2. Materials and methods
Lionfish were hand-collected off Little Cayman Island in January–August 2013. On the day of deployment, each fish was anesthetized, measured (mm total length; TL), and fitted with 20 cm of monofilament line secured to its lower jaw. Fish were held for ≥ 2 h to ensure tethers were secure, transported to field sites, and attached to lead weights (13:00–15:00 h). The following morning (07:30–09:00 h), fish missing from cleanly broken tethers were recorded as predation events, and the remaining lionfish were euthanized. Controls for tethering effects comprised three fish tethered in tanks for 24 h and video surveillance of fish in the field.
From 9 to 15 lionfish were tethered at intensely culled fore reef sites where spearfishers had removed lionfish approximately monthly for 3 years (Frazer et al., 2012; n = 3), rarely culled fore reef sites (n = 3), and back reef seagrass sites with no record of culling (n = 4). At sites with Thalassia testudinum, lionfish were tethered at 3-m intervals along transects at ~ 1 m depth. At reef sites, lionfish were tethered 3–5 m apart on sand or hard bottom at depths of 4.5–9.0 m. Tethered fish could not access a refuge, and fish were not deployed when large piscivores were visible.
Data were analyzed with a one-way analysis of variance and three binary logistic regressions. The analysis of variance assessed differences in fish size among habitats. An initial binary logistic regression assessed total length as a covariate and temporal differences in potential for predation between trials at rarely culled reefs. Additional regressions tested for differences with total lengths and among habitats using either data from seagrass or rarely culled reefs as reference values. Pearson's, deviance and Hosmer–Lemeshow tests assessed goodness of fit for these regressions.