Survey methods: Chain transect; Large quadrat (20 x 40 m); Small quadrat (4 x 5 m)

Assessment of coral cover on the study reef began in 1974 (Glynn 1976) and continued to 2010. This was accomplished by fixed chain transect sites (n = 10 transects per survey period, 1974 to 2010) and by sampling fixed 1 m2 plots (n = 11 plots per survey period, 1994 to 2010) as well as a single 4 × 5 m2 plot (n = 1, with the 20 m2 plot subdivided into 20 1 m2 quadrats, 1984 to 2010). Chain transects were established along the ~350 m long forereef base zone between 3 and 5 m depth, relative to the mean lower low water (MLLW) tidal datum, in the mid-1970s from strictly random site locations determined from an Uva Island reef grid map with numbered axes (Glynn 1976). A second set of 10 fixed transects was located upslope from the reef base in 1985 to monitor denser coral cover at slightly shallower depths following the 1982-1983 mortality event (Glynn et al, 2014, Fig. 1C). Most benthic sampling was conducted near, but not within the fish transects (see ‘fish species counts’) to minimize diver presence. During each of 20 survey periods, 10 to 20 chain transects were sampled. Coral species live cover and substrate type (e.g. rubble, turf algae, sponges) were enumerated according to the chain links (73 links m-1) resting on these categories. The chain transect sampling was designed to determine the advance or retreat of live coral cover along the deep reef base and slope zones. Coral monitoring was carried out at different times during the 5 to 7 d sampling periods, usually 2 to 3 d before the fish monitoring.

The 1 m2 plots were located at the northern end of the reef in the reef slope and base zones representative of the fish transect substrate cover. The benthic composition and cover of the 1 m2 plots were drawn by divers underwater and then digitized using a flat - bed scanner, Adobe Photoshop and ImageJ software. Percent coral cover was determined by dividing the number of pixels representing coral within a quadrat by the total number of pixels in that quadrat. For each year of sampling of the 4 × 5 m2 plot, all 20 1 m2 quadrats were mapped and their composition and cover determined as in the 1 m2 plots above; these values were averaged to determine the total mean percent coral cover for each survey period. The 4 × 5 m2 plot was established by R. H. Richmond in the reef slope zone immediately following the 1982- 1983 El Niño event when coral cover on the forereef was reduced to near-zero values. This location was selected to monitor live coral recovery; it is situated at 1 to 2 m depth, between and slightly overlapping fish transects, Transects 1 and 2 (see Glynn et al, 2014, ‘Fish monitoring’). Table S1 in the Supplement at www.int-res.com/articles/suppl/m495p233_supp.pdf summarizes all coral sampling records with information on time of sampling, season and number of samples completed.