The Hunger Games – A glimpse into larval fish prey capture behavior in-situ in aquaculture

Date:

abstract

Abstract The notoriously inefficient feeding of early-stage larval fish has been speculated to drive their “critical period” of extreme mortality. Larval feeding success is constrained by the viscous hydrodynamic regime they experience at small size, which is ameliorated as larvae grow. However, how larval feeding success is affected by variation in environmental conditions is poorly understood.

Here, we investigated age-dependent effects of temperature, pH, O2 and turbulence on larval fish feeding in large mesocosms. We tracked these parameters and the strike rate of undisturbed larvae for 17 cohorts of Sparus aurata larvae throughout their ontogeny; collecting 268 high-speed videos, each ~ 5 min in length. We then constructed a pipeline of machine learning models to detect feeding events within this massive dataset.

In the mesocosms, strike rates were ~0.01 times lower than in the laboratory. These low rates led to high occurrences of false detections by our pipeline, requiring a manual review of its outputs; though still improving analysis times considerably. Additionally, we observed a repertoire of frequent abrupt behaviors, including c-start, and spitting of prey. Although more common than feeding behaviors, their relationship with larval growth and well-being is unclear. Statistical analysis of over 200 videos indicated that larval fish maintain an extremely low feeding rate. This happens regardless of environmental conditions and despite of high encounter rates with prey in our mesocosms.