The potential impact of bedform migration on seagrass communities in Torres Strait, northern Australia

ARTICLE IN PRESS Continental Shelf Research 28 (2008) 2188–2202

Contents lists available at ScienceDirect

Continental Shelf Research journal homepage: www.elsevier.com/locate/csr

The potential impact of bedform migration on seagrass communities in Torres Strait, northern Australia James J. Daniell a,, Peter T. Harris a, Michael G. Hughes a, Mark Hemer b, Andrew Heap a a b

Marine and Coastal Environment Group, Geoscience Australia, GPO Box 378, Canberra, Australia CSIRO Marine and Atmospheric Research, GPO Box 1538, Hobart, Australia

a r t i c l e in f o

a b s t r a c t

Available online 10 April 2008

Seagrass communities in the northwest of Torres Strait are known to disappear episodically over broad areas. Sediment mobility surveys were undertaken within two study areas during the monsoon and trade wind seasons, in the vicinity of Turnagain Island, to find out if the migration of bedforms could explain this disappearance. The two study areas covered sand bank and sand dune environments to compare and contrast their migration characteristics. Repeat multibeam sonar surveys were used to measure dune-crest migration during each season. Our results show that seagrass beds occur in the troughs of sediment-starved dunes, but no seagrass occurs in association with full-bedded dunes that are superimposed on large sand bank features. The coincidence of seagrass beds with the sediment-starved dunes is in spite of the fact that they migrate faster (0.59 m day1) than full-bedded dunes (0.13 m day1), which indicates that some other factor (other than dune migration rate) limits seagrass growth within Torres Strait. We suggest that seagrasses are unable to colonise full-bedded dunes because of the semi-continuously transported sand that characterises this environment. In contrast, the troughs of sediment-starved dunes experience only limited bedload transport and are less hostile for seagrasses. A conceptual model is presented to explain the occurrence of seagrass beds in relation to their proximity to migrating sand dunes. Based on our analysis, we conclude that the widespread dieback of seagrasses documented for the Turnagain Island region was not caused by dune migration. Crown Copyright & 2008 Published by Elsevier Ltd. All rights reserved.

Keywords: Torres Strait Bedforms Migration Seagrass

1. Introduction Seagrasses are known to occupy a wide range of sandy littoral habitats. In a review of seagrass habitats of northeast Australia, Carruthers, et al. (2002) identified four specific habitats: river estuaries, coastal, deep water, and reef, with each having a key ‘factor’ limiting their growth (runoff, physical disturbance, low light and low nutrients, respectively). The dynamics of tropical seagrasses are heavily influenced by weather patterns, flood and cyclone events. The influence of weather events on the seabed (disturbance) or the water column (turbidity/low light) is known to make the distributions of seagrass temporally and spatially variable (Marba and Duarte, 1995; Carruthers et al., 2002; Waycott et al., 2004). The presence of seagrass is, however, indicative of a seabed that has been stable for a period long enough for colonisation to occur (a rate that is variable among seagrass species). Seagrasses are also known to occur in areas with active subaqueous dunes

 Corresponding author. Tel.: +61 2 6249 9691/ 9915.

E-mail address: [email protected] (J.J. Daniell).

and/or sand banks. Dunes and sand banks are indicators of active bedload transport and their lateral movement poses a threat to the seagrass communities (Harris, 1989; Marba and Duarte, 1995; Walker et al., 1996). In Torres Strait, seagrass communities are an important habitat for dugong, sea turtles, and commercially important fish species (Long and Poiner, 1997; Long and Skews, 1997). Both seasonal and episodic seagrass dieback occurs in the Torres Strait. Mellors et al. (2008) observed that seasonal variation in seagrass cover was climate related. Generally, seagrass abundance increased during the northwest monsoon, possibly a consequence of elevated nutrients, lower tidal exposure times, less wind and higher air temperatures. Low seagrass abundance coincided with the presence of greater winds and longer periods of exposure at low tides during the southeast trade winds. Episodes of ‘widespread’ seagrass dieback have been reported from Torres Strait from the early 1970s but as recently as 1991, 1999 and 2001 (Long et al., 1997; Marsh et al., 2004). Seagrass dieback from the 1991 event was estimated at 1199 km2 within a study area of 4388 km2 (Long et al., 1997) and was reportedly due to high turbidity from flooding river(s) in PNG during a coincident ENSO event (Poiner and Peterken, 1996). A high turbidity event in 2001 was also

0278-4343/$ - see front matter Crown Copyright & 2008 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.csr.2008.03.036

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

reported in Marsh et al. (2004) as far south as Turnagain Island (but not observed during surveys in 1987, 1991, and 1996). Sediment movement has also been suggested as a mechanism for seagrass dieback in the Torres Strait (Johannes and MacFarlane, 1991). Re-colonisation after large scale losses of seagrasses communities in Australia is known to take up to a decade or more (Poiner and Peterken, 1996). This paper presents the results of dune migration surveys in Torres Strait, at locations known to have previously undergone seagrass dieback in an attempt to provide a link between habitat stability with the presence and migration of bedforms and thus asses the potential threat that bedforms pose to seagrass communities. The study uses repeat multibeam surveys to generate accurate, high-resolution digital elevation models (DEMs) of the seabed. Rates of dune migration are measured from overlapping DEMs and when combined with seabed surveys and data from oceanographic sensors allowed an assessment of the potential for bedforms to impact on seagrass communities within Torres Strait. This assessment is presented in the form of a conceptual model illustrating four key sedimentary environments within Torres Strait and their associated level of ‘threat’ to seagrass communities.

2. Study area Torres Strait is a shallow epicontinental seaway in northern Australia, about 150 km in width (from north to south) separating

2189

the Cape York Peninsula (northern Australia) from Papua New Guinea (Fig. 1). The topography of the strait is characterised by a shallow ridge of basement rock, containing numerous scattered islands, sand banks, and coral reefs. The strait is generally shallow with water depths rarely exceeding 25 m (Harris, 1988). Torres Strait experiences two distinct seasons. The trade wind season lasts for 7 months (May–November) and is characterised by relatively low rainfall and strong southeasterly winds. The monsoon season lasts for 5 months (December–April) and is characterised by relatively high rainfall and weaker northwesterly winds (Fig. 2; Wolanski, 1986). Semidiurnal tides propagate into Torres Strait from the Coral Sea in the east and diurnal tides propagate from the Gulf of Carpentaria in the west. However, only 30% of the tidal wave energy approaching from the east or west is transmitted through the strait due to friction and attenuation of the tidal waves by the complex bathymetry (Wolanski et al., 1988). Daily current patterns in Torres Strait are dominated by east–west flowing currents that attain surface speeds of up to 2 m s1 within narrow passages during spring tides (Wolanski et al., 1988) and broad areas occur across central Torres Strait where peak spring tidal current speeds exceed 0.85 m s1 (Harris, 1994). Superimposed on the tidal currents are wave-induced currents, storm surges, ocean currents, and wind-driven currents. While tides produce the strongest near-bed currents (Wolanski and Thompson, 1984; Wolanski, 1986), the relatively weak (0.2 m s1) wind-driven currents also play a key role in

Fig. 1. Bathymetry of the Torres Strait region (Turnagain Island is marked Tu. Is).

Fig. 2. Average fortnightly wind vectors from Horn Island for the years 2000–2005. Pale grey shades indicate the months of the monsoon season (December–April). The two vectors coloured black indicate the vectors corresponding to the monsoon and trade wind surveys.

ARTICLE IN PRESS 2190

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Fig. 3. Bathymetry in the vicinity of Turnagain Island (Tu. Is.) and locations of multibeam sonar surveys.

controlling the net direction of bedload transport and dune migration in Torres Strait (Harris, 1989, 1991; Hemer et al., 2004; Saint-Cast and Condie, 2006). During the monsoon season, the wind-driven residual flow is towards the east, and during the trade wind season it is towards the west. Saint-Cast and Condie (2006) noted that the longer trade wind season with its stronger winds provides the impetus for a net westwards flow of water through Torres Strait of approximately 0.13 Sverdrups (130,000 m3 s1). Inter-annual variability in the wind field has also been shown to create significant variability in flow through the strait during the monsoon and trade wind seasons (Saint-Cast and Condie, 2006). This variability in flow is predicted to influence the net transport of sediment (and assumed dune migration) and highlights the potential of ‘extreme’ events to influence bedload transport patterns in the region (Margvelashvili et al., 2008). Saint-Cast (2008) concluded that turbidity in Torres Strait is expected to vary seasonally but out of phase with the seasonal circulation pattern due to the seasonal flux of water and sediment through the strait. Hydrodynamic modelling predicted that turbidity levels in Torres Strait peak at the end of the monsoon, due to an influx of fine sediment and turbid water coming from the Gulf of Carpentaria. Conversely, turbidity levels in Torres Strait were likely to be low at the end of the trade wind season due to the flushing of the strait. The strong tidal currents within Torres Strait create conditions suitable for the formation of bedforms. Bedform height and migration rate are directly coupled with the amplitude and frequency of habitat disturbance (Marba and Duarte, 1995) and therefore knowledge of bedform migration rates within Torres Strait is vital for understanding their potential to impact on seagrass communities. Sand banks, dunes and ribbons are considered to be important bedform facies within the Torres Strait (Harris, 1988), and each of which has different migration characteristics. Dunes are asymmetrical, transverse ridges of sand, typically of medium to coarse grade sediment with the steeper (lee) face facing the direction of migration. Within the Torres Strait their heights typically range from 0.05 to 5 m (typically 1/3–1/4 water depth). Dunes tend to be two-dimensional (straight crested) for a low speed just above the threshold of grain movement (Guy et al., 1966). For higher current speeds they become three-dimensional (sinuous crests). Full-bedded dunes occur where there is an abundance of sediment and show a continuous cover of sediment between dunes (Dalrymple and Rhodes, 1995). Sediment-starved dunes occur where sediment is sparse, and are characterised by regions of hard substrate in

between dunes. Sand dunes in Torres Strait are known to reverse their orientations with the seasonal changes in wind direction (Harris, 1989; Daniell et al., 2006) and their orientations reflect low-frequency bedload transport patterns in the region. Sand banks are common and significant features in the estuaries, coasts and continental shelves around the world and are the largest of the bedforms. Their location depends on the presence of sediments and currents (tidal or otherwise) that have the capability of periodically mobilising the sediment. Sand banks in Torres Strait are concentrated in the northwest (Fig. 3) and are co-located within areas of known seagrass dieback (Long et al., 1997). The sand banks in this area are variable in form, but are generally barchan-shaped. The banks are typically thousands of meters long, reach heights of 5–10 m, and have dunes superimposed upon them (Daniell and Hughes, 2007). The sediment within sand banks is generally finer than would be expected with local current strength and are held within the bank through ebb and flood dominated channels (Kenyon et al., 1981; Belderson et al., 1982). Using satellite imagery, sand banks in Torres Strait have been shown to migrate westwards at a rate of 15 m yr1 (Daniell and Hughes, 2007). Sand ribbons are bedforms indicative of sediment starved environments with strong tidal flows (Kenyon, 1970). They are low relief, elongate sand strips of sand, that rest on a coarser substrate with their long axis aligned parallel to flow (Allen, 1968). Dunes are often developed upon the sand ribbons, depending upon the availability of sufficient sand supply and current speed, with sediments migrating along the axis of sand ribbon (Kenyon, 1970). The distribution of sand ribbons in Torres Strait is poorly known due to their low relief making them difficult to observe in satellite imagery or aerial photos. Interspersed between bedforms in Torres Strait are hardgrounds, seagrass and algal reef habitats (Heap et al., 2005). Surface sediments in Torres Strait comprise a mixture of locally derived carbonate and relict siliciclastic material (Harris, 1988). Sand and gravel clasts dominate the carbonate fraction. The relatively large proportion of sand and gravel is attributed to the high energy conditions experienced though much of Torres Strait (Harris, 1988).

3. Materials and methods To understand seasonal variations in dune mobility, two surveys were undertaken in Torres Strait on board the RV James

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Kirby during the monsoon (March) and trade wind seasons (October) of 2004. Two areas with different bedform morphologies were targeted for study in the northwest of Torres Strait (Fig. 3). Both areas were located within an area of known seagrass dieback (Long et al., 1997). During both seasonal surveys each study area was mapped twice using multibeam sonar (Table 1). Area A was selected because it contained sand dunes (‘sand dune field’ facies of Harris 1988) in close proximity to seagrass and algae habitats to the southwest of Turnagain Island. The bathymetry of Area A (Fig. 4) is characterised by two flat topped platforms at 5 m depth along its east and west margins, separated by a shallow channel (10–11 m depth). The platforms are interpreted as submerged reefs due to the karst features on their surfaces (Heap et al., 2005). Dunes, typically on the order of 1–2 m high, are superimposed on the eastern reef platform. Apart from the dunes, the seabed in Area A typically forms a thin veneer of sediment which supports seagrass and algae habitats. Area B encompassed a sand bank to the southeast of Turnagain Island (Fig. 5) around which very little seagrass occurred. The key features of the bank (as described by Heap et al., 2005) are: (1) a triangular shape that bifurcates into two arms towards the east; (2) superimposed dunes up to 6.5 m high—the largest dunes being found towards the centre and east of the bank; and (3) smaller dunes (o0.25 m) superimposed on the stoss slopes of the larger dunes. Away from the sand bank unconsolidated sediment is sparse. Data from oceanographic multi-sensors (acquiring temperature, pressure, turbidity, salinity, photosynthetically active radiation, oxygen, and fluorescence data), current meters, video, Table 1 Dates for the commencement of each multibeam sonar survey Season

Area

Survey

Date commenced

Monsoon Monsoon Monsoon Monsoon Trade wind Trade wind Trade wind Trade wind

A A B B A A B B

1 2 1 2 1 2 1 2

29 March 2004 14 April 2004 30 March 2004 12 April 2004 12 October 2004 26 October 2004 10 October 2004 22 October 2004

2191

sediment samples, cores and water samples were acquired during the two surveys (Heap et al., 2005; Daniell et al., 2006). Repeat multibeam sonar surveys were also undertaken at the start and end of each survey (14 days apart) to measure changes in bathymetry and to estimate the rates of dune-crest migration. The interval between surveys also ensured that a full spring/neap tide cycle was recorded by the current meters. Synoptic meteorological data were acquired from Horn Island (101350 S, 1421130 E), approximately 110 km south of Turnagain Island (Fig. 1), to characterise seasonal changes in wind strength and direction.

3.1. Underwater video and grab samples Video footage and grab samples of the seabed were collected to characterise the substrate, morphology, and benthic biota in the study areas. The video camera was lowered to the seabed and recorded a minimum of 3 min of footage for each sample station. The type and distribution of seagrass was determined from the grab samples and interpreted from the video.

3.2. Current meters and oceanographic sensors Co-located current meters and oceanographic sensors were deployed away from the sand bank within Area B (Fig. 5). Due to the particular interest in seabed processes, processing and analysis of current meter data is focussed on data obtained near the seabed. The current meters (RD Instruments Workhorse Sentinel 600 kHz ADCPs) were configured such that the centre of first bin was 1.14 m above the sensors. This corresponded to approximately 1.6 m above the seabed. Measurements were taken every 10 min over the duration of the deployments. The Seabird Electronics SBE 19 oceanographic sensors measured conductivity, temperature, dissolved oxygen, photosynthetically active radiation (PAR), pressure, fluorescence, and turbidity. Samples were acquired for 20 s every 10 min over the duration of the deployment. PAR is a measure of the amount of light available for photosynthesis. Estimations of bedload transport were made using the method of Gadd et al. (1978). This method was chosen as it was previously shown to show a good agreement with dune celerity measurements in the region (Heap et al., 2005).

Fig. 4. Bathymetry of Area A, southwest of Turnagain Island. Distribution of seagrasses from grab and video data are shown. Numbers indicate the number of species found at that location (as per Tables 2 and 3). Large numerals indicate sparse seagrass coverage and small numerals size indicate very sparse.

ARTICLE IN PRESS 2192

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Fig. 5. Bathymetry of Area B, southeast of Turnagain Island. Distribution of seagrasses from grab and video data are shown. Numbers indicate the number of species found at that location (as per Tables 2 and 3). Large numerals indicate sparse seagrass coverage and small numerals indicate very sparse. MS and TW indicate the locations of current meters and oceanographic sensors during the monsoon and trade wind seasons.

3.3. Multibeam sonar The application of repeat echo sounder and/or side scan surveys to dune mobility studies is not a new area of research with studies dating back to 1965 (Jones et al., 1965; Hawkins and Sebbage, 1972; Bokuniewicz et al., 1977; McCave and Langhorne, 1982; Harris, 1989; Berne et al., 1993). However, the development of high-frequency multibeam sonar and differential GPS have allowed the production of accurate, high-resolution DEMs of the seabed (Hughes Clarke et al., 1996). The use of DEMs in sediment mobility studies is only a relatively recent application of this technology (Bartholoma et al., 2004; Duffy and Hughes Clarke, 2005; Knappen et al., 2005; Weinberg and Hebbeln, 2005). A ResonTM 8101 multibeam sonar was used to acquire bathymetry data of the survey areas. The transducer emits 101 acoustic beams of 1.51  1.51 with up to 20 pings per second, covering a swath width of up to seven times the water depth. A TSS dynamic motion sensor, mounted on the ship’s centre line, was used to correct for the pitch, roll, and heave. As the survey plan included repeat seabed surveys to monitor changes in topography, a high degree of positional accuracy was needed. An Ashtech G12 differential GPS was considered adequate for the survey, with an accuracy of 0.4 m RMS in the X and Y plane (www.ashtech.com). The data from all the peripheral sensors was recorded using the Reson 6042 Ver. 7.2 software. The acoustic signals were corrected for temperature and salinity of the seawater that were measured during the surveys using an Applied Microsystems Ltd. SV PLUSTM acoustic velocity profiler. The acoustic velocity of seawater was measured at 153971 m s1 and was consistent through the water column and between the survey areas. Sea level observations were recorded from pressure sensors co-located with the current meters (Heap et al., 2005, Daniell et al., 2006), and used to correct the bathymetric data over the survey period. Multibeam mapping of the study site was conducted in water depths of 3–12 m at a speed of 5–6 knots. The raw multibeam data files were converted to Extended Triton-Elics (XTF) format. All post-processing of the multibeam sonar data, including data editing, tide corrections and sound velocity profile corrections, was undertaken using the CarisTM HIPS/SIPS software suite. Bathymetric soundings were visually inspected and data spikes were removed to create a level and clean data set relative to mean sea level. The error of estimation for vertical soundings reported here is estimated to be 70.10 m. Final DEMs were produced using Caris with a 0.5 m cell size.

3.3.1. Dune-crest migration Repeat multibeam surveys enabled measurements of dunecrest mobility. To make the measurements, the locations of dune crests were digitised for each of the surveys. In order to provide a rigorous method for determining the precise location of the dune crests, an aspect algorithm was applied to each DEM. The aspect algorithm calculated the azimuth of the slope for each pixel within each DEM. The boundary between the lee and stoss sides of the sand dunes were characterised by a change in aspect of 1801. The dune crests were thus easily identified and then digitised using the aspect algorithm as a guide. Measurements of sand dune-crest migration were undertaken by quantifying the distance the dune crests had moved between the pairs of DEMs collected during the monsoon (March) and trade wind (October) season surveys. Points at 2 m intervals were defined along each dune crest for the first bathymetric survey. For each point, pairs of lines with azimuths +901 and 901 from that azimuth of the dune crests were defined. Dune-crest migration was measured by the shortest intersection of the two lines with the dune crests from the 2nd bathymetric survey. The length and direction of the lines provided a measure of dune-crest migration distance and azimuth. The differences between crest positions between each survey were much less than the dune wavelength, indicating that error associated with having missed an entire wavelength of migration was avoided. Crest migration analysis was also undertaken to compare crest migration between survey seasons (using the first monsoon season survey and the second trade wind season survey). The introduction of errors into the dune migration analysis either through the crest digitising process or the acquisition of the multibeam data sets (i.e. dynamic errors in DGPS positioning) are estimated to be 72 m. As a result, dune migration vectors less than 2 m are not considered reliable and were excluded from the data analysis. For the purpose of this study, a distinction is made between crest migration and bedform migration. Changes in sand dune crest position are good indicators of current activity, but are not necessarily good indicators of bedform migration since bedform migration infers that the centre of gravity of the bedform has moved (Harris, 1989). If bedforms are migrating in the direction expected from their geometry (i.e. towards their lee face) then changes in crest position and centre of gravity are assumed to be similar. As dunes in Torres Strait are known to reverse their orientations, then the position of the crest will change much more

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

rapidly than the centre of gravity (Harris, 1989). Migration of dune crests, therefore, provide an indication of recent bedload transport during reversal and should not be interpreted specifically as dune migration. A measurement of net erosion (ve values) and deposition (+ve values) can be obtained from repeat bathymetric surveys by subtracting a DEM (that occurred earlier in time) from second DEM. Dune and bank migration over time can then be inferred from the difference in water depths between two surveys. To estimate the rate of migration of the sand bank in Area B, the outline of the bank was digitised for all four DEMs. Comparing the changes in position of the outline of the sand bank over time was used to estimate the migration of the sand bank over the 6 months between survey seasons. This result was then compared to results obtained from other sand banks in Torres Strait region reported in Daniell and Hughes (2007).

3.4. Synoptic meteorological observations Hourly wind speed and direction data were obtained for Horn Island from October 1995–November 2004. The meteorological data was used to characterise seasonal variations in wind strength and direction. These variations are known to influence residual currents, dune orientation, and the net advection of sediment in Torres Strait (Wolanski et al., 1988; Harris, 1989; Hemer et al., 2004; Saint-Cast and Condie, 2006).

2193

4. Results 4.1. Seagrass surveys From a limited range of benthic environments, eight species of seagrass were observed within 34 of 69 camera stations and 11 of 71 sediment grabs in the vicinity of Turnagain Island (though not all samples were located within study areas A and B). This number is comparable the study of Sheppard et al. (2008) who targeted a wider range of habitats and observed 11 species within Torres Strait. Cymodocea serratula (occurring in 25 video and four grabs) and Halophilia ovalis (occurring in 18 video and one grab) were the most common species. Within Area A, 25 out of 42 camera stations (Table 2) and 10 out of 43 grab samples (Table 3) showed seagrass. Samples with seagrasses were evenly distributed throughout the study area though samples without seagrasses were generally found in the channel between the carbonate platforms. Seagrasses were observed in the sediment starved areas between dunes on the eastern carbonate platform. Seagrass, however, was not observed on the dunes. Within Area B, four out of seven camera stations (Table 2) and one out of 10 grab samples (Table 3) showed seagrass. Seagrass was present in samples collected from the trough area between the limbs of large sand bank. No samples were recovered with seagrasses on top of the sand bank itself. Seagrass occurred in 59% of samples within Area A compared to 40% in Area B. Area B also showed a lower diversity of

Table 2 Seagrass species observed during camera deployments (monsoon season survey) Station

Lon

Lat

Eastings

Northings

Cov

Can

Sub

Ben

8CAM05 11CAM08 13CAM10 14CAM11 19CAM14 21CAM16 22CAM17 23CAM18 24CAM19 28CAM23 30CAM25 35CAM30 37CAM32 38CAM33 40CAM35 43CAM38 44CAM39 45CAM40 46CAM41 47CAM42 48CAM43 49CAM44 50CAM45 51CAM46 52CAM47 53CAM48 54CAM49 55CAM50 56CAM51 57CAM52 68CAM56 69CAM57 70CAM58 73CAM61

142.34157 142.25433 142.22978 142.20310 142.34260 142.35025 142.35058 142.35140 142.35 142.22490 142.22908 142.22910 142.22368 142.22397 142.22688 142.21977 142.21825 142.22183 142.22535 142.21670 142.21452 142.21563 142.21290 142.21148 142.21397 142.21555 142.21760 142.21932 142.21568 142.21765 142.21252 142.21415 142.21628 142.21657

9.56697 9.57622 9.57450 9.56532 9.59178 9.59092 9.59163 9.59297 9.58333 9.57243 9.56698 9.57117 9.56713 9.57075 9.57185 9.57163 9.57190 9.57540 9.57422 9.57538 9.57520 9.57405 9.58327 9.57358 9.56687 9.56833 9.57110 9.57045 9.57098 9.56763 9.57708 9.57592 9.57890 9.57673

647233.33 637653.77 634959.86 632035.05 647335.69 648175.71 648211.63 648301.05 648151.57 634425.05 634886 634886.54 634293.23 634323.64 634642.61 633862.30 633695.36 634086.93 634473.74 633523.87 633284.67 633406.95 633103.72 632951.64 633227.55 633400.40 633624.33 633813.37 633413.64 633631.17 633064.42 633243.78 633476.40 633509.08

8942177.7 8941190.8 8941390.7 8942416.2 8939433.5 8939525.3 8939446.7 8939298.1 8940364.8 8941621.6 8942222.6 8941759.2 8942208.1 8941807.7 8941684.9 8941712.0 8941682.8 8941294.3 8941423.4 8941298.5 8941319.3 8941446.0 8940427.5 8941499.6 8942240.6 8942078.6 8941771.5 8941842.7 8941785.5 8942155.2 8941112.1 8941239.8 8940909.4 8941149.3

sp sp vs me sp sp sp sp vs vs sp sp vs vs sp vs vs vs vs vs vs vs sp vs sp sp vs vs sp vs vs sp vs vs

Low Low Low Low Low High Low High Low Low High High Low Low High Low Low High Low Low Low Low High Low High High Low Low High Low High High Undet Undet

Sandy Rubble Sandy Sandy Rubble Rubble Sandy Rubble Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Sandy Rubble Rubble

me me vs me me sp vs sp vs vs sp vs vs vs sp vs vs vs vs vs vs vs vs vs sp sp vs vs me vs vs me vs vs

Total

Cs

Cr

Hd

         

Ho

Hs

   

  

Hu

Si

Th

Total



 



5 4 2 4 2 1 2 2 3 2 3 3 1 2 1 2 2 1 1 1 1 1 4 3 5 5 4 2 4 2 1 2 4 2

 

  

   

 

   

 





  

   

      

  

     25

    

    

   

 



  1

10

18

14



 

2

13

1

Cov: coverage of seagrass; sp: sparse (10–25%); vs: very sparse (0–10%); me: medium (25–50%); Can: canopy height; undet: undetermined; low: typically less than 5 cm; high: typically less than 10 cm; Sub: substrate; Ben: coverage of benthos (as per cov); Cs: Cymodocea serrulata; Cr: Cymodocea rotunda; Hd: Halophilia decipiens; Ho: Halophilia ovalis; Hs: Halophilia spinulosa; Hu: Halodule uninervis; Si: Syringodium isoetifolium; Th: Thalassia hemprichii.

ARTICLE IN PRESS 2194

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Table 3 Seagrass species observed in sediment grab samples (monsoon season survey) Station

Lon

Lat

Eastings

Northings

8GR07 30GR26 33GR29 35GR31 40GR36 43GR39 47GR43 49GR45 67GR58 70GR61 72GR63

142.34018 142.22898 142.22822 142.22923 142.22702 142.21975 142.21670 142.21557 142.21290 142.21628 142.21205

9.56642 9.56738 9.56865 9.57117 9.57180 9.57163 9.57532 9.57402 9.57575 9.57890 9.57857

647080.9928 634874.8659 634790.9467 634900.8112 634657.9932 633860.1064 633523.8983 633400.3788 633106.6494 633476.4014 633012.2573

8942239.087 8942178.413 8942038.267 8941759.198 8941690.394 8941712.027 8941305.154 8941449.352 8941259.074 8940909.424 8940947.555

Total GR

Cs

Cr

Hd

Ho

Hs



Hu

Si

Th

Total







4 1 1 1 2 1 1 1 2 3 1

   

 

  

 



4

1

2

1

2

3

  4

1

Cs: Cymodocea serrulata; Cr: Cymodocea rotunda; Hd: Halophilia decipiens; Ho: Halophilia ovalis; Hs: Halophilia spinulosa; Hu: Halodule uninervis; Si: Syringodium isoetifolium; Th: Thalassia hemprichii.

seagrasses with stations only having zero to two species present. Area A had eight stations where three or more species were present. Seagrass coverage in both areas was typically sparse or very sparse and the canopy height typically low. The lower diversity and density of seagrass within Area B is assumed to be a result of the greater depths at that site (and therefore lower levels of light for photosynthesis) compared to Area A. 4.2. Current meters and oceanographic moorings Both current meter deployments exceeded 15 days and hence recorded at least one spring/neap cycle. Average current strengths were similar for both seasons with mean current velocities of 36.03 and 39.68 cm s1, respectively (Table 4 and Fig. 6). Net drift at both current meter stations was directed towards the west. Bedload transport, however, was directed toward the east, indicating the presence of asymmetric currents in the study area with average bedload transport measurements of 1.92 and 3.2 (  102 g cm1 s1). Unlike the average current strength, the turbidity recorded during each season was significantly different. Turbidity was 10 times greater during the monsoon season than the trade wind season (Fig. 6). In response to the high turbidity measurements during the monsoon season the PAR measurements (i.e. the amount of light available for photosynthesis) were also significantly lower than for the trade wind season. 4.3. Dune migration Although the dune migration results are reported here in relation to the monsoon and trade wind surveys, the timing of the ‘monsoon’ survey actually took place during the transition from the monsoon to the trade wind season. Therefore, our data on dune migrations from both monsoon and trade wind seasons reflect the prevalence of strong southeasterly winds. The results of the dune migration calculations (both average and upper 10th percentile) are summarised in Table 5. 4.3.1. Dune migration during the transition from the monsoon season to the trade wind season (March 2004) For the monsoon survey, the wind record for Horn Island shows that over the 2 weeks preceding and during the first part of the survey (Fig. 7, Table 6), the wind was dominantly from the northwest with average speeds of 16.9 km h1. On approximately 25 March, the wind changed direction from the northwest to then come from the southeast with an average speed of 21.9 km h1.

Table 4 Results from ADCP deployments for the monsoon and trade wind seasons

Deployment start Deployment duration Speed min. (cm s1) Speed mean (cm s1) Speed max. (cm s1) Average disp. day1 (km) Azimuth (1N) Q mean (  102 g cm1 s1) Q total (  104 g cm1) Q direction (1N) Threshold exceedance (%)

Monsoon season

Trade wind season

29 March 2004 31 days 16.329 36.032 76.578 3.84 123.6 1.92 1.55 73.45 42.8

8 October 2004 16 days 18.01 39.68 88.41 7.21 88.8 3.20 2.14 69.71 38.20

Bedload transport is represented by Q. Threshold exceedance is the number of times during the current meters deployment that bedload transport occurred.

The first multibeam sonar surveys for both Areas A and B show dunes facing to the east, resulting from east directed wind generated currents during the monsoon season. However, by the time the second multibeam survey was completed 2 weeks later, the dunes had started to reverse their asymmetry. Consequently, the dune migration analysis revealed westwards migration for all the dune crests in Area A (an average of 0.59 m day1; see Fig. 8). The highest migration rates of 1.0 m day1 for all the surveys were recorded within Area A at this time on dunes located towards the centre of the sand bank. Dune crests in Area B exhibited a westward migration direction except for a small region of eastward-migrating dunes on the northwest margin of the bank (Fig. 9). Dune migration rates in Area B (average of 0.31 m day1) were less, on average, than those from Area A. A boundary trending east–west across the northwest margin of the sand bank separates the two areas of opposing, east and west dune migration zones (Fig. 9). The existence of the two opposing dune migration zones indicates the presence of mutually evasive ebb–flood dominated channels in which the tidal asymmetry induces a residual circulation of sand around the sand bank.

4.3.2. Dune migration during the trade wind season (October 2004) The first multibeam survey completed in October 2004 in Area A revealed that all dunes were facing west (Fig. 10). Two weeks later, the second multibeam survey showed that this westward orientation had not changed. The dune migration analysis within Area A indicated the existence of discrete regions of east and west

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

2195

Fig. 6. Current strength (cm s1), turbidity (formazin turbidity units) and PAR (W m2) measurements from Area B during the monsoon and trade wind seasons. Figures A, C, and E are from the monsoon season and B, D, and F are from the trade wind season.

Table 5 Results from crest migration calculations, both average and upper 10th percentile are shown Crest migration (average)

Monsoon survey Area A Monsoon survey Area B Trade wind survey Area A Trade wind survey Area B Between seasons Area A Between seasons Area B

Crest migration (upper 10%)

For survey period (m)

Per day (m day1)

For survey period (m)

Per day (m day1)

8.35 4.38 2.74 1.84 29.26 11.90

0.59 0.31 0.19 0.13 0.13 0.05

14.1 7.6 3.4 3.3 48.7 26.0

1.0 0.54 0.24 0.23 0.22 0.12

Survey period for the monsoon and trade wind surveys is 14 days, 216 days elapsed between the two survey seasons.

migration dunes. The dunes on the eastern platform migrated west and dunes within the central channel migrated to the southeast. An average migration rate of 2.74 m (0.19 m day1) was calculated for the 2-week period (1/3 that of the monsoon season survey). Thus mutually evasive ebb–flood transport zones exist also in Area A. In Area B, the first multibeam survey for the trade wind season revealed the presence of westward facing dunes over most of the sand bank, apart from a small zone of east facing dunes on the northwest margin (Fig. 11), indicating a mutually evasive ebb– flood dominated sand transport regime. The dune migration analysis indicated an average migration rate of 0.13 m day1 (Fig. 11).

4.3.3. Dune migration during the trade wind season (7-month period) Crest migration analysis between the two survey seasons revealed that the majority of the dunes migrated to the west

within both study areas, consistent with the expected winddriven residual currents forced by the trade winds during the 7-month period (Figs. 12 and 13). The highest migration distances, 48.7 m over the 7-month period, were recorded in Area A. The average rate of dune migration within Area A was also approximately 2.5 times that of Area B (0.13 m day1 compared to 0.05 m day1, respectively). Within Area B, a zone of eastward migrating dunes (5–10 m over the time period) was observed on the northwest margin of the sand bank. The presence of eastward migrating dunes on the northwest margin of the bank over the 6-month period supported the observations made during the monsoon and trade wind surveys of a persistent, mutually evasive ebb–flood channel system. The overall migration of the sand bank in Area B was estimated by tracing the area covered by the bank and comparing its extent between the monsoon and trade wind seasons (Fig. 14). The only part of the sand bank that appeared to move any significant distance over the 6 months was the northeastern arm of the sand bank, which migrated 15 m to the west. This result is similar to

ARTICLE IN PRESS 2196

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Fig. 7. Progressive vector plot of wind data for the 2 weeks before and during the monsoon season survey. The start of the monsoon season survey is marked by the large red circle. Origin of the plot (0, 0) marks the start of the data plot.

Table 6 Basic statistics from synoptic meteorological observations at Horn Island

Average direction (1N) Average wind speed (km h1) Maximum wind speed (km h1)

Monsoon before survey

Monsoon during survey

Trade wind before survey

Trade wind during survey

277.7 16.9 29.9

138.7 22.0 38.9

116.5 23.4 40.1

113.2 22.0 35.0

Fig. 8. Monsoon season dune-crest migration within Area A measured using repeat multibeam sonar surveys. Approximately 14 days elapsed between surveys.

rates of migration of other sand banks in Torres Strait (average of 14.4 m yr1) measured by Daniell and Hughes (2007). Biggest differences between crest locations occurred in the centre and northwest arm of the bank where erosion and deposition of up to 73.5 m (average 71 m) were recorded (Fig. 15). These areas coincide with regions of highest dune mobility (Fig. 13).

5. Discussion The aim of this project was to attempt to quantify the migration of sand dunes and sand banks located in close proximity to seagrass beds that have been known to disappear episodically. The two key questions being addressed by this

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

2197

Fig. 9. Monsoon season dune-crest migration within Area B measured using repeat multibeam sonar surveys. Approximately 14 days elapsed between surveys. The boundary between east (blue vectors) and west (red vectors) migrating dunes is marked in purple.

Fig. 10. Trade wind season dune-crest migration within Area A measured using repeat multibeam sonar surveys. Approximately 14 days elapsed between surveys. The boundary between east and west migrating dunes is marked in purple.

Fig. 11. Trade wind season dune-crest migration within Area B measured using repeat multibeam sonar surveys. Approximately 14 days elapsed between surveys. The boundary between east and west migrating dunes is marked in purple.

ARTICLE IN PRESS 2198

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Fig. 12. Dune migration between survey seasons within Area A. Some dune crests have been omitted from calculations to prevent interference from dunes that have migrated over one wavelength.

Fig. 13. Dune migration between survey seasons within Area B. Some dune crests have been omitted from calculations to prevent interference from dunes that have migrated over one wavelength. The boundary between east (blue) and west (red) migrating dunes for the two survey seasons is shown in magenta.

Fig. 14. Map showing changes in location of sand bank within Area B. The area outlined in blue marks the extent of the sand bank during the monsoon season survey. The area outlined in red marks the extent of the sand bank during the trade wind season survey.

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

2199

Fig. 15. Difference grid for Area B, all measurements are in meters. Grid was generated by subtracting the first survey of the monsoon season from the last survey of the trade wind season. Blue shades represent regions of net erosion. Red shades represent regions of net deposition.

research are: (1) Under what conditions can seagrass beds coexist with migrating sand bodies? and (2) Are there any aspects of sand dune or sand bank migration that might explain the observed widespread dieback of local seagrass beds?

5.1. The coexistence of seagrass with bedforms In some regions, seagrasses have been observed to grow upon the flanks of sand banks, thereby stabilising the sediment and halting their migration (e.g. Spencer Gulf, South Australia; Gostin et al., 1988). Our underwater video has shown, however, that seagrass beds are not established upon the stoss or lee slopes of mobile sand dunes or sand banks in the two study areas. This is assumed to be due to the dunes being in highly energetic environments that undergo high rates of reworking due to strong tidal currents and thus are difficult environments to colonise. The threshold exceedances were 42.8% and 38.2%, respectively, indicating that bedload transport was occurring regularly through the currentmeter deployments. These values are also considered to underestimate the true rates of bedload transport as they do not include the additional influence of local wave activity. The trough areas between sand dunes in Area A have, however, been shown to be colonised with seagrasses. The dunes within Area A migrated 30 m to the west during the 7 months between surveys. Given their average wavelength 62 m, the original trough areas would be completely buried by and overturned once in every 14.4 months. The duration of burial is related to the width of the body of the dunes (31 m), which would require about 7.2 months on average to move past a fixed point. The height of the seagrass canopy also influences the potential impact of migrating dunes. As the canopy height of the seagrasses in the study is low, it is considered to be negligible in these calculations. This implies that seagrass beds in the troughs of Area A dunes are able to either re-establish themselves after having been destroyed by burial, or else they are able to survive burial and regenerate after the body of the dune has moved past. The sand dunes superimposed on the sand bank in Area B are less dynamic than those in Area A (i.e. they migrated only 11.9 m over the 7 months on average) but underwater video showed that no seagrass was growing at any location on the sand bank nor on dune lee or stoss slopes or dune troughs. If dune migration rate has not precluded seagrass growth on the bank or dune surfaces, then some other process must be responsible. A key

difference between the dunes in Areas A and B is the sedimentstarved character of Area A compared with the full-bedded dunes of Area B. Dunes in Area A are sediment-starved, meaning that individual dunes are separated by flat trough areas having little or no mobile sediments. Area A dune troughs experience no significant bedload transport regardless of current strength due to the paucity of sand and hence seagrasses might be able to grow due to the relatively stable seabed. Area B dunes, in contrast, are full-bedded, meaning the lee slope of one dune merges seamlessly with the stoss slope of its neighbour. The troughs of full-bedded dunes are mantled by a layer of mobile sand that is mobilised during each tidal cycle. Seagrasses may not be able to colonise this habitat because the substrate is being regularly reworked (40% of the time) and thus is a hostile environment for seagrasses to colonise. The only part of Area B to support seagrass beds was the trough areas separating sand banks. These are relatively sedimentstarved areas and relatively stable given the slow overall migration rate of the sand bank (15 m over a 7-month period). Hence, the inter-bank troughs of mutually evasive ebb–flood tidal channels may provide a more suitable habitat for seagrasses compared to the banks themselves.

5.2. Conceptual model of seagrass bed stability in relation to migrating dunes From the above analysis, habitats suitable for seagrass colonisation in relation to mobile sand dunes and sand banks in Torres Strait can be viewed as a continuum, ranging from ‘hostile’ environments on the bodies of migrating dunes through to low energy, sandy seabeds subject to limited bedload transport (Fig. 16). Dunes and troughs of full-bedded dunes are not colonised, probably due to the hostility of a semi-continuously reworked substrate (Fig. 16a). The seagrass habitat most proximal to migrating dunes is the dune trough habitat between sedimentstarved dunes (Fig. 16b). The average migration rate of sedimentstarved dunes in Area A (0.13 m day) and their average wavelength (62 m) suggests that trough areas are buried once in every 14.4 months. At a greater distance from dunes is the interbank trough habitat flanking large sand bank features, in which some seagrasses have been found (Fig. 16c). Such troughs may be mutually evasive tidal channels, from which bedload transport

ARTICLE IN PRESS 2200

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Fig. 16. Conceptual model of risk to seagrasses from dune migration.

diverges leaving a stable, sandy-gravel substrate suitable for seagrass. However, the depth, strong currents and exposure of basement rock or hard pavement in such areas may not be optimal for seagrass establishment or growth (evident by sparse coverage and low canopy heights). The migration rates of sand banks (15 m year1) in combination with the average spacing between sand bank limbs (1500 m, within Area B) indicates an average frequency of burial of only once every 100 years. With greater distance from sand dunes and sand banks, the current strength decreases and seagrasses are not threatened by burial from migrating bedforms. However, broad areas of Torres Strait seabed are believed to be comprised of a hard limestone pavement having little if any mobile sediment cover (Harris, 2001). Given that ideal seagrass habitat includes a sediment layer for seed germination and establishment of a root system, such limestone pavement areas may not provide an optimal habitat for seagrass. An ‘ideal’ seagrass habitat in presented in Fig. 16d and shows a low-energy environment with an optimum amount of sediment cover.

5.3. Seagrass dieback in relation to bedform migration Based on the conceptual model for seagrass habitat as a function of proximity to mobile sand dunes (Fig. 16), it is evident that seagrass beds in dune troughs are the most vulnerable to burial (once or twice a year on average). Seagrass beds within interbank channel troughs and beds that are distal to migrating dunes and sand banks are only vulnerable to burial at sub-century time scales. Therefore, we conclude that the widespread seagrass dieback events documented for the region, which have occurred on a sub-decadal time period, around Turnagain Island are not attributed to the migration of bedforms. Given that the degree of substrate mobility can be key factor limiting seagrass growth, then other types of sand deposits such as sand ribbons could restrict seagrasses in Torres Strait. Sand ribbons can have a range of morphologies, from elongate trains of mobile dunes to flat, mobile sand sheets (Kenyon, 1970). The spatial distribution of sand ribbons in the area around Turnagain Island is not known but they have been documented from sidescan sonar imagery collected in other parts of Torres Strait

ARTICLE IN PRESS J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

(Harris, 1988). Further research is required to determine the distribution of sand ribbons in the Turnagain Island area to eliminate these deposits as a possible factor in the seagrass dieback events. Measurements of current strength, turbidity and PAR during both the monsoon and trade wind seasons provides an insight into more regional scale hydrodynamic process active in the Torres Strait. During both surveys, mean current strengths were approximately equal, bedload transport calculations also only differed by 50%. Turbidity, however, was 10 times greater during the monsoon season than the trade wind season. PAR, in response to the elevated turbidity was comparatively low during the monsoon season. An influx of sediment from the Gulf of Carpentaria has been suggested to explain this phenomenon (Saint-Cast, 2008) and provides a mechanism for a regional scale turbidity event in the Torres Strait.

6. Conclusion Repeat multibeam sonar surveys combined with current meter deployments provided data on dune mobility in an area known to experience seagrass dieback in Torres Strait, northern Australia. Seagrass beds were observed in the troughs of sediment-starved dunes at the western end of Turnagain Island, in which the trough region is characterised by flat limestone draped with patches of coarse sand and gravel lag deposits. In contrast, no seagrass beds were observed on full-bedded dunes associated with a sand bank deposit at the east end of Turnagain Island. We attribute the absence of seagrass beds in the latter area to the influence of the semi-continuous bedload transport that would occur over the entire area of full-bedded dunes (but not in the troughs of sediment-starved dunes). The highest rates of dune mobility were measured for sediment-starved dunes during the transition from the monsoon to the trade wind season (0.59 m day1). The rapid rates of crest migration were a result of east facing dunes (at the end of the monsoon season) readjusting to a new hydrodynamic regime with the onset of the trade wind season. Hydrodynamic conditions experienced during the trade wind season survey resulted in dune migration rates of around 0.19 m day1. These migration rates, in conjunction with dune wavelengths of around 62 m, suggest seagrass beds are buried once every 14.4 months. By comparison, the slow rates of migration for much larger sand bank features (15 m year1) in conjunction with wavelengths of 1500 m, suggest seagrass beds are buried in the interbank troughs areas only once every 100 years. A conceptual model is proposed to illustrate different seagrass habitats as a function of dune migration and sand supply. Seagrasses in closest proximity to full-bedded sand dunes are at the highest risk of burial and abrasion, followed by troughs of sediment-starved dunes, and with the troughs between sand banks and interbank regions being the least hostile for seagrasses. Based on this analysis, we conclude that episodes of widespread seagrass dieback are not related to burial and death caused by the migration of dunes. However, other low amplitude sand ribbon deposits are known to occur in Torres Strait and their migration may be a factor in limiting the distribution of seagrasses. Further research is required to map sand ribbons and to measure their influence over the occurrence of seagrass beds. High levels of turbidity observed at the end of the monsoon season do not appear to be a result of local current activity. Instead, an influx of sediment from the Gulf of Carpentaria is a more likely scenario (Saint-Cast, 2008). This type of event may provide a more likely mechanism for regional scale seagrass

2201

dieback rather that dunes mobility based on the results of this study.

Acknowledgements Financial support for the work presented here was provided by Geoscience Australia, Torres Strait Cooperative Research Centre and Queensland Department of Primary Industry and Fisheries. The Bureau of Meteorology Climate and Consultancy Office provided the synoptic meteorological data sets from Horn Island. The authors wish to thank the staff of the sedimentology laboratory at Geoscience Australia for timely and efficient production of sediment composition and texture data. We are grateful to Prof. Bob Henderson (School of Earth Sciences, James Cook University) for permission to use the RV James Kirby, vibrocoring equipment and the Reson 8101. The efforts of the other participants in the Torres Strait surveys (Simon Kerville, Lyndon O’Grady, Dr Fredrick Saint-Cast, Franz Villagran, Mike Sexton, Jack Pittar, Ron Battersby and Kevin Hooper) made the survey program a success and acquired key data sets for TSCRC research. Published by permission of the Chief Executive Officer, Geoscience Australia. References Allen, J.R.L., 1968. The nature and origin of bedform hierarchies. Sedimentology 10, 161–182. Bartholoma, A., Ernstsen, V.B., Flemming, B.W., Bartholdy, J., 2004. Bedform dynamics and net sediment transport paths over a flood-ebb tidal cycle in the Gra˚dyb channel (Denmark), determined by high-resolution multibeam echosounding. Danish Journal of Geography 104 (1), 45–55. Belderson, R.H., Johnson, M.A., Kenyon, N.H., 1982. Bedforms. In: Stride, A.H. (Ed.), Offshore Tidal Sands, Processes and Deposits. Chapman & Hall, London, pp. 27–57. Berne, S., Castaing, P., Le Drezen, E., Lericolais, G., 1993. Morphology, internal structure, and reversal of asymmetry of large subtidal dunes in the entrance to Gironde Estuary. Journal of Sedimentary Petrology 63 (5), 780–793. Bokuniewicz, H.J., Gordon, R.B., Kastens, K.A., 1977. Form and migration of sand waves in a large estuary, Long Island Sound. Marine Geology 24, 185–199. Carruthers, T.J.B., Dennison, W.C., Longstaff, B.J., Waycott, M., Abal, E.G., McKenzie, L.J., Lee Long, W.J., 2002. Seagrasss habitats of northeast Australia: models of key processes and controls. Bulleting of Marine Science 71 (3), 1153–1169. Dalrymple, R.W., Rhodes, R.N., 1995. Estuarine dunes and bars. In: Perillo, G.M.E. (Ed.), Geomorphology and Sedimentology of Estuaries, Developments in Sedimentology, vol. 53. Elsevier, Amsterdam, pp. 359–422. Daniell, J., Hughes, M.G., 2007. The morphology of barchan-shaped sand banks from western Torres Strait, northern Australia. Sedimentary Geology 202, 638–652. Daniell, J., Hemer, M., Heap, A., Mathews, E., Sbaffi, L., Hughes, M., Harris, P., 2006. Biophysical processes in the Torres Strait marine ecosystem II—Survey results and review of activities in response to CRC objectives. Geoscience Australia record 2006/10, Canberra, 210pp. Duffy, G.P., Hughes Clarke, J.E., 2005. Application of spatial cross correlation to detection of migration of submarine dunes. Journal of Geophysical Research 110, F04S12. Gadd, P.E., Lavelle, J.W., Swift, D.J.P., 1978. Estimates of sand transport on the New York shelf using near-bottom current meter observations. Journal of Sedimentary Petrology 48, 239–252. Gostin, V.A., Belperio, A.P., Cann, J.H., 1988. The Holocene non-tropical coastal and shelf carbonate province of southern Australia. Sedimentary Geology 60, 51–70. Guy, H.P., Simons, D.B., Richardson, E.V., 1966. Summary of alluvial channel data from flume experiments 1955–1961. US Geological Survey Professional Paper. US Geological Survey, 92pp. Harris, P.T., 1988. Sediments, bedforms and bedload transport pathways on the continental shelf adjacent to Torres Strait, Australia-Papua New Guinea. Continental Shelf Research 8, 979–1003. Harris, P.T., 1989. Sand dune movement under tidal and wind-driven currents in a shallow marine environment: Adolphus Channel, northeastern Australia. Continental Shelf Research 9, 981–1002. Harris, P.T., 1991. Reversal of subtidal dune asymmetries caused by seasonally reversing wind-driven currents in Torres Strait, northeastern Australia. Continental Shelf Research 11, 655–662. Harris, P.T., 1994. Muddy waters. The physical sedimentology of Torres Strait. In: Choat, J.H., Bellwood, O., Saxena, N. (Eds.), Recent Advances in Marine Science and Technology, vol. 95. Proceedings of the PACON Symposium, July 1994, Townsville, Australia, pp. 149–160.

ARTICLE IN PRESS 2202

J.J. Daniell et al. / Continental Shelf Research 28 (2008) 2188–2202

Harris, P.T., 2001. Environmental management of Torres Strait: a marine geologist’s perspective. In: Gostin, V.A. (Ed.), Gondwana to Greenhouse: Environmental Geoscience—An Australian Perspective, vol. 21. Geological Society of Australia Special Publication, Adelaide, pp. 317–328. Hawkins, A.B., Sebbage, M.J., 1972. The reversal of sand dunes in the Bristol Channel. Marine Geology 12, M7–M9. Heap, A.D., Hemer, M., Daniell, J., Mathews, E., Harris, P.T., Kerville, S., O’Grady, L., 2005. Biophysical Processing in the Torres Strait Marine Ecosystem—post cruise report. Geoscience Australia record 2005/11, 112pp. Hemer, M.A., Harris, P.T., Coleman, R., Hunter, J., 2004. Sediment mobility due to currents and waves in the Torres Strait-Gulf of Papua region. Continental Shelf Research 24, 2297–2316. Hughes Clarke, J.E., Mayer, L.A., Wells, D.E., 1996. Shallow-water imaging multibeam sonars: a new tool for investigating seafloor processes on the coastal zone and on the continental shelf. Marine Geophysical Researches 18, 607–629. Johannes, R.E., MacFarlane, J.W., 1991. Traditional fishing in the Torres Strait Islands. CSIRO Division of Fisheries, Hobart, Australia. Jones, N.S., Kain, J.M., Stride, A.H., 1965. The movement of sand waves on Warts Bank, Isle of Man. Marine Geology 3, 329–336. Kenyon, N.H., 1970. Sand ribbons of the European tidal seas. Marine Geology 9, 25–39. Kenyon, N.H., Belderson, R.H., Stride, A.H., Johnson, M.A., 1981. Offshore tidal sandbanks as indicators of net sand transport and as potential deposits. Special Publication of the International Association of Sedimentologists 5, 257–268. Knappen, M.A.F., van Bergen Henegouw, C.N., Hu, Y.Y., 2005. Quantifying bedform migration using multibeam sonar. Geo-Marine Letters 25, 306–314. Long, B.G., Poiner, I.R., 1997. Seagrass communities of Torres Strait, Northern Australia. Marine Research, CSIRO. Report: MR-GIS 97/6. Long, B.G., Skews, T., 1997. The potential influence of land runoff on distribution and abundance of seagrass on the reefs in Torres Strait. Marine Research, CSIRO. Report: MR-GIS 97/6. Long, B., Skewes, T., Taranto, T., Thomas, M., Isdale, P., Pitcher, R., Poiner, I., 1997. Seagrass dieback in Northwestern Torres Strait. CSIRO Division of Marine Research Report MR-GIS 97/6. Marba, N., Duarte, C.M., 1995. Coupling of seagrass (Cymodocea nodosa) patch dynamics to subaqueous dune migration. Journal of Ecology 83, 381–389. Margvelashvili, N., Saint-Cast, F., Condie, S., 2008. Numerical modelling of the suspended sediment transport in Torres Strait. Continental Shelf Research, this volume, doi:10.1016/j.csr.2008.03.037.

Marsh, H., Lawler, I., Kwan, D., Pollock, K., Alldredge, M., 2004. Dugong distribution and abundance in Torres Strait. Australian Fisheries Management Authority Torres Strait research program final report R01/0895. McCave, I.N., Langhorne, D.N., 1982. Sand dunes and sediment transport around the end of a tidal sand bank. Sedimentology 29, 95–110. Mellors, J.E., McKenzie, L.J., Coles, R.G., 2008. Seagrass-watch: engaging Torres Strait Islanders in marine habitat monitoring. Continental Shelf Research, this volume, doi:10.1016/j.csr.2008.03.041. Poiner, I.R., Peterken, C., 1996. Seagrasses. In: Zann, L.P., Kailola, P. (Eds.), The State of the Marine Environment Report for Australia. Technical Annex: 1. Great Barrier Reef Marine Park Authority, Townsville, Australia, pp. 40–45. Saint-Cast, F., 2008. Multiple time-scale modelling of the circulation in Torres Strait, Australia. Continental Shelf Research, this volume, doi:10.1016/ j.csr.2008.03.035. Saint-Cast, F., Condie, S., 2006. Circulation modelling in Torres Strait. Geoscience Australia record 2006/18, Canberra, 88pp. Sheppard, J.K., Carter, A.B., McKenzie, L.J., Coles, R.G., 2008. Spatial patterns of subtidal seagrasses and their tissue nutrients in the Torres Strait, northern Australia: implications for management (this volume). Walker, D.I., Carruthers, T.J.B., Marrison, P.F., McComb, A.J., 1996. Experimental manipulation of canopy density in a temperate seagrass meadow: effects on sediments. In: Kuo, J., Phillips, R.C., Walker, D.I., Kirkman, H. (Eds.), Seagrass Biology. University of Western Australia, Perth, pp. 117–122. Waycott, M., McMahon, K., Mellors, J., Calladine, A., Kleine, D., 2004. A Guide to the Tropical Seagrasses of the Indo-Pacific. James Cook University, Townsville, Australia. Weinberg, C., Hebbeln, D., 2005. Impact of dumped sediments on subaqueous dune, outer Weser Estuary, German Bight, southeastern North Sea. Geo-Marine Letters 25, 43–53. Wolanski, E., 1986. A review of the physical oceanography of Torres Strait. In: Haines, A., Williams, G.C., Coates, D. (Eds.), Torres Strait Fisheries Seminar, Port Moresby, 11–14 February 1985. Australian Government Publishing Service, Canberra, pp. 275–291. Wolanski, E., Thompson, R.E., 1984. Wind-driven circulation of the northern Great Barrier Reef continental shelf in Summer. Estuarine and Coastal Shelf Science 18, 291–314. Wolanski, E., Ridd, P., Inoue, M., 1988. Currents though Torres Strait. Journal of Physical Oceanography 18, 1535–1545.