Nitrifying Bacteria — AOB, NOB, and Comammox

Every aquarist who has ever started a tank has waited on these organisms, usually without knowing their names. They are the "biological filter" — the invisible bacteria that turn the ammonia your fish and decaying food produce into nitrate, the far less toxic end product you remove with water changes. The weeks of waiting before a new tank is safe to stock, the dreaded nitrite spike, the reason an old established tank shrugs off a slug of ammonia that would crash a fresh one: all of it is the story of these few slow-growing microbes building up to a working population.

Nitrifiers are chemoautotrophs. They get their energy not from light or from eating other organisms but from the chemical oxidation of reduced nitrogen, and they use that energy to build biomass out of CO2 — much as a plant does with sunlight. EcoSym splits the job across three cooperating-and-competing groups, because lumping them together would erase exactly the things hobbyists watch for.

The two-step relay, and the one-cell shortcut

Classical nitrification is a relay race between two groups of specialists:

   ammonia  ──AOB──►  nitrite  ──NOB──►  nitrate
   (toxic)            (toxic)            (much safer)

   comammox: ammonia ─────────────────► nitrate   (one cell, no nitrite handed off)

• AOB — the ammonia-oxidising bacteria (Nitrosomonas-type) — take ammonia to nitrite. This step consumes the most oxygen and releases acid, which is why heavy nitrification slowly eats a tank's buffering and drives pH down.
• NOB — the nitrite-oxidising bacteria (Nitrobacter and Nitrospira) — finish the job, taking nitrite to nitrate. This step is roughly charge-neutral and consumes much less oxygen per unit of nitrogen.
• Comammox — complete-ammonia-oxidising Nitrospira — do the whole conversion inside a single cell, ammonia straight to nitrate, never handing a free nitrite molecule to the water (Daims et al. 2015).

The first two steps add up to the same overall reaction as the third. The point is who does the work and when, because that partition is what produces the patterns aquarists recognise:

• The new-tank nitrite spike. Anyone who has cycled a tank has watched nitrite climb to a few mg/L through weeks two to four and then suddenly collapse. It happens because NOB lag AOB by one to three weeks — they cannot get established until AOB have built up a nitrite supply to feed on, and free ammonia poisons NOB more than it poisons AOB. Comammox, also held back by free ammonia at this stage, stays scarce. So ammonia gets converted to nitrite faster than nitrite gets converted onward, and nitrite piles up until NOB finally catch up. A single lumped "nitrifier" would never reproduce this — its nitrate would appear the instant ammonia went in.
• Mature tanks run cleaner. In an established, lightly stocked tank, comammox quietly takes over the quiet surfaces — the glass, the substrate, the leaves (more on why, and on where it does not take over, below). Because it skips the nitrite hand-off entirely, a mature tank oxidises ammonia all the way to nitrate with no measurable nitrite in between.
• Nitrite is genuinely dangerous. Nitrite enters fish and invertebrates through the same gill pathway they use to take up chloride, then locks up the oxygen-carrying pigment in their blood — "brown blood disease" in fish, the equivalent in shrimp. Shrimp take harm at a few tenths of a milligram per litre; snails tolerate several times more. The comammox-dominated mature tank keeps nitrite at noise level; the AOB-dominated new tank spikes it. Both outcomes fall out of the kinetics rather than being scripted.
• Oxygen and pH timing. The acid release and the bulk of the oxygen demand belong to the ammonia-to-nitrite step, not spread evenly across the whole conversion. Keeping the steps separate puts the oxygen trough and the pH dip where they actually happen during cycling.

Shared biology

Both AOB and NOB are slow growers next to ordinary decomposer bacteria — think a doubling time of a day or two rather than hours — and they get very little energy from their chemistry, so they build biomass sparingly. Both are obligate aerobes that fade under hypoxia, both are persistent and hard to kill outright, and both tolerate a wide span of temperature and salinity. This combination — slow to build, slow to die — is exactly why a biological filter takes weeks to establish and then, once established, is remarkably durable.

Each group keeps both a free-floating (planktonic) population and surface-attached biofilm populations. Planktonic cells settle onto surfaces steadily, favouring rough substrates like ceramic and sand over smooth glass, and once embedded in the biofilm's extracellular polymeric substance matrix they almost never let go. That one-way drift from floating to attached, over the first few weeks, is the cycling period new-tank keepers wait out — and it is why the filter media and substrate, not the water, carry your biological filter.

Three things make the attached life worth it:

• Shade from light. Nitrifiers are photoinhibited, but it takes genuinely bright light to bite. Lab measurements put the half-response for ammonia oxidisers around full-spectrum daylight intensity — they shrug off the dim light reaching a shaded substrate or filter (unaffected up to ~15 µmol m⁻² s⁻¹; tolerant to ~200), and only strong overhead light meaningfully throttles a free-floating or fully-exposed cell. The nitrite oxidisers (NOB) are the sensitive ones, suppressed at intensities the ammonia oxidisers tolerate — which is why bright tanks can hold a little nitrite even after they seem cycled. A cell buried under a biofilm is shaded further still: how much depends on biofilm maturity — a brand-new film adds only what the surface texture itself blocks (porous ceramic blocks most light, bare glass almost none), while a mature film with its slime matrix and overlying algae cuts incoming light to a small fraction. Combined with the steady drift onto surfaces, this is why nitrifiers end up concentrated on the shaded, oxygen-rich substrate and media — not in the lit water column. (The model also lets nitrifiers living in the substrate breathe oxygen from the overlying water rather than suffocating in the anoxic deep pore, matching the real pattern that sediment nitrification is confined to the thin oxic surface layer.)
• Shelter from grazers. Free-floating nitrifiers are eaten freely by nanoflagellates and ciliates, which graze all bacteria. Cells locked into a mature biofilm matrix are largely out of reach. Without this refuge, slow-growing nitrifiers would be grazed to extinction by fast-breeding protists within weeks. The bigger threat in a stocked tank is the metazoan grazing crew — snails, shrimp, copepods, scuds — which rasp and filter biofilm far harder than protists do. The cells that survive them are the ones living inside the substrate: down in the interstitial spaces of sand and soil, and in the dense basal layers of a film, where a snail's radula or a shrimp's mouthparts simply cannot reach. In a soil or sand tank the model treats most of the nitrifying community on the floor substrate (~85 %) as living in this grazer-proof zone, the same for ammonia and nitrite oxidisers because the two sit side by side in the same micro-colonies. Cells on the glass walls get no burial credit — there is nothing to bury into, which is exactly why snails keep the glass nearly clean while the substrate stays alive with bacteria. This matters most for the nitrite oxidisers: they are the only thing that consumes nitrite, so if a growing population of snails and shrimp grazed them away, residual nitrite would have nowhere to go and the tank would read a small but stubborn nitrite level forever — exactly the failure the substrate refuge prevents. A bare-bottom tank has no such hideout, so its nitrifiers rely on the biofilm matrix alone.
• Richer food at the surface (for AOB). Surface-attached AOB sit in a still boundary layer where ammonia runs richer than the open water — a perceived concentration fed by settled-detritus mineralization, pore-water proximity on sand and soil, and the nitrifiers' own activity. AOB on a fertile sand surface can keep working even when a water test reads near-zero ammonia. NOB get no equivalent boost: nitrite is a fleeting intermediate, not something detritus steadily produces, so NOB simply read the open-water nitrite.

The three guilds at a glance

They share a niche but play different hands:

During cycling this plays out as: AOB rapidly drain ammonia into nitrite; nitrite accumulates because NOB have not built up yet and comammox is suppressed by the high ammonia; then NOB establish, nitrate starts filling, and the nitrite spike collapses. In a settled tank all three idle along at a steady state, comammox's share climbing month by month as ammonia falls, and bulk nitrite sits near zero. A test kit then reads zero ammonia, zero nitrite, and slowly climbing nitrate: the textbook "cycled" tank. The exact growth rates, affinities, and oxygen and inhibition thresholds behind all of this are tabulated in the Parameter Reference.

Why comammox takes over the quiet surfaces

Surveys of established freshwater filters — home aquaria, recirculating fish systems, even treated drinking-water pipes — consistently find comammox Nitrospira outnumbering the classical AOB and NOB combined (Bartelme et al. 2017; Sauder et al. 2017; Pinto et al. 2016). The reason is a clean affinity trade-off. Comammox bind ammonia tens to hundreds of times more tightly than AOB do, but they grow more slowly and are more easily poisoned by free ammonia. That makes them losers at the high ammonia of a cycling tank, where raw growth speed wins and AOB pull ahead. But as a tank matures and ammonia drops to a trace, AOB find themselves starved far below the level they need to run flat-out, while comammox are still feeding comfortably. Below a crossover of roughly a tenth of a mg/L of ammonia-nitrogen, comammox actually grow faster than AOB despite their slower top speed, and they gradually inherit the surface.

Which surface, though, is the interesting part. What decides the contest is not how much ammonia is dissolved in the tank but how much arrives at each bacterium, and those are different numbers. On still glass, ammonia has to diffuse across an unstirred film before anything can eat it; inside a filter, a pump delivers it. So in the simulator a filter's media are a high-ammonia niche even when the tank's water reads zero — and the fast growers own them. In one cycled, filter-dependent tank at two months, comammox holds 87% of the nitrifying biomass on the glass and 22% inside the filter. Load that tank heavily and comammox is squeezed off the media almost entirely, to under a third of a percent, while it gains ground on the sand — because the busy filter has made the rest of the tank hungrier. The filter does not just house bacteria; it sorts them. See Filters.

This is why keepers notice that old, mature tanks handle an ammonia slug faster and more cleanly than freshly cycled ones, even when the visible bacterial population looks similar: the mature tank has accumulated a comammox community that processes steady loads with no nitrite transient, whereas a young AOB-dominated one still throws a brief nitrite burst when challenged.

Where the model and the surveys part company

Read the surveys again and you will notice they sampled filters, and found comammox dominant there. The simulator says the opposite about filters specifically, and it is worth being clear that this is a real disagreement rather than a subtlety of wording.

The model's only mechanism for separating these guilds is the ammonia each one perceives, so anything that raises perceived ammonia — flow — must favour the fast grower. Reality is thought to have a second mechanism, and it is a place rather than a rate: a thick biofilm with a starved, dimly-oxygenated core, which is exactly the sort of address a slow, patient Nitrospira like comammox would inherit. Push an oxygen microelectrode into a laboratory nitrifying biofilm and oxygen runs out at about 150 µm, with ammonia still in plentiful supply below that (Okabe, Satoh & Watanabe 1999).

For a long time this page said the simulator lacked that interior because it tracks heterotrophic bacteria as one water-column population rather than as residents of a particular surface — and that giving them a home on the media would grow the missing slime. We went and measured that, and it is not true. The result is interesting enough to be worth setting out.

A biofilter's bacteria are limited by their food, not by their address. The same arithmetic that kills the surface-area story (Filters) applies to the heterotrophs: their standing population settles where growth balances death, and that is set by the dissolved organic carbon the tank supplies. Moving a population from one bookkeeping column to another does not feed it. We can put a hard number on the headroom: strip away ninety-nine percent of everything that eats or kills the tank's heterotrophs — far more shelter than living on a surface could ever confer — and the population grows by a factor of 1.26. The absolute ceiling, with every predator and every virus removed, is 1.50. A film thick enough to have an interior needs ten to forty times more.

So the heterotrophs would move onto the media and spread themselves thin, the film would roughly double, and it would still be far too thin to stratify. The obstacle was never the pool structure. It is that an aquarium is a very poor place to be a bacterium: this simulator's biofilter processes about 0.14 g of nitrogen per square metre per day, five to twelve times below the load a working aquaculture biofilter carries. Thin films are what small meals buy.

And here is the part we cannot tell you. To convert what the model knows — bacterial nitrogen per square centimetre — into a thickness in micrometres, you must divide by the density of a biofilm. The literature puts that density somewhere between 1 and 40 milligrams per cubic centimetre, a fortyfold spread, and nobody appears to have measured it on aquarium media at all. Run the model's own film through that range and it comes out anywhere from 24 µm to nearly a millimetre. Earlier versions of this page quoted "about 9 µm" as though it were a measurement. It was one end of a bracket, and the bracket is wider than the question.

What is genuinely wrong with the model's biofilm is its composition, not its thickness. Feed a real biofilm nothing but ammonia — no organic carbon at all — and half of the cells in it are still heterotrophs, living on the waste and corpses of the nitrifiers (Kindaichi, Ito & Okabe 2004). The simulator's filter film is essentially all nitrifiers. That is a real error, and it is one we have chosen not to fix, because fixing it would move every number in the model and change none of the answers anyone asks it for: the film would not meaningfully thicken, the media would not run out of room (the crowding ceiling is nowhere near binding, and a ceiling counting all bacteria rather than just nitrifiers would bind less still), and comammox would not come back to the filter.

So trust the model's direction — flow favours AOB and NOB, starvation favours comammox — and treat its claim about which physical object ends up comammox-dominated as an artefact. The niche separation is real and emergent; its street address is not. Closing the gap needs a bigger meal, not a better filing system.

Why nitrite creeps back in an oxygen-starved tank

There is a second, separate way nitrite reappears — not the new-tank lag, but an oxygen problem in an otherwise cycled tank. NOB need more oxygen than AOB to run at full tilt, so when oxygen dips, NOB falter first. This is counter-intuitive, since converting nitrite to nitrate actually consumes less oxygen per unit of nitrogen than the ammonia step does — but how much oxygen a reaction needs and how tightly an organism can scavenge oxygen are two different things, and NOB are the more oxygen-hungry of the two (Wiesmann 1994; Sin et al. 2008). At a night-time oxygen low around 2 mg/L, AOB keep humming while NOB are already running short, and the gap widens as oxygen falls further (Kits et al. 2017).

The hobbyist signature is unmistakable: a tank that cycled cleanly months ago starts showing a persistent low-grade nitrite reading after one of these changes:

• Overstocking or heavy feeding — fish and decomposer respiration drain oxygen faster than the surface can replace it, especially overnight.
• Higher temperature — warmer water both demands more oxygen from metabolism and holds less of it.
• A heavily planted tank in the dark — plants stop making oxygen at night but keep respiring, pulling oxygen down sharply for a few hours each cycle, so NOB lose a window every night and steady-state nitrite settles above zero.
• Reduced surface movement — a clogged sponge filter, an oily surface film, or a flow pointed away from the surface, all cutting gas exchange.

The fix is the one experienced aquarists already reach for — more aeration or surface agitation — and the simulator agrees: improving gas exchange in any of these cases collapses the steady-state nitrite back toward zero without changing the nitrifier population at all. Because comammox sit between AOB and NOB in their oxygen demand, comammox-dominated mature tanks ride out moderate oxygen dips better than fresh AOB-and-NOB filters — one more reason an established planted tank is more forgiving of overnight oxygen swings than a bare fishless-cycled one.

Iron and pH: the hidden cycling bottleneck

The enzymes nitrifiers use to oxidise ammonia and nitrite are both iron-based — the ammonia monooxygenase carries a di-iron core, and the nitrite-oxidising enzyme is built around an iron-sulfur cluster — so nitrifiers carry an unusually heavy iron demand, a touch heavier than ordinary phototrophs (Wagner et al. 2002; Pearson et al. 2011). In most planted tanks this never matters, but it can quietly stall a cycle in the wrong conditions.

The trap springs in tanks with little dissolved organic matter and a high pH — bare-bottom setups on hard tap water, with no soil or driftwood. There, dissolved iron oxidises to insoluble rust within minutes once pH drifts above 8, and the oxidation accelerates steeply with pH: every unit of pH multiplies the rate roughly a hundredfold (Stumm & Morgan 1996). With no dissolved organic matter around to hold iron in a protected, usable form, the water's iron pins at a vanishingly low level and the nitrifiers sit chronically iron-starved through the whole cycling period. The model reproduces this: bare hard-water cycles drag out well past the textbook three-and-a-half weeks, and an aggressively ammonia-dosed hard-water tank can stall the nitrite spike outright at a few mg/L, with high pH holding iron demand above supply.

Real planted tanks rarely fall into this trap, because soil and driftwood release humic organics that keep iron chelated and available, and substrate iron oxides leak a slow supply. Keepers fishless-cycling in bare tanks, though, sometimes report exactly the slow or stuck behaviour the model predicts — and a small dose of chelated iron at the start of cycling noticeably shortens the lag.

Two well-documented cycling failures fall out of the same kinetic chain:

1. Iron lock-out plus ammonia self-poisoning. Without iron, AOB stall; with AOB stalled, ammonia keeps building; with ammonia high, free ammonia stays above the level that poisons NOB; nitrite plateaus and refuses to clear. The standard forum remedy — a big water change — works in the model too, because fresh tap water restores both iron and buffering: the same cycle that sticks indefinitely without water changes finishes in about two weeks with weekly ones.
2. The late-cycle pH crash. Nitrification steadily consumes a tank's carbonate buffer, and once the buffer is gone, pH falls sharply toward the point where dissolved CO2 takes over (around 5.6). AOB drop outside their tolerance band and the cycle stalls. Aquarists know this as the "second wall" of fishless cycling, or as one face of "old tank syndrome"; the model reproduces it for any unbuffered bare tank pushed through enough nitrification.

The fourth ammonia oxidiser: the archaea EcoSym leaves out

There is a whole second domain of life running the ammonia-to-nitrite step that the three guilds above leave out entirely: the ammonia-oxidising archaea (AOA). Archaea look like bacteria down a microscope but sit on a separate branch of life, and their version of the ammonia-oxidising enzyme, while related, is its own tool. AOA were only recognised as nitrifiers in 2005, and the surprise since has been how much of the world's nitrification they actually run — in most soils and across the open ocean they outnumber the ammonia-oxidising bacteria by one to two orders of magnitude (Könneke et al. 2005; Leininger et al. 2006).

Two of their traits matter to aquarists, and both point at the same corner of the map the model currently leaves blank:

• They feed at trace levels. AOA are extreme oligotrophs, with among the tightest ammonia affinities known — comparable to, or tighter than, comammox. Wherever ammonia is scarce, they compete hard.
• They work in acid. Because the enzyme's real substrate is dissolved ammonia gas rather than the ammonium ion, and that gas grows vanishingly scarce as pH drops, ordinary AOB stall out below about pH 6. AOA — and especially dedicated acid-loving lineages such as Nitrosotalea — keep oxidising ammonia down to roughly pH 4–4.5, and in genuinely acidic soils they carry essentially all of the nitrification (Lehtovirta-Morley et al. 2011; Gubry-Rangin et al. 2011).

This is the grain of truth behind a common observation: a soft-water or blackwater tank sitting at pH 5.5 still, eventually, stops accumulating ammonia. EcoSym does not model AOA, so its three bacterial guilds hand in a nitrification rate that falls to zero right around pH 6 (see why cycles stall on pH). That makes the simulator deliberately pessimistic for very acidic tanks: it shows ammonia processing arresting where a real acid tank would instead settle into a slow, low-capacity archaeal nitrification.

Three honest caveats keep this in proportion. Most of the hard evidence for acidophilic AOA is from soils rather than aquaria; even where they do run, the rate is genuinely sluggish; and in a planted acid tank, plant uptake usually strips ammonia faster than any nitrifier would anyway. So the missing pathway bites hardest in an unplanted, acidic, fish-in tank — precisely the setup where the model is currently least trustworthy about ammonia. Closing the gap — most likely by adding AOA as a fourth, acid-tolerant, oligotrophic guild alongside the existing three — is a known item on the model's roadmap.

Why nitrifiers are biofilm dwellers

Put the three advantages together — shade from light, shelter from grazers, and (for AOB) richer food at the surface — and it is obvious why aquarium nitrifiers are overwhelmingly sessile. The model reproduces the real-world pattern: free-floating AOB and NOB dwindle over the first weeks as light and grazing wear them down, while surface populations slowly build on glass, sand, and ceramic until the biological filter is established and the tank is finally safe to stock.

It is also what ranks every method of giving a new tank a head start. Moving a surface out of an established aquarium (filter media, gravel, a plant) moves the bacteria along with it, while moving a bucket of that tank's water carries a negligible nitrifier inoculum, because there were never many nitrifiers suspended in the water to begin with. Del'Duca et al. (2019) measured exactly that split in a tilapia recirculating system: biofilm bacterial abundance climbed continuously while water-column abundance did not change significantly. See Starting the Cycle for the full ranking and the evidence behind each method.

That same split decides what happens when the ammonia runs out. An attached biofilm survives starvation by idling; a suspended population does not. See Starved Biofilters and Dormancy.

Further reading

• Starved Biofilters and Dormancy — what a starved biofilter does, why an idle filter still works, and why NOB comes back last
• Heterotrophic Bacteria — the fast decomposers that recycle organic matter and compete with nitrifiers for oxygen
• Nitrogen Cycle — where nitrification sits in the full journey of nitrogen through the tank
• Starting the Cycle — what each way of seeding a new tank is actually worth, and why carried-over water is worth nothing
• Iron Cycle — why iron scarcity can throttle cycling, and how soil and chelators relieve it
• Anaerobic sediment microbes — the denitrifier and the other buried guilds that remove nitrogen from the system
• Bacteria overview — how the bacterial guilds divide the work
• Parameter Reference — every rate, affinity, and threshold behind this page, with citations

Key references

• Bartelme, R.P., McLellan, S.L. & Newton, R.J. (2017). Freshwater recirculating aquaculture system operations drive biofilter bacterial community shifts around a stable nitrifying consortium of ammonia-oxidizing archaea and comammox Nitrospira. Frontiers in Microbiology 8, 101.
• Daims, H. et al. (2015). Complete nitrification by Nitrospira bacteria. Nature 528, 504–509.
• Del'Duca, A. et al. (2019). Variability of the nitrifying bacteria in the biofilm and water column of a recirculating aquaculture system for tilapia (Oreochromis niloticus) production. Aquaculture Research 50, 2229–2238. (Biofilm bacterial abundance rises continuously while water-column abundance does not change significantly — the measurement behind "nitrifiers are on the surfaces, not in the water".)
• Gubry-Rangin, C. et al. (2011). Niche specialization of terrestrial archaeal ammonia oxidizers. PNAS 108, 21206–21211.
• Kits, K.D. et al. (2017). Kinetic analysis of a complete nitrifier reveals an oligotrophic lifestyle. Nature 549, 269–272.
• Könneke, M. et al. (2005). Isolation of an autotrophic ammonia-oxidizing marine archaeon. Nature 437, 543–546.
• Lehtovirta-Morley, L.E. et al. (2011). Cultivation of an obligate acidophilic ammonia oxidizer from a nitrifying acid soil. PNAS 108, 15892–15897.
• Leininger, S. et al. (2006). Archaea predominate among ammonia-oxidizing prokaryotes in soils. Nature 442, 806–809.
• Kindaichi, T., Ito, T. & Okabe, S. (2004). Ecophysiological interaction between nitrifying bacteria and heterotrophic bacteria in autotrophic nitrifying biofilms as determined by microautoradiography-fluorescence in situ hybridization. Applied and Environmental Microbiology 70, 1641–1650. (A biofilm fed ammonia and no organic carbon at all is still ~50 % nitrifiers and ~50 % heterotrophs by cell count; the heterotrophs live on nitrifier-derived organic matter.)
• Okabe, S., Satoh, H. & Watanabe, Y. (1999). In situ analysis of nitrifying biofilms as determined by in situ hybridization and the use of microelectrodes. Applied and Environmental Microbiology 65, 3182–3191. (Oxygen penetrates ~150 µm into a ~200 µm autotrophic nitrifying biofilm, with ammonium still in excess below that. The 300 µM figure is the ammonium in the medium during microelectrode profiling; the film itself was grown at 3.6 mM. Note that the paper attributes the halt in ammonia oxidation at ~75 µm to where the nitrifiers physically sit, not to oxygen running out — it is not, in its own terms, a demonstration that the film is oxygen-limited.)
• Pearson, A.J. et al. (2011). Trace-metal requirements of nitrifying bacteria. (Iron acquisition in nitrifiers.)
• Pinto, A.J. et al. (2016). Metagenomic evidence for the presence of comammox Nitrospira-like bacteria in a drinking water system. mSphere 1, e00054-15.
• Sauder, L.A. et al. (2017). Cultivation and characterization of Candidatus Nitrosocosmicus and comammox Nitrospira from a municipal wastewater system. ISME Journal 11, 1142–1157.
• Sin, G. et al. (2008). Modelling nitrite in nitrification processes: oxygen affinities of ammonia- and nitrite-oxidisers. Water Research 42, 1313–1324.
• Stumm, W. & Morgan, J.J. (1996). Aquatic Chemistry, 3rd ed. Wiley-Interscience.
• Wagner, M. et al. (2002). Microbial community composition and function in wastewater treatment plants. Antonie van Leeuwenhoek 81, 665–680.
• Wiesmann, U. (1994). Biological nitrogen removal from wastewater. Advances in Biochemical Engineering/Biotechnology 51, 113–154.