Summary Although ecosystem structure and function can be influenced by both bottom up (primary productivity) and top-down (predation) processes (ref), top-down processes (predation) have been demonstrated to account for over 70% of the variation in the some temperate kelp communities (Halpern et al in press). Removal of key predators such as fish and lobsters, has been shown to have the potential to result in trophic cascade effects in which grazers may take over the system, producing massive changes to ecosystem structure. (Babcock et al 1999, Shears et al 2003, Steneck et al 2004). Locally, it has been suggested that predation by western rocklobster Panuluris cygnus may be particularly important in the structuring of local ecosystems. Abundances of P.cygnus have been shown to be strongly negatively correlated with the abundance of large (>2cm) trochid gastropods (Edgar 1990) and adult P.cygnus are known to feed readily on a wide range of locally abundant crustaceans and molluscs (James and Tong 1997, James et al 2001, Power et al 2005). Despite this, understanding about the linkages between lobster predation and local macroinvertebrates populations is still currently very limited. Here we contrast differences in the rates of predation on locally abundant blue mussels Mytilus edulis (Lamarck 1819), using fished and unfished areas as the principal source of contrast. Methods Predation rates on mussels, Mytilus eduli, were measured by adapting predator exclusion methods first described in Edgar 1990. To test the hypothesis that large predators, including P.cygnus, may be controlling abundances of molluscs, caged and uncaged mussels were deployed inside and outside fished reserves at Marmion, Rottnest Island and Jurien Bay. At each of the three locations, 12 similar sized mussels (70mm total length) were attached to a series of 8 plastic mesh grids, with 4 of the grids being fully enclosed to prevent predator access and 4 of the grids remaining open to allow predator access. (Fig. 1) 8 grids were attached to concrete blocks using cable tiestm and blocks deployed in approximately 4m water depth on seagrass substrate immediately adjacent high relief limestone reef. At Marmion and Jurien Bay, 3 blocks containing attached mussels were deployed outside reserves and 2 blocks with mussels were deployed inside reserves. At Rottnest Island 3 blocks were deployed both inside and outside of reserves. After a deployment period of seven days all grids were retrieved and the mussels on each plastic grid recorded as either alive, dead or missing

Twenty-nine sites have been surveyed: 13 in Marmion Marine Park (one near the sewage outfall and the remaining forming transects across a depth gradient with 4 inner, 4 mid and 4 outer), 6 in Cockburn Sound (5 forming a transect along an anthropogenic stress gradient at the northern end and 1 at the southern end), 2 localities between these regions at City Beach and Hall Bank, 1 within the Swan River and 7 at Jurien Bay to the north. At each site, physical factors of bottom water temperature, depth, water column light attenuation and the structure of sediment such as presence of ripples, depth of anaerobic sand and compactness were recorded. In addition, a water column integrated sample was collected and chlorophyll a quantified. Five transects of 5x1m were established at each site, ten 30mm diameter cores to 50mm depth and three 150mm diameter cores to 100mm depth were randomly collected within the transect, and all macrofauna collected using a small rake to search the sediment. Macrofauna were identified, counted and weighed. The small cores were sectioned into 0-2cm and 2-5cm depths and the size fractions from all 10 relating to 1 transect pooled and subsampled: one was analysed for organic carbon and nitrogen and a second for benthic microalgae biomass as chlorophyll a. One of the large cores was dry sieved through a set of 15 sieves from 0.063 to 8.0 mm pore size to determine proportions of grain size. One of the remaining cores was washed through 9 sieves from 8.0 to 0.5mm and biomass of infauna of each size class determined and daily secondary production calculated.