Thesis Evaluating Indicators for Monitoring Riparian Rehabilitation Success in the Upper Hunter Region of NSW, Australia
Citation: Gollan, J. 2008. Evaluating Indicators for Monitoring Riparian Rehabilitation Success in the Upper Hunter Region of NSW, Australia. PhD. PhD. School of Environmental and Rural Sciences. University of New England.Abstract:
This thesis was a response to scientific concern that riparian rehabilitation in Australia has become a multi-million dollar industry, yet outcomes are poorly evaluated, if at all. The situation prevailing in the Upper Hunter region, New South Wales, is probably no exception. Reasons for this summation include that: (1) the ecology of riparian systems is poorly understood in Australia, which limits our ability to predict how ecosystems will respond to revegetation, and (2) there are limited baseline data with which to make informed decisions about rehabilitation thresholds and targets. According to published management plans for the Upper Hunter, the goals of re-establishing riparian vegetation are to improve native biodiversity, ecosystem function and ecological health of the catchment.
One option to assess outcomes of rehabilitation and ensure resources are not wasted is through the use of ecological indicators in monitoring strategies. Indicators that are used for monitoring rehabilitation success refer to pre-determined successional trajectories. Thus, indicators need rigorous testing and validation before they can be recommended for monitoring programmes. A robust indicator must be clearly distinguishable from natural variability. Hence, the overall aim of this thesis was to evaluate a range of responses, for evaluating outcomes of riparian rehabilitation in the Upper Hunter region.
In the first component of the research, 23 response variables were selected based on two broad types: (1) terrestrial invertebrate abundance, diversity and composition, and (2) ecological processes mediated by terrestrial invertebrates. A chronosequence of sites was used to test the hypothesis that variables in rehabilitated (planted) areas would become increasingly dissimilar to those in unplanted areas and increasingly similar to those in mature riparian woodland. A fully nested design was used to examine natural variability at two spatial scales (tens of kilometres [sub-catchment scale] and tens of metres [patch scale]). Web-building spider abundance and species richness were the only two response variables that showed promise as ecological indicators at the sub-catchment scale. All other response variables did not show a clear trend across the chronosequence since there was significant natural variability among locations of the same seral stage. Variables that did not show a clear trend were subsequently rejected as sub-catchment scale indicators for progress-monitoring programmes. Monitoring that takes a site or project-specific approach will need to account for variability at the patch scale during evaluation.
In the second stage of the research, 19 environmental variables known to influence patterns of invertebrate distribution were selected based on two broad types: (1) vegetation, and (2) soil. Environmental variables were subsequently tested for correlation with ranked seral stages of vegetation to identify potential confounding factors in a chronosequence approach. A data-mining approach was then used to investigate relationships between response variables (the potential indicators from stage 1) and environmental variables. Results revealed no consistent patterns between different invertebrate groups or species and environmental variables as they were related to different suites of environmental variables. For example, responses of beetle communities were correlated with soil moisture, whereas leafhoppers showed strong relationships with soil condition such as phosphorus concentration. Notably, soil moisture and soil attributes were not correlated with ranked seral stages of vegetation and considered likely confounding factors when using a chronosequence approach using beetle or leafhopper responses.
The third stage of the study compared the reliability of data collected by community volunteers and those of professional scientists. This was seen as important aspect of the research because companies and regional organisations are depending on data collected by community volunteer groups. In some cases, volunteer labour has been proposed as a way of complementing or substituting data collected by professional scientists. However, professional scientists have been wary of relying too heavily on data collected by untrained people because it could be unreliable or collected in an imprecise manner. Using measures of habitat structure and herbivore leaf damage, comparisons were made between the data collected by scientists with those collected by volunteers, and benchmark scores or control standards. While agreement between benchmark scores and the scores of observers (scientists and volunteers combined) was generally good, when scientist and volunteer data were analysed separately, data collected by scientists (as a collective group) were more accurate than data collected by volunteers. However, the concordance between individual observers varied widely for different data variables. Thus, data based on vegetation structure and herbivore leaf damage collected by some volunteers were just as valid as those collected by scientists. There was also evidence that different structural variables were scored more accurately than others. For example, percentage foliage cover in the mid and lower storeys were scored more accurately than amounts of coarse woody debris. Importantly, and at least with the types of data tested, there appears no justification to ‘condemn’ all data collected by volunteers. Data validity depends not only on the particular variable being measured, but also on the experience and capacity of the individual observer.
In the final stage of the study, the data were revisited to develop an informed operational monitoring programme to evaluate rehabilitation outcomes in the study region. The costs of collecting information about response variables in different successional stages and variance estimates (from stage 1), were used to identify optimal sampling designs for abundance and species richness of each invertebrate group using cost–benefit optimisation. Correlated variables were also compared to identify the more economical in each pair as a cost-saving measure. Strong correlations between many soil attributes suggested that measuring only one or two of the soil attributes would be cost-effective, as they would be expected to reflect the behaviour of other soil variables. In correlation tests between different invertebrate taxa, none of the taxa were strongly correlated to each other signifying that where changes in or estimates of species richness are used to guide management decisions, trends revealed in one taxon may not indicate similar trends in others.
While the question, ‘Why monitor the outcomes of rehabilitation?’ is relatively easy to answer, solutions to other important questions such as, ‘Which variables should be monitored, how should monitoring be undertaken, and who should be doing the monitoring?’ are definite challenges for natural resource managers in the Upper Hunter region. Final determination ultimately rests with the clarity of the project’s goals. Nonetheless, closely aligned to clear project goals is an ability to deal with constraints that will determine which variables can be measured economically, as well as the availability of suitable personnel to acquire, analyse and interpret data. This study has provided natural resource managers with greater insight into the ecology of the Upper Hunter region’s riparian system and how certain biodiversity components have responded to riparian revegetation. The study has also provided guidance as to which variables should be monitored and how, given the constraints of time, finance and staff expertise and availability. Monitoring the outcomes of rehabilitation is a complex task that will require collaborative relationships between natural resource managers and research scientists, both of whom differ in their approach to the questions (as above). As a disciplinary group, research scientists might respond to those questions by drawing on complex theories of how a system and organisms should respond to rehabilitation and then seek the optimum in experimental design. Natural resource managers on the other hand, are likely to be under pressure to simply ‘get the job done’—whether scientists are involved or not. In my opinion, an effective rehabilitation initiative, which necessarily includes assessing outcomes, will need to find the common ground between the two approaches that exist between natural resource managers and research scientists. Ultimately a solution that satisfactorily meets the expectations of both parties will need to be sought.