimise environmental outcomes is not yet available.
Perhaps most importantly, the studies to state focus exclusively on the ability of stormwater
harvesting to reduce stormwater volumes and peak flows, along with pollutant loads. Given the
nature of stormwater harvesting systems, their ability to restore other parts of the flow regime, such as
subsurface flow and low flows, is likely to be limited and will require careful integration with other
stormwater management measures such as retention and infiltration systems.
For example, it may be beneficial to include a component of ‗active detention‘ storage in harvesting
storages, allowing water to slowly trickle out into downstream infiltration-based systems, thus
maximising the efficiency of those infiltration systems, and providing storage to help reduce peak
flows from subsequent storms. Unfortunately, research on the interaction of stormwater harvesting
and stormwater retention and infiltration techniques is virtually non-existent, leaving designers and
practitioners with a lack of guidance and tools to help them integrate technologies of each type.
Unfortunately, however, little guidance exists on how to integrate stormwater harvesting with other
stormwater management techniques. Even the modelling of such systems is limited (Elliott, et al.,
2010; Lee, et al., 2008), particularly given the problems of likely dispersed nature of such techniques
throughout the urban catchment.
In addition to the inability of current tools to consider interactions between harvesting and other
WSUD systems such as infiltration and bioretention, there is a deficiency in the ability to represent the
influence of scales and spatial arrangement of systems and their interactions on flow regimes at a
given point within the catchment. This is a critical prerequisite to optimising the extent and location of
source control techniques in relation to larger-scale end-of-pipe systems. For example, the
widespread use of rainwater tanks connected to biofiltration systems at the allotment scale might be
effective in restoring flow regimes at the catchment scale. One attempt was made recently by
Endreny and Collins (2009) who assessed the impact of distributed, clustered, and single
(centralized) infiltration systems with the use of MODFLOW (a finite element model). Their results
showed that centralised infiltration systems resulted (perhaps not surprisingly) in more pronounced
groundwater mounding than did homogenously distributed at-source systems. Hardy et al. (2004)
developed a model to assess the effect of distributed rainwater storage (UrbanCycle), but did not
include consideration of interactions with low-flow hydrology. Significant advances in modelling of
urban hydrology has been developed by Rodriguez and others (Rodriguez, et al., 2008; Rodriguez, et
al., 2005) in the form of the distributed hydrological model, URBS. However, we are yet to see these
research tools being used in regular practice to determine the optimal arrangement of systems
throughout a catchment (based on an assessment of impacts on receiving water hydrology).
Despite this lack of empirical data, it would seem reasonable to hypothesise that restoring
hydrological water balances at more local scales would consequently help to restore catchment-scale
flow regimes towards their natural condition. It is hoped that the various scales at which the Cities as
Catchments project will undertake its research and demonstration will help to test this hypothesis.
Most other attempts to improve ecological condition of streams in urbanised catchments have focused
on reach-scale enhancement of physical habitat(Brown, 2000). Unfortunately, ecological effects of
such habitat enhancement often are not assessed. In almost all cases where assessments were
done, changes in biotic composition were small, with only a few taxa colonizing new habitat (Davis, et
al., 2003; Larson, et al., 2008; Purcell, et al., 2002; Sudduth & Meyer, 2006; Suren & McMurtrie, 2005;
Walsh & Breen, 2001).
A few studies have shown some limited success is possible in certain circumstances. Charbonneau
and Resh (1992) reported significant improvements in the composition of urban stream animal
assemblages following restoration, but that restoration involved both reach-scale habitat improvement
and catchment-scale actions such as the removal of sewage pollution from the creek. Larned et al.
(2006) successfully displaced an exotic submerged macrophyte from weed beds with a native
macrophyte species in a Christchurch stream, although this change in macrophyte species
composition had no effect on the degraded macroinvertebrate assemblage in the stream, and
subsequent attempts to replicate this success in other streams were less successful, and not
universally appreciated by residents (Suren, 2009). Groffman et al. (2005) showed that retention
structures in severely degraded urban streams acted as hot spots for nutrient processes. However,
such small-scale studies of nutrient retention need to be assessed against the efficiency of in-
catchment nutrient retention, and such a comparison has not yet been made. Furthermore retention
structures are often unsustainable in the highly modified flow regime of urban streams (Booth, 2005;
Frissell & Nawa, 1992).
The weight of evidence suggests that urban stream ecosystems are generally limited in their
ecological capacity by the impacts of urban stormwater runoff (Paul & Meyer, 2001; Walsh et al.,
2005b). Attempts at restoration of stream ecosystems by in-stream or riparian habitat improvement
are, therefore, likely to fail because they do not match the scale of the restoration action to that of the
constraining impact (Hobbs & Norton, 1996; Lewis et al., 1996). This situation is more strongly true for
urban (than rural) catchments because links between the catchment and the stream are more
pronounced.
The paradigm in urban stream management to date has been to consider riparian replanting activities
and broader catchment scale management such as provision of wetlands, entirely separately.
However impacts on stream ecology are a result of a complex interaction between catchment level
effects on water quality and hydrology, and local effects on organic matter supply. There is strong
evidence for an effect of riparian vegetation on stream macroinvertebrates, but this is contingent on
the degree of urbanisation in the upstream catchment. The actual mechanism for the interaction
between catchment and local scale impacts appears to be complex, but is likely to be mediated by
organic matter supply and retention, and the degree to which urbanisation has altered other habitat
variables such as sediment composition (Roy, et al., 2006). In most cases riparian replantings are
relatively small relative to the area of the catchment, and local scale restoration is simply not
adequate to mitigate for the majority of impacts on streams (Newall & Walsh, 2005; Suren &
McMurtrie, 2005).
Our review of the major impacts of urbanisation (Table 6) clearly indicates that the impacts of
urbanisation on streams are a product of catchment level and local scale impacts. The majority of
ecological components of interest are primarily affected by hydrologic and water quality impacts
occurring at a catchment scale as a result of stormwater. Although local riparian restoration can
restore organic matter supply, the retention and availability of that organic matter will be greatly