Monitoring and Restoration of natural temperate grasslands in South Australia

Abstract

The Natural temperate grasslands of South Australia face significant ecological threats, suffering from severe degradation. The invasion of exotic species and agricultural practices have impacted the vegetation composition and soil characteristics of grasslands, leading to grassland ecosystem degradation. To address these threats, assessing the condition of degraded grasslands to identify areas that require interventions, such as restoration efforts, is imperative. Furthermore, a deeper understanding of the factors limiting the growth of native species and favouring exotic species in degraded grasslands is needed to contribute to management interventions oriented toward conserving temperate grasslands in South Australia. This project conducted a remote sensing classification using Sentinel-2 multi-spectral time series to develop a random forest model that discriminated three levels of condition based on the cover of cover of the native perennial graminoid iron-grass (Lomandra effusa). In addition, the model also discriminated woodland and patches dominated by annual exotic wild oat (Avena barbata), the main invasive exotic species. The results indicated that spectral bands and vegetation indices from the dry and rainy seasons at 10 m resolutions produced the best classifications (Overall accuracy = 67 %, kappa coefficient = 56 %). The dry season dataset emerged as the most informative, showing that differences in phenological properties among vegetation classes were critical for their discrimination. This analysis contributes to the condition assessment by mapping classes associated with vegetation attributes. Maps that inform the condition of vegetation could provide critical information to prioritise management actions in degraded areas. Even though remote sensing classifications provide insight into the condition of vegetation, they are limited in terms of transferability in space and time. This project developed a repeatable assessment to monitor the condition of temperate grasslands by modelling a continuous variable like iron-grass cover. The model condensed seasonal spectral reflectance patterns associated with the perennial growth strategy from iron-grass. These patterns were extracted from the 2022 NDVI Sentinel-2 satellite time series at 10 m resolution. The model performed well (R2=0.63), producing an RMSE of 0.09 ± 0.03) for 2022. The application of the model to previous years (2019-2021), with different rainfall distributions and NDVI time series, produced similar RMSE values, resulting in high correlations between the site-based cover and predictions (R >0.75) and between predictions (R > 0.85). The consistency of the model performances over multiple years evidences the reliability of the analysis across time, despite rainfall variability. Hence, the analysis enabled the mapping of vegetation cover trends over time without collecting field-based data each year. This analysis contributes to detecting subtle patterns in plant community composition associated with the cover of native species, thus providing an instrument for monitoring condition over multiple years and environmental conditions. The composition of plant communities in degraded grasslands is subject to a series of environmental factors associated with climatic variability. In temperate grasslands, precipitation fluctuations caused by extreme climatic events like drought and high rainfall are expected to modify plant and soil microbial community composition and relationships. This project conducted a precipitation manipulation experiment to explore the effects of precipitation reductions in the short term. The occurrence of high rainfall events in the second year of the study provided new insight into the effects of soil water content increments to plant and soil microbial communities. Reductions in precipitation decreased wild oat frequency, while increments in rainfall increased native perennial spear grass (Austrostipa spp.) frequency. Precipitation reductions had little effect on soil microbial community composition and structure. In contrast, higher rainfall in the second year of the study significantly shifted the composition of fungal and bacterial communities, with a more evident effect on fungal richness and diversity. These results prove that short-term precipitation fluctuations can influence temperate grasslands, highlighting the role of increments in rainfall as a driver for plant and microbial community composition and structure. Aligning restoration and weed control campaigns in temperate grasslands to rainfall events would be of most benefit in the upcoming decades. Degraded grasslands exhibit a range of properties associated with agricultural practices that have lasting effects (legacies), favouring the dominance of exotic species and hindering the reestablishment of native species. Recent research highlights the critical role of soil microbes and plant interactions in successful restoration efforts. However, the impact of established techniques, such as topsoil removal and emerging microbial inocula methods, on soil microbial communities in natural environments remains to be thoroughly investigated. This project conducted a glasshouse experiment exploring the effects of topsoil removal and microbial filtrates from different soil origins on the growth of the native wallaby grass (Rytidosperma auriculatum) and on soil microbial communities. The results demonstrated the effectiveness of topsoil removal in mitigating soil legacies and emphasized its importance for the successful application of microbial filtrates from remnant grassland soil. Although microbial filtrates did not produce sizeable shifts in microbial communities, subtle changes in their composition may be the cause of increased growth of wallaby grass. Incorporating microbial filtrates into restoration practices and applying them to tubestock in nurseries holds promise for improving restoration outcomes in degraded temperate grasslands. Grazing is recognized as a primary driver of floristic and edaphic changes, highlighting the need to consider its manifold impacts. Livestock grazing and dung depositions are often associated with the dominance of exotic species. However, little is known about the effects of the dung from native herbivores on plant growth and its potential implications for restoration. This project incorporated a glasshouse experiment where sheep and kangaroo dung were added to pots with wild oats and wallaby grass. The results showed that kangaroo dung addition can improve wallaby grass aboveground biomass and negatively affect wild oat. In addition, the experiment reported no effects from sheep dung. The results suggest that kangaroo dung effects may be related to its chemical and biological characteristics rather than its nutrient composition. The positive effects of kangaroo dung found in this study suggest that kangaroo dung could be used as fertilizer during tubestock preparation or planting in restoration projects. The outcomes of this project improve our understanding of degraded temperate grasslands. Notably, this project demonstrates the effectiveness of remote sensing techniques to map and monitor vegetation condition in degraded grasslands. Furthermore, this project highlights the importance of precipitation fluctuations, microbial filtrates and dung addition in ecological restoration. Incorporating this project's findings into the management of temperate grasslands should lead to a better knowledge of the level of vegetation degradation and the improvement of management interventions aimed at conserving grasslands.