Water Quality

A culture unit is a small, heavily loaded chemical reactor with animals in it. Oxygen enters from the atmosphere or from an aeration device and leaves through respiration by fish, bacteria and, in ponds, algae. Nitrogen enters as feed protein and leaves as ammonia, is oxidised to nitrite and then nitrate, and departs through water exchange, plant uptake or denitrification. Carbon dioxide accumulates. Alkalinity is consumed. Solids build up unless removed. Each of these processes has a rate, and the farm is stable only while removal rates match production rates.

The parameters that matter are not independent. Temperature sets oxygen solubility and metabolic rate at once. pH determines what fraction of total ammonia is present in its toxic un-ionised form, and pH itself is held by alkalinity, which nitrification consumes. Carbon dioxide suppresses pH and impairs oxygen uptake at the gill even when dissolved oxygen appears adequate. A water quality problem is almost always a chain, and the visible symptom is rarely the first link.

Water quality is the most common proximate cause of loss in fed aquaculture, and the most common ultimate cause of disease. Sub-lethal conditions do not kill animals directly; they suppress appetite, impair growth, damage gill and skin tissue and depress immune competence, after which an opportunistic pathogen that was already present becomes an outbreak. Investigating such an outbreak as a pathogen problem alone reliably misses the cause.

It also determines the economics of the system. The point at which water quality can no longer be held defines carrying capacity, which defines how much biomass the capital investment carries, which defines whether the farm returns anything. Treatment capacity and production capacity are the same quantity described from two directions.

Dissolved oxygen

Oxygen is the parameter with the shortest time constant, which is why it dominates operational attention. Solubility falls as temperature rises and as salinity rises, precisely when metabolic demand is highest. Demand comes from the animals, from bacteria oxidising organic matter and ammonia, and in ponds from algal and sediment respiration at night. Because the dissolved inventory in a heavily stocked unit represents only minutes of consumption, oxygen is managed by continuous supply rather than by storage.

Ammonia and its speciation

Total ammonia nitrogen exists in equilibrium between the ionised ammonium form and the un-ionised ammonia form. Only the un-ionised fraction crosses the gill membrane readily, and it is the toxicologically important species. That fraction rises with pH and with temperature, so an identical total ammonia measurement can be harmless in one water and damaging in another. Reporting total ammonia without the accompanying pH and temperature is reporting an incomplete measurement.

Nitrite and nitrate

Nitrite is the intermediate of nitrification and is toxic through its interference with oxygen transport in the blood. Its effect is strongly modified by chloride concentration, which competes for the same uptake pathway, so nitrite tolerance differs greatly between fresh and saline waters. Nitrate is the end product and is far less toxic, but it accumulates in systems with low water exchange and is the parameter that eventually forces either exchange or denitrification.

pH, alkalinity and carbon dioxide

Alkalinity is the water's capacity to resist pH change, supplied mainly by bicarbonate. Nitrification consumes alkalinity, and respiration adds carbon dioxide which forms carbonic acid, so an intensively loaded system drifts downward in both alkalinity and pH unless buffer is replaced. When alkalinity is exhausted, pH becomes unstable, nitrification slows, and the system can move quickly from apparently fine to failing.

Solids

Suspended and settleable solids matter in their own right, through gill abrasion and irritation, and indirectly because they are a reservoir of oxygen demand and of ammonia that is released as they decompose. Removing solids early, before they break down into fine particles, is one of the highest-value interventions in any intensive system.

Diel variation in ponds

In any water body with a significant algal population, oxygen, carbon dioxide and pH swing on a daily cycle: photosynthesis raises oxygen and pH through the afternoon and stops at dusk, after which respiration alone drives oxygen down and carbon dioxide up until dawn. The minimum oxygen and the maximum un-ionised ammonia therefore occur at different times of day, and a sampling regime that only visits mid-morning sees neither.

Source water

Groundwater is often low in oxygen, supersaturated with nitrogen or carbon dioxide, and may carry iron, manganese or hydrogen sulphide requiring aeration and oxidation before use. Surface water brings variable turbidity, agricultural residues and pathogens. Characterising the source across seasons, rather than from a single sample, is a fundamental site decision rather than a commissioning detail.

A working monitoring regime distinguishes parameters that must be watched continuously from those that need periodic measurement, and it defines in advance what each reading will trigger.

1. Instrument dissolved oxygen and temperature continuously in every unit that cannot survive a short interruption, with an alarm path that reaches a person who can act.
2. Measure total ammonia nitrogen, nitrite and pH on a fixed schedule, and always record pH and temperature alongside ammonia so speciation can be computed.
3. Track alkalinity as a routine parameter rather than as a diagnostic reached for after a problem, because it is the early warning for pH instability.
4. Sample at the time of day when the parameter is worst, not at the time of day that suits the round: before dawn for oxygen, mid-afternoon for pH and un-ionised ammonia.
5. Sample from the outlet as well as the inlet of a culture unit, since the difference between them is the load the animals are actually imposing.
6. Calibrate probes on a schedule and against known standards; an uncalibrated continuous reading is worse than no reading because it is trusted.
7. Plot the record. A parameter walking slowly in one direction across two weeks is the signal, and a table of numbers hides it.

Species differ substantially in their tolerances, and the differences are not merely quantitative. Air-breathing species tolerate oxygen conditions that would be fatal to salmonids. Euryhaline species carry chloride that alters nitrite risk. Larvae and juveniles are consistently less tolerant than adults of the same species, and animals recovering from handling are less tolerant than the same animals a week later. A single farm-wide standard applied across life stages will be wrong for several of them.

On the engineering side, the design question is not what the treatment train achieves at steady state but what it achieves during a transient. Nitrifying populations grow slowly, so a step increase in feeding produces an ammonia and then a nitrite excursion before the biofilter catches up. Aeration and oxygenation, by contrast, respond in seconds. Matching the ramp rate of feeding to the response rate of the slowest process is the practical expression of this.

Gas supersaturation deserves separate mention as an engineering-created hazard. Pressurised leaks on suction pipework, or heating water without degassing, can drive total dissolved gas above saturation and produce gas bubble disease even when every conventional parameter looks correct.

• Manage the whole set of parameters together. Correcting one in isolation frequently moves another in the wrong direction.
• Control the input rather than chasing the output: the feed rate is the primary lever on ammonia, solids and oxygen demand alike.
• Remove solids as early and as intact as possible, before they disperse and become dissolved load.
• Maintain alkalinity deliberately in any system where nitrification is doing the nitrogen work.
• Change conditions gradually. Aquatic animals tolerate a wider range of stable conditions than they tolerate rapid movement within a narrower range.
• Keep an independent means of measurement available, so that a suspicious continuous reading can be checked rather than debated.
• Treat every water quality investigation as a search for the chain, not the symptom.

• Characterise the source water fully and across seasons before designing anything that depends on it.
• Commission and mature the biological treatment before stocking, and prove it against a simulated nitrogen load.
• Increase feed in measured steps, holding each step until the water response has been observed and has settled.
• Log every parameter with date, time, unit and the identity of the person taking the reading.
• Keep aeration redundancy sized for the failure case rather than the operating case, with an independent power path.
• Review the record at fixed intervals rather than only when something goes wrong, so the trend is seen while it is still slow.
• Retain water samples when an unexplained event occurs, so that a later analysis has material to work with.

Beyond the standard parameter set, intensive low-exchange systems accumulate constituents that routine testing does not look for: dissolved organic carbon, humic and coloured compounds, trace metals concentrated by evaporation, and metabolites that influence flavour. Off-flavour compounds produced by certain actinomycetes and cyanobacteria are the most commercially significant of these, and they are managed by purging in clean water before harvest rather than by treatment of the production water.

A second advanced theme is the microbial ecology of the water itself. Systems can be operated to favour a stable, largely harmless microbial community rather than to minimise total bacterial counts, on the reasoning that a sterilised water column is rapidly recolonised by whatever arrives first. Managing organic carbon availability and avoiding repeated disinfection of the whole system are the practical levers.

Finally, continuous instrumentation changes what water quality management can be. When a parameter is sampled once a day, management is reactive by definition. When it is logged continuously, the rate of change becomes visible, and the rate of change is usually a better predictor of an incident than the value itself. Building control logic on trend rather than on threshold is the direction that instrumentation makes available.