Dataset: Whelk surveys

Study system:  Our study included 4 sites along the Oregon coast: Strawberry Hill (44.250°N-124.115°W) and Yachats Beach (44.319°N-124.109°W), located on Cape Perpetua, and Fogarty Creek (44.837°N-124.0587°W) and Boiler Bay (44.832°N-124.061°W) located on Cape Foulweather (Fig.1). Cape Perpetua is a rocky headland adjacent to a wide continental shelf offshore that promotes retention of propagules such as larvae and phytoplankton (Menge et al. 2015). As a result, Cape Perpetua intertidal sites are characterized by high phytoplankton productivity and high recruitment of invertebrates (Menge et al. 1997, 2004, 2015). Cape Foulweather, in contrast, is characterized by a narrower offshore continental shelf, which leads to reduced retentiveness, lower invertebrate abundance, and high macrophyte abundance. Prior to the onset of SSWD, densities of P. ochraceus could be as high as 8 individuals m-2 at Cape Perpetua and 4 individuals m-2 at Cape Foulweather sites (Menge et al. 2016).

Surveys of subordinate predator distribution 
In the absence of the keystone predator, we predicted that whelk predators would shift toward foraging lower down on the shore in response to reduced consumptive pressure and increased resource density. We tested for changes in the vertical distribution of whelks from April-September of 2014 (before and after peak SSWD incidence in P. ochraceus) and in January-July of 2015 (when incidence was low and adult densities had been fully reduced) using vertical transect surveys. In our surveys, we measured the density of whelks, Leptasterias spp., and any P. ochraceus within five intertidal zones at four sites. These zones were defined by ecology rather than absolute tide height, and included (from low to high on the shore): the upper algal zone dominated by algae and barnacles, the lower edge of the mussel bed, the middle section of mussel bed, the upper edge of the mussel bed, and the high zone dominated by barnacles and fucoid algae. In some cases where M. trossulus was abundant within the upper algal zone, we conducted additional quadrats to target this zone specifically.

At each of four sites (Boiler Bay, Fogarty Creek, Strawberry Hill, and Yachats Beach, see Fig. 1), we conducted  surveys using five replicate transect lines that ran vertically along the shore. These transect lines were marked at both ends with stainless steel lag screw. Within each transect, we measured density in one 0.25 m2 quadrat within each of the five ecological zones, recording the distance of each quadrat along the transect line for consistency over temporally repeated sampling. Because we expected to see the greatest changes due to SSWD in the lower ecological zones where P. ochraceus had previously been abundant, we measured predator density in additional quadrats in the upper algal zone and the lower edge of the mussel bed. This was done using a horizontal transect line that ran perpendicular from the low point of each vertical transect line described above forming an inverted "T" shape. Along each of the five replicate horizontal transect lines, we included four quadrats in the upper algal zone and four in the lower edge of the mussel bed. 
We conducted these surveys to track changes in the vertical distribution of subordinate predators and changes in their overall abundance.  We also took a photograph of every quadrat monitored in order to have a record of prey availability for each set of predators counted. Surveys were mostly conducted during morning low tides and, when possible, we attempted to finish all four sites within a single tide series. We recorded the tidal height of each quadrat in July 2015 relative to Mean Low Low Water (MLLW) using a laser level and stadia rod to measure the height of each plot above reference measurements at the water line in centimeters (Pincebourde et al. 2008).

Subordinate predator removal experiment
To assess the effects of subordinate predators in the absence of the keystone, we conducted a factorial removal experiment at two intertidal sites located on Cape Perpetua, Oregon (Strawberry Hill and Yachats Beach, see Fig. 1). We predicted that whelk predators would affect establishment of the dominant mussel, Mytilus californianus, by consuming the mid-successional prey species that facilitate its recruitment. As such, we chose to follow prey dynamics from a mid-successional stage by placing plots where there was abundant cover of the mussel Mytilus trossulus and several barnacle species. This mid-successional community is where we expected to see the greatest effects of subordinate predators and the largest changes in community structure following SSWD. We originally examined the effects of two groups of subordinate predators, gastropod whelks Nucella canaliculata and N. ostrina (W) and the smaller sea star Leptasterias spp. (L), in a factorial design including four treatments:  +W +L, -W +L, +W -L, and -W -L. However, Leptasterias spp. were rare in our plots, and treatments were combined to include control (+W) and whelk removal (-W) treatments only (see data analysis section below). It is important to note that our experiment tested the effects of subordinate predators at reduced P. ochraceus densities, rather than comparing their effects in the presence or absence of P. ochraceus.

We followed prey community structure over time at 5 replicate plots (10 when treatments were combined) within each site at the upper edge of the low zone. The main treatment plots were 0.25 m2 in size and corners were marked with stainless steel lag screws. Each main plot was surrounded by four additional subplots adjacent to each plot side that were meant to act as a buffer for the main plot. We monitored plots either bi-weekly or monthly as tides permitted from experimental initiation in June 2014 through May 2015. At each monitoring, we counted the total number of each subordinate predator species (N. canaliculata, N. ostrina, and Leptasterias spp.), as well as any less common predators, such as the whelk Nucella lamellosa, in all plots. When possible, we conducted a full monitoring with counts and removals of predators in the main plot, the four adjacent subplots, and in the corners between subplots. On some occasions, particularly during winter months with limited site access, we only monitored the main plots. All removal of whelks and Leptasterias spp. was conducted using forceps, and removed predators were relocated away from the plot area. In plots without removals, we mimicked the use of forceps in the plot while counting subordinate predators to limit the possibility that the physical action of predator removal would influence our results. Although SSWD caused declines in Pisaster ochraceus densities, it did not extirpate the species entirely from our sites. We recorded and relocated any P. ochraceus within each plot and any adult and juvenile P. ochraceus from a 3m radius around each plot.