Camera field-of-view and fish abundance estimation: A comparison of individual-based model output and empirical data

Camera technology is evolving rapidly as computing and sensor technology reduces in size and increases in processing speed. This convergence of high speed computing and optics has resulted in the creation of full-spherical cameras and their application in marine environments has begun. Thus, there is a need to understand how large increases in camera field-of-view (FOV) impacts data collection and statistical inference, particularly when dealing with long-term time-series data. Data collected in this survey consisted of a single year of sampling and were insufficient to describe relationships for single species, therefore we species were grouped based on lineage and behavior into snappers, jacks, and mixed-reef groups. The empirical and theoretical models presented here demonstrate an asymptotic relationship between single-camera MaxN and four-camera MaxN counts for snappers and to a lesser extent the mixed reef groups composed primarily of groupers. In contrast the jack group showed a linear relationship between counts. This result suggests that accuracy of the counts can be improved by increasing camera field-of-view to 360° or full-spherical. Habitat complexity was demonstrated to be positively related to fish counts but had no effect on the relationship between camera FOV and fish counts. This could be the result of several things, 1) fish behavior is more critical in driving aggregative behaviors that modify the relationship between FOV and fish counts, 2) abundances observed in this region are not high enough for habitat effects to have impacted the relationship, or 3) a mixture of these effects. Further exploration of these relationships is needed that would improve sample sizes, more observations at high counts, and that establish species specific relationships. This work provides empirical footing to further develop the relationship between MaxN-type estimators relative to camera FOV and additionally includes insight into the effects of habitat. Critically, this work provides a method to relate counts produced from 360° systems to those produced with restricted FOV.