Oral Presentation Australian Society for Limnology Conference 2016

Long-term inundation regimes drive floodplain wetland persistence at the landscape scale: informing environmental water management (#35)

Rachael F Thomas 1 2 , Senani Karunaratne 1 , Richard T Kingsford 2 , Jessica Heath 1 , Sharon Bowen 1
  1. NSW Office of Environment and Heritage (OEH), Sydney South, NSW, Australia
  2. Centre for Ecosystem Sciences, School of Biological, Earth and Environmental Sciences, UNSW, Sydney, Australia

Inland floodplain wetlands in the Murray-Darling Basin depend on highly variable river flows and flooding regimes characterised by unique variations in magnitude, timing, duration and frequency. Vegetation responses are dependent on relationships among the flow pulse (annual), the previous sequence of flows (five years) and the long-term flow regime (decades). Understanding these relationships at the landscape scale is critical for effective environmental flow management. We use novel methods to map flooding regime extents (duration, frequency and time since last flood) in the Macquarie Marshes at three different ecologically relevant timescales: flow pulse, five year intervals and decadal long-term inundation regimes. We measure floodplain wetland persistence within two time periods: 1991-2008 and 2008-2013, corresponding to vegetation type mapping (1991, 2008 and 2013). At the shortest time scale, we identify flow pulses using river flow data and then for each flow pulse, we spatially aggregate inundation maps derived from Landsat observations calculating cumulative area (ha) of floodplain inundated and its inundation duration (days). For five yearly intervals before 2008 and 2013, we calculate inundation pulse frequency and duration metrics (mean, standard deviation, median, minimum and maximum). For long-term inundation, we focus on the 20 year interval before 2008 and 2013, calculating time since last flood observation, inundation pulse frequency and duration metrics. We use the random forests algorithm for classification to identify the variable importance of inundation co-variates in explaining vegetation persistence over short (5 year) and long (20 year) time periods. Long-term inundation covariates: time since last flood observation, 20 year average and standard deviation duration (days) and 20 year inundation pulse frequency are the most useful co-variates explaining vegetation persistence between 1991 and 2008 and between 2008 and 2013, compared to five year inundation co-variates. Long-term water management is critical for ensuring the persistence of a floodplain wetland vegetation mosaic.