Managing grazing to increase ground cover in rangelands: using remote sensing to detect change

Introduction

With the introduction and expansion of pastoralism throughout the Australian rangelands, significant landscape degradation occurred, primarily a result of overgrazing by both livestock and introduced rabbits, in combination with extended periods of drought (Greene and Tongway 1989; McKeon 2004). Degradation includes soil erosion (wind, sheet and gully), growth of invasive native shrubs, and an overall decline in perennial grass content and pasture quality. This degradation reduced the productive capacity of grazing land and the provision and support of important ecosystem services including air and water quality, biodiversity, and climate mitigation (Pickup 1996; MacLeod and McIvor 2006; Stafford Smith et al. 2007). Furthermore, recent research indicates that in the absence of widespread adoption of land management practices that promote soil carbon sequestration and retention in the rangelands, rangelands will become a considerable source of greenhouse gas emissions, and landscape function will continue to be significantly and irreversibly constrained in the future (Chappell et al. 2019; Viscarra Rossel et al. 2024).

Given rangelands occupy 81% of Australia’s land mass and support approximately 30% of Australia’s livestock production (Bell et al. 2014; Foran et al. 2019), improving grazing management and reversing degradation in these areas can have important economic and ecological impacts (Bell et al. 2014). In recent years, there has been a shift in societal preferences, public policy and industry values towards multifunctional landscapes that deliver agricultural productivity, alongside environmental and societal outcomes (Briske et al. 2020; Witt et al. 2021). For example, environmental stewardship targets are now embedded in many industry frameworks, including the Australian Beef Sustainability Framework (ABSF 2024), Sheep Sustainability Framework (SSF 2024), National Farmers Federation 2030 roadmap (NFF 2018), and are also an essential component of Australia’s commitment to meeting the United Nations Sustainable Development Goals (UN SDGs).

To meet national and global sustainability goals and maximise access to emerging ecosystem service markets, there remains an ongoing need to identify management practices that can achieve improvements in land condition while maintaining agricultural production. Improving management of livestock grazing by matching stocking rates to feed availability (carrying capacity), providing periods of rest and recovery from grazing, and adapting grazing systems to seasonal conditions has been demonstrated to increase ground cover, productivity and plant diversity, and support ecosystem service provision including carbon sequestration in rangelands throughout the world (Byrnes et al. 2018; di Virgilio et al. 2019; McDonald et al. 2019a; Teague and Kreuter 2020). However, in southern Australian semi-arid rangelands, there have been only a small number of quantitative studies assessing the impact of grazing management systems, and in particular, few that have assessed multi-paddock grazing systems (Hacker and McDonald 2021). Those that have been undertaken were primarily one-off, fence line comparison studies across relatively small scales (e.g. Alemseged et al. 2011; Waters et al. 2017; McDonald et al. 2018, 2019b), the limitations of which are well recognised (Landsberg et al. 2003; Teague and Barnes 2017). To meet industry goals and demonstrate environmental stewardship, a robust, scalable, repeatable and management-sensitive method is required (Briske et al. 2020).

Ground cover (in the context of this paper, defined as the cover of the soil surface by living and non-living organic material including plants, litter, cryptogam, coarse woody debris and dung, for example) is an important indicator of land condition in the arid and semi-arid Australian rangelands (Beutel et al. 2021), and is correlated with many other indicators of ecosystem condition including biodiversity (Ludwig et al. 2004; Ward and Kutt 2009; McDonald et al. 2018), higher soil carbon stocks (Orgill et al. 2017; Wang et al. 2018; Gray et al. 2022), reduced runoff and water erosion (Greene et al. 1994), reduced wind erosion (Leys et al. 2023), landscape function (Tongway and Hindley 2004) and productivity (Pickup 1996). However, detecting changes in ground cover associated with changed management can be challenging because of the large scale of rangeland grazing properties, low rainfall and variable climate, the variability in soil type, vegetation and grazing patterns across management areas, and slow timeframes over which changes occur (Bastin et al. 2024). Remote sensing has the potential to overcome some of these challenges and to provide cost-effective, objective data across larger spatial and temporal scales (Beutel et al. 2019; Retallack et al. 2023), thus supporting producer decision making and enabling adaptive management by monitoring key indicators of landscape condition (Jansen et al. 2022).

In monitoring the impact of management change on ground cover, particularly in rangeland environments where climate variability is a primary driver of change (Pickup 1996), it is important to separate the impact of management from landscape and climatic drivers (Bastin et al. 2012; Barnetson et al. 2017). One approach to achieve this is by comparing ‘treatment’ areas with dynamic benchmark areas (‘controls’) of comparable land types, and that are subject to similar climatic factors (e.g. Bastin et al. 2012; Zhang and Carter 2018; Donohue et al. 2022). This approach has been successfully used to demonstrate improvement in ground cover from management change (e.g. Beutel et al. 2021; Bartley et al. 2023; Bastin et al. 2024). However, no studies have examined effects of intensive multi-paddock grazing management with flexible stocking rates matched to carrying capacity at property-scale, across multiple properties and using this dynamic benchmark remote sensing approach in southern Australian rangelands.

This study aimed to quantify any change in ground cover and assess the potential to achieve environmental outcomes through changed grazing management (adoption of multi-paddock grazing and flexible stocking rates matched to carrying capacity) in NSW semi-arid rangelands. In addition, the study aimed to identify which land types and in what preceding rainfall conditions changes in ground cover were most apparent.

Methods

Study location

This study was undertaken on seven extensive grazing properties (7100–17,000 ha) in NSW semi-arid rangelands in three key bioregions – the Darling Riverine Floodplain, Cobar Peneplain and Murray Darling Depression (Fig. 1, IBRA7 2012). Average annual rainfall across the region ranges from 500 mm in the east to less than 200 mm in the west (BOM 2024). Average temperatures across the region range from 24—33°C in January to 9–15°C in July (BOM 2024). Vegetation across the seven properties is predominantly native open grassland, chenopod shrubland, and Acacia, Casuarina, Eucalyptus or Callitris woodland, with some clearing of invasive native shrubs and annual cropping (winter cereal and fodder crops) on the Cobar Peneplain.

Fig. 1.

Location of study sites (S1–S7) in the NSW rangelands and the Interim Biogeographic Regions of Australia (IBRA7 2012). The red outline indicates the boundary of the Australian rangelands.

Results

Average median ground cover across the seven study sites between 1989 and 2023 ranged from 43% (S2) to 72% (S4), with significant temporal variability (depending on site, a difference of approximately 50–80% between maximum and minimum cover over this period) precluding the ability to detect any obvious trends over this period (Table 2, Fig. 2). Overall, ground cover was approximately 15% higher in winter than summer (Table 2).

Fig. 2.

Median seasonal ground cover (%) across each site and associated benchmark area between 1989 and 2023. The vertical dotted line represents the date at which grazing management began to change across each site.

Prior to implementation of management changes, ground cover across five of the seven study sites was decreasing relative to that of the benchmark areas (Fig. 3). Following the management change of flexible stocking rates and introducing rest in the grazing system, a significant positive change in ground cover (ranging from 2 to 7% from the time of management change, P < 0.05), as determined by a significantly different slope between the pre- and post-management periods, was observed at four of the seven study sites (S2, S3, S5, S6; Figs 3 and 4). A nonsignificant positive change was observed at a fifth site (S1, P = 0.45; Fig. 4). At S4, there was no significant difference overall, although this site had confounding impacts of a clearing and an occasional cropping cycle and had the highest average ground cover of all the sites. At S7, there was no significant change in ground cover in the 20 years from management change relative to the benchmark area.

Fig. 3.

The trend of difference in total ground cover (%) between the study site and benchmark areas (GC_S−B). The straight lines represent the median slope of the trend pre- and post-management change (Pohlert 2023). The vertical dotted line represents the date at which grazing management began to change across each site. Seasonal (3 monthly) rainfall totals are represented by blue bars on secondary axis.

Fig. 4.

The median (±95% confidence intervals) slope (analysed as Sen’s slope, Pohlert 2023) of the trend of the difference between the study site (S1–S7) and benchmark areas pre- and post-management change. Positive values represent increasing cover on the property relative to the benchmark and negative values vice versa. Where the confidence interval overlaps zero, this indicates no significant trend in the change in ground cover between site and benchmark areas.

Trends reported here are a simplified representation of temporally dynamic time series. When examined at a finer temporal resolution, trends over time are nonlinear and at times are comprised of shorter-term reversals in trends. Assessment of the linear trends for each land type within a property also indicate that the direction and magnitude of change was dependent on land type (Supplementary Fig. S1). There was a weak negative association between the trend in GC_S−B post-management change with the normalised difference water index (NDWI), total standing dry matter (TSDM) and green reflectance (r² = −0.36, −0.15 and −0.16 respectively), potentially indicating that less productive land types had the greatest positive response to the management change (Supplementary Fig. S2).

Analysis of the effect of preceding 12 month rainfall on GC_S−B and the trend of GC_S−B pre- and post-management change (Fig. 5a) showed that in wet years the trend in GC_S−B was negative across all sites (significant at S1 and S7; Fig. 5b). This indicates that ground cover in the benchmark areas increased more (or decreased less), than the study sites during these periods. There was no significant difference in the trend in GC_S−B pre- and post-management change at any study site during dry periods, but in intermediate rainfall years the trend in GC_S−B was significantly more positive post-management change at S2 (Fig. 5a).

Fig. 5.

Comparison of (a) the trend of the slope of difference in ground cover between the study site and benchmark (GC_S−B) and (b) median ground cover GC_S−B between pre- (yellow) and post- (blue) management change for sampling periods where preceding 12 month rainfall was in the lowest (Dry), middle (Intermediate) and highest (Wet) tercile of rainfall years during the analysis period. Black dots indicate individual data points for the slope (a) and median difference (b) for each sampling period, and the boxes represent the inter quartile range of data. Significant differences (P < 0.05) of the mean between pre- and post-management change are indicated by an asterisk. (same as previous caption)

Discussion

This study demonstrated the capacity for improved grazing management (matching stocking rate to carrying capacity and providing regular periods of long rest from grazing) to increase ground cover and associated ecosystem services across four of the seven heterogeneous, dynamic rangeland grazing sites examined in this study. These results are supported by findings elsewhere in Australian rangelands. Using remote sensing in the Great Barrier Reef Catchment, Queensland, Bartley et al. (2023) demonstrated a ground cover increase at some, but not all sites that had adopted regenerative grazing management (i.e. planned rotational grazing). The potential for high-intensity rotational or multi-paddock grazing systems (>20 paddocks) in conjunction with flexible stocking rates to improve ground cover has rarely been tested in Australia, at scale, and under ‘real world’ conditions where the practical implications and seasonal adaptations that land managers face are incorporated. Results of previous smaller scale, cross fence studies comparing ground cover between two or more grazing management treatments in the southern Australian rangelands are also consistent with our findings, demonstrating potential improvements in ground cover in these managed grazing systems (Alemseged et al. 2011; Waters et al. 2017; McDonald et al. 2019b). However, in other areas of Australia, benefits of intensive rotational grazing systems are less clear. For example, in Queensland, Hall et al. (2014) found no difference in total ground cover between intensive rotational grazing, extensive rotational grazing and continuous grazing systems. Similarly, a synthesis of evidence related to sustainable grazing management practices for northern Australia found little benefit of complex multi-paddock grazing systems (O’Reagain et al. 2014). Across Australia, a recent meta-analysis of grazing intensity and strategy comparisons revealed significantly higher ground cover under lower intensity grazing and in grazing systems that incorporate periods of rest (e.g. rotational grazing) compared with continuous grazing systems (McDonald et al. 2023), however this review did not examine the nuances of different grazing systems and the interaction of grazing intensity and strategy. Across the four sites where a significant improvement in ground cover relative to that of the benchmark sites was observed from the time of management change, the increase was up to 7%. This increase is consistent with the reported increase in cover from the ‘Enterprise Based Conservation’ project in western NSW (which reported change in ground over ranging from 0.63 to 2.69% after 5 years of destocking or managing grazing to maintain cover; Moss et al. 2012).

Each site, and each land type within each site, experienced differing patterns in the response of total ground cover following management change, likely reflecting the complexity and considerable variability in land types, rainfall, climate patterns and grazing management approaches across properties in western NSW. The variable response across sites and land types is consistent with previous multi-site studies in NSW rangelands (e.g. Waters et al. 2017). This inconsistency in the observed response highlights the challenges of detecting and measuring change and recommending best management to achieve environmental outcomes in the rangelands, and may account for the often variable and conflicting results reported in grazing management studies (Briske et al. 2008). We were not able to confidently predict where (i.e. land type) and when (i.e. preceding 12 month rainfall) the changes (if any) would be most apparent with the approach and landscape variables used in this study, although there was some indication that increased ground cover in response to the grazing management change was greatest in areas of lower productivity (i.e. lower NDWI, TSDM and greenness). Alternative analysis methods that explored additional predictor variables or increased spatial resolution may reduce the variability and provide better prediction of the key landscape determiners of change. The considerable variability in the GC_S−B and the trend of GC_S−B over time at individual sites (Figs 2 and 3), demonstrates the value and importance of longer term temporal studies to understand impacts of management changes in this environment. To better understand and report change, for example for use in environmental accounting, we recommend frequent measurements (at least seasonal to capture the temporal variability), at property scale and where possible minimising the impact of climatic variability by comparing against dynamic benchmarks that are subject to similar land type and climate constraints.

Response times are often slow in rangeland environments and are driven by rainfall events. Previous studies have indicated it can take over 20 years for arid and semiarid regions of Australia to respond to management changes (e.g. Bartley et al. 2023). It is also pertinent to recognise that in this study, the year of management change represents when these property managers started their new grazing management approach. In reality it takes years to fully implement and refine the grazing management strategy at the property scale, for example, installing fencing, water points and logistics of bringing together livestock (and associated classes) in large groups to rotationally graze. Across all properties, management changes over the 10–20 years from initial change were incremental and were continuing at the time of this study. Therefore, it is possible that the benefits of the changed grazing management are not yet fully apparent. It is also important to consider that with overall low productivity and high variability in ground cover for this region, we are unable to determine if the change in ground cover that has been achieved, and the rate of change, is able to be sustained over longer timeframes. A reference ecosystem or site able to act as a benchmark, like those used in the National Restoration Standards (Standards Reference Group SERA 2021), could potentially provide an indication of maximum attainable ground cover, however to the best of our knowledge these are not documented for the semi-arid rangelands of NSW. The benefit of the approach taken with this study, is that this low-cost temporal evaluation can be undertaken regularly, and in the future at a meaningful scale in these extensively grazed rangeland systems. With the emergence of online tools and platforms to access remotely sensed ground cover data, the approach taken in this study has the potential to be integrated and automated to provide producers real-time information to aid their decision making in managing their livestock to maintain or improve ground cover and demonstrate environmental outcomes. There is a continued drive for ecosystem service markets, supported by robust, repeatable, transparent and accurate methods of monitoring and measuring change in response to management actions. This is of particular importance in rangeland environments where distance, scale, variability in season and landscapes across properties adds additional challenges to monitoring and measuring change. Importantly, for environmental accounting, this approach is management sensitive, outcome driven and reflects the current climatic conditions, thereby meeting integrity standards and minimising the impact of any metrics primarily associated with patterns of extreme dry (e.g. El Nino and associated droughts) or wet (e.g. La Nina). With continued development of remote sensing technologies, including the use of different sensors, Light Detection and Ranging (LiDAR) and Unmanned Aerial Vehicles (UAVs), and additional indicators of condition, any response of ground cover and other ecosystem services to grazing management can be quantified and demonstrated with greater precision, potentially increasing access to environmental markets in rangeland grazing systems.

Conclusion

The study has highlighted that livestock producers can potentially achieve improvements in ground cover and associated ecosystem services through changes in grazing management (management of stocking rates and introduction of planned rest into grazing systems). However, the magnitude and direction of response observed may be dependent on specific grazing management decisions, the timing of monitoring, land type and preceding rainfall during which the management change is evaluated. With increasing pressure for livestock producers to demonstrate sustainable management, and emergence of environmental markets rewarding producers for provision of ecosystem services, this study supports the used of remote sensing using biophysically similar, dynamic benchmarks (Bastin et al. 2012; Zhang and Carter 2018; Donohue et al. 2022) to detect and measure change and support the adoption of management of the timing and intensity of livestock grazing in variable semiarid rangeland environments.

Supplementary material

Supplementary material is available online.