Dataset: Sea fans - nearest neighbor

Methodology from DOI: 10.1016/j.jembe.2017.12.015

Study sites and microhabitat characterization:

The distribution of octocorals was evaluated in July and August 2015, and July 2016, using surveys completed on shallow fringing reefs (8–10-m depth) at White Point, Europa Bay, Yawzi Point, Tektite reef, and East Cabritte (Fig. 1). Hard substrata composed of igneous rock, dead scleractinian colonies, or crustose coralline algae were codified in a categorical scheme identifying three types of microhabitats based on the predominant orientation of their surface: vertical (~ 80–110°, relative to horizontal), inclined (~ 35–60°), and horizontal (~ 0–20°). Octocoral distribution was evaluated among these microhabitats, with microhabitat location of the colonies determined by the position of their holdfast (Fig. 2). Surfaces of live scleractinians were excluded from these analyses, as octocorals were not found attached to live coral tissue, and therefore, we inferred that scleractinian tissue was not a suitable substratum for settlement by octocorals.

Octocoral distribution:

The spatial distribution of octocorals on benthic surfaces can be affected by multiple mechanisms, including microhabitat availability, substratum selection by larvae (Lasker and Coffroth, 1983), and post-settlement success (Lasker and Kim, 1996). To evaluate the role of microhabitat availability in determining the distribution of octocorals, the abundances of vertical, inclined, and horizontal microhabitats were quantified using 20-m transects placed at 8–10-m depth at sites where octocorals are common (Rogers and Teytaud, 1988, Lenz et al., 2015). Quadrats (0.25 m2) subdivided into 25 equal sub-squares (each 0.01 m2) were placed at random positions along each transect, and each sub-square was evaluated from multiple perspectives for the dominant type of non-living substrata composing vertical, inclined, or horizontal surfaces.

The analyses for microhabitat availability were constrained to hard, non-living substrata because octocorals do not settle on sediments. Shallow habitats along this portion of the coast of St. John typically have scleractinian cover of ~ 4% (as in 2012 [Edmunds, 2013]), which is equivalent to one sub-square of the subdivided quadrats used for the present analysis. When such low coral cover was divided among all 25 sub-squares in the 0.5 × 0.5 m quadrat, it was rare for a sub-square to be scored as dominated by live scleractinians. The dominant orientation of non-living surfaces was determined based on the qualitative abundance of hard, non-living substrata in each microhabitat category, but the results were scored by planar area with a resolution of 4% (i.e., one sub-square). With this scheme, a sub-square could be scored as dominated by vertical surfaces if the estimated three-dimensional area of the vertical surface was larger than the estimated three-dimensional area of horizontal and inclined surfaces, even though vertical surfaces were not always fully visible in planar view. It was not possible to quantify the limitations of this estimation procedure, as actual three-dimensional area could not be measured, and the relative bias in the technique is unique to each quadrat and dependent on the absolute area of vertical, horizontal, and inclined surfaces. Emerging technologies like structure-from-motion (Leon et al., 2015) offer potential to refine this approach.

The distribution of arborescent octocorals among microhabitats was evaluated using band transects (2 × 20 m) placed haphazardly along a constant isobath (8–10 m) at each site, within which octocorals were scored for microhabitat based on the surface to which their holdfasts were attached. The focus was placed on arborescent species because these are the most common octocorals in the region (Williams et al., 2015, Lenz et al., 2015). Octocorals were identified to species (Gorgonia ventalina), or genus (Eunicea spp., Pseudoplexaura spp., Muricea spp., Plexaurella spp., Antillogorgia spp., Plexaura spp., and Pterogorgia spp.), and colonies were summed by taxon within each microhabitat. To test the hypothesis of non-random octocoral distribution among microhabitats, the observed distribution among microhabitats was compared with the distribution expected based on the proportional abundance of each microhabitat. These analyses were completed first, for all octocorals (i.e., pooled among taxa), and second, for the most common taxa. Observed and expected abundances were compared with χ2 tests, and where minimum expected frequency for this analysis was not achieved (i.e., 5 [Yates et al., 1999]), taxa were pooled into a single category.

Nearest neighbor distances:

The distances between nearest arborescent octocoral colonies were measured in each microhabitat to evaluate crowding among octocorals. Nearest neighbor distances were measured by haphazardly selecting one octocoral colony (the focal colony) in each microhabitat, and then recording the shortest straight-line distance to the nearest neighboring octocoral. The size class for this analysis was defined as ≤ 6-cm height based on three criteria. First, we selected the smallest colonies at which genus level identification was possible. Second, small colonies allowed the identification of nearest neighbors by proximity of any portion of their holdfasts. The detection of nearest neighbors became ambiguous as colonies grew larger (i.e., > 6-cm height), which caused their holdfasts to meet and coalesce. Third, the focus on small colonies provided a balance between a tractable sample size and a sampling of an early life stage that is more likely to reflect the effects of substratum choice rather than post-settlement mortality. Limiting our analysis in this portion of the study to small colonies also had the advantage of reducing the likelihood of colony size affecting nearest neighbor distance, as it appears to do when nearest neighbors are evaluated by proximity of their branches (Gambrel and Lasker, 2016). Distances between nearest neighboring octocorals were compared among microhabitats using one-way ANOVA for each taxon.

Colony size:

To test for fitness-related consequences of octocoral distribution, the sizes of colonies of Gorgonia ventalina were compared among microhabitats. This species was selected because it is common on shallow reefs around St. John (Lenz et al., 2015), and its two-dimensional morphology is tractable to quantification through planar area (cm2). Colonies of G. ventalina with holdfasts in one of the three microhabitats were haphazardly selected (n = 15 colonies microhabitat− 1) at 10-m depth, and their planar area measured using photographs (Sony Cyber-shot Exmor R camera with 18.2-megapixel resolution). Each photograph included a scale bar, and the colonies were gently held against a flat plastic mesh to obtain a complete planar image. Photographs were analyzed using ImageJ 1.48v software (Schneider et al., 2012) and colony sizes were compared among microhabitats using one-way ANOVA.

All statistical analyses were conducted in RStudio version 3.2.4 (R Core Team, 2016), and the statistical assumptions of ANOVA (for normality and homoscedasticity) were tested through graphical analyses of the residuals.