LINKS BETWEEN RAIN, SALINITY, AND PREDATION IN A ROCKY SUBTIDAL COMMUNITY

Physical control of predation appeared likely in subtidal habitats of fjords in southwestern New Zealand because they intersect a shallow lo w-salinity layer (LSL) where marine predators might be excluded by low -salinity stress. We investigated links between predator distributions , predation intensity, rainfall, the depth of the LSL, and sessile inv ertebrate community structure at two sites in Doubtful Sound, New Zeal and. Sessile invertebrate communities living on the rock walls of the fjord showed striking patterns of vertical zonation corresponding to t he depth of the LSL. Barnacles and mussels, Mytilus edulis galloprovin cialis, were most abundant at the 1-3 m depth range commonly bathed by the LSL, while sponges, bryozoans, and ascidians dominated space belo w it (6-18 m depth). Mussel abundance declined sharply with depth at b oth sites, from maxima of 52-80% cover in the LSL (3 m) to <2% cover j ust below the LSL (6 m). To test the hypothesis that the LSL influence s the lower limit of the mussel zone by excluding predators and reduci ng predation from mobile invertebrates living below it, we measured pr edator densities and tracked predation on transplanted mussels, M. e. galloprovincialis, in the LSL (2.5 m depth) and at 3-4 depths below. P redators were most abundant just below the lower boundary of the LSL d uring February and April 1993. No transplanted mussels were consumed i n the LSL at 2.5 m, but 20-80% of them were consumed in experiments co nducted below it, indicating that the LSL represents a spatial refuge from predation. The shallowing of the LSL to 1.5 m depth during low ra infall in November 1993 provided an additional opportunity to evaluate the influence of the LSL on predation. Repeated predator surveys show ed that sea stars and urchins moved up with the ascending LSL to creat e a pattern of high predator abundance at 2.5 m depth in November, sug gesting that the shallow zone previously buffered from predation by th e LSL was not a predation refuge in November. Rainfall and salinity at 2.5 m depth were inversely related. A significant regression of rainf all one day prior to salinity profiling and the depth of the LSL bound ary (30 mg/g salinity) explained 61.5% of the variation in the depth o f the boundary, which represented the maximum depth of the predation r efuge. The regression enabled us to predict the depth of the predation refuge over a 1.5-yr period beginning in October 1992. The average de pth of the refuge was 2.35 m, ranging from 1.5 to 5.6 m as a function of rainfall. Daily tidal variation increased the vertical excursion of the refuge by 0.9-1.7 m. Distinguishing between a partial predation r efuge where the LSL occurred at a given depth for a part of a day, and a complete refuge where a depth occurred in the refuge for the entire day, we estimated that the 2.5-m habitat was a complete refuge during 10% of the 18-mo period and a partial predation refuge 90% of the tim e. The dose correspondence between the average depth of the predation refuge (2.35 m) and the depth of maximum mussel cover (3.0 m) suggests that the mussel zone is restricted to the LSL by high predation at th e lower boundary of the layer. The alternative hypothesis, that the lo wer limit of the mussel zone (6 m depth) was determined by a failure o f mussel recruitment to this depth, was tested by a l-yr recruitment e xperiment. Mussels recruited to the lower edge of the mussel zone at 6 m, although overall levels of mussel recruitment during May 1993-Apri l 1994 were too low to maintain the observed zonation pattern. By incr easing the spatial extent of the low-salinity layer, we speculate that changes in climatic conditions leading to increased freshwater runoff may alter salinity gradients, bathymetric distributions of predators, and the structure of shallow subtidal communities, especially in fjor ds.