In eastern Canada, the destruction of kelp beds by dense aggregations (fronts) of the omnivorous green sea urchin, Strongylocentrotus droebachiensis, is a key determinant of the structure and dynamics of shallow reef communities. Recent studies suggest that hydrodynamic forces, but not sea temperature, determine the strength of urchin-kelp interactions, which deviates from the tenets of the metabolic theory of ecology (MTE). We tested the hypothesis that water temperature can predict short-term kelp bed destruction by S. droebachiensis in calm hydrodynamic environments. Specifically, we experimentally determined relationships among water temperature, body size, and individual feeding in the absence of waves, as well as among wave velocity, season, and aggregative feeding. We quantified variation in kelp-bed boundary dynamics, sea temperature, and wave height over three months at one subtidal site in Newfoundland to test the validity of thermal tipping ranges and regression equations derived from laboratory results. Consistent with the MTE, individual feeding during early summer (June-July) obeyed a non-linear, size- and temperature-dependent relationship: feeding in large urchins was consistently highest and positively correlated with temperature <12°C and dropped within and above the 12–15°C tipping range. This relationship was more apparent in large than small urchins. Observed and expected rates of kelp loss based on sea temperature and urchin density and size structure at the front were highly correlated and differed by one order of magnitude. The present study speaks to the importance of considering body size and natural variation in sea temperature in studies of urchin-kelp interactions. It provides the first compelling evidence that sea temperature, and not only hydrodynamic forces, can predict kelp bed destruction by urchin fronts in shallow reef communities. Studying urchin-seaweed-predator interactions within the conceptual foundations of the MTE holds high potential for improving capacity to predict and manage shifts in marine food web structure and productivity.

Supplement 1 Methods 1. Collection and maintenance of urchins Urchins used in Experiment 1 and Experiment 2 were hand collected by divers at depths of 3 to 6 m in the barrens at Bread and Cheese Cove (BCC) in January, June, and July 2012. They were transported in large containers filled with seawater to the Ocean Sciences Centre (OSC) of Memorial University of Newfoundland. Upon arrival at the OSC (<5 hours after collection), urchins were transferred to 330-L holding tanks supplied with ambient flow-through seawater pumped in from a depth of ∼5 m in the adjacent embayment, Logy Bay, and sorted by size. All individuals with a test diameter of 40 to 60 mm that clung or displaced readily in the tanks, indicating that the podia functioned normally, were kept for the experiments. This size class was chosen because individuals of this size are sexually mature (Himmelman 1986, Raymond & Scheibling 1987, Munk 1992) therefore eliminating potential behavioral differences between mature and non-mature individuals, and it was the most frequent size class at times of collection. Each holding tank contained 200 urchins. Urchins used in Experiment 1 were not fed because urchin feeding in eastern Canada at the time the experiment was conducted (January) is typically low (Scheibling & Hatcher 2007, P. Gagnon, unpublished data) and feeding them could have altered metabolic activity and behavior. Urchins used in Experiment 2 were fed every two days with 25 g (wet weight) of freshly collected Alaria esculenta blades cut into pieces of ∼2.5 x 2.5 cm to standardize hunger levels at a time of year (June and July) when feeding in eastern Canada markedly increases (Scheibling et al. 1999, Gagnon et al. 2004, Lauzon-Guay & Scheibling 2007, Frey & Gagnon 2015). Urchin feces and unconsumed kelp were removed from the holding tanks every two days. Water temperature in the holding tanks prior to trials in Experiment 1 and Experiment 2 was measured with a temperature logger with a precision of ±0.5 °C (HOBO Pendant; Onset Computer Corporation). It averaged 4.1 °C (±0.2) and 10.0 °C (±0.9), respectively.