Nutrient Cycling

From 18 November to 31 December 1965, the forest on one small watershed at Hubbard Brook, New Hampshire, was cut, and herbicide kept anything from growing back. The stream draining it soon carried far more nitrate than the stream of an uncut watershed nearby. The nitrogen had been there all along, cycling between soil and trees; removing the trees changed where it went. To see that, you keep accounts for an element the way a bank keeps accounts for money.

Pools and fluxes: a budget is bookkeeping with a boundary

A pool (or stock) is the amount of an element held in one compartment at one time: nitrogen in soil organic matter, carbon in wood, phosphorus in lake sediment. For an ecosystem it is given as mass per area, such as kg N per hectare. A flux is a transfer between pools, or across the edge of the system, given as mass per area per time, such as kg N ha^-1 yr^-1. A pool is a bank balance; a flux is a deposit or a withdrawal.

Before adding anything, draw the boundary (what counts as inside) and choose the interval (a year, a growing season). Then one rule holds:

change in stock = inputs - outputs (for a stated boundary and interval)

residence time = stock / flux through the pool

Residence time is the average time an atom spends in a pool when inflow and outflow balance. This is the mass balance from energy flow, except that energy leaves as heat while atoms are reused.

Worked problem. Take an invented forested watershed; the numbers are illustrative. All fluxes are in kg N ha^-1 yr^-1, and the boundary runs from the canopy down to bedrock. Assume rates hold steady through the year and that water, with the nitrogen dissolved in it, leaves only by the stream.

Inputs: atmospheric deposition 8, biological fixation 6. Outputs: stream export of dissolved N 3, gas loss by denitrification 2.

1. Net change. Inputs = 8 + 6 = 14. Outputs = 3 + 2 = 5. Change in stored N = 14 - 5 = +9 kg N ha^-1 yr^-1. The watershed is accumulating nitrogen, in wood and soil.
2. Internal cycling. The soil organic N pool is 4,500 kg N per hectare, and gross mineralisation (decomposers releasing organic N as ammonium) is 90 kg N ha^-1 yr^-1. Residence time = 4,500 / 90 = 50 years. That internal flux is about six times the external inputs (90 / 14 = 6.4), yet it never appeared in step 1. Summed over the whole watershed, nitrogen leaving the soil pool enters the plant pool: an output of one pool is an input of another, and the two cancel. Cancelling is an accounting result, not evidence that internal turnover is small; here it is the largest flux.
3. Disturbance. Now cut the forest and suppress regrowth. Plant uptake stops, but mineralisation and nitrification continue, and nitrate, a mobile anion that soil holds weakly, washes out. Suppose stream export rises to 40 while the other terms stay fixed. Outputs = 40 + 2 = 42, and change = 14 - 42 = -28 kg N ha^-1 yr^-1, a net loss.

The invented rise in step 3 shows the direction of the real Hubbard Brook result, not its size. Gene Likens, F. Herbert Bormann and colleagues reported it in Ecological Monographs in 1970: from the cut, herbicide-treated watershed, stream losses of nitrate and of cations such as calcium rose well above those of an uncut reference watershed. Their small-watershed approach relied on bedrock that holds little water, so a weir on the stream measured nearly everything leaving.

The carbon cycle: fast exchanges laid over slow stores

The large non-rock pools of carbon are the ocean, the atmosphere, living plants, soil and fossil fuels. Two biological fluxes dominate the fast cycle. Photosynthesis by plants and phytoplankton moves CO2 into organic matter (the Calvin cycle has the chemistry). Respiration by plants, animals and above all decomposers returns it as CO2 (oxidative phosphorylation has the mechanism). In the Northern Hemisphere you can see them out of phase: atmospheric CO2 falls each summer, when photosynthesis outpaces respiration, and rises each winter.

The ocean exchanges CO2 with the atmosphere across its surface. Dissolved CO2 forms bicarbonate and carbonate, so the ocean holds far more carbon than the air. Fossil carbon is organic matter buried over millions of years. Left alone it is a pool with almost no outflow; burning it opens a large flux from a slow pool into the fast ones.

Soil carbon shows why budgets need long records. In a soil-warming experiment in a mid-latitude hardwood forest, Melillo and colleagues reported in 2017 that over 26 years of heating, the warmed soils went through phases of carbon loss alternating with phases of no detectable loss. A short record would have misled.

The nitrogen cycle: microbes run most of the transformations

The atmosphere is about 78 percent N2, but its triple bond puts it out of reach of most organisms. The cycle is a series of redox steps, nearly all carried out by microbes:

• Fixation reduces N2 to ammonia, using the enzyme nitrogenase. Some bacteria and archaea do it, free-living or in symbioses such as rhizobia in legume root nodules. Lightning and fertiliser manufacture also fix nitrogen.
• Ammonification releases ammonium from organic nitrogen as decomposers break down dead matter (amino-acid nitrogen is covered in protein turnover and amino acid catabolism).
• Nitrification oxidises ammonia through nitrite to nitrate and requires oxygen. The textbook version uses two groups: ammonia oxidisers (bacteria, and archaea such as the marine isolate Könneke and colleagues reported in 2005) and nitrite oxidisers.
• Denitrification reduces nitrate or nitrite, in low-oxygen soils and sediments, to gases that can include NO, N2O and N2.
• Leaching carries nitrate in drainage water to streams and rivers.

Two discoveries have changed that list. One is anammox (anaerobic ammonium oxidation), which joins ammonium and nitrite into N2 without oxygen:

NH4+ + NO2- -> N2 + 2 H2O

In 2003 Marcel Kuypers and colleagues showed that anammox bacteria in the Black Sea water column just below the oxic zone carry out this reaction, so denitrification is not the only biological route from fixed nitrogen back to N2. The other is comammox. In 2015 two groups, van Kessel and colleagues and Daims and colleagues, each reported in Nature enriched Nitrospira that perform complete ammonia oxidation to nitrate within one organism. In 2017 Kits and colleagues isolated one of them, Nitrospira inopinata, in pure culture and showed that it oxidised ammonia to nitrate stoichiometrically. Nitrification names two chemical steps; how many organisms perform them is a separate question.

Fixation has a new case too: in 2024 Coale and colleagues interpreted UCYN-A, a nitrogen-fixing cyanobacterium inside one marine alga, Braarudosphaera bigelowii, as an early-stage nitrogen-fixing organelle.

Phosphorus and limitation: the Experimental Lakes Area

Phosphorus has no major gas phase. It enters ecosystems mostly by weathering of phosphate-bearing rock, and much of it ends in marine sediments; only uplift of those sediments over geological time returns it to land. In soil, phosphate binds to mineral surfaces (sorption) and is released again (desorption), microbes store some in their biomass and use phosphatases to free phosphate from organic matter, and plants take up soluble orthophosphate. Every nucleotide carries it (nucleotide metabolism). No gas phase does not mean it never moves through the air: Barkley and colleagues (2019), sampling aerosols at Cayenne, French Guiana, and modelling their transport, reported that African aerosols carry phosphorus to the Amazon and tropical Atlantic.

A nutrient is limiting when adding it, and not the others, increases production. The direct test is to add it to a whole ecosystem. A PNAS profile records that David Schindler was invited in 1968 to head an experimental-lakes project in northwestern Ontario, where small lakes were set aside for whole-lake experiments. Lake 226 was divided into two basins. Both received nitrate and sucrose (a carbon source), and one also received phosphorus. In his 1974 Science paper, only the phosphorus basin became highly eutrophic (rich in nutrients and dense with algae); the other stayed as before.

From 1990 to 2005 Lake 227 received phosphorus fertiliser but no nitrogen fertiliser. Schindler and colleagues, analysing 37 years of it in 2008, reported that it stayed highly eutrophic: nitrogen fixation sustained phytoplankton biomass in proportion to the phosphorus supply.

Phosphorus limitation on land is not only tropical. Hou and colleagues’ 2020 synthesis found significant phosphorus limitation in 46.2 percent of 652 field phosphorus-addition experiments in natural ecosystems and croplands, including outside the tropics. In the 436 experiments in natural terrestrial ecosystems, phosphorus addition raised aboveground plant production by a mean of 34.9 percent.

What a budget can and cannot show

A budget that balances is not proof that each term is right. Worked check. Global Carbon Budget 2024 estimated, for 2023, emissions E = 11.1 ± 0.9 GtC yr^-1, atmospheric accumulation A = 5.9 ± 0.2, ocean sink O = 2.9 ± 0.4 and land sink L = 2.3 ± 1.0 (± one standard deviation). The residual is:

B = E - A - O - L = 11.1 - 5.9 - 2.9 - 2.3 = 0.0 GtC yr^-1

Check: 5.9 + 2.9 + 2.3 = 11.1. The zero comes from rounding; the unrounded source figure is about -0.02. The land sink is uncertain by nearly half its value, so errors in two terms could cancel and still give zero.

Budgets usually fail to close, for four reasons:

• Unmeasured fluxes. N2 lost by denitrification is hard to detect against an atmosphere that is mostly N2. Rock can be an input: Morford and colleagues estimated 1.6 to 10.7 kg N ha^-1 yr^-1 from weathering at three Klamath Mountains forest sites on nitrogen-rich mica schist.
• Sampling frequency. Li and colleagues (2019), subsampling a daily watershed record, found that the uncertainty of estimated nutrient export depended on how often samples were taken.
• Tracer bias. Mohr and colleagues (2010) found that adding 15N2 as a gas bubble, which does not fully equilibrate with water, underestimated fixation by 60 percent in a 12-hour incubation of Crocosphaera watsonii cultures.
• Compensating errors, as in the carbon check above.

Design problem

A lake manager wants to know whether to control nitrogen or phosphorus to reduce algal blooms. Design the test, and say what it can and cannot show.