What level of mortality should trigger a health investigation on a fish farm?

There is no mortality percentage that is valid for every species, life stage and production system. A health investigation should begin when mortality departs from the expected level for that stock, rises rapidly or appears alongside other abnormal signals. The useful trigger is therefore a farm-specific threshold built from the farm’s own data, not a number borrowed from another industry.

Why would a universal threshold be misleading?

Background mortality varies with species, age, season, stocking density, production unit, water quality, recent handling and production phase. Ten deaths do not carry the same meaning in a stock of 500 fish and one of 100,000. The same rate may also represent an acute event over a few hours or a slow accumulation over several weeks.

Studies in Atlantic salmon show that mortality patterns vary between cohorts, periods and production areas. They demonstrate the value of analysing mortality time series, but their thresholds cannot simply be transferred to another species or farm. Chapter 1.4 of the World Organisation for Animal Health’s Aquatic Code likewise includes mortality and production records among passive-surveillance information and calls for investigation of suspicious health events.

To interpret a signal, always define:

• the stock and epidemiological unit concerned: tank, cage, pond or recirculating system;
• the population genuinely at risk, using the most reliable count or biomass available;
• the calculation window: daily, weekly and rolling cumulative figures;
• which fish are counted, including deaths, removed moribund fish and unexplained losses where relevant;
• events that may temporarily change the expected level, such as transfer, grading, vaccination or a thermal episode.

The mortality rate and mortality curve are more informative than a raw count, provided that the numerator, denominator and time window remain consistent.

Which signals warrant immediate investigation?

A percentage may still be low while representing an important change. At minimum, start an immediate check when any of the following appears:

• a sudden increase compared with the previous days or weeks;
• deaths clustered in one tank, cage, area or size class;
• several production units affected at nearly the same time;
• lethargy, reduced feeding, loss of balance, rapid ventilation or visible lesions;
• a recent change in oxygen, temperature, flow, feeding or behaviour;
• persistent, low-level mortality that exceeds the stock’s usual variation;
• suspicion of a regulated, emerging or rapidly spreading disease.

A succession of small losses can reveal chronic disease or a lasting environmental problem. Conversely, an acute technical failure may kill fish before distinctive clinical signs develop. The absence of a high mortality rate therefore excludes neither disease nor an urgent husbandry problem.

How can a farm establish internal alert levels?

Use historical data for each production type to establish a baseline by species, life stage, season and unit. Medians or expected curves can help, but they should be reviewed whenever the system changes. Link each alert level to a written action:

1. Watch: confirm the count and stock number, check water parameters and recent events, and increase observation.
2. Investigate: notify the health manager or veterinarian, compare affected and unaffected units, examine fish and plan diagnostic sampling.
3. Emergency: immediately secure oxygenation and critical system functions, restrict movements and activate the appropriate biosecurity and notification plan.

Statistical thresholds described in research, including receiver operating characteristic approaches, can help balance early detection against false alarms. They still need validation against the farm’s own data. A threshold set too high delays diagnosis; one set too low, without a confirmation rule, makes repeated alerts easy to ignore.

What should the first investigation document?

Preserve a precise timeline: stock size, daily and cumulative deaths, affected units, observed signs, feed intake, treatments, transfers and handling. Record temperature, dissolved oxygen, pH, total ammonia nitrogen and nitrite, followed by parameters specific to the system. The FAQ on water parameters to measure when fish are sick explains this initial check in more detail.

Remove and record dead fish, but do not discard all potentially useful evidence before seeking advice. Fresh, untreated moribund fish often provide more diagnostic information than decomposed carcasses. Agree sample selection, packaging and transport with the veterinarian or laboratory. Avoid blind treatments that may obscure lesions, interfere with testing and delay identification of the cause.

Systematically compare an affected unit with an apparently healthy one. This comparison helps distinguish a shared factor — water, feed, equipment or an intervention — from a more localised focus. Record corrective actions and their effects as well: improvement after oxygen is restored does not remove the need to establish why the incident occurred.

Finally, an internal investigation threshold does not replace statutory reporting duties. If a regulated or unusual disease is suspected, contact the responsible aquatic-animal veterinarian and competent authority promptly under the applicable rules. Vetofish can support the farm in characterising the signal, examining fish and arranging diagnostic testing. The aim is not to wait for a spectacular number, but to identify a credible departure from normal stock performance early enough to act.