Both active and passive human interactions with reef fish communities are increasingly recognized to cause fish behavioural changes. However, few studies have considered how these behavioural adaptations impact standard reef survey techniques, particularly across natural gradients of interest to ecologists and reef managers. Here we measure fish abundance, biomass and minimum approach distance using stereo-video surveys to compare the effects of bubble-producing open-circuit scuba vs near-silent closed-circuit rebreathers. Surveys extended across a shallow to upper-mesophotic gradient on the fringing reefs of Utila, Honduras, to explore how the effects of diver gear choice vary with depth. For most fish families we recorded similar abundances and biomass with the two diving techniques, suggesting that open-circuit transects are generally appropriate for surveying western Atlantic reefs similar to Utila with regular tourist diving but no spearfishing. Despite no overall significant difference in fish abundance or biomass, we identified several fish families (Labridae, Pomacentridae, Scaridae) that allowed closed-circuit rebreather divers to approach more closely than open-circuit divers. In addition, smaller fish generally allowed divers to approach more closely than larger fish, and in most cases divers could approach fish more closely on mesophotic than shallow reefs. Despite these significant differences in approach distances, their magnitude suggest they are unlikely to affect reef fish detectability during normal fish surveys for most families. Our findings highlight the importance of considering variation in fish behavioural adaptations along natural gradients such as depth, which otherwise has the potential to cause biases when surveying by traditional monitoring programmes.