Fish after dark: monitoring the hidden half

A husbandry round during opening hours captures only part of aquarium life. After lights-out, some species rest while others explore, forage, reproduce or move to different shelters. Subtle aggression and technical disturbance may also emerge. Night monitoring does not turn every inactive fish into a clinical case. It provides a time-based reference that helps teams separate expected behaviour from a new change worth investigating.

Why night observations strengthen welfare assessment

A study at two US public aquariums applied a modified Five Domains assessment to exhibit tanks on 16 occasions. Indicators covered nutrition, physical environment, health and behavioural interactions. Total welfare scores were high, but the authors emphasised continuous monitoring because public aquarium exhibits are structurally and socially complex and may include predator and prey species in the same restricted environment.

The work shows why resource measures alone are insufficient. Temperature, water quality, volume and shelter matter, but teams must also observe how animals use those resources. Space use, inter-individual distance, feeding and social interactions may change markedly with lighting and visitor activity.

Night observations can answer practical questions. Does an animal use a shelter only after visitors leave? Does a supposedly quiet species begin foraging after dusk? Is one zone avoided when pumps change regime? Does an isolated fish follow a stable resting pattern, or has that pattern recently shifted?

Describe behavioural states without overcalling sleep

For fish, “sleep-like behaviour” is often the appropriate term when brain activity is not measured. Inactivity, a characteristic posture, reduced responsiveness and rebound after deprivation can support interpretation, but video alone cannot establish a physiological sleep state.

The experiment by Bauhus and colleagues illustrates this distinction. Three-spined sticklebacks were recorded automatically under infrared illumination and their activity states were classified with a statistical model. Fish spent about 19% of their time in a low-activity, sleep-like state, mostly between 21:30 and 06:30. At 29–32 days after experimental tapeworm exposure, infected fish showed more of this state than controls, especially at night.

Those findings concern one species, one parasite and a controlled experimental system. They do not justify diagnosing infection in a display fish that remains still. They do demonstrate that health changes can alter a temporal pattern, making comparison with the animal’s or group’s own baseline more useful than a universal activity threshold.

Monitor without creating a second daytime

Cameras should see the animals without illuminating their night. Infrared lighting is suitable only after checking the species’ spectral sensitivity and confirming there is no visible light leakage. Reflections, bubbles, suspended particles and high-contrast décor can create false detections. Before automation, manually annotated sequences should be compared with software outputs.

Framing should cover functional areas rather than create an attractive portrait: shelters, territorial interfaces, feeding locations, outlets and narrow passages. Large exhibits may need several cameras. Video, lighting, feeding and building-management clocks should be synchronised.

Short repeated samples may be more useful than continuous footage nobody can review. Fixed windows at the start, middle and end of the night allow comparisons across days. Maintenance, décor changes, unusual noise, system faults and animal introductions must be logged because they may explain apparent behavioural shifts.

Choose species-relevant measures

A useful ethogram separates observable facts from interpretation. It may record tank position, speed, orientation, distance to the group, shelter use, interactions, visible ventilation and response to an event. Categories should be defined before analysis and illustrated with shared examples so different keepers apply them consistently.

Expected patterns depend on ecology. A diurnal species may reduce activity without becoming motionless; a nocturnal species may start exploring before lights-out. Individuals within a group may also differ. Repeated change under comparable conditions provides stronger evidence than an isolated dramatic clip.

Night measures should be linked with daytime information: individual consumption, body condition, water quality, lesions, veterinary interventions and social changes. More chasing after dark may prompt a review of refuge distribution, but it does not identify the cause. Lower activity may reflect rest, temperature, stress, pain or disease and requires proportionate investigation.

Protect the biological night

Monitoring is not a reason to maintain permanent illumination. Photoperiod, transitions and darkness are environmental resources. Indicator lights, screens, safety lamps and after-hours maintenance can create residual illumination that staff cease to notice. Measuring at tank level and viewing the exhibit from the animals’ perspective can reveal these sources.

When a change is proposed, retain a baseline period and modify one factor at a time. The goal is not maximum activity; it is a rhythm and behavioural repertoire compatible with the species’ biology. Enrichment that appears useful by day may become a monopolised resource or visual trap at night, so its actual use needs observation.

Turn recordings into decisions

Define escalation rules before watching the footage. A persistent departure from baseline, repeated exclusion from a resource, injury risk or a change paired with appetite or ventilation should prompt direct assessment. A one-off change with a clear maintenance explanation may simply need documentation and follow-up.

Data retention should match the question. Keep representative baseline and event sequences, annotations, time stamps and relevant system data rather than unlimited raw video. Access controls are needed when cameras may capture visitors or staff, and local privacy requirements should be reviewed before installation.

Automated tracking can support screening, but it must be validated for each view and species. Occlusion, schooling, reflections and similar body patterns can break individual identification. Human review remains essential for welfare interpretation.

Conclusion

The night should not remain a blind spot in public aquarium welfare assessment. Planned observations reveal space use, interactions and low-activity states hidden from daytime rounds. Their value depends on non-disturbing equipment, a species-specific ethogram and longitudinal comparison. No isolated sequence replaces clinical examination or husbandry context.

Vetofish can help aquariums define indicators, validate a video protocol, interpret changes alongside health data and turn observations into measurable adjustments to exhibits and care routines.