Why Fish Die at Dawn — Overnight Oxygen

Written 2026-09-29

In a heavily planted tank, night flips oxygen from a surplus to a drain. Once the lights go off, plants stop photosynthesizing and start consuming oxygen too, right alongside the fish and filter bacteria, while the water only trades gas with the air slowly. So oxygen keeps falling from lights-off all the way to just before lights-on — dawn is the low point of the whole day. Running this simulator on a 60cm tank with only a low-agitation canister filter, no air pump, and the heater at 27°C for 90 days, the daily oxygen low landed at 8am — one hour before lights-on — on 54 of 60 days, and that chronic dawn dip eventually killed all 6 of the oxygen-sensitive otocinclus catfish (between day 41 and day 89), while the tank’s 22 neon tetras were completely untouched. Adding one small air pump (surface agitation 0.3) to the same tank let all 28 fish survive the full 90 days.

Why dawn — oxygen only ever goes down at night

During the day, lit plants pump out oxygen through photosynthesis and the water sits close to saturation (the most it can hold at that temperature). The instant lights go off, plants stop making oxygen and start consuming it through respiration, just like the fish and the filter bacteria always do. Gas exchange at the surface — the only thing refilling that oxygen — is comparatively slow, especially with a low-agitation filter like a canister and no air pump running. So oxygen falls continuously from lights-off until the moment lights come back on, and that last moment is the lowest point of the whole cycle.

Temperature plays a part too. By this simulator’s saturation formula, water at 25°C can hold up to 8.18mg of oxygen per liter, but at 27°C that drops to 7.87mg. Warmer water simply has less room for oxygen to begin with, so a warmer tank makes the dawn low point that much riskier.

What we actually ran — only the sensitive species died, one by one

We set up a standard 60cm tank with only a mid-size canister filter (low surface agitation, a common choice in planted tanks) and no air pump, heater held at 27°C. We planted it densely with Rotala rotundifolia and hygrophila, let it fully cycle, then stocked 22 neon tetras together with 6 otocinclus catfish on day 30, and watched for 90 days. Otocinclus has the lowest oxygen tolerance in this catalog (0.7, versus the default 1.0).

Ammonia and nitrite never came near a dangerous level across the whole run — nitrite peaked at just 0.04ppm on a test-kit scale, basically never rising at all — so the water chemistry itself was stable. Even so, the otocinclus died one at a time from day 41 to day 89 until all 6 were gone, and the only cause logged each time was low oxygen. Oxygen fell every night after lights-off and bottomed out right before lights-on — counting the daily low across the 60 days after stocking, 54 of them hit bottom at 8am, one hour before the 9am lights-on. Oxygen at the moment of each death ranged 4.36–4.89mg/L: below otocinclus’s own low-oxygen threshold (5.2mg/L) but above the default threshold (4.0mg/L) — exactly the band that only catches sensitive species. The tank’s 22 neon tetras (tolerance 1.0, threshold 4.0mg/L), sharing the same water, lost none across all 90 days.

The fix — one small air pump is enough

We changed exactly one thing in the same tank — added an air pump, raising surface agitation from 0 to 0.3 (this tank injects no CO₂, so there was no CO₂ to lose by doing it). All 28 fish, otocinclus and neon tetra alike, survived the full 90 days with zero low-oxygen deaths. The air pump churns the surface, speeding up how fast the water trades gas with the air — the amount of oxygen consumed overnight didn’t change, but the tank now has enough margin to make it to dawn.

Watch for this combination in particular: ① a low-agitation filter (canister or undergravel) with no air pump, ② a heater set near the top of a species’ comfortable range (which lowers the oxygen ceiling), and ③ an oxygen-sensitive species like otocinclus. Stack all three and the water can look perfectly fine on a test kit while the tank quietly runs a dangerous stretch every single night.

The button below is the tank from this guide with the small air pump added, where all 28 fish survived.

The numbers in this guide come from the Water Kiwoogi simulator model; real tanks can differ.