From Lenz et al (2015) Coral Reefs:

For the local-scale (~5 km) analysis, octocoral densities were measured along the south shore of St. John, in the Virgin Islands National Park (VINP) and Biosphere Reserve (Rogers and Teytaud 1988; Rogers et al. 2008).Within this area, coral reef community structure has been studied since the 1950s (Randall 1961; Collette and Earle 1972) and in a systematic manner since 1987 (Rogers et al.1991; Rogers and Beets 2001; Miller et al. 2006; Edmunds 2013). Six sites on shallow reefs (7-9 m depth) were selected between Cabritte Horn and White Point in 1992 (Fig. 2a, with sites restricted to hard substrata but otherwise identified based on randomly selected coordinates. These sites have been censused annually to measure the percent cover of macroalgae, scleractinians, and a combined category of crustose coralline algae, algal turf,and bare space (CTB; Edmunds 2013).

At each site, photoquadrats (0.25 m2) were recorded at random positions along a fixed transect each year, with 20-m transects and 17-20 photoquadrats site-1 between 1992 and 1999, and 40-m transects and 40 photoquadrats site-1 from 2000 to present (Edmunds 2002, 2013). Before 2000, photoquadrats were recorded with a NikonosTM V camera (fitted with a 28-mm lens, two Nikonos SB 105 strobes, and Kodachrome 64 film) mounted on a quadrapod that held the camera perpendicular to the seafloor (Edmunds 2002, 2013). In 2000, the method was upgraded to digital images using first a 3.3 megapixel camera (2000-2006, Nikon Coolpix 990) and then a 6.1 megapixel camera (2007-2012, Nikon D70). The camera framer remained unchanged throughout the sampling, and the images allowed objects C10 mm in diameter to be resolved. While photoquadrats from the six sites (archived at http://mcr.lternet.edu/vinp) have been recorded annually,photoquadrats in the present analysis were sampled at 2- to3-yr intervals: 1992, 1994, 1997, 1999, 2002, 2004, 2007,2009, and 2012 (n = 1630 images). Logistical constraints precluded analyzing the photoquadrats every year for octocoral abundance, and by subsampling every 2-3 yr, it was possible to effectively capture temporal trends in octocoral abundance. With this temporal resolution, we completed a coarse-grained analysis of changes over time in octocoral abundance at each site.

For each photoquadrat, octocoral colonies were counted based on the presence of their holdfasts within the framer. In some cases, other organisms, including large octocorals,obscured the holdfasts, or holdfasts were hidden in crevices. Most octocorals could not be distinguished to species in the photographs, in part because taxonomic identification requires inspection of sclerites, and therefore, analyses were constrained to genera [Antillogorgia spp. (formerly Pseudopterogorgia; Williams and Chen 2012), Briareum spp., Eunicea spp., Erythropodium spp., Gorgonia spp., Muricea spp., Muriceopsis spp., Plexaura spp., Plexaurella spp., Pseudoplexaura spp., and Pterogorgia spp.].Encrusting Erythropodium caribaeorum and the encrusting form of Briareum asbestinum were excluded from the analysis, which instead focused on arborescent octocoral(hereafter referenced as octocorals) that dominate octocoral communities in St. John. Of the eleven octocoral genera found on these shallow reefs, Eunicea spp., Plexaurella spp., Pseudoplexaura spp., and Plexaura spp. could not be distinguished from each other in the images when colonies were small (\12 cm tall), and therefore, members of these genera were categorized by their family (i.e., Plexauridae).

This retrospective analysis was augmented with in situ surveys in July and August 2013 that were used to quantify the accuracy and precision of the octocoral population census conducted using photoquadrats. Given the challenges of quantifying arborescent colonies in planar images, we did not expect perfect concordance between methods, but expected to detect a strong correlation between the approaches and quantify the underestimation associated with the photographic technique. To compare these census methods, octocorals at the six sites were censused with 0.25 m 217 -2 quadrats placed along the same40-m transect as was used for the photoquadrats (n = 40 site 219 -1). Photoquadrats were recorded, and in situ counts completed at each site within 5 d of each other. Quadrats for counts were placed at the same location where photoquadrats were recorded, but underwater logistics prevented perfect concordance between sampling areas. Densities of octocorals from the in situ counts and photoquadrats were tested for association and concordance (see Lenz et al, 2015) using site-specific means (n = 6).

Octocorals were also surveyed in 2012 and 2013 at Booby Rock, 1.3 km east of Cabritte Horn (Fig. 2a) where octocorals were very abundant (Fig. 1). Data from this site provided insight into the upper range of octocoral densities that occur along the south shore of St. John. At this location, octocorals were censused in situ using 20 quadrats (1.0 9 1.0 m) placed at random points along a 40-m transect running along the 7-9 m depth contour. As there were no historic data for Booby Rock, octocoral densities were not used in the contrast of octocoral abundances over time in St. John, although they were included in the regional-scale assessment.