What an aquarium water change does

Alkalinity is eaten by nitrification, and which way your pH goes is a feeding rate

Oxidising ammonium releases two hydrogen ions per nitrogen, which works out at about 7.1 mg of alkalinity as calcium carbonate destroyed per mg of ammonium nitrogen¹. In hobby units, one mg of nitrogen per liter per day destroys about 2.79 dKH a week². A 100 liter tank fed one gram of dry food a day is losing about 2 dKH a week if nothing plants it back².

Top-off adds alkalinity too, at your evaporation rate times your tap carbonate hardness. Set the two equal and there is a break-even feeding rate. On 5 dKH central Yerevan tap³ at 5 percent evaporation, a 100 liter tank with plants taking the nitrate back up breaks even at about a quarter of a gram of dry food a day². Almost everyone feeds two to four times that.

So in a normally fed tank with no water changes, general hardness rises and carbonate hardness falls at the same time. The tank drifts towards high hardness with no buffer, which is the combination nobody warns about. In genuinely soft water with a normal bioload the buffer collapses and the pH crashes. In hard water with a light bioload and heavy evaporation, both climb instead. A water change fixes either direction without you having to know which one you are in.

Dissolved organic matter accumulates, and no test kit can see it

Measured properly, in the closest system to a low water change aquarium that anyone has instrumented. Two zero-discharge recirculating systems were tracked with fluorescence spectroscopy: one tryptophan-like component, from protein, and four humic-substance-like components. Three of the humic components and the protein component rose steadily throughout operation, and much faster than in a flow-through aquarium run alongside as a control⁴.

The finding to sit with is the next one. Humic-like and tryptophan-like material matching the culture water was found in the blood of the fish, at higher levels in the closed-system fish, rising with how long the system had been closed⁴.

What it does in a tank: it feeds the fast growing heterotrophs that blanket a nitrifying biofilm and cut off its oxygen, it tints the water and absorbs light before the light reaches the plants, and it binds metals. It is not all bad. Blackwater keeping is built on the protective side of it⁵. The honest position is that this is a real, measurable accumulation with effects in both directions, and no hobbyist has any way to know which direction their tank is in, because there is no test.

A mechanical filter removes particles, not dissolved compounds. A biofilter eats the easy fraction and leaves the humic fraction, which is exactly the fraction that was measured accumulating. Carbon works and is exhausted in weeks. Protein skimming does not work in fresh water. A water change is the only routine tool a freshwater keeper has for this.

Metals accumulate as water re-use rises, and the Armenian version is not the obvious one

Three recirculating systems at 30, 70 and 1,500 liters per kilogram of feed per day were analysed for ten metals. Arsenic, iron, manganese, nickel and zinc all rose in the water as the exchange rate fell. Honest counter-finding from the same paper: this did not translate into fish tissue above food safety levels⁶.

The companion experiment is the striking one. Carp eggs were incubated in ultrafiltered water from a high accumulation system and from a low accumulation one. Ultrafiltration strips particles and micro-organisms, so only the dissolved chemistry differs. The high accumulation water produced higher egg and larval mortality, reduced and delayed hatching, and smaller lighter larvae, and it carried more arsenic, copper, manganese, nickel, zinc, potassium, magnesium, sodium, phosphorus and sulfur⁷.

The Armenian correction. The obvious story is that Yerevan tap water carries copper and top-off concentrates it. That is wrong at the first step: copper, zinc, lead, cadmium, iron and arsenic were at or below detection in almost all Yerevan tap samples³. The supply is clean. The risk is your own plumbing, because stagnation time raises dissolved metal from pipes and fittings⁸. Run the cold tap for a minute or two before drawing water for a shrimp tank, and never draw from the hot side. After that, the arithmetic applies, because copper has no route out of a tank except a water change.

Nitrogen that is not nitrate, and phosphate that is not in the water

The hobby's nitrogen diagram starts at ammonia. The step before it is heterotrophs breaking protein into peptides, amino acids and urea, which is dissolved organic nitrogen. That is where most nitrogen in a tank spends part of its life, and no hobby kit sees it. An ammonia kit does not, a nitrite kit does not, a nitrate kit does not.

Phosphate is the same trick in a different compartment. A zero reading says the water column has none. Sediment binds phosphate to iron oxides while it is oxidised, and releases it when the sediment goes anoxic and those oxides dissolve. Lake science calls this internal loading and it is why a lake keeps blooming for years after the external supply is cut⁹. Walstad sets out the same sequence for a submerged soil in her own substrate chapter¹⁰.

Where this stops short of an argument for a water change. Changing water does not touch the sediment reserve, and lowering the water column concentration may even steepen the release gradient. The useful conclusions are narrower: a phosphate reading of zero is not evidence the tank is phosphate poor, and the water change that is genuinely justified is the one after you disturb the substrate.

Trace element depletion, which is real and is a bad argument for a water change

Plants strip iron, manganese, boron, zinc, copper and molybdenum out of the water. In a closed system with no replacement those fall. The aquaponics literature treats iron supplementation as routine rather than exceptional, because fish feed does not supply enough in a form plants can use¹¹.

But changing water will not fix it here. Yerevan tap water is soft, volcanic in origin and near the detection limit for iron³. You cannot replace what the water does not contain. And the chemistry is against you anyway: in oxygenated water at aquarium pH, iron oxidises and precipitates within hours, which is why fertilisers use chelates. The fix for trace depletion is dosing.

Walstad's counter-argument is a good one and belongs here. Fish food contains every element plants require, on the grounds that living things are chemically similar, and sediment micronutrient concentrations can be around a hundred thousand times those in the water above¹⁰. In a soil tank, trace elements come from the substrate and the food, and the water change has nothing to do with it. That is why this ranks tenth and not fourth.

Medications, and the one that never leaves

Most aquarium medications leave by exactly three routes: degradation, activated carbon, and a water change. Some degrade in days. Some do not. Copper-based treatments have no degradation route at all, because copper is an element. It stays until it is physically removed or bound into the substrate, from where it can come back.

Three consequences. Finishing a course of medication means removing the medication, and the label instruction is not a formality. Anything copper-based makes a tank permanently doubtful for shrimp. And two treatments a fortnight apart, in a tank that gets no water changes, are two treatments present at once.

Marked Medium because the mechanism is certain and no persistence measurements for hobby medications in aquarium conditions were found.