Modelling the potential conservation values of farm dams in South-Eastern Tasmania

Zhao, F ORCID: 0000-0002-7933-9511 2020 , 'Modelling the potential conservation values of farm dams in South-Eastern Tasmania', PhD thesis, University of Tasmania.

 Preview
PDF (Whole thesis)

| Preview

Abstract

Farm dams in Tasmania are the most strictly regulated in Australia. Dam construction needs a permit if the capacity is over one megalitre or on stream. In 2015, there were 9,404 registered dams in the State, including both hydro and non-hydro dams. However, the exact number of existing dams is not clear, as dams are constructed without registration and many do not require registration. The potential effects of physical, chemical and biological attributes of dams on biodiversity has been rarely investigated in Australia.
The present study determines the attributes of dams that make them useful for the conservation of biodiversity at the regional and local scale. Its objectives were to determine: the accuracy of imagery segmentation algorithms in dam identification on both satellite imagery and aerial photographs; the effects of water quality, water source, dam types, and physical attributes on the biodiversity of farm dams; and the attributes of dams that promote nature conservation values.
The distribution and physical attributes of farm dams in the case study area were identified on Google Earth Pro satellite imagery dated 20 June 2015. The development history of farm dams in the area from 1969 to 2015 was mapped from historical aerial photographs and historical satellite imagery from Google Earth Pro. The potential impacts of these farm dams on local stream flow and evaporation were estimated based on historical climate data from Bureau of Meteorology, including rainfall data from 1969 to 2015, evaporation data from 1986 to 2015 and water flow data from 1971 to 2015.
The Mean Shift Algorithm, Watershed Algorithm and Morphological Profiles Based Segmentation Algorithm from Orfeo ToolBox in the QGIS were applied to imagery of the study area. These algorithms were used to evaluate their accuracy in dam identification on both PlanetScope satellite imagery (3m resolution, dated at 17 July 2016 and 26 September 2016) and aerial photographs (0.1m resolution, dated at 4 April 2016).
Surveys were conducted on 104 random selected farm dams in south-eastern Tasmania, within the NRM South region. Physical data, some chemical data and species data were collected in the field. Water samples and species samples were collected for lab assessment. Global non-metric multidimensional scaling was performed to ordinate species composition data in four dimensions using the default option in DECODA. Agglomerative classification in Minitab 18 was used to perform cluster analysis for all species. Multiple-regression analyses between species richness variables and other variables were also conducted in Minitab 18.
Weighted endemism and corrected weighted endemism were used to estimate the species richness and distribution of locally rare species in the surveyed farm dams. Multiple-regressions between endemism scores and environmental variables were conducted to determine variables that effect locally rare species in the dams. Species rarity and endemism at State level were also analysed based on farm dam species data and species occurrence data from the Natural Values Atlas of Tasmania. The effect of environmental variables on these species were analysed.
Aside from the 22 registered dams in the Orielton Rivulet sub-catchment, another 239 non-registered dams were identified. Non-registered dams have small surface area, were small vegetation coverage in surrounding area, constructed off stream, over ten years old and with clear seasonality. The increased dam surface area had a negative impact to local stream flow. The predicted daily evaporation of dam water accounted for almost 200% of daily stream flow.
All the three algorithms had very low accuracy in identifying dams from PlanetScope satellite imagery. The Watershed segmentation algorithm had better results than the other two algorithms but could only identify a limited number of dams. Results from aerial photographs were more accurate than satellite imagery with all three algorithms, but over-segmentation issuesstill existed in the final segments. While spatial resolution of imagery influences the accuracy of dam identification, biophysical variables and surrounding landscapes also significantly affected the result.
Conductivity, turbidity and sodium were correlated with non-native plant species and native plant species richness, while phosphorous related to total plant species richness and invertebrate species richness. Conductivity was negatively related to frog species richness. Dissolved oxygen and nitrogen affected the distribution of locally rare frog species, such as Litoria ewingii. Composition of all species was significantly affected by conductivity, dissolved oxygen and sodium. Conductivity and sodium also correlated with plant species composition.
Few relationships were found between water source types of dams and species richness and composition. Total species composition was impacted by spring feed dams. Dams that depended on water from other dams had negative effects on the presence of rare/endemic plant species and positive effects on non-native plant species richness. Dams that were designed for stock water supply, domestic use and crop/irrigation related to native plant species and non-native plant species richness. Firefighting, recreational and stock dams were related to plant species composition. Presence of rare/endemic plant species was affected by stock and domestic use dams.
Bank height, age and altitude were correlated with plant species, invertebrate species and total species composition. Both seasonality and slope were related to plant species composition. Water depth related to invertebrate species and total species composition. Frog species richness was affected by depth and bank height. Depth also impacted the overall species richness. There was no relationship between physical attributes and the presence of rare/endemic species, but seasonality, dam size, and slope affected the distribution of State scale regionally rare plant species. Seasonality, dam size and depth were related to native plant and non-native plant species richness.
Permanent dams that were designed for domestic use with surface areas over 1000 m$$^2$$ and clean water are more likely to have higher plant species richness and native plant species richness. If these dams do not depend on water from other dams, they have a higher chance to have rare or endemic plant species. Dams with deeper water, low banks and low conductivity may have more frog species than other dams. When the water in these dams has lower level of nitrogen, regionally rare frog species are more likely to be observed. It is therefore possible to design dams that can provide essential habitat for many species and significantly contribute to local biodiversity conservation.

Item Type: Thesis - PhD Zhao, F farm dams, dam distribution, historical development, conservation values, rare and endemic species, dam attributes https://doi.org/10.25959/100.00035852 Copyright 2020 the author View statistics for this item