Census design
We used a hierarchical sampling design (Noda 2004) to examine the context dependence of population dynamics of C. dalli andG. furcata . We performed an 18-year intertidal sessile assemblage census in the rocky intertidal zone at five sites (Sites 1–5) along the Pacific coast of eastern Hokkaido, Japan (Fig. 1). Within each site, five 4000-cm2 sampling plots (n = 25) were established on near-vertical rock walls at semi-exposed locations. Each plot was 50 cm wide by 80 cm high, and the vertical midpoint corresponded to the mean tidal level. The vertical range of each plot was divided into four equal quadrats (upper, upper middle, lower middle, and lower). In each quadrat, 40 grid points were placed on the rock surface at 5-cm intervals in both the vertical and horizontal directions excluding the edges, and the sessile species occupying each grid point were identified at a fixed observation point and recorded. Coverage data were collected annually in summer (July or August) from 2002 to 2019.
Quadrats in which C. dalli and G. furcata were observed continuously during the census period were used for further analyses. A total of 33 quadrats at four different heights (9 upper, 14 in upper middle, 7 lower middle, and 3 lower) from 15 rocks (4 in Site 1, 4 in Site 2, 2 in Site 3, and 5 in Site 4; Fig. 1) met the above conditions. The median coverage of C. dalli and G. furcata , average coverage of other species, and total coverage of other species for all quadrats in each year during the census period were 32.5%, 15%, 0.1%, and 5% respectively (Fig. 2).