**Methodology from Gambrel, B. and Lasker, H.R. 2016: Interactions in the canopy among Caribbean reef octocorals. Marine Ecology Progress Series. 546: 85–95, doi: 10.3354/meps11670**

To further examine the spatial distribution of colonies and potential competition among them, nearest neighbor data were collected along the belt transects from each site. Each transect had 20 sampling points arranged at the corners of every 1 m2 quadrat along the first 9 m of each transect. The octocoral colony closest to each sampling point was selected, identified,and the distance between its base and the base of its nearest branching octocoral neighbor was measured to the nearest centimeter (Fig. S1 in the Supplement). The height, width, and length of each colony was measured to the nearest centimeter to calculate cross-sectional area (height × length) and volume (height × width × length), and the proximity of the colony’s branches to nearby octocorals was also noted. Due to the water flow and the resulting oscillation of colony branches, measurements were made when the branches were vertical in the water column to optimize the precision of our measurements.To increase sample sizes, an additional 9 m transect parallel to the other 5 was sampled at each site.

The nature of the spatial distribution of octocorals at each site was determined from the nearest-neighbor data following Clark & Evans (1954). Observed and expected mean distances between the octocoral neighbors were calculated using the total distance between neighbors, sample size (120 pairs of octocorals per site) and the density of octocorals at each site(calculated from the belt transect data). The ratio (R)of the observed and expected (given a randomly distributed octocoral community) mean distances between octocoral neighbors describes the octocoral distribution at each site, where R = 1 denotes a random distribution, R < 1, an aggregated distribution,and R > 1, a uniform distribution (Clark & Evans1954). The significance of R was determined by analyzing the standard variate of the normal curve (c), since the measured distances between neighbors in a randomly dispersed community are expected to follow a normal distribution.

The effects of colony−colony proximity on colony size were assessed by correlating the distance between neighbors at the base with the sum of their sizes (Pielou 1962). If competition affects growth, then the closer the organisms are, the smaller their expected sizes will be (Pielou 1962). Implicit in these analyses is the notion that size is both an indicator of resource use and of success in acquiring resources.

The relationships among the distance between octocoral neighbors at the base (divided into 3 distance groups to make the data categorical: 5−14 cm, 15−24 cm and 25−34 cm), branch proximity and site were analyzed using a hierarchical log-linear test in SPSS. The relationship between the distance between neighbors and branch proximity was further analyzed in a separate log-linear test in SPSS.