The stoichiometry of the three pathways was set out by Ebeling et al. (2006) and their place in suspended-growth systems by Hargreaves (2006). Algae assimilate ammonia in the light, consuming carbon dioxide and producing oxygen. Nitrifying bacteria oxidise ammonia to nitrite and then to nitrate; they grow slowly, yield little biomass, and consume oxygen and a large amount of alkalinity. Heterotrophic bacteria, when supplied with organic carbon, assimilate ammonia directly into new cells. They grow fast and respond within hours, but they produce many times more biomass per unit of nitrogen than nitrifiers do, and they also consume oxygen and release carbon dioxide. Table 1 compares the pathways and the management response each calls for.

Avnimelech (1999) reasoned that because bacteria build their cells with a fairly constant proportion of carbon to nitrogen, a supply of carbohydrate matched to the ammonia to be removed would drive heterotrophic uptake. Molasses, sugar, starch and cereal brans have all been used (Avnimelech, 2009). The same effect can be had by lowering the protein content of the feed, since protein is the source of the nitrogen.

In practice the pathways coexist and their balance shifts through a crop. Algae dominate a newly filled pond in sunlight. As feeding rises, bacteria overtake them and the water turns from green to brown. Nitrifiers establish over a period of weeks and, once present, take a growing share of the ammonia, so that mature systems accumulate nitrate and need less added carbon than the heterotrophic model implies (Hargreaves, 2013). Operators who continue to add carbohydrate at the initial rate produce more solids, and consume more oxygen, than the water chemistry requires.

The price of converting nitrogen to biomass is that the biomass remains in the water. Suspended solids rise through the crop, and with them the oxygen demand of the water itself. Ray et al. (2010) showed that removing part of the solids with simple settling chambers improved shrimp production in superintensive raceways, and Schveitzer et al. (2013) found that performance was best at intermediate floc concentrations, with very high levels harming the animals. A biofloc system is therefore managed within a band of solids, monitored with settling cones or by total suspended solids, and not simply allowed to thicken.

Respiration and nitrification both acidify the water and deplete alkalinity, and in closed systems the loss is not made good by incoming water. Regular additions of bicarbonate, carbonate or hydrated lime are standard, and their relative merits were compared by Furtado et al. (2011). Start-up is the most hazardous period. Ammonia-oxidising bacteria usually establish before nitrite oxidisers, and nitrite can accumulate to harmful concentrations in the interval. The toxicity of both ammonia and nitrite to L. vannamei increases as salinity falls (Lin and Chen, 2001; Lin and Chen, 2003), which matters because much biofloc production takes place inland in low-salinity water.

Flocs are eaten. Using nitrogen isotopes in the Belize ponds, Burford et al. (2004) showed that flocculated material contributed measurably to the nitrogen retained by the shrimp, and Wasielesky et al. (2006) found that juveniles grew better and converted feed more efficiently in floc-rich water than in clear water on the same diet. The nutritional value of flocs varies with the carbon source and the microbial community, and they supplement formulated feed without replacing it (Emerenciano et al., 2013). The practical gain is that feeds of somewhat lower protein content can be used, and that nitrogen is recycled once before it leaves the system.

The first mechanism is exclusion. White spot syndrome virus and other major pathogens enter farms with water, with wild crustacean carriers and with infected postlarvae (Lightner, 2011; Flegel, 2012). A lined pond or tank that is filled once with disinfected water, stocked with postlarvae from tested broodstock and then closed has far fewer routes of entry than a pond exchanging water daily. This is a property of low exchange and good biosecurity, not of bioflocs as such, but biofloc management is what makes low exchange workable at high density. The argument is strong and follows from first principles of epidemiology. It also implies a corollary that is often overlooked: once a pathogen is inside a closed, densely stocked system, there is little to slow it.

The second mechanism is an effect of the flocs on the shrimp or on the pathogens. Bacterial cell wall components are recognised by the innate immune system of crustaceans, and shrimp living in bioflocs are continuously exposed to them. Ekasari et al. (2014) reported higher values of several immune indicators and better survival after challenge with infectious myonecrosis virus in shrimp that had been fed bioflocs grown on different carbon sources, and Kim et al. (2014) found increased expression of immune-related genes in postlarvae reared in biofloc water. These are laboratory results over short periods. They show that an effect exists; they do not show how large or how lasting it is in a commercial crop.

Effects on bacterial pathogens have also been proposed. Floc bacteria may compete with vibrios for nutrients and attachment sites, some accumulate poly-β-hydroxybutyrate, which has shown protective activity against pathogenic vibrios in model organisms, and some interfere with the quorum sensing that regulates vibrio virulence (Defoirdt et al., 2007; De Schryver et al., 2008). Against this must be set the fact that bioflocs are an organically rich habitat in which vibrios also grow well. Acute hepatopancreatic necrosis disease, caused by particular strains of Vibrio parahaemolyticus (Tran et al., 2013), has prompted interest in biofloc and related mature-water systems as a control measure. The supporting evidence is largely from small-scale challenge tests and farm observation, and controlled field comparisons are lacking. The honest summary is that bioflocs probably help through biosecurity, may help through immune stimulation and microbial competition, and cannot be relied upon to protect a crop once a virulent pathogen has entered.