Freshwater Biology (1998) 39, 171–178

APPLIED ISSUES

Influence of invasive macrophytes on channel

morphology and hydrology in an open tropical
lowland stream, and potential control by riparian
shading
S. E. BUNN
Centre for Catchment and In-Stream Research, Faculty of Environmental Sciences, Griffith University, Nathan, Queensland
4111, Australia

P. M . D AV I E S
Department of Zoology, The University of Western Australia, Nedlands 6907, Australia

D . M . K E L L A W AY
Centre for Catchment and In-Stream Research, Faculty of Environmental Sciences, Griffith University, Nathan, Queensland
4111, Australia

I . P. P R O S S E R
Cooperative Research Centre for Catchment Hydrology, CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601,
Australia

S U M M A RY

1. The catchments of many tropical lowland streams in far north Queensland have been
extensively cleared for the cultivation of sugar cane to the extent where very little of the
native riparian vegetation remains. Stream channels are often choked by a matrix of
introduced pasture grass (Brachiaria mutica, or para grass) and accumulated sediment
from cropland erosion.
2. Detailed transects across Bamboo Creek, a fourth order cane-land stream, revealed an
estimated sediment load of 20 000 t km–1. This has resulted in an estimated 85%
reduction in the predicted bankful discharge of the original stream channel. Channel
capacity has been reduced from 2.3 times to 0.3 times the predicted Q50 flood discharge
of 140 m3 s–1.
3. Shade cloth treatments of 50% and 90% across the stream were used to mimic the
effect of shading by riparian vegetation. Three months of shading resulted in a
substantial reduction in the height and standing biomass of para grass in both shade
treatments, compared to open plots (0% shade). The most dramatic effect was in the
90% treatment, where a mean reduction of 63% in height and 52% in total biomass was
recorded. This was despite high net primary production of para grass in the open plots
of 2.8 g dry wt m–2 day–1, which resulted in a overall increase of 11% and 28% in plant
height and total biomass, respectively.
4. These data suggest that restoration of native riparian vegetation will be an effective
long-term means of controlling invasive macrophytes in disturbed cane-land stream
© 1998 Blackwell Science Ltd 171
172 S.E.Bunn et al.
channels. Reduction of excessive macrophyte growth and the mobilisation of
accumulated sediment are essential to the restoration of natural hydrological and
ecological processes.

Introduction
Riparian vegetation has a controlling influence on influence of para grass and accumulated sediment
ecosystem function in most forested streams and on channel morphology and hydrology in a tropical
rivers, primarily by reducing solar radiation and effec- lowland stream; and (ii) to determine the effectiveness
tively limiting in-stream primary production (Boston of shading as a means of controlling invasive riparian
& Hill, 1991; Cummins et al., 1995). As a consequence, weeds such as para grass. The effects of para grass
food webs in most forested streams are thought to be on stream ecosystem processes, and the ecological
largely dependent on terrestrial inputs of carbon (e.g. implications of riparian restoration are discussed in
Hynes, 1975; Vannote et al., 1980; Bunn, 1993; Cummins Bunn et al. (1997).
et al., 1995). Reduction of shading of streams, as a
result of removal of riparian vegetation, often triggers
prolific growth of aquatic and semiaquatic macro- Materials and methods
phytes (Brookes, 1994; Gore, 1994; Wade, 1994). Study area
Although increased nutrient inputs arising from catch-
ment clearing are also likely to stimulate in-stream We conducted our study along a fourth order reach
plant production, light availability is considered to be (™ 500 m) of Bamboo Creek, a lowland stream near
the primary factor limiting macrophyte distribution Innisfail in the wet tropics of far north Queensland,
and abundance (Canfield & Hoyer, 1988). Australia (145°59.89E, 17°349S). Approximately 63% of
Dense stands of macrophytes act as an effective trap the 800 ha catchment above the study area has been
for sediments and can lead to dramatic changes in converted to sugar cane which is often grown up to
stream morphology and hydrology (Wade, 1994). In the high bank, and very little native riparian vegetation
addition to the obvious loss of benthic habitat, high remains. The riparian zone is now dominated by para
respiration rates of living plants, together with micro- grass, which grows prolifically along stream margins
bial respiration of unconsumed detritus in stream and often extends into and over the channel.
sediments, may affect stream benthic communities
through reduced availability of dissolved oxygen and
Effect of para grass on channel morphology and
other associated changes in water quality (Bourg &
hydrology
Loch, 1995; Forstner, 1995; Bunn et al., 1997).
Many tropical lowland streams in far north Three cross-sectional measurements of channel profile
Queensland, Australia, have catchments extensively and sediment depth were made at three of the five
cleared for the cultivation of sugar cane so that very sites selected for the shade experiment. The transects
little native riparian vegetation remains. Para grass within sites were separated by at least 10 m and the
(Brachiaria mutica Forssk. Stapf) is one of several weeds distance between sites was at least 100 m (see below).
that has become well established in these disturbed Sediment depth was recorded at 2 m intervals along
riparian zones (Bunn et al., 1997). The species was the cross section using a 3 m pole, which was pushed
originally introduced to this region in 1884 to control into the substrate until the point of first resistence,
bank erosion, and in recent years has been actively which marked the boundary between noncohesive
promoted as a ponded pasture grass (Middleton, 1991). and cohesive materials. All cross sections were refer-
Unfortunately, it has become a major weed in disturbed enced to the high bank at the edge of the floodplain, the
stream channels in northern New South Wales and stream bed and surface water level using a surveyor’s
Queensland (Sainty & Jacobs, 1988), and in the wet- dumpy level.
lands of the wet–dry tropics of Australia (Humphries The hydrological implications of the reduction in
et al., 1994). channel capacity were examined in two ways, by
The aims of this study were (i) to describe the comparing (i) bankful discharge, and (ii) the capacity
© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178
 Control of invasive macrophytes in a tropical stream 173

to convey a major flood with an annual recurrence frames (™ 14 m 3 7.4 m) were erected over each of the
interval (ARI) of 50 years, with and without sediment plots (3 treatments 3 5 sites) in early August 1995.
and para grass. Bankful discharge was estimated as These spanned the stream channel and were strained
mean velocity (m s–1) 3 mean area (m2). Mean vel- by steel posts to keep them above the para grass,
ocity was calculated using Manning’s formula: v 5 about 1.5 m above the water surface. Treatments within
R2/3 s1/2n–1, where R 5 mean hydraulic radius (m), sites were deliberately arranged, using five of the six
s 5 channel slope, and n 5 Manning roughness possible spatial configurations, to ensure that any
coefficient (Chow, 1959). The mean hydraulic radius potential confounding upstream–downstream effects
(R 5 area/wetted perimeter) was calculated from were stratified across treatments. Scaffold planks were
each of the nine cross sections. A channel slope used to minimise trampling of the plots during con-
value for Bamboo Creek of 0.03 was used, and struction of the wire frames, however some trampling
roughness coefficients (n) were estimated to be 0.04 of para grass occurred around the edges. Although
without para grass, based on a clean winding shade cloth was fastened only to the 50% and 90%
channel (Newbury & Gaboury, 1993), and 0.1 with treatments, the level of disturbance in the open control
para grass, based on dense grass beds (Prosser & plots was similar. The width of the wetted stream
Slade, 1994). Mean bankful channel width was channel was measured at each of the 15 treatment/
calculated using the high bank measurements from site plots.
the nine cross-sections.
The Rational Method (Pilgrim, 1987) was used to
Data collection
estimate the discharge of a flood with an ARI of
50 years: Q50 5 0.278 C5050IdtA, where C50 5 runoff Five quadrats (1 m2) of para grass were randomly
coefficient for 50 years, A 5 catchment area (km2), collected from along the stream, before the shade cloth
50I 5 average rainfall intensity (mm h–1) for design treatments were erected, to provide a measurement of
dt
duration (dt), and ARI of 50 years. A C50 value of pre-manipulation standing biomass, and to estimate
0.8 was determined from information presented by the number of post-manipulation quadrats needed to
Pilgrim (1987) using values recorded for Innisfail. A detect a significant change with a statistical power of
modified version of the Friend Formula was used to 0.8 or greater, and therefore minimise the likelihood
determine the time of concentration for flood duration of a Type II error (Cohen, 1988). Para grass was
for Bamboo Creek (dt 5 2.512 h). There are insufficient cropped with hand-shears and separated into two
gauging data from the wet tropics to estimate floods batches: an upper standing crop of green stems, blades
of other ARIs with any confidence. and shoots, and a lower matrix of stems above the
dense root mass. The upper and lower batches from
each quadrat were washed, dried in an oven at 80 °C
Design of shading experiment
for 43 h, and weighed. Below-ground biomass was
The experiment was undertaken using two shade cloth not collected.
treatments (50% and 90%) to mimic the effect of Calculations of statistical power indicated that 8–10
shading by a riparian tree canopy. An open treatment replicates would suffice to detect a ‘medium’ reduction
(0% shade) was used as a control. The two shade in biomass (30–50%) with a β of 0.8 (Table 1; see
treatments span the range of values of below-canopy Cohen, 1988; Dallal, 1988). Based on these calculations,
light for nearby forested streams in similar size catch- 10 samples (5 from each bank) of upper and lower
ments. This is based on measures of canopy cover, biomass of para grass were randomly collected on 12
and estimates of above and below canopy light (ter November 1995, from each of the three treatments at
Steege, 1993). The mean percentage of below:above each site. Samples were washed, dried and weighed
canopy light (6 1 SE, n 5 8) for nearby forest streams as described above.
was 31.3 6 2.6% (unpublished data). In addition to above ground biomass, para grass
Each of the three treatments was positioned at five height was measured before (31 July to 2 August
randomly chosen sites along the stream. The distance 1995), during (28 September 1995) and after (12 Nov-
between treatments within sites was at least 10 m, and ember 1995) the shading experiment. Five random
the distance between sites was at least 100 m. Wire measures of stem height of para grass on each bank
© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178
174 S.E.Bunn et al.
Table 1 Sample sizes required to detect reductions in upper
(320.9 6 46.1 g m–2) and lower (726.9 6 174.9 g m–2) standing
biomass of para grass, with statistical power (β) ù 0.8, based
on initial collections from Bamboo Creek in August 1995
(mean 6 1SE)

Sample size required

Fig. 1 Example of a cross section of the Bamboo Creek

Reduction Upper biomass Lower biomass channel, showing para grass and the degree of sedimentation.

30% 20 12
40% 12 7
50% 8 5
60% 6 4
70% 5 4

were made for all treatments at all sites with a

graduated pole. In the middle sampling period, meas-
urements were made beneath the shade cloth.

Data analysis

A oneway ANOVA was used to compare para grass

biomass (upper and lower) among shade treatments,
using Bank and Site as blocks. There was no significant
Bank or Site effect, and post hoc pair-wise comparisons
of treatment means were made to identify significant
differences in biomass. This included comparisons
with the mean values for the pre-shade treatment
biomass.
A split plot ANOVA design was used to compare
Fig. 2 Mean biomass (6 1 SE) of the upper and lower
para grass height among treatments, over the three components of para grass from 0, 50 and 90% shade
time periods. Replicates were nested within treatment, treatments on Bamboo Creek after three months of shading
blocked for both site and bank. Error 1 (replicates (n 5 10). The initial (pre-shade) biomass of each component is
within treatment) was used to test effects of shade also given (n 5 5). Significantly different means are labelled
with different letters (a–c).
treatment, site and bank. The effect of time and the
time–treatment interaction were tested using a pooled
46 m3 s–1. Channel capacity was reduced from 2.3
error term, including all other interaction terms with
to 0.3 times the predicted Q50 flood discharge of
time (error 2).
140 m3 s–1.

Results
Effect of shading on para grass biomass and height
Effects on channel morphology and hydrology
The initial mean dry weight of para grass before
Soft sediment occupied ™ 20 m2 of the Bamboo Creek shading was 1.05 6 0.11 kg m–2 (n 5 5), with most of
channel cross sections, which translates to an accumu- the biomass consisting of the lower matrix of stems
lated load of ™ 20 000 t km–1 (Fig. 1). Mean wetted (0.73 6 0.08 kg m–2 dry wt) (Fig. 2). After three months,
channel width was only 1.3 6 0.05 m (n 5 15), com- total biomass had increased in the open (0% shade)
pared with stream bankful width of 15.7 m. plots to 1.34 6 0.05 kg m–2 dry wt, with a rate of net
The accumulated sediment resulted in an 85% reduc- primary production of 2.81 g dry wt m–2 day–1. The
tion in the predicted bankfull discharge from 316 to slight decrease (15%) in biomass of the lower matrix
© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178
 Control of invasive macrophytes in a tropical stream 175
Table 2 Oneway ANOVAs comparing (a) upper and (b) lower
biomass of para grass among treatments after 3 months of
shading. Site and Bank were included as blocks

Source df MS F P

(a) Upper biomass

Treatment 2 781133.4 83.6 ,0.0001
Bank 1 6157.6 0.7 0.43
Site 4 15145.7 1.6 0.20
Error 22 9339.7
(b) Lower biomass
Treatment 2 205483.0 25.6 ,0.0001
Bank 1 2410.2 0.3 0.59
Site 4 4122.6 0.5 0.73
Error 22 8017.0

was countered by a substantial increase in production

of the upper layer (3.84 g dry wt m–2 day–1) (Fig. 2).
After three months of shading, para grass biomass
Fig. 3. Mean height (6 1 SE) of para grass from 0, 50 and 90%
was significantly lower in the 50% and 90% shade
treatment plots on Bamboo Creek, before (August), during
treatment compared with the open control (Table 2, (September) and at the end of the shading experiment
Fig. 2). By far the greatest effect was observed in the (November) (n 5 10). Significantly different means are labelled
upper layer under 90% shade, where the biomass was with different letters (a–c).
only 22% of that in the open controls. In this treatment,
Table 3 Split plot ANOVA comparing para grass height
very little of the remaining biomass was made up of among treatments (0, 50 and 90% shade) and times (0, 2 and 3
green shoots, and much of the lower layer appeared months of shading). Site and Bank were included as blocks
to be dead stems.
The ANOVA indicated a highly significant inter- Source df MS F P

action (time 3 shade treatment) in para grass height Treatment 2 12.76 440.0 ,0.0001
(Table 3, Fig. 3). Although there was an 11% increase Site 4 0.67 2.2 0.012
in mean height in control plots (0% shade) over the Bank 1 0.68 23.5 ,0.0001
three month period, this was not significantly different Error 1 142 0.03
Time 2 4.59 228.5 ,0.0001
from the initial values (Fig. 3). However, shading Time*treatment 4 3.34 166.1 ,0.0001
resulted in a 26% reduction in mean grass height Error 2 294 0.02
under the 50% treatment, and a 63% reduction in
mean height under the 90% treatment after three
thought to explain most of the variation in the distribu-
months. Most of this change had occurred after the
tion and abundance of macrophytes (e.g. Canfield &
first two months of shading (Fig. 3).
Hoyer, 1988; Wade, 1994). Extensive clearing of ripar-
ian vegetation in the tropical cane-lands of northern
Discussion Australia has created ideal conditions of light, water
and nutrients for invasive macrophytes such as the
Effect of para grass on hydrological and ecological
introduced para grass. The resulting proliferation of
processes
plant growth has clearly had a dramatic effect on
Native riparian vegetation controls fundamental eco- channel morphology, and flood capacity. Channels
system processes in forested river systems. Removal choked by aquatic macrophytes inevitably become
of this vegetation provides a graphic demonstration sites of sediment deposition, reducing the effective
of the degree to which plant production in stream channel capacity (Humphries et al., 1994; Smith et al.,
channels is controlled by light. Although catchment 1995). These hydrological and morphological changes
and riparian clearing is often accompanied by signi- substantially increase the incidence of flood events,
ficant nutrient inputs to streams, shading alone is which must translate into a significant economic cost
© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178
176 S.E.Bunn et al.

to cane farmers through erosion of top soil and direct in northern Australia where introductions of other
loss of crops. exotic pasture grasses (e.g. Hymenachne amplexucaulis
In addition to the direct effects on channel morpho- and Echinochloa polystachya) are actively promoted.
logy and hydrology, macrophyte invasion and sedi- These species may pose a greater threat to streams
mentation together have a pronounced influence on and wetlands than para grass because they can grow
aquatic ecosystem function (Bunn et al., 1997). Aside in up to 1.2 m of water and can be up to 50% more
from the obvious direct loss of aquatic habitat, high productive (Wildin & Chapman, 1987; Humphries
benthic respiration from accumulated organic matter et al., 1994). Lonsdale (1994) suggested that introduced
has resulted in limited oxygen penetration of fine bed pasture species, such as these, have more than an 80%
sediments, and few benthic organisms were recorded chance of becoming major weeds. Riparian revegeta-
by Bunn et al. (1997). Macrophytes apparently contrib- tion offers many other ecological benefits to streams.
ute very little to the food webs of streams and rivers, These include a source of leaves, fruits and arthropods
neither directly through herbivory nor indirectly via for aquatic food webs, trapping of sediment and
a detrital pathway (France, 1996). This seems to be nutrients from overland flow, reduction in water tem-
particularly true for C4 plants, including para grass peratures and restoration of terrestrial habitat for
(Forsberg et al., 1993; Bunn et al., 1997). Much of this wildlife (Bunn, 1993).
unconsumed plant biomass remains in the stream bed.

Fate of accumulated sediment

Effectiveness of riparian shading
Although riparian shading can effectively reduce mac-
The experimental data obtained in this study suggest rophyte growth, the restoration of ecological and
that riparian shading may be a highly effective means hydrological processes in these streams cannot be
of controlling invasive macrophytes such as para grass achieved until the vast store of accumulated sediment
in disturbed stream channels. Although the para grass (estimated at 20 000 t km–1) is removed from the chan-
was not killed in shaded plots during the three month nel. Complete removal of macrophyte cover would
period, plant biomass in both shade treatments was leave an exposed organic clay substrate. Clays can be
significantly reduced, despite very high rates of pro- mobilised by boundary shear stresses of 20 N m–2
ductivity in open plots (2.81 g dry wt m–2 day–1). (Chow, 1959), which in Bamboo Creek would be
Although plot size was approximately that of the achieved by an average flow depth over the clay of
canopy cover offered by a single riparian tree, plant ™ 7 cm. Consequently, small floods would be capable
viability beneath shaded areas could have been main- of mobilising the stored sediment.
tained by vegetative growth outside. Therefore, we Mobilisation of accumulated sediment, and down-
would expect riparian shading to be more effective stream transport of this material could, however, pose
than this at greater spatial scales. The potential effects problems for coastal marine ecosystems (Yellowlees,
of physical and allelopathic interference of macrophyte 1990). There are 38 km of lowland stream choked with
growth by established trees may also enhance long- para grass in the lower Johnstone River catchment,
term control. We were unable to continue the experi- giving an estimated total sediment storage of 760 000
ment through the summer wet season because shade t. It is reasonable to assume that it would take at least
frames were likely to have been lost during high flows. 10 y to mobilise the sediment, giving an annual
Macrophyte growth is likely to be further suppressed sediment load of 38% of the 200 000 t load transported
during winter (June to August) when incident light to the coast during a large flood, and 14% of the
levels are lower than during the study period. 560 000 t annual load of sediment delivered to streams
Shading by vegetation offers a long-term, cost- from cropland erosion (Hunter, in press).
effective and ecologically sound means of controlling
aquatic weeds, compared with chemical and mechan-
Acknowledgments
ical methods (Wade, 1994; Smith et al., 1995). The latter
are often less effective because of the likelihood of re- We gratefully acknowledge the assistance of staff from
invasion, or weed-by-weed replacement (Harris, 1988; the Queensland Department of Natural Resources,
Hobbs & Mooney, 1993). This is particularly important especially Rob Lait, John Amprimo, Tom McShane,
© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178
 Control of invasive macrophytes in a tropical stream 177

Paul Nowlan, and others at the South Johnstone and Stream Ecosystems (eds C.E. Cushing, K.W.
Research Station. Andy Page, from the Johnstone Cummins and G.W. Minshall), pp. 1–80. Ecosystems of
Catchment Centre in Innisfail, assisted with the selec- the World 22. Elsevier, Amsterdam.
tion of the study site and liaison with local land Dallal, GE (1988) DESIGN: a supplementary module for
Systat and SYGRAPH. Systat Inc., Evanston, IL.
owners. Jim Gangi gave us permission to work at the
Forsberg, BR, Araujo-Lima, CARM, Martinelli, LA,
study site and provided valuable information on past
Victoria, RL & Bonassi, JA (1993) Autotrophic carbon
conditions in Bamboo Creek and the occasional sight-
sources for fish of the Central Amazon. Ecology, 74,
ing of large reptiles. Chris Marshall assisted with the 643–652.
shade cloth experiment. Deborah Weekes calculated Forstner, U (1995) Non-linear releases of metals from
changes in channel capacity and flood frequency, aquatic sediments. Biogeodynamics of Pollutants in Soils
Thorsten Mosisch estimated forest canopy light and Sediments; Risk Assessment of Delayed and Non-Linear
regimes, and Michael Arthur assisted with the data Responses (eds W. Soloman and W.M. Stigliani), pp. 89–
analysis. This project was funded by the Land and 96. Springer. Berlin.
Water Resources R & D Corporation under the France, RL (1996) Stable isotopic survey of the role of
National Riparian Lands Program grant to SB macrophytes in the carbon flow of aquatic foodwebs.
(GRU17). Vegetatio, 124, 67–72.
Gore, JA (1994) Hydrological change. The Rivers Handbook.
Hydrological and Ecological Principles. Vol. 2 (eds P.
Calow and G.E. Petts), pp. 33–54. Blackwell Scientific
Publications, Oxford.
References
Harris, P (1988) Environmental impacts of weed control
Boston, HL & Hill, WR (1991) Photosynthesis-light insects. Bioscience, 38, 542–548.
relations of stream periphyton communities. Limnology Hobbs, RJ & Mooney, HA (1993) Restoration ecology
and Oceanography, 36, 644–656. and invasions. Nature Conservation 3: Reconstruction of
Bourg, ACM & Loch, JPG (1995) Mobilization of heavy fragmented ecosystems (eds D.A. Saunders, R.J. Hobbs
metals as affected by pH and redox conditions. and P.R. Erhlich), pp. 263–276. Surrey Beatty & Sons,
Biogeodynamics of Pollutants in Soils and Sediments; Risk Chipping Norton, NSW.
Assessment of Delayed and Non-Linear Responses (eds W. Humphries, SE, Groves, RH & Mitchell, DS (1994) Plant
Soloman and W.M. Stigliani), pp. 115–127. Springer, invasions: homogenising Australian ecosystems.
Berlin. Conservation Biology in Australia and Oceania (eds C.
Brookes, A (1994) River channel change. The Rivers Moritz and J. Kikkawa), pp. 149–170. Surrey Beatty &
Handbook. Hydrological and Ecological Principles. Vol. 2 Sons, Chipping Norton, NSW.
(Eds P. Calow and G.E. Petts), pp. 55–75. Blackwell Hunter, HM (1997) Nutrients and suspended sediment
Scientific Publications, Oxford. discharge from the Johnstone River catchment during
Bunn, SE (1993) Riparian–stream linkages: research needs cyclone Sadie. Cyclone Sadie Flood Plumes in the Great
for the protection of in-stream values. Australian Barrier Reef Lagoon: Composition and Consequences. Great
Biologist, 6, 46–51. Barrier Reef Marine Park Authority Workshop Series,
Bunn, SE, Davies, PM & Kellaway, DM (1997) No. 22, pp. 1–8.
Contributions of sugar cane and invasive pasture grass Hynes, HBN (1975) The stream and its valley.
to the aquatic food web of a tropical lowland stream. Verhandlungen der Internationalen Vereinigung für
Marine and Freshwater Research, 48, 173–179. Theoretische und Angewandte Limnologie, 19, 1–15.
Canfield, DE & JrHoyer, MV (1988) Influence of nutrient Lonsdale, WM (1994) Inviting trouble: introduced pasture
enrichment and light availability on the abundance species in northern Australia. Australian Journal of
of aquatic macrophytes in Florida streams. Canadian Ecology, 19, 345–354.
Journal of Fisheries and Aquatic Sciences, 45, 1467–1472. Middleton, CH (1991) History of para grass in
Chow, VT (1959) Open-Channel Hydraulics. McGraw-Hill, Queensland. Probing Ponded Pastures Workshop.
New York. University of Central Queensland, Rockhampton, 4 pp.
Cohen, J (1988) Statistical Power Analysis for the Behavioral Newbury, RW & Gaboury, MN (1993) Stream Analysis and
Sciences, 2nd edn. Lawrence Erlbaum Associates. Fish Habitat Design. Newbury Hydraulics Ltd, Gibsons,
Hillsdale, New Jersey. British Columbia.
Cummins, KW, Cushing, CE & Minshall, GW (1995) Pilgrim, DH (1987) Australian Rainfall and Runoff: A Guide
Introduction: An overview of stream ecosystems. River to Flood Estimation. Australia, Institution of Engineers.

© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178

178 S.E.Bunn et al.
Prosser, IP & Slade, CJ (1994) Gully formation and the Canadian Journal of Fisheries and Aquatic Sciences, 37,
role of valley floor vegetation, southeastern Australia. 130–137.
Geology, 22, 1127–1130. Wade, PM (1994) Management of macrophytic vegetation.
Sainty, GR & Jacobs, SWL (1988) Waterplants in Australia. The Rivers Handbook. Hydrological and Ecological
Sainty & Associates, Sydney. Principles. Vol. 2 (eds P. Calow and G.E. Petts), pp. 363–
Smith, C, Youdan, T & Redmond, C (1995) Practical 385. Blackwell Scientific Publications, Oxford.
Wildin, JH & Chapman, DG (1987) Ponded pasture systems –
aspects of restoration of channel diversity in physically
capitalising on available water. Queensland Department
degraded streams. The Ecological Basis for River
of Primary Industries, Rockhampton. 35pp.
Management (Eds D.M. Harper and A.J.D. Ferguson).
Yellowlees, D (1990) Land Use Patterns and Nutrient
John Wiley & Sons, Chichester.
Loading of the Great Barrier Reef Region. Proceedings
ter Steege, H (1993) HEMIPHOT. a Program to Analyze of the Workshop held at the James Cook University of
Vegetation Indices, Light and Light Quality from North Queensland, 17–18 November, 1990. 234, pp.
Hemispherical Photographs. The Tropenbos Foundation, (Sir George Fisher Centre for Tropical Marine Studies,
Wageningen, The Netherlands. James Cook University of North Queensland).
Vannote, RL, Minshall, GW, Cummins, KW, Sedell, JR
& Cushing, CE (1980) The river continuum concept. (Manuscript accepted 7 August 1997)

© 1998 Blackwell Science Ltd, Freshwater Biology, 39, 171–178