This was a manipulative field experiment conducted during June – August 2013 to determine whether lionfish affect native fishes in the areas surrounding coral patch reefs.  Sixteen patch reefs were paired based on similarity in location, size, benthic composition, initial fish community, and surrounding habitat, then one reef in each pair was randomly assigned to either low-lionfish or high-lionfish treatments.  A pair of divers removed all lionfish from the eight low-lionfish reefs during weekly visits (resulting mean lionfish density on low-lionfish reefs was 0.01 to 0.03 lionfish/square meter).  The eight remaining reefs were maintained as high-lionfish treatment reefs, with divers adding lionfish to the reefs to keep densities within the naturally occurring range observed in Rock Sound (resulting mean lionfish density on high-lionfish reefs was 0.12 to 0.29 lionfish/square meter.  

A pair of divers returned to each reef weekly for 7 weeks to census the native fish communities on the main reefs and in three different habitat types in the surrounding areas.  During all surveys, divers counted all fish less than 15 cm TL and all piscivores of any size.  Around each main reef, selected between one and four satellite coral heads were haphazardly selected to survey.  Satellite coral heads were small structures comprised of one to several coral colonies, ranging in size from 0.31 to 4.7 meters squared and located between 3.8 and 25.1 m from the main reefs.  In addition to being smaller than the main reefs, satellite heads also supported neither resident lionfish nor native piscivores.  To document native fish communities in open sand and seagrass areas without any coral heads or other hard structures, two 35 m transects were run outward from each main reef in randomly selected directions.  Divers surveyed fish at 5-m intervals within 2 square meter plots around the transect tapes.  To obtain a standardized measure of fish communities at set distances from each reef that accounted for habitat type, patch size, and initial fish community, I set-up small (0.5 meters squared) identical standardized habitat units (SHUs) consisting of 9 empty queen conch (Strombus gigas) shells with their apertures facing upward.  All reefs had one row of 4 SHUs located at 7-m intervals extending away from the reef.  Half of the reefs (4 high-lionfish and 4 low-lionfish reefs) also had a second row of 2 SHUs located at 15 m and 31 m from the reef and extending outward in the opposite direction from the first row.  During the final week of the experiment, in addition to visually surveying the fish communities on the SHUs, a pair of divers placed all conch shells in dry bags and brought them up to a boat. Each shell was shaken in a bucket of seawater to remove and count any individuals hiding in the shells.

Note that five of the 16 main coral patch reefs had high abundances of small, juvenile grunts (Haemulon spp.), which formed large schools of approximately 100 - 300 individuals above the reefs.  Counts of these juvenile grunts are not considered to be precise or accurate.