An Oxygen Sag in a Small River

At A. DO is 7.4 mg dm⁻³, close to air saturation for warm water, and BOD is 1.8 mg dm⁻³. This is a clean forest stream with little degradable organic matter and a diverse sensitive fauna (18 EPT taxa).

At B. Two hundred metres below the outfall the effluent has mixed into the stream. The organic matter is oxidised by aerobic bacteria, and this respiration consumes oxygen faster than reaeration can replace it, so DO falls to 2.0 mg dm⁻³. The BOD of 42 is not a separate pollutant but a measure of the same organic load still available to be oxidised. Only pollution-tolerant animals remain; tubificid worms are classic indicators of organic enrichment, and EPT taxa disappear.

At C and D. Downstream, the easily degradable fraction has been mineralised, so the oxygen demand falls and DO begins to recover through reaeration from the surface and photosynthesis by algae. Recovery is a balance between the remaining BOD and the rate of reaeration, which is slow in a sluggish stream, which is why it takes kilometres rather than metres.

Why the fish died at C, not B. The kill followed three weeks without rain, which cut the flow and therefore the dilution, deepening the sag and moving its lowest point downstream. A single daytime reading of 4.8 at C conceals diurnal swings: at dawn, algal and bacterial respiration can pull DO far below the daytime figure, into the range that kills sensitive fish. The effluent also carries ammonia, and its nitrification consumes further oxygen. Fish are mobile and tend to avoid the outfall zone, so mortality shows up where they congregate, in the deeper pools near C, not where the worst water is.

Recovery is real but incomplete. BOD at D (3.1) is still about 70% above the reference (1.8), and EPT richness is 12 against 18 upstream, only about two-thirds of the reference community. Taxa with poor dispersal and high oxygen demands take years to recolonise, so the fauna lags behind the chemistry. Calling D recovered would be premature.

Much of the apparent recovery is dilution, not treatment. The load is 1200 × 0.6 = 720 m³ of effluent per day at 250 g m⁻³, which is 180 kg BOD per day. Dry-season flow of 0.05 m³ s⁻¹ is 4320 m³ per day, so full mixing would add about 42 mg dm⁻³, which is exactly the BOD measured at B. The river is simply too small for this load, and the sag will worsen in a drier year.

Design weaknesses. (i) One sampling occasion cannot describe a system that varies with flow and season. (ii) Three grabs at a site are pseudo-replicates; they estimate analytical and local variation, not site variation, so there is no true replication of the impact. (iii) The upstream reference is a forest stream that differs from the downstream reaches in shade, substratum and temperature, so any difference is confounded with habitat. (iv) Grab samples miss diurnal oxygen swings and storm events. (v) Only BOD and DO were measured; ammonia, nitrate, phosphate, turbidity and the effluent flow itself are missing, so the load cannot be audited.

Improvements. Sample in both dry and wet seasons, replicate sites within each reach, add two independent reference streams of similar size, log DO continuously over 24 hours including pre-dawn, measure ammonia and the discharge flow, and identify invertebrates to a standard index level.

The load. 1200 t of fruit bunch per day × 0.6 m³ per tonne = 720 m³ of effluent per day. At 250 g m⁻³ this is 720 × 250 = 180 000 g, that is 180 kg BOD per day.

Wetland capacity. Constructed wetlands treating this kind of effluent remove of the order of 50-100 kg BOD per hectare per day, and less while establishing; a surface-flow wetland receiving 250 mg dm⁻³ would also risk surface clogging and anaerobic conditions. At a generous 100 kg per hectare per day, 0.5 ha handles about 50 kg per day, roughly 28% of the load. Even an unrealistic 250 kg per hectare per day gives 125 kg per day, about 69%, and that ignores seasonal failure. So the wetland is short by roughly a factor of four on optimistic assumptions and by much more in practice; it cannot be the sole remedy.

The real fault is upstream of the wetland. The tertiary pond is silted, and the discharge is 250 mg dm⁻³ against a limit of 20. Restoring and desludging the pond system to meet 20 mg dm⁻³ would cut the load to 720 × 20 = 14.4 kg per day, which at 0.05 m³ s⁻¹ adds only about 3.3 mg dm⁻³ to the river, within its assimilative capacity.

Other options. Land application of the treated effluent to the estate as a fertiliser, at a calculated nitrogen and BOD loading rate, is a genuine use rather than a discharge. Biogas capture from the anaerobic ponds reduces both the load and the methane emissions. A constructed wetland is then appropriate only as a polishing step, sized as load divided by the measured removal rate.

Recommendation. Repair and monitor the pond system to meet the licence limit, monitor BOD and DO upstream and downstream at least monthly with continuous DO logging, and use a wetland only for polishing, sized on the measured residual load.