Ecol Res (2011) 26: 123–131

DOI 10.1007/s11284-010-0767-2

O R I GI N A L A R T IC L E

Omar L. Arnaiz • Andrea L. Wilson

Robyn J. Watts • Mark M. Stevens

Influence of riparian condition on aquatic macroinvertebrate

communities in an agricultural catchment in south-eastern Australia

Received: 24 January 2010 / Accepted: 18 September 2010 / Published online: 16 October 2010
 The Ecological Society of Japan 2010

Abstract Riparian vegetation is known to affect aquatic indices (abundance, Shannon diversity, SIGNAL and
macroinvertebrate communities through contributions SIGNAL2). Macroinvertebrate communities were sig-
of organic matter and shading. Despite the widespread nificantly different in stream reaches from different
degradation of riparian vegetation in Australia, there are riparian condition categories (ANOSIM; p < 0.05).
relatively few studies examining the effect of changes in Our results indicate that efforts to rehabilitate riparian
riparian vegetation on in-stream macroinvertebrate vegetation may have a positive effect on in-stream biota
assemblages on individual catchments. In particular, even when implemented at a relatively small scale by
information is lacking on the responses of macroinver- individual landholders.
tebrate communities in catchments dominated by agri-
culture, where farms that are managed at the paddock Keywords Ecological condition Æ Riparian vegetation Æ
scale result in riparian vegetation condition varying over Stream community Æ Rapid assessment Æ Biotic indices
relatively short distances. In this study, macroinverte-
brate assemblages were assessed from 12 reaches along a
25-km section of a small agricultural stream in south- Introduction
eastern Australia. Riparian condition was assessed using
in-stream coarse woody debris (CWD) levels and the It is well recognised that riparian vegetation influences
rapid appraisal of riparian condition (RARC) index, a stream macroinvertebrate communities through the
numerical system for categorising the health of riparian provision of organic matter and by shading the stream.
areas that incorporates sub-indices reflecting habitat Organic matter derived from riparian sources provides
continuity, vegetation cover, plant debris levels, native habitat and food to the macroinvertebrate community
vegetation dominance, and other indicative features. (O’Connor 1991, 1992; McKie and Cranston 1998;
There was a significant positive correlation between Flory and Milner 1999; Hession et al. 2003). Shade from
RARC scores and macroinvertebrate taxon richness riparian vegetation can indirectly influence potential
(p < 0.01), and also between CWD scores and macro- food and habitat sources through limiting light avail-
invertebrate taxon richness (p < 0.05). In contrast, ability and subsequent primary production (Bunn et al.
there was no significant correlation observed between 1997, 1999; Mackay and Marsh 2005). Shade can also
riparian condition and the other macroinvertebrate influence stream macroinvertebrate communities by
reducing water temperature, which can impact sensitive
species by limiting their biological functions, or elimi-
O. L. Arnaiz Æ A. L. Wilson nate them entirely from the ecosystem (Davies and
Institute for Land, Water and Society, Nelson 1994; Rutherford et al. 1997, 2004).
Charles Sturt University, Locked Mail Bag 588,
Wagga Wagga, NSW 2678, Australia Due to the strong association between riparian
vegetation and aquatic macroinvertebrate assemblages, it
R. J. Watts is likely that differences in more broadly defined indices
Institute for Land, Water and Society, Charles Sturt University, of riparian condition that include a vegetation compo-
PO Box 789, Albury, NSW 2640, Australia
nent will also correlate to differences in the composition
M. M. Stevens (&) of stream macroinvertebrate communities. Indeed, many
E H Graham Centre for Agricultural Innovation studies have reported different macroinvertebrate com-
(Industry & Investment NSW and Charles Sturt University), munities occurring in association with different riparian
Yanco Agricultural Institute, Private Mail Bag, zone habitats. However, these findings are often at fine
Yanco, NSW 2703, Australia
E-mail: [email protected] microhabitat scales, which can be difficult to extrapolate
124

up to larger scales (Downes and Hindell 2000), at very pasture species and often by abundant emergent mac-
large landscape scales where there are differences in the rophytes such as Typha sp. and Phragmites australis.
composition of macroinvertebrate communities between Reaches in better condition were either remnants or
different catchments (e.g. Ormerod et al. 1993; Weigel older revegetation projects (>15 years) with riparian
et al. 2000), or between reaches categorised by major zones up to 20 m wide and consisting of Eucalyptus
land-use changes (Scarsbrook and Halliday 1999; Danger camaldulensis with a sparse under storey of Acacia
and Robson 2004). dealbata, mixed native/exotic pasture ground cover and
In many agricultural catchments, the majority of patchily distributed macrophytes. A stratified random
riparian land is owned or managed by private land- approach was used when selecting study reaches to
holders. While some factors controlling riparian condi- ensure that reaches with similar riparian condition were
tion operate at spatial and temporal scales beyond the not located successively, but were instead interspersed
influence of these land managers, management practices throughout the length of the study area (Fig. 1).
adopted at the paddock scale by private landholders can
significantly influence the condition of riparian vegeta-
tion (Jansen and Robertson 2001a; Jansen et al. 2003). Sampling
There is currently a poor understanding of aquatic
community responses to changes in riparian condition Riparian zone assessments, macroinvertebrate sampling,
within agricultural catchments occurring at this reach in-stream coarse woody debris (CWD) assessments and
scale, despite grazing management practices having been environmental monitoring were all conducted during the
shown to have a significant impact on riparian zone later part of March 2005 (austral autumn) following
condition (Jansen and Robertson 2001a), riparian bird a slightly drier-than-average summer (rainfall 13%
communities (Jansen and Robertson 2001b; Martin below the long-term average). Water depths at this time
et al. 2006) and frog communities (Jansen and Healey varied from 8 to 60 cm and stream width varied from
2003). approximately 3 to 8 m, well below bankfull capacity
This study was conducted on Umbango Creek in (Page and Carden 1998). The 12 study reaches were each
south-eastern Australia. Umbango Creek is in an agri- assessed using the rapid appraisal of riparian condition
cultural catchment with a relatively uniform land-use index (RARC, Jansen et al. 2003) to provide a quanti-
pattern (grazing) but with varying land management and tative measure of ecological condition. The RARC con-
riparian zone management by different landholders. The sists of five sub-indices: habitat continuity, vegetation
aquatic macroinvertebrate community was compared cover, plant debris density, dominance of native vegeta-
across reaches with varying riparian conditions within tion, and additional indicative features. Each of these
this single stream. The aim of this study was to examine sub-indices comprises a number of indicator variables,
the relationships between riparian condition at the reach and is scored out of 10 to give a maximum score of 50. As
level and aquatic macroinvertebrate community rich- recommended by Jansen et al. (2003), the RARC meth-
ness, diversity, abundance and composition. odology was adapted to the conditions specific to the
study. In this study, 500-m reaches of creek were assessed
to reflect the management units used in this catchment.
Materials and methods Within each study reach, four evenly spaced 40-m-long
and 10-m-wide transects perpendicular to the direction of
Study area flow were assessed. The mean of the four transects of
each study reach was used to calculate the final RARC
Umbango Creek flows into Tarcutta Creek, which is a score. Maximum RARC scores of 50 could not be
tributary of the Murrumbidgee River in south-eastern achieved even in undamaged areas of the Umbango
Australia (Fig. 1). The stream is comprised of alternating Creek catchment due to the structure of the native plant
erosional and depositional reaches and carries a sandy communities in the area, and we therefore redefined the
bedload similar to other Murrumbidgee tributaries in the categories identified by total RARC scores. The four
area (Page and Carden 1998). The lower slopes and categories were: <15 very poor, 15–25 poor, 25–40
floodplain of Umbango Creek have been cleared of native average, and >40 good/excellent. The RARC system
vegetation and are dominated by agricultural land use, was considered suitable for this project as it has been
predominantly grazing. As a result of this land use, extensively tested in south-eastern Australian agricul-
riparian vegetation is discontinuous along the length of tural catchments that are similar to Umbango Creek, and
the stream, although relatively homogeneous at the has been shown to be a good indicator of the biodiversity
paddock scale. A lowland section of the stream was selected and functioning of riparian zones (Jansen et al. 2003).
for this study because there was little change in elevation Coarse woody debris within each study reach was
over its length (slope 0.2%, Page and Carden 1998). assessed as a measure of aquatic habitat complexity.
Twelve reaches in Umbango Creek were selected for CWD was measured using a rapid appraisal score where
this study. The reaches were selected to represent a range no woody debris scored 0, scattered small quantities
of riparian conditions. Poorer reaches lacked native scored 1, regular large quantities scored 2, and log jams
riparian vegetation and were dominated by exotic approximately the width of the creek scored the maximum
 125

Fig. 1 Location of the study

area and the distribution of
study reaches. Umbango Creek
flows northwards prior to
joining Tarcutta Creek. Dotted
lines indicate boundaries
between contiguous study
reaches (reaches 1 and 8,
and 6 and 10)

of 3. CWD levels were assessed using four transects per were upstream within the study reach. Conductivity,
study reach at the same location as the RARC transects. dissolved oxygen, pH and turbidity were recorded using
Transects were 10 m wide, perpendicular to the direction a Quanta Hydrolab (AquaLab Scientific Pty. Ltd.,
of flow and extended across the whole channel. The final Sydney, Australia). Flow and depth were measured
score for CWD at each study reach was calculated as the using a flow meter (TSE Inc. Michigan, USA). All
mean of the four transect scores. measurements were taken mid-channel between 9 a.m.
All physico-chemical and macroinvertebrate sam- and noon. Temperature was recorded over a period of
pling was conducted within 50 m of the downstream end 48 h at 15-min intervals using data loggers (Hastings
of each 500-m-long study reach. This point was con- Data Loggers, Port Macquarie, Australia; Tiny Tag and
sidered to represent the sum of the characteristics that Tiny Talk models).
126

Quantitative macroinvertebrate sampling was Additionally, the study reaches were grouped into
undertaken using Surber samplers with a 20 · 20 cm riparian condition classes according to their classifica-
base and 250-lm mesh. Five replicate samples were tion within our RARC categories and their overall
taken within each study reach approximately 10 m apart numerical scores were compared using ANOVA in order
longitudinally within the channel. All Surber sites were to ensure groupings were discrete before undertaking
chosen to have a similar substrate consisting of medium multivariate community analysis. CWD scores were also
to coarse (250–750 lm) sands with small quantities of compared across riparian condition classes using the
silt and small cobbles. Each additional microhabitat Kruskal–Wallis test.
(submerged overhanging bank vegetation, CWD and/or Transformed (Y = ln (y + 1)) taxon-specific abun-
aquatic macrophytes) present at the sampling area dance data for each study reach were used in the mul-
within the study reach was sampled using a sweep net tivariate analysis. The Bray–Curtis similarity coefficient
with a 950-cm2 mouth size and 250-lm mesh. Three was used to calculate similarity matrices for all com-
1-m-long ‘sweeps’ of each microhabitat were completed. munity analyses and a non-metric multi-dimensional
Faunal samples were preserved in 70% ethanol for scaling (NMDS) ordination plot was used to visualise
transport to the laboratory where they were sorted un- similarities between macroinvertebrate communities
der a microscope at 20· magnification and identified to occupying the study reaches. Analysis of similarities
family level in most instances. Exceptions were the (ANOSIM) was used to determine the statistical signi-
Nematoda, Oligochaeta and Collembola, which were ficance of any compositional differences between riparian
identified to phylum, class and order, respectively. Mites condition classes (Clarke and Warwick 1994). Macro-
were identified as Acarina only, due to very low numbers invertebrate assemblage differences between riparian
throughout the study area. classes were further explored by determining the taxa
that contributed most to community differences deter-
mined by ANOSIM. For this purpose, similarity per-
Statistical analyses centage analysis (SIMPER) was used. All components
of the multivariate analysis were conducted using
The relationships between the RARC scores and the PRIMER Ver. 5.2.9 software (PRIMER-E Ltd. 2002).
physico-chemical variables were explored using Pearson Populations of key taxa identified using SIMPER were
correlation in order to reduce the potential for coincid- compared across riparian condition classes using
ing physico-chemical changes obscuring the cause for ANOVA.
any differences in macroinvertebrate communities
(McKie and Cranston 1998). Several indices were cal-
culated to describe the macroinvertebrate communities Results
of each study reach. Taxa richness and total abundance
were assessed by calculating the total number of taxa Riparian condition and macroinvertebrate diversity
(Surber plus sweep samples, and also Surber samples and abundance
alone) and the number of individuals collected per reach
(Surber samples only), respectively. Macroinvertebrate RARC scores for the 12 study reaches ranged between
community evenness and richness were assessed in 5.39 and 35.33 (Table 1). Study reaches with the highest
combination using the Shannon diversity index, calcu- RARC scores were only within the ‘average’ condition
lated from the richness and abundance data (Surber range, despite being the most intact riparian reaches
samples only) for each study reach using the equation: within the study area. These low scores reflect the
H¢ = Ri qi(ln qi) (PRIMER-E Ltd. 2002). To further narrow width of remnant and restored trees, the scat-
explore possible correlations between riparian condition tered sections of exotic trees (e.g. willows), and the
and indices that describe macroinvertebrate community ground cover being dominated by exotic pasture
structure, SIGNAL2 scores (Chessman 2003) were cal- species. Significant differences in total RARC scores
culated using the Surber data to measure the dominance between riparian condition classes were identified by
of pollution-tolerant or pollution-sensitive taxa in the ANOVA (df = 2, 9, F = 88.143, p < 0.001), and a
macroinvertebrate assemblages. The data from both post hoc LSD test demonstrated that all three catego-
Surber samples and sweep samples were pooled to gen- ries were distinct from each other (p < 0.001). The
erate presence/absence data for each study reach, and three riparian classes represented at the study sites are
unweighted SIGNAL scores were then calculated for hereafter referred to as ‘very poor’, ‘poor’ and ‘average’
each reach using this presence/absence data (Chessman reaches.
1995). SIGNAL differs from SIGNAL2 in that taxa are Average CWD scores for each reach in the study area
not given a weighting in regard to their level of pollution were generally low, however scores were relatively vari-
tolerance. able within each riparian condition class. CWD scores
The relationships between the RARC scores, the five differed significantly between all riparian condition
RARC sub-indices and CWD scores, and aquatic classes, with higher mean CWD scores in the average
macroinvertebrate indices were examined by Pearson reaches and progressively lower amounts of CWD in the
correlation using SPSS for Windows (SPSS Inc. 2006). poor and very poor reaches (Kruskal–Wallis p = 0.015).
 127

Table 1 Rapid appraisal of riparian condition (RARC) index scores, including sub-index scores, for 12 study reaches in Umbango Creek,
south-eastern Australia

Study reach Habitat Cover Debris Native plants Features Total Riparian condition class

1 1.25 4.38 1.07 0.83 1.07 8.60 Very poor

2 0.63 4.58 2.86 1.25 0.71 10.03 Very poor
3 0.00 4.54 0.71 0.63 2.50 8.38 Very poor
4 0.00 3.96 0.36 0.00 1.07 5.39 Very poor
5 7.19 5.63 3.93 3.33 1.78 21.86 Poor
6 7.81 5.42 3.93 2.71 1.78 21.65 Poor
7 4.06 4.79 2.50 2.09 5.00 18.44 Poor
8 5.94 4.58 2.86 2.29 4.29 19.96 Poor
9 8.75 7.71 5.36 4.58 5.71 32.11 Average
10 9.38 6.46 7.50 5.62 5.71 34.67 Average
11 8.13 8.33 6.07 4.59 8.21 35.33 Average
12 8.13 6.67 5.36 3.11 3.93 27.20 Average

Sub-indices are scored out of 10, totals out of 50. Locations of study reaches are shown in Fig 1. Riparian condition classes are: <15, very
poor; 15–25, poor; 25–40, average; >40 good/excellent

Physico-chemical variables varied considerably There was a significant positive correlation between
between study reaches (Table 2). There was no signifi- both indices of taxon richness and the amount of in-
cant correlation between RARC scores and any of the stream CWD present in the study reaches (Table 3).
physicochemical variables (p > 0.05). However, there However, macroinvertebrate total abundance, Shannon
was a clear trend for maximum, minimum and mean diversity index and sensitivity (SIGNAL2 and un-
water temperatures averaged across reach categories to weighted SIGNAL scores) were not significantly corre-
decrease with improved riparian conditions. lated to in-stream CWD scores.
A total of 3397 individuals representing 40 different
families or higher level taxa were identified from the
Surber samples. An additional six taxa were present in Community composition
the qualitative sweep samples. The most abundant taxa
in the macroinvertebrate samples were chironomids The majority of invertebrate taxa identified in this study
(1221 individuals) and corixids (461 individuals), which were present in reaches of all three riparian condition
occurred in all reaches sampled. With the exceptions of classes. When a taxon was present exclusively in samples
Chironomidae, Baetidae, Caenidae, Leptophlebiidae from ‘poor’ or ‘very poor’ riparian condition reaches, it
and Corixidae, all taxa at each study reach had less than was invariably represented by only a single individual.
ten individuals on average. Of the more abundant Curculionidae (Coleoptera), Nepidae (Hemiptera) and
families, only the Chironomidae, Baetidae and Corixi- Tasimiidae (Trichoptera) were unique to ‘poor’ condi-
dae had more than ten individuals in more than one tion reaches, whilst Tetragnathidae (Araneae), Hirudi-
study reach. nidae (Hirudinea) and Corduliidae (Odonata) were
Shannon diversity index scores ranged from 1.36 to unique to ‘very poor’ riparian condition reaches. Taxa
2.31. SIGNAL2 and unweighted SIGNAL scores were unique to reaches in the ‘average’ riparian condition
also quite variable and all study reaches had low scores class were more abundant, and included Glossiphonii-
ranging from 3.23 to 4.35 and 2.88 to 4.20 for the two dae (Hirudinea, ten individuals), Elmidae (Coleoptera,
indices, respectively. five individuals), and Calamoceratidae (Trichoptera,
four individuals).
Significant differences were found between the macro-
Correlations between macroinvertebrate indices invertebrate assemblages present in reaches belonging
and riparian condition to different riparian condition classes when analysing
community data with ANOSIM (global R = 0.396,
RARC scores and macroinvertebrate taxon richness p = 0.002). Pair-wise comparisons of macroinvertebrate
were significantly positively correlated for both the total assemblages from riparian classes showed a significant
number of taxa per study reach and the total number of difference between ‘very poor’ and ‘average’ reaches
taxa from the Surber samples only (Table 3; Fig. 2). (R = 0.688, p = 0.029), and also between the ‘poor’ and
Each of the sub-indices of the RARC was also signifi- ‘average’ reaches (R = 0.375, p = 0.029). The ‘very
cantly correlated to both the total number of taxa and poor’ and ‘poor’ reaches were not significantly different
the total number of taxa found in the Surber samples (R = 0.208, p = 0.114). The NMDS plot illustrates this
alone (Table 3). There was no significant correlation pattern with the ‘poor’ reaches plotting between the ‘very
between RARC scores and macroinvertebrate abun- poor’ and ‘average’ ones (Fig. 3).
dance, Shannon diversity index, SIGNAL2 or Six taxa were identified by SIMPER analysis to
unweighted SIGNAL scores. contribute 5% or more of the dissimilarity between the
 128

‘average’ and ‘very poor’ riparian condition classes

Mean

16.3
16.1
15.5
16.9

16.2
14.8
17.0
14.2

13.1
16.4
15.0
13.4
based on the macroinvertebrate abundance data.
Ecnomidae and Ancylidae were significantly more
abundant in ‘average’ condition riparian reaches, whilst
Hydropsychidae were less abundant (ANOVA,
Temperature (C)

Max.

19.5
17.9
20.2
20.4

18.3
17.5
20.7
18.9

15.9
17.9
20.2
14.9
p < 0.05). None of these taxa were significantly corre-
lated with the RARC or its sub-indices (p > 0.05). The
remaining three taxa, Oligochaeta, Corixidae and
Baetidae, failed the homogeneity of variance test
Min.

9.8
13.8
13.8
12.7
14.3

14.3
12.5
14.3
10.9

14.2
12.2
11.9
(Levene’s statistic) despite transformation.

Discussion
Depth (m)

Several studies have found differences in aquatic macro-

0.10
0.20
0.15
0.47

0.08
0.18
0.15
0.20

0.60
0.20
0.25
0.35
Table 2 Coarse woody debris and physico-chemical variables recorded for 12 study reaches in Umbango Creek, south-eastern Australia

invertebrate assemblages associated with differing

riparian conditions at the reach scale within a single
stream. Danger and Robson (2004) studied stream
Flow velocity

macroinvertebrate community responses to a change

from forested to pasture-dominated riparian zones
across a single transition between these riparian habitat
(m/s)

0.24
0.03

0.04

0.11
0.14
0.18
0.16

0.01
0.18
0.01
0.18

types. They found that the macroinvertebrate commu-

–

nity of the pasture-dominated reach was significantly

different to that in the forested reach, and that samples
taken near the transition in riparian vegetation repre-
Turbidity

sented an overlap between the two communities.

(NTU)

160.0
81.5

46.5
80.2

47.1
41.7
21.8
57.1

89.1
47.5

56.6
3.5

Scarsbrook and Halliday (1999) also compared macro-

invertebrate communities across single transitions from
pastoral land use to forested riparian zones in three New
Zealand streams. Shifts in the macroinvertebrate com-
oxygen (mg/l)

munity occurred as little as 50 m into forested reaches

with the faunal composition comparable to reference
Dissolved

sites a further 250 m into the reach. This suggests that

10.81

11.10

10.95
7.85
7.18

9.70
8.66
7.88

7.50
7.32
6.30
7.59

aquatic macroinvertebrate communities can respond to

Location of study reaches is shown in Fig. 1. CWD scores range from 0 to 3

changes in riparian vegetation over relatively short dis-

tances. However, in studies conducted across single
transitions on individual streams, a variety of contextual
Conductivity

issues have the potential to influence the results

(Downes and Hindell 2000; Townsend et al. 2004).
(mS/cm)

Small-scale variability in geomorphology and hydro-

0.485
0.466
0.751
0.356

0.431
0.460
0.408
0.491

0.489
0.457
0.393
0.403

logical regimes prevents catchment-level patterns from

being reliably detected in studies conducted on a single
‘representative reach’ per river or catchment (Townsend
et al. 2004). This limitation logically extends to studies
pH

7.7
7.3
7.4
7.6

7.6
7.4
7.3
7.7

7.2
7.7
7.4
7.6

involving differences between a single pair of adjacent

reaches.
In Umbango Creek, differences in aquatic macroin-
CWD

vertebrate communities were found to correlate with

0.25
1.25
0.00
0.00

1.00
1.00
1.25
1.00

1.50
2.50
1.50
3.00

changes in the condition of riparian habitats. Impor-

tantly, these differences in community composition
CWD Coarse woody debris

occurred repeatedly over multiple reaches within the

Riparian condition class

same stream, each with different riparian conditions

despite largely consistent agricultural land use across the
entire study area. Reaches with better riparian vegeta-
tion condition had higher taxonomic richness. Different
and reach #

riparian condition classes also supported significantly

Very poor

Average

different macroinvertebrate assemblages. These shifts in

riparian condition from reach to reach are related to
Poor

10
11
12

landholder paddock-scale management practices, such

1
2
3
4

5
6
7
8

9
 129

Table 3 Pearson correlation analysis of the total numbers of macroinvertebrate taxa and the rapid appraisal of riparian condition index
scores (RARC), RARC sub-indices and in-stream coarse woody debris scores

Total number of taxa from Surber Total number of taxa from Surber
plus sweep net samples samples only

R p R p

RARC 0.845 *** 0.806 **

Habitat 0.708 ** 0.641 *
Cover 0.782 ** 0.863 ***
Native plants 0.757 ** 0.703 *
Debris 0.726 ** 0.703 *
Features 0.964 *** 0.906 ***
In-stream CWD 0.628 * 0.619 *

* p £ 0.05; ** p £ 0.01; *** p £ 0.001

Fig. 3 NMDS plot of macroinvertebrate communities from stream

reaches in the three riparian condition classes. Surber samples only.
Data transformed to Y = ln (y + 1), Bray–Curtis similarity.
Fig. 2 Correlation between rapid appraisal of riparian condition
Symbols represent individual study reaches
(RARC) index scores and the total number of taxa recorded at each
site (Surber plus sweep net samples). Pearson correlation p < 0.001
A recent study conducted in the Gellibrand River
as the fencing out of livestock from riparian zones and catchment of southern Victoria by Becker and Robson
land clearing history. (2009) provides an interesting comparison with our
Several studies have demonstrated that riparian con- results. Their study compared native forest, willow-
dition changes related to agricultural management result dominated, and revegetated sites, but did not include
in altered terrestrial fauna communities (e.g. Jansen and unrevegetated sites with minimal riparian vegetation, a
Robertson 2001b; Jansen and Healey 2003; Martin et al. more extreme situation that we found in the Umbango
2006). The driving mechanism for terrestrial community Creek catchment. In accordance with our results, Becker
change is commonly identified as habitat change, and and Robson (2009) found that water temperature was
terrestrial habitat differences occurring from reach to highest at recently revegetated sites with high incident
reach are a likely mechanism driving changes in aquatic light intensity, and that invertebrate abundance did not
macroinvertebrate communities in Umbango Creek. A differ significantly between site categories (defined in
particularly common land-use pattern in this system is their study by existing riparian vegetation type and in our
grazing by domestic livestock. This can have a significant study by RARC scores). In addition, Becker and Robson
influence on riparian habitats because stock grazing and (2009) found that macroinvertebrate community com-
trampling activity is typically concentrated around water position differed between some site categories during
sources (Trimble and Mendel 1995; Robertson 1997). In autumn, but not during spring. Revegetated sites re-
particular, this form of habitat damage can affect the mained significantly different to forest sites despite some
adult stages of aquatic macroinvertebrates and thus have sites having been revegetated up to 8 years previously.
a direct influence on adjacent stream communities. The effects of riparian condition on both in-stream
Harrison and Harris (2002) associated reductions in food and habitat availability are generally closely
macroinvertebrate abundance and richness with reduc- related, and as such it is difficult to separate their effects
tions in available bank and littoral habitats caused by on the macroinvertebrate community (Robertson and
livestock activity. Milner 2001). For example, in Australian streams where
130

riparian vegetation has been degraded, areas in better Scarsbrook and Halliday (1999) suggest discontinu-
condition have been shown to contribute to greater ous riparian vegetation could mitigate some of the
habitat complexity (Robertson and Rowling 2000) and effects on stream biota associated with pastoral land
higher quantities of palatable food (Bunn et al. 1999; use. We concur that there are benefits for the aquatic
Read and Barmuta 1999). While identifying the in- macroinvertebrate community if native riparian vege-
stream drivers of macroinvertebrate composition was tation is conserved or restored, even as relatively dis-
not the intention of this study, significant differences in continuous narrow buffers along reaches. As most
CWD quantities existed between reaches of different streams pass through multiple privately owned proper-
riparian condition classes. In line with other Australian ties, rehabilitation efforts can often only be undertaken
studies (O’Connor 1991, 1992; McKie and Cranston at the reach, paddock or property scale. This can lead
1998, 2001), this suggests that CWD may be a key to patchy longitudinal connectivity of riparian areas,
component in structuring aquatic macroinvertebrate with significant variation in the riparian resources
communities. available for macroinvertebrate communities. Our re-
Macroinvertebrate community changes that are sults show that rehabilitation of riparian vegetation can
associated with shifts in riparian vegetation at the potentially have a positive effect on in-stream biota
paddock-scale have been attributed to differences in locally, even if rehabilitated areas are discontinuous and
water quality. Storey and Cowley (1997) found macro- located within a catchment dominated by agricultural
invertebrate community composition in pastoral streams land use. Similar research on the macroinvertebrates of
approached that of forested control streams only when agricultural streams in other catchments would provide
pastoral streams were surrounded by remnant forest, a useful test of this finding. These studies could readily
and then only from a point approximately 600 m be undertaken in conjunction with stream rehabilitation
downstream from the riparian transition. This commu- projects.
nity-composition change was due to changes in water
temperature and its effect on dissolved oxygen concen- Acknowledgments We are grateful for assistance in the field from
trations, in addition to apparent changes in the in- Sascha Healy, Mark Wilson and Steven Sass. The authors also
thank the landholders of Umbango Creek catchment who allowed
stream processing of nutrients and dissolved solids access to their properties.
(Storey and Cowley 1997). A number of researchers
have proposed that the distribution of some aquatic
macroinvertebrates and related alterations in commu- References
nity composition occurring at the paddock-scale are due
to changes in water temperature (e.g. Davies and Nelson Becker A, Robson BJ (2009) Riverine macroinvertebrate assem-
1994; Bunn et al. 1999; Marsh et al. 2005). Water tem- blages up to 8 years after riparian restoration in a semi-rural
peratures in Umbango Creek showed very limited vari- catchment in Victoria, Australia. Mar Freshw Res 60:1309–
ability across riparian condition classes, and this 1316
Bunn SE, Davies PM, Kellaway DM (1997) Contributions of sugar
resulted in no significant relationships between water cane and invasive pasture grass to the aquatic food web of a
temperature and macroinvertebrate indices or commu- tropical lowland stream. Mar Freshw Res 48:173–179
nity composition being identified in this study. This low Bunn SE, Davies PM, Mosisch TD (1999) Ecosystem measures of
level of variability may be due to the length of study river health and their response to riparian and catchment
reaches used, which were limited by the size of the land degradation. Freshw Biol 41:333–345
Chessman BC (1995) Rapid assessment of rivers using macroin-
management units in Umbango Creek catchment, many vertebrates: a procedure based on habitat-specific sampling,
of which were only around 500 m in length. Davies et al. family level identification and a biotic index. Aust J Ecol
(2004) proposed that reaches of approximately 1000 m 20:122–129
at an appropriate level of shading (dependent upon Chessman BC (2003) SIGNAL 2—a scoring system for macro-
invertebrates (‘Water Bugs’) in Australian rivers. Monitoring
geographic region) would be required for water tem- River Health Initiative. Report 31, Commonwealth of Australia,
peratures to reach equilibrium in most first-order Canberra
streams. Additionally, sampling for our study was done Clarke KR, Warwick RM (1994) Change in marine communities:
during mid-autumn (late March). Water temperatures an approach to statistical analysis and interpretation. Natural
Environment Research Council, Bournemouth
that exceed the tolerance levels of sensitive macroinver- Danger AR, Robson BJ (2004) The effects of land use on leaf-litter
tebrate species are likely to occur only in the warmer processing by macroinvertebrates in an Australian temperate
months of the year (Johnson and Jones 2000; Ruther- coastal stream. Aquat Sci 66:296–304
ford et al. 1997; 2004). In our study, lower water tem- Davies PE, Nelson M (1994) Relationships between riparian buffer
peratures (averaged across reach categories) correlated widths and the effects of logging on stream habitat, invertebrate
community composition and fish abundance. Aust J Mar
with higher values for the ‘cover’ component of the Freshw Res 45:1289–1305
RARC scores, which reflect higher levels of shading. Davies P, Cook B, Rutherford K, Walshe T (2004) Managing high
Water temperature did not correlate closely with stream in-stream temperatures using riparian vegetation. River man-
depth; intermediate sites in the ‘poor’ RARC category, agement technical guideline no. 5, Land and Water Australia,
Canberra
which also had intermediate water temperatures had, on Downes BJ, Hindell JS, Bond NR (2000) What’s in a site? Varia-
average, shallower water than either the ‘very poor’ or tion in lotic macroinvertebrate density and diversity in a
‘average’ sites. spatially replicated experiment. Austral Ecol 25:128–139
 131

Flory EA, Milner AM (1999) Influence of riparian vegetation on O’Connor NA (1991) The effects of habitat complexity on the
invertebrate assemblages in a recently formed stream in Glacier macroinvertebrates colonizing wood substrates in a lowland
Bay National Park, Alaska. J North Am Benthol Soc 18:261– stream. Oecologia 85:504–512
273 O’Connor NA (1992) Quantification of submerged wood in a
Harrison SCC, Harris IT (2002) The effects of bankside manage- lowland Australian stream system. Freshw Biol 27:387–395
ment on chalk stream invertebrate communities. Freshw Biol Ormerod SJ, Rundle SD, Lloyd EC, Douglas AA (1993) The
47:2233–2245 influence of riparian management on the habitat structure and
Hession WC, Johnson TE, Charles DF, Horwitz RJ, Kreeger DA, macroinvertebrate communities of upland streams draining
Marshall BD, Pizzuto JE, Velinsky DJ (2003) Ecological ben- plantation forests. J Appl Ecol 30:13–24
efits of riparian reforestation in urban watersheds. In: Clar M, Page KJ, Carden YR (1998) Channel adjustment following the
Carpenter D, Gracie J, Slate L (eds) Protection and restoration crossing of a threshold: Tarcutta Creek, southeastern Australia.
of urban and rural streams. American Society of Civil Engi- Aust Geogr Stud 36:289–311
neers, Reston, pp 373–382 PRIMER-E Ltd. (2002) PRIMER 5 for Windows, ver. 5.2.9.
Jansen A, Healey M (2003) Frog communities and wetland con- PRIMER-E Ltd, Plymouth
dition: relationships with grazing by domestic livestock along Read MG, Barmuta LA (1999) Comparisons of benthic commu-
an Australian floodplain river. Biol Conserv 109:207–219 nities adjacent to riparian native eucalypt and introduced
Jansen A, Robertson AI (2001a) Relationships between livestock willow vegetation. Freshw Biol 42:359–374
management and the ecological habitats along an Australian Robertson AI (1997) Land-water linkages in floodplain river sys-
river floodplain. J Appl Ecol 38:63–75 tems: the influence of domestic stock. In: Klomp N, Lunt I (eds)
Jansen A, Robertson AI (2001b) Riparian bird communities in Frontiers in ecology: building the links. Elsevier, Oxford,
relation to land management practices in floodplain woodlands pp 207–218
of south-eastern Australia. Biol Conserv 100:173–185 Robertson AL, Milner AM (2001) Coarse particulate organic
Jansen A, Robertson A, Thompson L, Wilson A (2003) Develop- matter: a habitat or food resource for the meiofaunal commu-
ment and application of a method for the rapid appraisal of nity of a recently formed stream? Arch Hydrobiol 152:529–541
riparian condition. River management technical guideline no. 4, Robertson AI, Rowling RW (2000) Effects of livestock on riparian
Land and Water Australia, Canberra zone vegetation in an Australian dryland river. Regul Rivers
Johnson SL, Jones JA (2000) Stream temperature responses to Res Manag 16:527–541
forest harvest and debris flows in Western Cascades, Oregon. Rutherford JC, Blackett S, Blackett C, Saito L, Davies-Colley RJ
Can J Fish Aquat Sci 57:30–39 (1997) Predicting the effects of shade on water temperature in
Mackay SJ, Marsh N (2005) Interaction between hydraulic habitat small streams. N Z J Mar Freshw Res 31:707–721
and riparian vegetation in controlling aquatic macrophyte Rutherford JC, Marsh NA, Davies PM, Bunn SE (2004) Effects of
growth. In: Rutherfurd ID, Wiszniewski I, Askey-Doran M, patchy shade on stream water temperature: how quickly do
Glazik R (eds) Proceedings of the 4th Australian stream man- small streams heat and cool? Mar Freshw Res 55:737–748
agement conference. Linking rivers to landscapes. Department Scarsbrook MR, Halliday J (1999) Transition from forest to native
of Primary Industries, Water and Environment, Hobart, pasture land-use along stream continua: effects on stream eco-
pp 391–397 systems and implications for restoration. N Z J Mar Freshw
Marsh N, Bunn S, Rutherford JC (2005) Restoring the water Res 33:293–310
temperature of small streams with riparian vegetation: an SPSS Inc. (2006) SPSS for Windows, ver. 14.0.2, SPSS Inc.,
experimental study from southeast Queensland. In: Rutherfurd Chicago
ID, Wiszniewski I, Askey-Doran M, Glazik R (eds) Proceed- Storey RG, Cowley DR (1997) Recovery of three New Zealand
ings of the 4th Australian stream management conference: rural streams as they pass through native forest remnants.
linking rivers to landscapes. Department of Primary Industries, Hydrobiologia 353:63–76
Water and Environment, Hobart, pp 403–414 Townsend CR, Downes BJ, Peacock K, Arbuckle CJ (2004) Scale
Martin TG, McIntyre S, Catterall CP, Possingham HP (2006) Is and the detection of land-use effects on morphology, vegetation
landscape context important for riparian conservation? Birds in and macroinvertebrate communities of grassland streams.
grassy woodland. Biol Conserv 127:201–214 Freshw Biol 49:448–462
McKie B, Cranston PS (1998) Keystone coleopterans? Coloniza- Trimble SW, Mendel AC (1995) The cow as a geomorphic agent—a
tion by wood-feeding elmids of experimentally immersed woods critical review. Geomorphology 13:233–253
in south-eastern Australia. Mar Freshw Res 49:79–88 Weigel BM, Lyons J, Paine LK, Dodson SI, Undersander DJ
McKie B, Cranston PS (2001) Colonisation of experimentally (2000) Using stream macroinvertebrates to compare riparian
immersed wood in south eastern Australia: responses of feeding land use practices on cattle farms in southwestern Wisconsin.
groups to changes in riparian vegetation. Hydrobiologia J Freshw Ecol 15:93–106
452:1–14