Riparian vegetation in an agricultural landscape: assessment and analysis of condition and change

© 2013 Dr. Christopher Stuart JonesVegetation assessment is done using a wide range approaches depending on the assessment aims and constraints. Two approaches to sampling with vastly different targets, designs, and costs, are snapshot (single time period) or longitudinal (multiple time period) studies. Each approach has a range of strengths and weaknesses for vegetation assessment that allow interrogation of different research objectives. The collection and analysis of data within each sampling approach contains an array of challenges for any given study. The research reported here uses a case study of riparian vegetation in northern Victoria, Australia to explore the use of snapshot and longitudinal surveys for understanding riparian vegetation attributes and the effectiveness of different management actions in an agricultural landscape in northern Victoria. Riparian vegetation is a valuable component of the landscape that acts as the interface between aquatic systems and terrestrial environments. Healthy riparian zones are important for maintaining aquatic ecosystems by providing habitat, shade, banks stability, sediment filtration and nutrient processing. They also support terrestrial ecosystems through provision of ecosystem services. Riparian vegetation is valued for agricultural and recreational uses, which has led to high levels of degradation worldwide. In Victoria, extensive clearing and uncontrolled stock access continues to be a major pressure on riparian vegetation condition. The impact of livestock grazing on riparian systems is predominantly negative. There is a global wealth of knowledge on grazing management, but many gaps exist on the effectiveness of management in different systems and the methods used to evaluate this. Field vegetation surveys remain an essential component of conservation management. Riparian vegetation contains gradients in resources, structure and function, which provides a challenge to survey techniques. Therefore, this research used a quantitative sampling approach designed to capture variation along the gradients of distance from the creek edge, distance along the creek and seasonality. These data were used to characterise the variation in vegetation composition across all sites, in order to inform a new method of sampling riparian vegetation based on delimited zones. This zoned approach was then used in combination with published benchmarks of vegetation condition to evaluate livestock grazing management in this system. Vegetation condition assessments record various vegetation attributes as measures of the value, quality or condition of a site. Comparisons to published benchmarks suggested that the study sites had low cover of all native ground cover life forms, especially medium tufted graminoids and herbs. However, tree canopy cover was at or above expected levels for most sites. Results indicated that livestock grazing was associated with an increase in cover of bare ground and exotic annual herbs and a decrease in cover of native shrubs, herbs and native tufted graminoids. In addition, the effects of grazing varied with distance from the creek. Grazing exclusion through fencing and revoking grazing licences may have benefits for native vegetation condition, but the positive effect of grazing on reducing exotic perennial graminoid cover should also be considered. Evaluating management against vegetation change over time is superior to that of a single point in time as it allows identification of the rate and trajectory of change. Assessment of change requires multiple comparable surveys through time but these data are expensive and slow to acquire. In the absence of these data, historical data sets can provide valuable insights into past vegetation condition and structure. However, historical data collection methods that were not designed for the purposes of monitoring change will produce data not directly comparable to data from more detailed methods. The comparison of unequal datasets is common in ecology and many other fields, but the most appropriate method for comparing unequal data sets is not clearly established. This thesis addresses this issue through applying two methods, one existing and one novel, and comparing their performance in predicting vegetation change through time in order to inform management. The two alternative methods used different approaches to reinterpret coarse historical data. The novel method used data calibration via a double sampling approach, whereas existing methods used Bayesian prior distributions. Data calibration was superior to the existing method in two out of three examined vegetation variables. However, calibration methods required more field surveys and more complicated data analysis. Grazing management indicators were as good or better predictors of vegetation change than a selection of site and environmental variables. Fencing and grazing licence changes had varied effects on different vegetation attributes across sites. The data analyses used were effective tools for evaluating change and effects of management, but made uncertain predictions due to high variation between sites and the thorough accounting of data uncertainties in the models used