Dataset: Urchin and Abalone Density Responses

Experiment

To test whether and at what distance the presence of Pycnopodia can reduce prey density, we performed an underwater caging experiment at three urchin barren sites approximately six km east of Sitka in February 2023: Ellsworth Cut (57.036, -135.280), Harris Island (57.033, -135.277), and Whale Park (57.033, -135.255). Each experimental array consisted of a central cage with four 4 m long radial transect lines (Figure 1). We constructed cages (30 x 30 x 15 cm; l x w x h) using a PVC frame covered in ~1cm Vexar mesh fastened with zip ties that could be opened underwater to add a Pycnopodia. We attached a 4m long lead line to each corner of the cage, forming a plus pattern (Figure 1). We simultaneously deployed four of these arrays at each site in two blocks. In each block, one cage served as an experimental treatment (with a Pycnopodia) and the second cage served as a control (an empty cage). We placed all arrays in areas with high sea urchin density, hard rocky substrate, and low rugosity, and the cages within a block were ~20-30 m apart.

Once the four cages and attached lead lines were deployed, but before adding sea stars to the treatment cages, we surveyed the initial densities of pinto abalone and red and green sea urchins using 0.25m^2 quadrats at metre marks 0, 0.5, 1, 2, and 3.5 along each lead line. The experiment began when we sealed a Pycnopodia (9.5-19.5 cm radius, lab acclimated for >2 weeks) into the experimental treatment cage and the control cage was sealed with a dive weight. To each lead line of all cages, we attached a yellow nylon “kelp line” each with 9 blades attached at distinct meter marks. At 15-30 minutes after sea stars were deployed, we resurveyed the grazer densities in the same quadrat locations. We then left the array overnight and surveyed grazer densities again at approximately 24 hours. After the 24 hour surveys, we removed the arrays.

Analyses

First, we calculated the change in sea urchin density from the start of the experiment [as density at each time point - density before start] for each quadrat location and species (abalone, red or green urchins), so that positive and negative values represent increases and decreases in density (respectively) after 30 mins or 24 hours of deployment. We dropped pinto abalone (H. kamtschatkana) from analyses due to low counts in density surveys. We tested the main and interactive effects of  Pycnopodia treatment and continuous distance from the cage on the change in density of grazers in each quadrat using four mixed effects hierarchical linear models, one for each urchin species and time point combination, fit using the lmer() function in the lme4 package in R. When possible, we included site, block (nested within site), and array (nested within site, block, and treatment) as random factors to account for the non-independence of quadrats within a given array, and arrays within a block, and blocks within a given site. Array was dropped from the red urchins at 30 minutes model and site and array were dropped from the green urchins at 30 minutes model to avoid singularity errors while keeping the maximal random effects structure justified by the design. We again included site, block (nested within site), and array (nested within site, and block, and treatment), and transect (nested within site and array) as random factors. 

We then calculated the net effect of the sea star at 24 hours on sea urchin density as the difference in the average urchin density in the sea star treatment minus its paired control treatment (i.e., the differences between treatments in each block). In other words, for each block and at a given distance from the cage at 24 hours we calculated: [avg. density in sea star treatments at each distance - avg. density in paired control treatments at each distance]. We tested the main and interactive effects of continuous distance from the cage and species (red or green urchins) on the net sea star effect using a linear model (lm()) in base R. We originally performed a model that included site, block and array as random factors matching the model construction as above, but were forced to simplify the model to avoid singularity errors. We then ran follow-up, individual linear fits (lm()) for each species separately to obtain the equation for each line (i.e., net effect = intercept + (slope * distance from cage)). Finally, we calculated the radius of the ‘halo of influence’ of the sea star by setting the net effect to zero and solving for the distance to the cage (i.e., the distance at which the sea star effect was no longer detectable). We calculated the area of influence using pie*(radius2), with the radius being defined as the distance from the cage statistic solved from the previous equation. This gives us an approximate measure of how far from an inactive sea star we can expect sea urchin density to be suppressed for at least 24 hours, even when kelp is present in an urchin barren.