• grazing;
  • indicators;
  • rapid appraisal;
  • restoration;
  • riparian zone


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Grazing by introduced ungulate livestock is a major form of land use over large parts of Australia. Due to the tendency of stock to concentrate around water, riparian zones and wetlands are heavily impacted by grazing. However, little is known about how effects on riparian habitats vary spatially and with management regimes. We investigated how livestock affected riparian habitats on the Murrumbidgee River in south-eastern Australia.
  • 2
    A rapid appraisal index of the ecological condition of floodplain riparian habitats was developed. This measured habitat continuity and extent, vegetation cover, bank stability, soil structure, quantity of fallen debris, dominance of natives vs. exotics, and the presence of indicative species. The method could be readily adapted for use on other floodplain rivers with extensive riparian habitats.
  • 3
    Riparian condition was scored at 138 sites along 620 km of the Murrumbidgee River on private properties (n = 77), in State Forests (n = 27) and on Crown Land (n = 34). Riparian condition declined significantly with increasing grazing intensity and also with distance upstream in the upper half of the floodplain.
  • 4
    Stocking rate, distance upstream, relative periods of paddock rest and grazing, proportion of bank accessible to stock, and the presence of off-river water in the paddock, accounted for 76% of the variance in riparian condition.
  • 5
    Most riparian habitats on the Murrumbidgee River and other rivers in the Murray-Darling Basin are privately owned. Thus exclusion of the grazing industry from the riparian zone is not practical. However, lowered stocking rates, particularly in the upper parts of the catchment, resting of paddocks to allow recovery from grazing, and the provision of off-river watering points could all be used to improve riparian habitats.
  • 6
    Exotic plants are ubiquitous, occurring even where grazing has been excluded for many years. Thus restoration of riparian habitats will require weed removal even in areas not used by livestock.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Riparian habitats are important regional sites supporting high levels of biodiversity and have significant effects on material fluxes between terrestrial and riverine ecosystems (Naiman & Decamps 1997). Because they are at the boundary of terrestrial and aquatic systems, riparian areas are powerful indicators of catchment quality (Rapport et al. 1998).

Human settlement has always been focused on rivers, and human activity is often a major determinant of riparian structure and function (Dynesius & Nilsson 1994). The grazing and trampling activities of domestic livestock have had a particularly pervasive influence on riparian habitats (Fleischner 1994; Trimble & Mendel 1995).

In Australia, grazing by introduced ungulate livestock is the major land use over 60% of the land surface (Wilson 1990). Since European settlement, riverine landscapes and wetlands have been viewed by Australian farmers as watering points for stock. Riparian and wetland habitats and areas around artificial watering points in pastoral regions suffer greater impacts from domestic and feral grazing herds than do dryland habitats because stock concentrate around water sources (Robertson 1997; James, Landsberg & Morton 1999). These effects are exacerbated during hot, dry, periods of the year and during drought years, when water becomes scarce (Robertson 1998; James, Landsberg & Morton 1999).

Introduced livestock and grazing management practices are among the most widespread agents of chronic modification to land–water interfaces in Australia (McComb & Lake 1988; Wilson 1990; Walker 1993; Morton, Short & Barker 1995; Robertson 1998). Stock can have direct impacts on the geomorphology of riparian habitats (Trimble & Mendel 1995) and on vegetation and water quality (Robertson 1997). They indirectly affect faunal communities and the biogeochemistry of floodplain habitats by altering habitat structure and patterns of production in and around wetlands (Robertson 1998).

Previous investigations of the effects of livestock on riparian habitats have compared stock-exclusion plots with nearby areas where livestock are present (Kauffman & Krueger 1984; Fleischner 1994; Humphrey & Patterson 2000). We have shown, using this method, that, at scales of hundreds of metres, livestock have major negative impacts on the vegetation and soils of river banks in the Murrumbidgee River and its tributary streams in the Murray-Darling Basin of south-eastern Australia (Robertson & Rowling 2000). While these studies have highlighted the ways in which livestock alter riparian structure and function, there have been no investigations of catchment-scale variation in riparian habitat function with different stocking rates and livestock management regimes.

Here we describe a rapid appraisal method for the assessment of the ecological condition of riparian areas. We used this approach to investigate how the effects of livestock on riparian habitats vary at scales of hundreds of kilometres along the length of the Murrumbidgee River. We assessed which stock and land management practices contributed most to catchment-scale variation in the condition of river banks.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study area

The study area comprised a 620 km section in the middle reaches of the Murrumbidgee River, one of the larger rivers of the Murray-Darling Basin in south-eastern Australia (Fig. 1). The change in elevation along this section of the river is only 100 m, and the floodplain varies in width from hundreds of metres to several kilometres. Most of the river channel has a sandy–silty substrate and meanders occur along the entire section studied.


Figure 1. The study area showing the Murrumbidgee River in relation to other rivers in the Murray-Darling Basin. Also shown are major towns and locations referred to in the text.

Download figure to PowerPoint

The area has a Mediterranean climate, with hot dry summers (average maximum 31·2 °C in January) and cold damp winters (average maximum 12·5 °C in July). Average annual rainfall is 585 mm at Wagga Wagga but varies from 714 mm at Gundagai at the eastern edge of the study area to 366 mm at Hay on the western edge of the study area (Fig. 1).

The floodplain vegetation is dominated by the native river red gum Eucalyptus camaldulensis Dehnh., which defines the limits of flooding in the area. Thus the riparian zone of the Murrumbidgee River was defined as that area which was (or would have been prior to clearing) vegetated with river red gum. The region has been settled since the 1830s, has been extensively cropped and grazed over the last 150 years, and is now the primary agricultural region of Australia (Crabb 1997). Riparian habitats have been highly modified at local and regional scales by clearing, grazing by livestock and the introduction of exotic plant species (Margules & Partners Pty Ltd, P. J. Smith Ecological Consultants & Department of Conservation Forests and Lands Victoria 1990).

Ecological condition and rapid appraisal

Condition refers to the degree to which human-altered ecosystems diverge from local semi-natural ecosystems in their ability to support a community of organisms and perform ecological functions (cf. Karr 1999). In this study we used the term ecological condition to encompass both geophysical and biological attributes (Table 1).

Table 1. Functions of the riparian zone at different levels of organization, the components of the riparian ecosystem that perform those functions, and the indicators of the function used in this study (CWD = coarse woody debris)
Reduction of erosion of banksRootsTree cover
  Bank stability
Sediment trappingRoots, aquatic CWDTree cover
  Aquatic CWD
Controlling stream microclimate/discharge/water temperaturesRiparian forestTree cover
Filtering of nutrientsVegetation, soil, leaf litterGround cover vegetation
  Leaf litter cover
  Soil structure
Provision of organic matter to aquatic food chainsVegetationVegetation cover
  Leaf litter cover
Retention of plant propagulesTerrestrial CWD, leaf litterTerrestrial CWD
  Leaf litter cover
Maintenance of plant diversityRegeneration of dominant species, presence of important species, dominance of natives vs. exoticsAmount of regeneration
  Damage to regeneration
  Presence of Phragmites
  Dominance of native vs. exotic vegetation
Provision of habitat for aquatic and terrestrial faunaAquatic CWD, terrestrial CWD, leaf litter, standing dead trees/hollows, riparian forest, habitat complexityAquatic CWD
  Terrestrial CWD
  Leaf litter cover
  Standing dead trees
  Vegetation cover
  Number of vegetation layers
Provision of biological connections in the landscapeRiparian forest (cover, width, connectedness)Vegetation cover
  Width of riparian forest
  Longitudinal continuity of riparian forest
Provision of refuge in droughtsRiparian forestTree cover

Owing to the spatial scales of human impacts on the landscape and the need for assessment of ecological change, there is an expanding field of research focused on rapid appraisal techniques to measure ecosystem condition or integrity (Boulton 1999; Fairweather 1999). In order to investigate variation in the structure and function of the riparian zone of the Murrumbidgee River, we used rapid surveys of the ecological condition of riparian habitats. This approach enabled us to investigate relationships between land use and riparian condition at 138 sites along hundreds of kilometres of river bank, and complements our studies at smaller scales (Robertson 1997; Robertson & Rowling 2000).

An index of riparian condition

A synthetic approach, incorporating indicators of geophysical and biological properties and processes, is likely to provide reliable estimates of ecological condition in riverine ecosystems (Fairweather 1999; Boulton 1999). Ladson et al. (1999) described an index of stream condition based on 18 indicators that measured alterations to the hydrology, physical form, streamside vegetation, water quality and biota of streams. Recently we developed and tested an index for the rapid appraisal of the ecological condition of floodplain wetlands (Spencer, Robertson & Curtis 1998) using a subset of the indicators proposed by Ladson et al. (1999). For this work, we took a similar approach and chose indicators to reflect functional aspects of the physical, community and landscape features of the riparian zone, as defined by Naiman & Decamps (1997) (Table 1). The index was made up of six subindices, each with a number of indicator variables, as shown in Table 2: (i) habitat continuity and extent (HABITAT); (ii) vegetation cover and structural complexity (COVER); (iii) bank and soil structure and stability (BANKS); (iv) standing and fallen debris (DEBRIS); (v) dominance of natives vs. exotics (NATIVES); and (vi) indicative species (SPECIES).

Table 2. Subindices (and their weighting in the final score) and indicators of the index of riparian condition, the range within which each was scored, the method of scoring for each indicator, and the number of measurements per site for each indicator (n)
Sub-indexIndicatorRangeMethod of scoring n
HABITAT (10/50)Width of riparian vegetation0–4Width standardized by channel width (CW): 0 = < 0·25 × CW, 1 = 0·25–0·49 × CW, 2 = 0·5–1·49 × CW, 3 = 1·5–2·9 × CW, 4 = > 3 × CW10
 Longitudinal continuity of riparian vegetation0–40 = < 40% vegetated bank, 1 = 40–64% vegetated bank, 2 = 65–79% vegetated bank, 3 = 80–94% vegetated bank, 4 = > 95% vegetated bank, with one point taken off for each significant discontinuity 1
COVER (10/50)Canopy cover0–30 = absent, 1 = 1–30%, 2 = 31–60%, 3 = > 60% cover 4
 Understorey cover0–30 = absent, 1 = 1–30%, 2 = 31–60%, 3 = > 60% cover 4
 Ground cover0–30 = absent, 1 = 1–30%, 2 = 31–60%, 3 = > 60% cover 4
 Number of layers0–30 = no vegetation layers to 3 = ground cover, understorey and canopy layers 4
BANKS(10/50)Bank stability0–40 = extreme erosion with little vegetation, 1 = extensive erosion, 2 = moderate erosion with banks held by discontinuous vegetation, 3 = isolated patches of erosion with good vegetation cover, 4 = stable banks with no undermining and good vegetation cover 4
 Aquatic woody debris0–40 = none, 1 = parts of trees and branches along 25% of the bank, 2 = few large intact trees along > 25% of the bank, 3 = many large intact trees along > 25% of the bank, 4 = numerous large intact trees along > 50% of the bank 1
 Soil structure0–20 = no structure or soil organisms, 1 = roots and voids present, 2 = roots and voids plus either soil organisms or horizons evident 3
DEBRIS (10/50)Leaf litter0–30 = none, 1 = 1–30%, 2 = 31–60%, 3 = > 60% ground cover10
 Standing dead trees0–10 = absent, 1 = present 4
 Terrestrial woody debris0–30 = none, 1 = small quantities, 2 = abundant but some removed, 3 = abundant with no signs of removal 4
NATIVES (5/50)Canopy0–1Cover score for natives divided by the cover score for all vegetation 4
 Understorey0–1Cover score for natives divided by the cover score for all vegetation 4
 Ground cover0–1Cover score for natives divided by the cover score for all vegetation 4
SPECIES (5/50) Eucalyptus camaldulensis regeneration0–20 = none, 1 = scattered, and 2 = abundant seedlings 4
 Damage to regeneration0–20 = all damaged, 1 = some damaged, 2 = no damage 4
  Phragmites australis 0–10 = absent, 1 = present 4

The contribution of each subindex score to the overall condition index value was weighted, as shown in Table 2. The subindices NATIVES and SPECIES were given a lower weighting than the other subindices. While it was considered that regeneration of the dominant overstorey species (river red gum) was important, regeneration of this species is highly patchy in space and time (Margules & Partners Pty Ltd, P. J. Smith Ecological Consultants & Department of Conservation Forests and Lands Victoria 1990) so the presence of regeneration at one particular point in time may not be critical. Also, it is not known how much damage to river red gum can be sustained by seedlings before survival is reduced. In the case of the subindex NATIVES, there are few data available on how exotic species that have invaded natural habitats may continue to perform the functions of the original vegetation. Thus, the indicator dominance of natives vs. exotics was also difficult to score in relation to ecosystem function, and the contribution of NATIVES to the total condition index was reduced (Table 2).

Each site was a 1-km section of the riparian zone on one side of the river. The estimates for each indicator were averaged for each site, scored and weighted (Table 2), then summed to give a total score for each site. Potential scores ranged from 0 (worst condition) to 50 (best condition). In order to summarize some of our results we grouped total condition index scores for surveyed sites into five categories: very poor condition < 25; poor condition ≥ 25 < 30; average condition ≥ 30 < 35; good condition ≥ 35 < 40; and excellent condition ≥ 40.

Selection of sites

There were three categories of land tenure along the river: private properties used for grazing and cropping, State Forests used for production forestry and some grazing, and Crown Land used for a variety of purposes, including support of travelling stock and for recreation. Sites were chosen along the length of the river in each of these tenure types, subject to availability (for example no State Forests existed on the river east of Currawarna; Fig. 1). On private properties, sites were chosen in paddocks that (i) were large enough to include at least 1 km of river bank and (ii) were subject to a range of grazing regimes. A total of 138 sites (77 on private properties, 34 on Crown Land and 27 on State Forests) was visited to give broad coverage along the 620 km of river between Gundagai and Hay (Fig. 1).

Survey methods

All surveys were conducted by a single observer who had previously completed a week-long training period. All measurements were taken in mid-summer, when the river was running high for irrigation. At each site, four 100 × 5-m transects were evenly spaced length-wise along the river bank. Within each of these transects we measured the abundance of terrestrial coarse woody debris (> 10 cm in diameter), abundance of river red gum seedlings (< 1 m tall), grazing and trampling damage to river red gum seedlings, presence/absence of standing dead trees and the understorey reed Phragmites australis (Cav.) Steud., vegetation cover within three layers (ground cover: grasses, herbs, reeds and sedges to 1 m tall; understorey: herbs, reeds, shrubs and saplings 1–5 m tall; canopy: trees > 5 m tall) of both natives and all vegetation combined, number of vegetation layers, and bank stability (Table 2).

At 10 evenly spaced points along the river bank at each site we measured river width, width of the riparian vegetation (on the side of the river being assessed) and leaf litter cover on the ground (Table 2). At three evenly spaced points on the river bank at each site, a trowel was used to dig a small hole and soil structure was assessed for soil organisms, roots and voids (air spaces) and layering. Finally, an estimate was made of the abundance of aquatic coarse woody debris for the entire 1-km site and a diagram made of the vegetation along the river bank to determine the length and number of any discontinuities in canopy cover. The bank was considered to be vegetated if the riparian vegetation was at least 5 m wide, and significant discontinuities were gaps of at least 50 m with no vegetation greater than 5 m in width. All measures were recorded as rank scores in the field, except width and longitudinal continuity of riparian vegetation, and channel width, which were estimated to the nearest 5 m in the field and converted to scores prior to analysis.

Other site variables

In order to investigate the relationships between riparian condition and land management practices, we recorded information on other site variables. For each 1-km survey site we recorded tenure type (private property, Crown Land, State Forest of New South Wales forest reserve), paddock size and length of river frontage per paddock (for private properties), presence of off-river water in the paddock (for private properties), land use (for private properties: grazing or cropping) and the proportion of the total length of river bank with naturally sheer cliffs, i.e. erosional banks (as a measure of the proportion of the site inaccessible to stock).

We also recorded the distance along the Murrumbidgee River from its downstream confluence with the Murray River (Fig. 1). Distance upriver is a proxy measure of variation in rainfall, with mean annual rainfall increasing from 366 mm at Hay, near the most downstream survey sites, to 714 mm at Gundagai, near the most upstream sites. In addition, there are changes in the geomorphology of the floodplain and farming practices with distance upriver. For instance, below Tom Bullen (Fig. 1) the active floodplain is bordered by extensive plains and agricultural activity often centres around irrigated crops away from the river bank, rather than grazing adjacent to the river. Upstream of Tom Bullen, the floodplain becomes progressively more confined and grazing and farming are often concentrated close to the river.

Stocking rates and management practices within each paddock on the private properties were determined by interviewing the owners and managers. Stocking rates were relevant to the time at which the condition scores were measured but may have varied in the past. Private properties on the Murrumbidgee River run both sheep and cattle. Stocking rates were standardized to dry sheep (wethers) equivalents (DSE) per hectare per annum, using conversion factors for different types of stock (Table 3). Stocking strategies were classified as set stocked (continuously grazed), graze > rest (grazed for more than half of the year) and graze < rest (grazed for less than half of the year).

Table 3. Dry sheep equivalents (DSE) for different types of stock under Australian conditions (after Denney, Ridings & Thornberry 1990)
Rams 2
Wethers 1
Ewes 1·5
Weaner lambs 1·5
Steers 9
Cows 8
Cows and calves15
Weaner calves 6

Stocking rate within floodplain paddocks may not be the best estimate of the impact of livestock on riparian habitats because stock concentrate their activities at land–water interfaces (Robertson 1997). Thus we used cowpat density as an index of cattle activity at each site (Lange & Willcocks 1978). To do this, cowpats were counted in three evenly spaced 100-m long transect lines placed perpendicular to the direction of river flow from the river bank at the point of bank-full capacity. Cowpats were counted within 1 m of the line on each transect. Cowpats may remain intact for up to 18 months in some areas (A. Jansen, personal observation), so cowpat density reflects cumulative cattle activity over such a period. Cattle were the dominant livestock in the riparian zone. Thus while sheep did occur at some sites, it was not considered necessary to develop a similar index for sheep.

Data analysis

The contribution of subindices to the total condition scores was determined using step-wise multiple regression (SPSS 1996). Variation in condition scores and cowpat densities was modelled using general linear models (SPSS 1996). The general approach taken with modelling was to begin with all relevant factors and interactions, then remove those interactions that did not contribute significantly to the explanation of total variance of the variable under investigation. If a factor was not significant in interactions or singly, it was also removed from the model. Squared factors were added to models to detect non-linearity in the response to each continuous factor and were then removed if non-significant. All models were tested to assess whether error variances were equal between groups and errors were normally distributed. Proportional measures were arcsin-square root-transformed before analysis. Both cowpat densities and stocking rates were significantly correlated with distance upstream (R = 0·356, n = 138, P < 0·01 and R = 0·840, n = 45, P < 0·01, respectively). However, multicollinearity between factors was small enough to cause a less than two-fold variance inflation for any slope estimated in any model.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Contribution of subindices to riparian condition index scores

All subindex scores increased with increasing riparian condition scores, but some subindices contributed less than others (Fig. 2). For instance, the average relative abundance of exotic species (see NATIVES) was high at all sites, almost regardless of the total condition score, and the BANK indicators had relatively low scores, even at sites with overall good condition. Regression of the individual subindices against the total score showed that the DEBRIS subindex accounted for 86·3% of the variance in the total score (F1,136 = 858·5, P < 0·001). The addition of each subindex contributed significantly to improving the proportion of variance accounted for in the total score, in the following order: COVER (R2 = 0·929, change F1,135 = 123·2, P < 0·001); HABITAT (R2 = 0·967, change F1,134 = 160·5, P < 0·001); SPECIES (R2 = 0·985, change F1,133 = 153·9, P < 0·001); BANKS (R2 = 0·998, change F1,132 = 688·5, P < 0·001); and NATIVES (R2 = 1, change F and P inestimable).


Figure 2. Mean scores for each subindex (HABITAT, COVER, BANKS, DEBRIS, NATIVES and SPECIES; see Table 2) for all riparian sites in each category (1 = very poor, 2 = poor, 3 = average, 4 = good and 5 = excellent) based on the total index of riparian condition (n = 138).

Download figure to PowerPoint

Condition index and site variables

Condition scores recorded along the Murrumbidgee River between Gundagai and Hay ranged from 12·8 to 43·1. Sites that had not been grazed for several years and lightly grazed State Forests scored at the top of this range, while heavily grazed paddocks with little remaining vegetation, woody debris and litter, and eroding banks, scored at the bottom. No sites scored near the theoretical maximum (50) for the index, mainly due to the widespread occurrence of exotic species along the river and the lack of aquatic coarse woody debris in many sections of the river.

Broad relationships between condition scores and tenure and grazing intensity indicated by cowpat densities are shown in Figs 3–5. The riparian zone in the majority of State Forest sites was in good or excellent condition, while the lowest condition index scores were recorded on private properties and at some sites on Crown Land (Fig. 3). Of the five sites on private properties that scored as excellent, two had not been grazed for 10 years and one had been grazed only occasionally for very short periods of time over the last 9 years. The condition scores for private properties, State Forests and Crown Land were all significantly negatively correlated with cowpat densities (Fig. 4). All subindices making up the total condition score were also significantly negatively correlated with cowpat densities (Fig. 5).


Figure 3. Frequency distribution of site scores in each condition category for the three land tenure types surveyed in this study (n = 138).

Download figure to PowerPoint


Figure 4. Index of riparian condition scores in relation to cowpat densities at each site for the three land tenure types (n = 138).

Download figure to PowerPoint


Figure 5. Subindex scores from all survey sites in relation to cowpat densities (n = 138).

Download figure to PowerPoint

To examine these patterns in more detail, we analysed condition scores in relation to tenure and grazing intensity, with distance upstream as a covariate. Examination of the data revealed a non-linear relationship between distance upstream and condition scores. Given the changes in geomorphology and land management practices that occurred at Tom Bullen, we performed separate analyses for sites above and below this point (Table 4). Upstream of Tom Bullen, there was no significant effect of tenure but condition declined significantly with both increasing cowpat density and distance upstream (overall R2 = 0·55). Downstream of Tom Bullen, cowpat density explained 77% of the variance in condition scores and there were no significant effects of either distance upstream or tenure. Condition declined significantly with increasing cowpat density (Table 4).

Table 4. General linear models analysis of condition scores in relation to cowpat densities (COWPATS) and distance up the Murrumbidgee River from its confluence with the Murray River (RKM). The analysis was done separately for sites above and below Tom Bullen (at 630 river kilometres) (see text for explanation). Also shown are slope estimates (B) and standard errors for each factor in the models. n = 138 sites
SourceType III SSd.f. F P Partial R2 B SE
Above Tom Bullen        
Corrected model 2954·23 2 57·860·000   
COWPATS 1458·03 1 57·110·0000·380 0·1350·018
RKM  540·24 1 21·160·0000·185−0·0250·005
Error 2374·3393     
Below Tom Bullen        
Corrected model 1199·21 1132·720·000   
COWPATS 1199·21 1132·720·0000·768−0·2310·020
Error  361·4440     

Cowpat densities and stocking rates

For paddocks that contained only cattle, and in which no cropping occurred, the relationship between stocking rates and cowpat densities was highly variable. However, 50% of the variance in cowpat densities could be explained by stocking rates in combination with distance upstream and the presence of off-river water in the paddock (Table 5). The factors paddock area, length of river frontage in the paddock, and the proportion of bank accessible to stock did not contribute significantly to the model.

Table 5. General linear models analysis of cowpat densities in relation to stocking rates ha−1 annum−1 (DSE), distance up the Murrumbidgee River from its confluence with the Murray River (RKM), and the presence of off-river water in paddocks (WATER). Data are for private properties for which stocking rate data were available and for paddocks used only for cattle grazing and excluding data for paddocks where the cattle were grain-fed. n = 45 sites
SourceType III SSd.f. F P Partial R2
Corrected model 24720·36 5 7·710·000 
DSE  6507·70 110·150·0030·206
RKM  1447·44 1 2·260·1410·055
WATER  6391·15 1 9·960·0030·203
WATER × DSE  2915·42 1 4·550·0390·104
WATER × RKM  5227·13 1 8·150·0070·173
Error 25017·4139   

Stocking rates and condition scores on private properties

There was a relatively strong relationship between cowpat densities and stocking rates (see above). However, stocking rates are easily managed, so we examined condition scores in relation to stocking rates. In analysing the data for private properties, we excluded cropped paddocks, and those where stock were routinely grain-fed. We first examined relationships between condition scores and individual management variables. Condition scores declined significantly with increased stocking rates (R2 = 0·25, F1,63 = 20·32, P < 0·001; B = −0·73, SE = 0·16), distance upstream (R2 = 0·27, F1,63 = 23·21, P < 0·001; B = −0·02, SE = 0·01) and the proportion of bank inaccessible to stock (R2 = 0·09, F1,63 = 6·21, P = 0·015; B = −0·07, SE = 0·03). Condition scores also differed significantly between stocking strategies (F2,63 = 4·50, P = 0·015), with sites in set stocked paddocks having significantly lower condition scores than those in graze < rest paddocks. Sites in graze > rest paddocks were intermediate. No significant relationship was found between condition scores and the presence of off-river water in paddocks (F1,63 = 0·097, P = 0·756).

General linear models analysis of the condition scores for these paddocks showed that 76% of the variance in scores could be explained, as shown in Table 6. The factors paddock area and length of river frontage did not contribute significantly to the model. Although the presence of off-river water in the paddock was not found to be related to condition scores overall, it was included in the final model due to significant interactions with other factors. Post-hoc tests showed that condition scores in paddocks without off-river water (estimated marginal mean = 22·5, SE = 2·3) were significantly lower than those in paddocks with off-river water (estimated marginal mean = 29·2, SE = 1·3; F1,40 = 11·07, P = 0·002).

Table 6. General linear models analysis of condition scores for private properties in relation to stocking strategy (STRATEGY where 1 = set stocked, 2 = graze > rest and 3 = graze < rest), presence of off-river water in the paddock (WATER where 0 = river only and 1 = off-river water available), the proportion of river bank at each site that was naturally erosional (PROP), distance up the Murrumbidgee River from its confluence with the Murray River (RKM) and stocking rate ha−1 annum−1 (DSE). n = 64 sites
SourceType III SSd.f. F P Partial R2
Corrected model 2317·9523 5·570·000 
STRATEGY  314·35 2 8·680·0010·303
WATER    0·50 1 0·030·8700·001
PROP  324·20 117·910·0000·309
DSE  181·15 110·010·0030·200
RKM  242·69 113·410·0010·251
RKM2  237·44 113·120·0010·247
STRATEGY × PROP  165·22 2 4·560·0160·186
STRATEGY × RKM  312·60 2 8·640·0010·302
WATER × PROP   80·22 1 4·430·0420·100
PROP × DSE  156·11 1 8·630·0050·177
DSE × RKM  143·44 1 7·930·0080·165
STRATEGY × WATER  421·76 211·650·0000·368
STRATEGY × WATER × RKM  508·94 3 9·370·0000·413
STRATEGY × WATER × PROP  103·74 2 2·870·0690·125
WATER × DSE × RKM  110·65 1 6·110·0180·133
PROP × DSE × RKM  139·04 1 7·6820·0080·161
Error  723·9940   


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Riparian condition index

Indices of condition or integrity need to be benchmarked against relatively pristine sites in order to provide a measure of variation from natural situations (Boulton 1999). The upstream portion of the Murrumbidgee River catchment includes two sites where livestock have been excluded for more than 50 years (Robertson & Rowling 2000). These sites served as good reference sites for establishing the scoring ranges for indicator variables in our riparian condition index.

Although at least one of these two reference sites was subject to the grazing and trampling activities of introduced livestock during the period since European settlement until they were fenced off (i.e. 1830s–1940s), they both have an intact and continuous overstorey of river red gum, a well developed understorey of silver wattle Acacia dealbata Link trees, a groundcover including Phragmites australis, significant coarse woody debris and litter cover, seedlings and saplings of the overstorey trees, and very stable banks (Robertson & Rowling 2000). They do, however, have significant growth of exotic blackberry Rubus fruticosus L. agg. in the understorey. Thus even these reference sites would not obtain a maximum score of 50 using our condition index.

The condition index used in this study was composed of a number of indicators of ecological function. We analysed the performance of related groups of indicators (subindices) relative to the total index scores using the approach of Steedman (1988). We found that all subindex scores correlated positively with the total index scores. However, two of the subindices, NATIVES (cover of natives relative to exotics) and BANKS (composed of the indicators bank stability, aquatic woody debris and soil structure), contributed relatively little to the discrimination of sites based on total condition scores. For NATIVES this was because exotic plant species were ubiquitous in the riparian zone. In the case of BANKS, most sites we surveyed had relatively poor soil structure, presumably owing to the long history of livestock activity in the riparian zone.

Rapid appraisal indices for assessing the ecological condition of streams and wetlands (Spencer, Robertson & Curtis 1998; Ladson et al. 1999) are usually designed to assist agency staff and citizens groups in monitoring environmental change. Indices of ecosystem health, condition or integrity have been employed to investigate relationships between in-stream habitats and land and water management practices world-wide (Rapport et al. 1998; Boulton 1999), but such an approach has not been employed previously for riparian habitats, although indices are available (Ladson et al. 1999). We have shown here that rapid appraisal indices are valuable tools in research that requires coverage of many sites over large spatial scales. The riparian condition index could be applied to many rivers with extensive riparian habitats world-wide, as long as care is taken to adjust the scoring of indicator variables against relevant reference sites or data (Brinson & Rheinhardt 1996; Boulton 1999). Care must also be taken to minimize confounding influences in such broad-scale studies (Manel, Buckton & Ormerod 2000).

The ecological condition of murrumbidgee riparian habitats

At the catchment scale, riparian condition generally decreased with distance upstream. Several factors covary with distance upstream, including climate and rainfall, land management practices, past history of clearing and management, flooding frequency and duration, geomorphology and susceptibility to erosion.

The mean January (summer) temperatures and annual rainfall downstream at Hay, and upstream at Gundagai, are 32·9 °C and 366 mm, and 31·6 °C and 714 mm, respectively. This climatic difference has affected land management practices as well as the likelihood of exotic plant invasions. In the lower part of the catchment stocking rates of cattle and sheep are lower than upstream and paddocks are less continually grazed. The floodplain in the downstream portion of the catchment has been less heavily cleared and forest cover on river banks is more continuous. This region of the catchment is downstream of the major irrigation channels (upstream of Tom Bullen), and thus while annual flows are reduced in the region, when high flows or floods do occur, they are more likely to be in the natural spring flood period (EPA 1997). As successful recruitment to river red gum populations depends on high soil moisture in the late spring (Dexter, Rose & Davies 1986), it is likely that river red gum recruitment may often be more successful in the downstream part of the catchment. Finally, below Tom Bullen the active floodplain is bordered by extensive plains, and agricultural activity often centres around irrigated crops away from the river bank, rather than grazing adjacent to the river. Upstream of Tom Bullen, the floodplain becomes progressively more confined and grazing and farming are often concentrated close to the river.

Along the catchment, riparian habitats on the Murrumbidgee River are in generally poor condition with regard to exotic plant invasions. All sites had exotic plant species in the ground cover, and where understorey was present it consisted mainly of exotic species. Exotic trees such as willow Salixbabylonica L. drop their leaves 6 months later than river red gums (Schulze & Walker 1997), thus shifting the timing of organic matter input to river channels. In addition, exotic trees may not provide the same kinds of habitat for fauna as do native tree species (Ladson et al. 1999). However, it is not known whether exotic understorey species such as blackberry may fulfil some of the functions of native species. Certainly some vegetation, whether exotic or native, must be better than none in preventing erosion, filtering nutrients and providing habitat for fauna.

Fifty per cent of sites on private property were in poor or very poor condition, owing to a lack of woody debris, standing dead trees and leaf litter, highly fragmented and narrow overstorey tree cover, erosion of river banks, abundant exotic understorey taxa and no regeneration or Phragmites. While these sites functioned as livestock pasture, they were unlikely to provide the normal functioning of riparian habitats (Naiman & Decamps 1997) because they lacked habitat for ground-dwelling and arboreal fauna, were likely to contribute less organic matter to aquatic food chains, and were unlikely to act as efficient sites for the interception of materials in overland flows. Many poor condition sites on private property had very little or no regeneration of the dominant canopy tree, river red gum. This suggests a poor prognosis for the future of the riparian zone in these areas: once the existing trees die, there may be very little to take their place, resulting in further fragmentation and loss of habitat.

Only 7% of riparian sites on private properties scored in excellent condition. These sites experienced no or very infrequent grazing. In addition all of these sites were in unfragmented forest with good vegetation cover and had abundant terrestrial and aquatic woody debris and leaf litter. However, even these sites had abundant exotic species of ground cover and understorey plants.

Most (25 of 29) riparian sites in State Forests were in good or excellent condition (Fig. 3). State Forests are managed for production of red gum timber, and are characterized by relatively dense and continuous stands of trees. The masses of coarse woody debris and fine litter are relatively high in these areas (Glazebrook & Robertson 1999). Livestock are used to reduce understorey fuel loads to reduce the risk of fire, but stocking rates are often two orders of magnitude less than on private properties (Robertson 1997).

Eleven of the 34 Crown Land sites were in poor or very poor condition. These sites were all parts of travelling stock routes used by drovers moving livestock during dry conditions. They have thus been subject to uncontrolled intense grazing and trampling pressure for some decades. In contrast, most of the Crown Land in good or excellent condition (17 of 34 sites) was contained in recreation reserves, used for camping and picnicking. While these sites have abundant exotic plant species, they are generally in well forested areas.

The influence of livestock and their management on riparian condition

Cowpat density, our index of livestock activity, explained a high proportion of the variance in condition index scores, suggesting that livestock have a significant influence on riparian condition. Cowpat density reflects only current grazing pressure and may be influenced by rainfall. As rainfall increased with distance upstream, as did cowpat densities, our indicator of grazing intensity is likely to be conservative. The condition index measured a variety of factors, some of which are the result of historical events, such as the clearing of floodplain forests and the removal of coarse woody debris from the river, much of which occurred last century (Crabb 1997). Riparian condition is thus not solely a direct result of current grazing practices (Turner 1999). However, current grazing intensity is likely to be strongly correlated with past management practices. For example, on the floodplain of the Murrumbidgee and other rivers nearby (Fig. 1), stocking rates are higher on paddocks that have been more heavily cleared of river red gum (A. Jansen, unpublished data). Current grazing intensity, plus past management practices that have led to those stocking rates being adopted, both contribute to modern riparian condition.

Although stocking rates explained a significant amount of the variance in cowpat densities, it is clear that the activity of livestock (in terms of cowpat densities within 100 m of the river bank) is not simply a function of stocking rates in riparian paddocks. Grazing animals concentrate their activity around watering points (Fleischner 1994; Robertson 1997), so paddock-level stocking rate data do not reflect accurately effective stocking rates in riparian zones of paddocks. It is clear that our measure of grazing intensity was much more directly related to riparian condition than stocking rates per se. However, paddock-level stocking rates are more easily measured and managed than livestock activity so are more useful in management terms.

On private properties, a large proportion of the variance in condition scores could be predicted from several variables that could be easily managed by farmers (see below). Although the complexity of the model relating these variables to riparian condition (Table 6) precluded detailed examination of the effects of each variable in interaction with all other variables, several conclusions could be drawn. Condition was better in paddocks with lower stocking rates, where there were periods of rest from grazing, where more of the bank was accessible to stock, and where off-river water was available. Condition declined with distance upstream.

Management implications and options

In the south-eastern region of the Murray-Darling Basin most riparian lands are privately owned, and the length of riparian habitats on the many rivers in the region precludes fencing as a solution to the impact of livestock on riparian condition. While the cattle and sheep grazing industries remain economically viable, restoration of riparian habitat on private properties rests with farmers and their management of livestock.

Many of the factors that had a significant influence on the condition scores for riparian sites on private properties along the Murrumbidgee River, such as stocking rates, stocking strategies and off-river water supplies, are all readily amenable to management by farmers. Thus some clear conclusions can be drawn about how grazing management practices could be altered to improve riparian condition in different parts of the Murrumbidgee and other similar catchments in the south-eastern Murray-Darling Basin, where livestock grazing is a major floodplain land use.

High stocking rates (up to 20 DSE ha−1 annum−1) in the upstream sections of the Murrumbidgee floodplain are impacting negatively on riparian zone structure and function and need to be reduced. It is clear from some sites in excellent condition in downstream regions of the river that 5–10 DSE ha−1 annum−1 is a more desirable stocking rate. Some farmers in the upstream regions do use these stocking rates so they are clearly acceptable.

Most farmers in the upstream section of the study area use set stocking. Experience with stocking strategies, where for instance paddocks are rested for several months then grazed at high stocking rates for a few weeks, in downstream, more arid portions of the study area (P. Milliken, personal communication) and in similar areas elsewhere (Elmore 1992), indicates that rest periods should be part of livestock management in the upstream higher rainfall portions of the Murrumbidgee floodplain. Some farmers utilize rest periods during flood events and this strategy could be readily extended to non-flood years.

The provision of alternative watering points in riparian paddocks would also reduce the impacts of grazing and trampling on riparian habitats. This is likely to be most important during the drier summer and autumn months, when livestock have the most impact on floodplain habitats (Robertson 1997). A few farmers have already recognized the benefits to both their stock and the river banks of providing off-river water, and have done so.

The proportion of river bank accessible by livestock has a significant modifying influence on the impact of stock on riparian condition (Table 6). The smaller the proportion of accessible (i.e. not naturally erosional) river bank in a paddock, the greater will be the impact of a certain stocking rate on riparian condition because stock will be concentrated on parts of the river bank to access water and seek shade. Thus farmers in the upper regions of the catchment will need to keep stocking rates lower in riparian paddocks that have a higher proportion of steep river banks.

The condition of riparian habitats generally responds rapidly to the exclusion of stock or rotations aimed at decreasing the impact of stock (Elmore 1992; Fleischner 1994; Robertson & Rowling 2000). However, because the condition of riparian habitats on the Murrumbidgee floodplain is not wholly related to current grazing regimes, riparian condition will not approach that of relatively pristine sites for some time, purely in response to alterations in livestock management. For instance, sites from which stock have been excluded for > 50 years still have problems with exotic plants (Robertson & Rowling 2000). Farm management at the paddock scale aimed at increasing the condition of riparian habitats will thus require a focus on weed removal in conjunction with livestock management. Many farmers already have a programme of noxious weed control but this needs to be extended to all farms as well as to Crown Land and State Forests.

However, other factors, operating at spatial and temporal scales greater than can be controlled by individual farmers, also have a major influence on riparian habitat structure and function in the Murrumbidgee and other floodplain rivers in the region. Recruitment events for the dominant riparian tree, the river red gum, are driven by floods that occur in the mid- to late spring (Bren & Gibbs 1986; Dexter, Rose & Davies 1986). Current river regulation in the Murrumbidgee and other floodplain rivers in the region to supply water for summer irrigated crops has reduced the frequency of flooding in spring (Maheshwari, Walker & McMahon 1995). River flow management at the catchment scale will be required to maintain recruitment of the dominant riparian tree species (EPA 1997).

Provision of adequate riparian habitat structure at the local or even catchment scale may not necessarily benefit fauna dependent on riparian zones but whose populations are controlled at regional scales (Saab 1999). For instance, bird communities in the eastern and southern Murray-Darling Basin are in decline in response to widespread reductions in habitat quantity and quality (Robinson & Traill 1996; Fisher 1997). In order to restore riparian bird communities, regional networks of farmers will need to manage livestock to restore riparian habitat at large scales across all rivers in the region (Robinson & Traill 1996).


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Rohan Rowling for help in developing and testing the condition index, and for performing most of the field surveys. We are also very grateful to the many land managers who gave us detailed information on their management practices and access to their properties. Comments from two anonymous referees and the editors greatly improved the manuscript. The research was supported by an Australian Research Council Collaborative Grant C29700044 to A. I. Robertson.


  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Bren, L.J. & Gibbs, N.L. (1986) Relationships between flood frequency, vegetation and topography in a river red gum forest. Australian Forest Research, 16, 357370.
  • Brinson, M.M. & Rheinhardt, R. (1996) The role of reference wetlands in functional assessment and mitigation. Ecological Applications, 6, 6976.
  • Boulton, A.J. (1999) An overview of river health assessment: philosophies, practice, problems and prognosis. Freshwater Biology, 41, 469479.
  • Crabb, P. (1997) Murray-Darling Basin Resources. The Murray-Darling Basin Commission, Canberra, Australia.
  • Denney, G.D., Ridings, H.I., Thornberry, K.J. (1990) An analysis of the variation in wool production between commercial properties from a survey of a wheat-sheep shire in New South Wales. Australian Journal of Experimental Agriculture, 30, 329336.
  • Dexter, B.D., Rose, J.J., Davies, N. (1986) River regulation and associated forest management problems in the River Murray red gum forests. Australian Forestry, 49, 1627.
  • Dynesius, M. & Nilsson, C. (1994) Fragmentation and flow regulation of river systems in the northern third of the world. Science, 266, 753762.
  • Elmore, W. (1992) Riparian responses to grazing practices. Watershed Management: Balancing Sustainability and Environmental Change (ed. R.J. Naiman), pp. 442457. Springer, New York, NY.
  • EPA (1997) Proposed Interim Environmental Objectives for NSW Waters. Inland Rivers. Environmental Protection Authority of New South Wales, Chatswood, Australia.
  • Fairweather, P.G. (1999) State of environment indicators of ‘river health’: exploring the metaphor. Freshwater Biology, 41, 211220.
  • Fisher, A.M. (1997) The distribution and abundance of avifauna in the Bathurst landscape: implications for conservation and land management. PhD Thesis. Charles Sturt University, Wagga Wagga, Australia.
  • Fleischner, T.L. (1994) Ecological costs of livestock grazing in western North America. Biological Conservation, 8, 629644.
  • Glazebrook, H.S. & Robertson, A.I. (1999) The effects of flooding and flood timing on leaf litter breakdown rates and nutrient dynamics in a river red gum (Eucalyptus camaldulensis) forest. Australian Journal of Ecology, 24, 625635.
  • Humphrey, J.W. & Patterson, G.S. (2000) Effects of late summer cattle grazing on the diversity of riparian pasture vegetation in an upland conifer forest. Journal of Applied Ecology, 37, 986996.
  • James, C.D., Landsberg, J., Morton, S.R. (1999) Provision of watering points in the Australian arid zone: a review of effects on biota. Journal of Arid Environments, 41, 87121.
  • Karr, J.R. (1999) Defining and measuring river health. Freshwater Biology, 41, 221234.
  • Kauffman, J.B. & Krueger, W.C. (1984) Livestock impacts on riparian ecosystems and streamside management implications: a review. Journal of Range Management, 37, 430438.
  • Ladson, A.R., White, L.J., Doolan, J.A., Finlayson, B.L., Hart, B.T., Lake, S., Tilleard, J.W. (1999) Development and testing of an index of stream condition for waterway management in Australia. Freshwater Biology, 41, 453468.
  • Lange, R.T. & Willcocks, M.C. (1978) The relation between sheep-time spent and egesta accumulated within an arid-zone paddock. Australian Journal of Experimental Agriculture and Animal Husbandry, 18, 764767.
  • McComb, A.J. & Lake, P.S. (1988) The Conservation of Australian Wetlands. Surrey Beatty and Sons, Sydney, Australia.
  • Maheshwari, B.L., Walker, K.F., McMahon, T.A. (1995) Effects of regulation on the flow regime of the River Murray, Australia. Regulated Rivers: Research and Management, 10, 1538.
  • Manel, S., Buckton, S.T., Ormerod, S.J. (2000) Testing large-scale hypotheses using surveys: the effects of land use on the habitats, invertebrates and birds of Himalayan rivers. Journal of Applied Ecology, 37, 756770.
  • Margules & Partners Pty Ltd, P. J. Smith Ecological Consultants & Department of Conservation Forests and Lands Victoria (1990) River Murray Riparian Vegetation Study. Murray-Darling Basin Commission, Canberra, Australia.
  • Morton, S.R., Short, J., Barker, R.D. (1995) Refugia for Biological Diversity in Arid and Semi-Arid Australia. Biodiversity Series, Paper No. 4. Biodiversity Unit, Department of Environment, Sport and Territories, Canberra, Australia.
  • Naiman, R.J. & Decamps, H. (1997) The ecology of interfaces: riparian zones. Annual Review of Ecology and Systematics, 28, 621658.
  • Rapport, D.J., Gaudet, C., Karr, J.R., Baron, J.S., Bohlen, C., Jackson, W., Jones, B., Naiman, R.J., Norton, B., Pollock, M.M. (1998) Evaluating landscape health: integrating societal goals and biophysical process. Journal of Environmental Management, 53, 115.
  • Robertson, A.I. (1997) Land-water linkages in floodplain river systems: the influence of domestic stock. Frontiers in Ecology: Building the Links (eds N. Klomp & I. Lunt), pp. 207218. Elsevier Scientific, Oxford, UK.
  • Robertson, A.I. (1998) The effect of cattle on wetlands. Wetlands in a Dry Land: Understanding for Management (ed. W.D. Williams), pp. 195204. Environment Australia and The Land and Water Resources Research and Development Corporation, Canberra, Australia.
  • Robertson, A.I. & Rowling, R.W. (2000) Effects of livestock on riparian zone vegetation in an Australian dryland river. Regulated Rivers: Research and Management, 16, 527541.
  • Robinson, D. & Traill, B.J. (1996) Conserving Woodland Birds in the Wheat and Sheep Belts of Southern Australia. RAOU Conservation Statement No. 10. Royal Australasian Ornithologists Union, Melbourne, Australia.
  • Saab, V. (1999) Importance of spatial scale to habitat use by breeding birds in riparian forests: a hierarchical analysis. Ecological Applications, 9, 135151.
  • Schulze, D.J. & Walker, K.F. (1997) Riparian eucalypts and willows and their significance for aquatic invertebrates in the River Murray. Regulated Rivers: Research and Management, 13, 557577.
  • Spencer, C., Robertson, A.I., Curtis, A. (1998) Development and testing of a rapid appraisal wetland condition index in south-eastern Australia. Journal of Environmental Management, 54, 143159.
  • SPSS (1996) SPSS, Version 7.5.1, SPSS for Windows. SPSS Chicago, IL.
  • Steedman, R.J. (1988) Modification and assessment of an index of biotic integrity to quantify stream quality in southern Ontario. Canadian Journal of Fisheries and Aquatic Sciences, 45, 492501.
  • Trimble, S.W. & Mendel, A.C. (1995) The cow as a geomorphic agent – a critical review. Geomorphology, 13, 233253.
  • Turner, M.D. (1999) Spatial and temporal scaling of grazing impact on the species composition and productivity of Sahelian annual grasslands. Journal of Arid Environments, 41, 277297.
  • Walker, K.F. (1993) Issues in the riparian ecology of large rivers. Ecology and Management of Riparian Zones in Australia (eds S.E. Bunn, B.J. Pusey & P. Price), pp. 3140. The Land and Water Resources Research and Development Corporation and the Centre for Catchment and In-stream Research, Griffith University, Canberra, Australia.
  • Wilson, A.D. (1990) The effects of grazing on Australian ecosystems. Proceedings of the Ecological Society of Australia, 16, 235244.

Received 25 October 1999; revision received 25 May 2000