Automated flow control of a multi-lane swimming chamber for small fishes indicates species-specific sensitivity to experimental protocols

In fishes, swimming performance is considered an important metric to measure fitness, dispersal, and migratory abilities. Swimming performance of individual larval fishes is often integrated into models to make inferences on how environmental parameters affect population-level dynamics (e.g., connectivity). However, little information exists regarding how experimental protocols affect the swimming performance of marine fish larvae. In addition, the technical setups used to measure larval fish swimming performance often lack automation and accurate control of water quality parameters and flow velocity. In this study, we automated the control of multi-lane swimming chambers for small fishes by developing an open-source algorithm. This automation allowed us to execute repeatable flow scenarios and reduce operator interference and inaccuracies in flow velocity typically associated with manual control. Furthermore, we made structural modifications to a prior design to reduce areas of lower flow velocity. We then validated the flow dynamics of the new chambers using computational fluid dynamics and particle tracking software. The algorithm provided accurate alignment between set and measured flow velocities and we used it to test whether faster critical swimming speed (Ucrit) protocols (i.e., shorter time intervals and higher velocity increments) would increase Ucrit of early life stages of two tropical fish species (4-10 mm standard length). The Ucrit of barramundi (Lates calcarifer) and cinnamon anemonefish (Amphiprion melanopus) increased linearly with fish length, but in cinnamon anemonefish, Ucrit started to decrease upon metamorphosis. Swimming protocols using longer time intervals (>2.5 times increase) negatively affected Ucrit in cinnamon anemonefish but not in barramundi. These species-specific differences in swimming performance highlight the importance of testing suitable Ucrit protocols prior to experimentation. Automated control of flow velocity will create more accurate and repeatable data on swimming performance of larval fishes. Integrating refined measurements into individual-based models will support future research on the effects of environmental change.

Methods
Please refer to the original manuscript for details on how the data were collected.

Regarding the swimming protocol data, the ISI Web of Knowledge was searched (Clarivate Analytics, Core collection search on 19.02.2019) using the term: ((swim* OR sust* OR prolong* OR burst*OR cruis* OR routin* OR Ucrit OR endur*) AND (early life stage* OR larv*) AND (marin* OR sea* OR brack*) AND (fish* OR teleost*))). The results were manually checked and 36 papers were identified were fish were swum in multi-lane swimming chambers.

The video files show recordings of fluorescent green polyethylene buoyant particles (850-1000 µm) inside the multi-lane swimming chambers at 10 and 40 cm s-1, respectively. The original recordings were processed using a Flowtrace algorithm (see original manuscript for further details) and inverted using Sony Vegas Pro 13.0 (VEGAS Creative Software). A still of the multi-lane swimming chamber, added to the background of the videos, allowed the evaluation and tracking of individual particles in each raceway. The video of the lower velocity (10 cm s-1) is provided at the original speed, whereas the video of the higher velocity (40 cm s-1) is slowed down to 0.25x speed.