Aquarium temperature: heating and cooling in Armenia

Heat, and it is the serious one here

Warm water holds less oxygen at exactly the moment the fish need more of it.

Yerevan has 82 days a year when the maximum passes 30 C. Armavir has 90, the Yerevan plain station has 90, and Meghri has 91¹. That is roughly a quarter of the year. The national record is 43 C, set in Yerevan and tied by Meghri on 1 August 2011¹. Gyumri and Dilijan are a different country for this purpose, and advice written for one is wrong for the other.

The oxygen curve, with an Armenian correction that nobody makes

Oxygen dissolves less readily in warm water, and the numbers for that are standard. What is not standard, and what changes every stocking recommendation you will read, is altitude Gas dissolves in proportion to its partial pressure above the water, and partial pressure falls with barometric pressure. Yerevan sits at about 887 hPa and Gyumri at about 846, against 1013 at sea level¹.

The table below is computed by us² from the Benson and Krause saturation equation as published in Standard Methods³, corrected for barometric pressure in the usual way.

Read the last column. A tank in Yerevan holds about 13 percent less oxygen than the identical tank at sea level, at every temperature. Gyumri loses about 17 percent². Every stocking table in every English or Russian aquarium book is written for sea level, so in Armenia they are all optimistic by roughly an eighth. Nobody mentions this, and it applies to every tank in the country.

The squeeze, in two numbers

Supply falls with temperature, as above. Demand rises with temperature, because a fish is an ectotherm and its metabolism follows the water. This has been measured on an aquarium fish. Cooper, Mueller and Eme acclimated neon tetra and black neon tetra at 26, 29 and 31 C and measured oxygen consumption. The temperature quotient across 26 to 31 C was 1.92 for neon tetra⁴, which means a ten degree rise very nearly doubles the oxygen the fish needs².

Take a Yerevan tank from 24 C to 30 C and put the two together.

That is the whole summer problem in one line. The fish needs about half as much again, and the water can hold about a tenth less. Nothing else about an Armenian summer needs explaining once that sentence is understood.

Lethal thresholds

Two different numbers get quoted for how hot a fish can survive, and the forum always quotes the wrong one.

CTMax, the critical thermal maximum, is measured by warming the water at a steady rate over minutes and recording where the fish loses equilibrium. It is an acute, short-exposure figure. CLMax or CLT, the chronic lethal maximum, is the temperature at which the fish is held for a long period and cannot live. The gap between the two is large, and the number that circulates in the hobby is almost always the acute one.

A neon tetra that survives a five minute ramp to 39 C cannot live at 33.5 C. The authors state the conclusion plainly: neither tetra can survive temperatures four to five degrees above their natural averages⁴. Peppered cory gain 0.55 C of upper tolerance for every degree of acclimation temperature⁶, which is the measured version of the true but limited idea that fish get used to things.

Air breathers are a real exception, and there is a mechanism

De Lima and colleagues tested two Amazonian loricariid catfishes under low oxygen in 2025. In Pterygoplichthys pardalis, the sailfin pleco, thermal tolerance was unaffected by the oxygen in the water so long as the fish could reach the surface, and dropped only when surface access was blocked. In Ancistrus dolichopterus, the bristlenose, whose air-breathing stomach is less capable, thermal tolerance fell away below the critical oxygen pressure even with air available⁷.

That is a published, mechanistic reason why plecos are hardy in hot water. It also says something more useful: "pleco" is not one animal. A sailfin is tougher than a bristlenose, and it is tougher for an anatomical reason you can point at. Bettas and gouramis do something related. Le, Burggren and Martinez-Bautista found adults of both Betta splendens and the blue gourami shifting from gill ventilation towards their air-breathing organ as the temperature climbed⁸. A labyrinth fish in hot water solves the oxygen problem by leaving the water for it.

The remedies, ranked honestly

Five ways to cool a tank, in order of what they buy for what they cost.

Every temperature figure below marked computed comes from one steady-state energy balance, stated here once so that it can be checked or argued with. A 100 L tank of 80 by 35 by 40 cm. Heat crosses the glass at an overall 5.5 watts per square meter per kelvin. The open surface loses heat by convection at 3.5 watts per square meter per kelvin and by evaporation, with the evaporation coefficient taken from the Lewis relation. Latent heat 2,440 kJ per kilogram. Barometric pressure 887 hPa, which is Yerevan¹. Saturation vapour pressure from Buck⁹. This is a model, presented as an order of magnitude, not a measurement¹⁰.

Add 25 watts of light and pump heat, which is realistic, and the no-fan case rises to 28.2 C and the clip fan case to 23.1 C. So in a working tank the fan is worth about five degrees.¹⁰ Two honest notes on that. The hobby figure of "a fan buys two to four degrees" is the marginal gain over an already-evaporating open tank, and it is entirely compatible with this model. And an open tank is already four degrees below the room before you do anything at all, which is why a cover glass is a handicap in July and a saving in January.

What the fan costs, and this is the part that gets left out. A clip fan on a 100 L tank evaporates about 3 liters a day, three percent of the tank, every day, for as long as the heat lasts. Over a thirty day hot spell that is roughly ninety liters, close to a whole tank volume¹⁰.

Evaporation removes water and leaves everything dissolved in it behind. Top up with tap water and every liter adds its full mineral load, and none of it ever leaves. Ninety liters of tap water topped into a 100 L tank over a month roughly doubles the dissolved solids if no water is changed in between¹⁰. Top up with rain, distilled or reverse osmosis water. Change water with tap water. Topping up and changing water are two different operations with two different liquids, and a fan in an Armenian August is how most people find that out the hard way.

The summer answer, in one paragraph. In Yerevan, plan for a room in the low thirties for about a quarter of the year. Site the tank out of the sun first, because that is free and it is the biggest single term. Take the lid off, because an open tank is already several degrees below the room. Put a fan across the surface, accept that it costs about three percent of the volume in evaporation every day, and top that up with mineral-free water rather than tap. Keep the surface moving all summer, because at 30 C the fish need about half as much oxygen again as at 24 and the water holds about a tenth less, before the thirteen percent Yerevan altitude penalty is even counted². And choose the fish for the room you actually have, rather than buying a chiller for the fish you wanted. If the honest answer is that this species cannot live in your flat in August, that is a real answer and it is much cheaper to hear it now.

Winter

How big a heater, and why the way it fails matters more than the way it works.

The hobby rule is one watt per liter. That is roughly right for a mid-sized tank in a cool room and wrong at both ends of the range, because heat loss follows surface area while volume follows the cube of length. A small tank needs more watts per liter than a large one, and every table below says so.

Computed with the same energy balance as the cooling section, holding the water at 25 C, at Yerevan pressure, with indoor humidity taken as 30 to 35 percent¹⁰. That is not a guess: Yerevan winter air heated to room temperature reaches only 14 to 22 percent relative humidity before the occupants add any moisture of their own¹. The two figures in each cell are open surface, then with a cover glass

In watts per liter that runs from 1.31 for a 54 liter tank in a 16 C room down to 0.46 for a 270 liter tank in a 22 C room¹⁰. The one watt per liter rule sits in the middle of that range and is not a bad rule, but it is a rule about one particular tank in one particular room.

The cover glass is the largest single lever in winter and it is nearly free. It roughly halves the requirement, because most of the loss from an open tank into dry Armenian winter air is evaporation rather than conduction. The same model says an open 100 liter tank at 25 C in a 16 C room evaporates about 1.6 liters a day; with a cover glass, about 0.3¹⁰. That is the exact mirror of the summer section. The lid is a heater in winter and a handicap in summer, so take it off in June and put it back in October.

One Armenian footnote on sizing. Yerevan's January mean is minus 2.4 C and Gyumri's is minus 7.8¹, but the tank is indoors, so neither number is the one that matters. Measure your own room at six in the morning in January and size the heater against that. A room that sits at 21 C in the evening and 15 C before dawn is a room whose heater must be sized for 15.

Four ways a heater fails

Why failing on is much worse than failing off

It is bounded on one side and not the other. A heater that stops cannot cool the tank below room temperature. A heater that will not stop is bounded only by its own wattage against the tank's heat loss.

The direction is wrong in three ways at once. Cold slows a fish down, so oxygen demand falls, and cold water holds more oxygen. A cooling tank moves towards a larger safety margin on every axis. A heating tank does the opposite of all three simultaneously, which is exactly the squeeze described above.

The clock is different. A tank that loses its heater in a heated flat drifts down a fraction of a degree an hour and takes most of a day to reach room temperature. That is enough time to notice. A small tank with a large stuck heater passes the lethal line in hours.

And here are the numbers. Computed, same model, room at 20 C, open surface, showing where the water settles with the heater stuck permanently on. Compare each cell against the measured chronic lethal limits above: 33.5 C for neon tetra⁴, 34.9 for peppered cory⁶.

Read along the top row. A 100 watt heater stuck on in a 54 liter tank settles at 32 C, which is bad and survivable. A 300 watt heater in the same tank settles above 50 C and passes 33 C in about two hours¹⁰. The commonest way to kill a tank with a heater is to buy a heater that is too big for it.

So: size the heater from the table, allow up to about half again for a cold snap, and stop. If a large tank in a cold room needs more than one heater's worth, fit two smaller heaters at opposite ends rather than one large one. Two heaters halve the consequence of a stuck thermostat, keep the tank warm when one of them fails off, and even out the temperature along the length of the glass. It costs a little more and it is the best value decision on this page. Then add a thermometer that is not part of the heater, and read it every day, because that is the only thing that catches a drifting controller.