Spatially explicit current and future threats to seagrass habitats in Australia

Brief description

This mapped dataset is a compilation of spatially explicit, nation-wide threats to seagrass based on current pressures and projected future climate change pressures. In addition, the value of this mapped dataset can potentially extend to assess threats to other coastal habitats. Current threats in this mapped dataset include urban/agricultural runoff, industrial pollution, sediment resuspension, port infrastructure and dredging, shipping accidents, oil and gas accidents. Future threats in this mapped dataset include modelled increase in sea surface temperature for 2070, modelled increase in total annual rainfall for 2070 and modelled increase in sea level rise for 2070. All threats in this mapped dataset are given as a single ArcGIS polygon shapefile composed of 10 x 10 km coastal grid cells.

Lineage

Maintenance and Update Frequency: notPlanned

Statement: Lineage Statement: 1. Risk estimation methodology For each threat we categorised the risk in each 10 x 10 km grid cell as High, Moderate, Low and No risk. 1.1. Urban/agricultural runoff This threat is divided into two threat layers: acute sediment and nutrient inputs and chronic sediment and nutrient inputs. 1.1.1. Acute sediment nutrient risk Method – This threat layer was derived by considering the catchment condition moderated by the likelihood of large pulses of flow along river channels as well as the total volume of the flow. Specifically, the disturbance of the catchment (as identified in the National Estuary Audit 2000, n=974 estuaries http://www.ozcoasts.gov.au/search_data/estuary_search.jsp) was used to describe catchment condition. As sediment and nutrient loads are strongly linked to catchment clearing and landuse, we assumed that catchments that were near pristine and largely unmodified would pose a low risk to seagrasses in terms of sediment and nutrient loads. Similarly, the highest risk would be from catchments which are extensively modified, with a moderate risk from those moderately modified. We considered that estuaries receiving very pulsed streamflow were more susceptible to acute nutrient and sediment loads. To determine the pulse regime, we compiled streamflow data from the Australian Bureau of Meteorology supplemented by the Western Australian Department of Water Data (bom.gov.au and water.wa.gov.au) which described the daily flows from the period 1990 -1999 from 241 stream gauging stations Australia-wide. Gauging stations within 250 km of the coast were ‘moved’ to the nearest point on the Australian coastline linked to the appropriate waterway, and estuaries matched with their nearest streamflow. We then calculated a pulse metric based on the number of days which daily streamflow was greater than 1SD above the mean daily streamflow (determined on ln(ML+0.01) of daily data for each gauging station). If the pulse metric was <25th percentile, then streamflow was more constant so acute risk assumed to be zero. If the pulse metric was within the 25th-75th percentile, the acute risk was assumed to be reduced and acute risk greatest for estuaries where the pulse metric >75th percentile. The risk of acute sediment and nutrient risk for each estuary was determined based on the catchment condition and pulse metric as summarised in Table 1, where 4 is high risk, 3 moderate risk and 2 low with one indicating no risk. Once the risk values were generated for each estuary point location, the spatial extent of the influence of the threat was considered based on annual streamflow. Areas with higher annual streamflow would have greater sediment and nutrient risks than those which received less annual streamflow. The annual flow data was derived from the same dataset as above and the metric defined as ln(annual flow, ML)). Areas receiving streamflow of 20 333 ML/yr or less, were in the lowest 25th percentile, and the spatial extent of impact was considered small. A medium extent of impact was assigned for flow between 20 333 ML/yr and 181 680 ML/yr (25th – 75th percentiles) and >181 680 ML/yr was assigned a large extent of impact. The spatial extent was estimated based on both the risk of acute sediment and nutrient risk in the estuary (1-4 above) and the streamflow category (Figure 1). For low risk cells a small streamflow generated no buffer, a moderate stream flow had a buffer of 1 10x10 km cell around the estuary at low risk, and the high stream flow generated a buffer of 2 10x10 km cells around the estuary. For moderate and high risk cells, the size of the buffer varied and the buffer dropped down one risk category. A small flow generated a buffer of 1 10x10 km cell around the estuary, a medium flow generated a buffer of 2 10x10 km cells and a high flow buffer of 4 10x10 km cells (Figure 1). 1.1.2. Chronic sediment nutrient input risk Method – The chronic sediment and nutrient input risk was derived following a similar approach as the acute risk, but the flow metric varied. We considered that waterbodies which received their streamflow more constantly throughout the year were more susceptible to chronic nutrient and sediment load. To categorise the hydrologic regime, we calculated a hydrologic metric based on the ratio of the mean daily flow and the monthly variance. If the hydrologic metric was <25th percentile, then streamflow more constant so chronic risk assumed to be greater. If hydrologic metric was >75th percentile, streamflow was more patchy, so chronic risk assumed to be reduced. No adjustment was made for estuaries where hydrologic metric was between 25th-75th percentiles. Once again the final risk in the estuary was estimated based on the catchment condition and the hydrolic metric as summarised in Table 1 and the buffer generated based on the annual flow as described for acute sediment and nutrient risk. 1.2. Industrial pollution risk Method - The industrial pollution layer was generated from the industrial class cover of the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) 2005-2006 land use map derived from an AVHRR satellite image (http://adl.brs.gov.au). This industrial pollution layer assumes that with more industrial landuse in a 10 x 10 km grid cell, the greater chance of industrial pollution reaching the marine environment, either through direct runoff, groundwater contamination or atmospheric deposition. In this approach, we only considered the grid cells that were adjacent to the coast, and not those further inland, hence the limitation is that we capture industrial pollution from direct run-off and groundwater contamination, but not from surface run-off from catchments further inland. The percentage of the terrestrial grid cell adjacent to the coast that contained industrial pollution was calculated, based on the number of pixels within each cell (total of 100). If the terrestrial grid cells adjacent to the coastal grid cell contained no industrial land-use, then it was considered to have no exposure to industrial pollution. If less than 2% of the grid cell was industrial this was categorised as low likelihood (=low risk), 2-10% was considered a moderate likelihood (=medium risk), and >10% a high likelihood (=high risk). Buffers were created adjacent to any moderate or high likelihood cells. Any marine grid cell adjacent to a high risk cell was considered a moderate risk, and those adjacent to a moderate risk cell were considered a low risk. If any grid cell was allocated more than one risk category, then the highest category was maintained. 1.3. Sediment resuspension risk Method – Resuspension data was derived from Geoscience Australia’s dataset “Percentage of the time that the Shields parameter exceeded 0.25”. The Shields parameter defines the bed shear stress required to initiate sediment movement. When it is >0.25, conditions on the seabed are highly mobile, hence there is more chance of resuspending sediments which can have a negative impact on seagrasses due to reductions in light. The percentage of the time that the Shields parameter exceeded 0.25 was determined from the Geological and Oceanograhic Model of Australia’s Continental Shelf (GEOMACS) model (Hughes et al., 2010, Hemer, 2006, Harris and Hughes, 2012). We predicted that with a greater percentage of time above the Shields parameter of 0.25 there would be a greater risk due to sediment resuspension. As there was no data to base the resuspension risk on we took the standard approach of assigning a low risk below the 25th percentile (0.8% of the time), a high risk above the 75th percentile (15.8% of the time) and a moderate risk between the two. There was no exposure and hence no risk to seagrass habitat from resuspension when the Shields parameter did not exceeded 0.25 at any time. 1.4. Port infrastructure and dredging risk Method - The threat to seagrass habitat from port infrastructure and dredging was assessed based on the locations of ports in Australia provided by the Australian Customs & Border Protection Service (http://data.gov.au/dataset/australian-ports). We predicted that there was a high risk to seagrass habitat when there was a port located in a grid cell, a moderate risk in cells adjacent to a high cell, and a low risk in cells adjacent to moderate. We considered that there was no exposure to the threat of port infrastructure and development and hence no risk in all other grid cells. 1.5. Shipping accidents risk Method - The threat to seagrass habitat from port infrastructure and dredging was assessed from the locations of ports in Australia provided by the Australian Customs & Border Protection Service (http://data.gov.au/dataset/australian-ports). We predicted that there was a high risk to seagrass habitat when there was a port located in a grid cell, a moderate risk in cells adjacent to a high cell, and a low risk in cells adjacent to moderate. We considered that there was no exposure to the threat of port infrastructure and development and hence no risk in all other grid cells. 1.6. Oil and gas accidents risk Method - The threat to seagrass habitat from oil and gas accidents was predicted from the location of oil and gas wells in the coastal environment. Gas pipelines were not considered as this information is restricted. The location of oil and gas production wells was sourced from GeoSciences Australia (http://dbforms.ga.gov.au/www/npm.well.search). We predicted that there was a high risk to seagrass habitat when shipping vessels passed through the grid cell, a moderate risk in cells adjacent to a high cell, and a low risk in cells adjacent to moderate adjacent cells. We considered that there was no exposure to the threat of shipping accidents and hence no risk in all other grid cells. 1.7. Increase in sea surface temperature risk Method – Different seagrass species have different temperature tolerances (Lee et al., 2007) and in Australia species are distributed across locations that have a broad temperature range (Kilminster et al., 2015). Therefore some locations, such as at the range edge may be more susceptible than other locations (Jordà et al., 2012). To employ a consistent and justifiable prediction for impacts of increased temperature we used published literature data and short-term responses to increased temperature. This may be an overestimate of the response, as we have no understanding about their ability to adapt to changing temperatures. The majority of studies on the effects of short-term temperature increases to seagrasses have shown negative effects with increases of 2°C or more (Seddon et al., 2000, Waycott et al., 2007, Collier et al., 2011, Moore et al., 2013, Thomson et al., 2015). Therefore, if the predicted temperature did not increase by 2°C we considered no risk, however this situation did not occur based on 2070 OzClim Mk3.5 model predicted sea surface temperature dataset (www.csiro.au/ozclim). Thus, a percentile approach was used for identifying the cut-offs for High > 3.2°C (75th percentile), Moderate 2.8-3.2°C and Low 2.8°C (25th percentile). 1.8. Change in rainfall risk Method - We predicted that a greater rainfall would lead to more sediment and nutrient delivery, more flooding and more low light events that could impact seagrasses (Collier et al., 2012) and hence either more acute or chronic sediment and nutrient risk to seagrasses. The predicted change in rainfall in 2070 dataset by OzClim GFDL-CM2.1 model (www.csiro.au/ozclim) was used and we focused on the period of the year that was considered the wet period, as the OzClim predictions provided predictions based on wet and dry periods. We classified this data as no risk if the rainfall was not predicted to increase, low risk if it increased up to 50 mm per year, moderate if it increased 50-100 mm and a high risk if it increased more than 100 mm. 1.9. Sea level rise risk Method - An increase in sea level can have a negative effect on seagrasses if the shoreline is hardened and they cannot colonise new habitats, also seagrasses can be lost on the deeper edge if light becomes limiting to growth (Waycott et al., 2007, Saunders et al., 2013). Saunders et al. (2013) modelled the impact of sea level rise on a large embayment in Queensland and found that the area of seagrass declined by 17% with a 1.1. m rise in sea level. Obviously these predictions are location specific but we used these as a guide to categorise the likelihood of the risk. Dataset on the projected departure from global mean (A1B scenario) at 2070 (mm) from 17 model simulations was used (http://www.cmar.csiro.au/sealevel/sl_proj_regional.html) to quantify sea level increase. If no increases were predicted, then no risk was assigned, less than 50 mm was low, 50-200 moderate, and > 200 mm a high likelihood. 2. Overall structure of the dataset file with the 10 threat layers All 10 threat layers were put together as one shapefile. In this shapefile, each 10 x 10 km grid cell/polygon will have the following attribute corresponding to a specific threat layer: 2070temp - increase in sea surface temperature risk, 2070seaL- sea level rise risk, 2070rn - change in rainfall risk, Industry - industrial pollution risk, Oilgas - Oil and gas accident risk, Port - port infrastructure and dredging risk, Resuspen - sediment resuspension risk, Shipping - shipping accident risk, ChrSedNut - chronic sediment nutrient load risk, AcuSedNut - acute sediment nutrient load risk. Each grid cell/polygon will have a risk value (high risk =4, medium risk=3, low risk=2 or no risk=1) for each of the 10 risk layers. Important Note: The risk values for the 10 threat layers were generated for all coastal grid cells with and without seagrass presence. In order to view risk for grid cells with seagrass, a seagrass presence / absence layer (Canto et al., 2016g, Canto et al., 2016b, Canto et al., 2016f, Canto et al., 2016a, Canto et al., 2016e, Canto et al., 2016c, Canto et al., 2016d) was added as indicated by the “SG” attribute. This is done by doing a query/filter function where grid cells with “SG value =1” are shown. Important Note: Figure and table cited is available in the pdf supplementray information document availbale as a separate download. 3. References CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016a. New South Wales Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016b. Northern Territory Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016c. Queensland Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016d. South Australia Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016e. Tasmania Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016f. Victoria Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility CANTO, R., UDY, J., MCMAHON, K., KILMINSTER, K., KENDRICK, G., ROELFSEMA CHRIS, M., SCANES, P. & WEST, G. 2016g. Western Australia Seagrass Habitat Map. Terrestrial Ecosystem Research Network - Australian Coastal Ecosystems Facility COLLIER, C. J., UTHICKE, S. & WAYCOTT, M. 2011. Thermal tolerance of two seagrass species at contrasting light levels: Implications for future distribution in the Great Barrier Reef. Limnology and Oceanography, 56, 2200-2210. COLLIER, C. J., WAYCOTT, M. & MCKENZIE, L. J. 2012. Light thresholds derived from seagrass loss in the coastal zone of the northern Great Barrier Reef, Australia. Ecological Indicators, 23, 211-219. HARRIS, P. T. & HUGHES, M. G. 2012. Predicted benthic disturbance regimes on the Australian continental shelf: a modelling approach. Marine Ecology Progress Series, 449, 13-25. HEMER, M. 2006. The magnitude and frequency of combined flow bed shear stress as a measure of exposure on the Australian continental shelf. . Continental Shelf Research, 26, 1258-1280. HUGHES, M., HARRIS, P. & BROOKE, B. 2010. Seabed exposure and ecological disturbance on Australia’s continental shelf: potential surrogates for marine biodiversity. Canberra: Geoscience Australia Record 2010/43. JORDÀ, G., MARBÀ, N. & DUARTE, C. 2012. Mediterranean seagrass vulnerable to regional climate warming. Nature Climate Change, 2, 821–824. KILMINSTER, K., MCMAHON, K., WAYCOTT, M., KENDRICK, G. A., SCANES, P., MCKENZIE, L., O'BRIEN, K. R., LYONS, M., FERGUSON, A., MAXWELL, P., GLASBY, T. & UDY, J. 2015. Unravelling complexity in seagrass systems for management: Australia as a microcosm. Science of the Total Environment, 534, 97-109. LEE, K. S., PARK, S. R. & KIM, Y. K. 2007. Effects of irradiance, temperature, and nutrients on growth dynamics of seagrasses: A review. Journal of Experimental Marine Biology And Ecology, 350, 144-175. MOORE, K., SHIELDS, E. & PARRISH, D. 2013. Impacts of varying estuarine temperature and light conditions on Zostera marina (eelgrass) and its interactions with Ruppia maritima (widgeongrass). Estuaries and Coasts, 37, S20–S30. SAUNDERS, M., LEON, J., PHINN, S., CALLAGHAN, D., O'BRIEN, K., ROELFSEMA, C., LOVELOCK, C., LYONS, M. & MUMBY, P. 2013. Coastal retreat and improved water quality mitigate losses of seagrass from sea level rise. Global Change Biology, 19, 2569-2583. SEDDON, S., CONNOLLY, R. & EDYVANE, K. 2000. Large-scale seagrass dieback in northern Spencer Gulf, South Australia. Aquatic Botany, 66, 297–310. THOMSON, J. A., BURKHOLDER, D. A., HEITHAUS, M. R., FOURQUREAN, J. W., FRASER, M. W., STATTON, J. & KENDRICK, G. A. 2015. Extreme temperatures, foundation species, and abrupt ecosystem change: an example from an iconic seagrass ecosystem. Global Change Biology, 21, 1463–1474. WAYCOTT, M., COLLIER, C. J., MCMAHON, K., RALPH, P. J., MCKENZIE, L. J., UDY, J. & GRECH, A. 2007. Vulnerability of seagrasses in the Great Barrier Reef to climate change. In: JOHNSON, J. E. & MARSHALL, P. (eds.) Climate Change and the Great Barrier Reef: A vulnerability assessment. . Townsville: Great Barrier Reef Marine Park Authority and Australian Greenhouse Office, Australia.

Notes

Credit
We acknowledge the Australian Centre for Ecological Analysis and Synthesis for funding to bring together the Seagrass Working Group (Kate O’Brien, Alastair Hirst, Gary Kendrick, Gregory West, Phillippa Bricher, Patricia von Baumgarten, James Udy, Jonathan Hodge, Michelle Waycott, Jeff Ross, Lynda Radke, Len McKenzie, Bill Dennison, Angus Ferguson, Paul Maxwell, Vanessa Lucieer, Peter Scanes and Jonathan Hodge) for developing the concept of this project. Additional support was provided through the Collaborative Research networks (CRN) Program, Funding Agreement CRN2011:05 for Kathryn McMahon, CSIRO Coastal Carbon Biogeochemistry Cluster and UQ-RSSC for Chris Roelfsema and Robert Canto.