Reports

Phosphorus Cycling in a Northern Hardwood Forest:Biological and Chemical Control

Abstract. Phosphorus is tightly conserved within the northern hardwood forestecosystems at Hubbard Brook, New Hampshire. Detailed analyses ofthe soil systemindicate that biological and geochemical processes, stratified within the profile,regulate phosphorus retention.

Of the six major plant nutrients, phos-phorus is typically one of the most tight-ly conserved within forested ecosys-

tems. Hydrologic exports of dissolvedinorganic phosphorus (DIP) with streamwater and deep seepage rarely exceed0.05 kg ha-' year-', whereas losses ofdissolved calcium, magnesium, potassi-um, nitrogen, and sulfur generally fall inthe range of 2 to 20 kg ha-l year-l (1).Forest plants, their mycorrhizal associ-ates, and other microorganisms can con-

tribute to phosphorus retention throughclose coupling of decomposition and up-

take processes. Such coupling has beenreported for moist tropical forests wheredense surficial root mats absorb largeportions of the nutrients released fromdecomposing litter (2). Phosphorus can

also be retained in forest soils throughgeochemical fixation on calcium, iron,and aluminum minerals (3). Thus, a thor-ough understanding of phosphorus cy-

cling and retention in forest ecosystemsrequires detailed study of the actions andinteractions of biological and geochemi-cal control mechanisms.For the past 1-5 years, scientists at the

Hubbard Brook Experimental Forest incentral New Hampshire have been con-

ducting intensive long-term studies offorest biogeochemistry, using the smallwatershed approach (1). The results oftheir work indicate that the phosphoruscycle within this 55- to 65-year-old north-ern hardwood forest ecosystem is bothdynamic and exceedingly tight. Plantstake up an estimated 12.5 kg of phospho-rus per hectare per year from the soil.About 1.5 kg ha-' year-' are stored inaccumulating biomass, and 11.0 kg ha-lyear-' are returned to the soil withaboveground and belowground litter,throughfall, and root exudates. Morethan 10.0 kg ha-' year-' of organic phos-27 JANUARY 1984

phorus are mineralized through decom-position processes in the soil. Weather-ing of primary soil minerals releases anadditional 1.5 to 1.8 kg of phosphorusper hectare to soil solutions each year(4).

Despite these large intrasystem fluxes,only 0.007 kg ha-' year-' of DIP are lostfrom these forests in stream water (5).Concentrations of DIP in HubbardBrook streams remain at < 1 ±g liter-lthroughout the year. These exports andconcentrations are among the lowest re-ported in the literature, and, in the ab-sence of other significant vectors of out-put, they indicate the strong degree towhich phosphorus is conserved withinthis forest ecosystem. Questions re-mained, however, concerning the degreeto which biological versus geochemicalprocesses contribute to phosphorus re-tention, and in 1977 we initiated studiesto address this problem. Some resultsare reported here.Hubbard Brook soils are generally

acid, well-drained spodosols (haplorth-ods) of sandy loam texture. Mature pro-files, undisturbed by windthrow, are typ-ically characterized by five distinct hori-zons [AO or forest floor, A2, B21h, B2ir,and Cx (6)], which differ markedly intheir chemical and physical characteris-

Table 1. Select characteristics of HubbardBrook soils. Values are the means from tenprofiles.

Soil Depth Loss on pHhorizon (cm) ignition (water)

AO 0 to 9 54.1 3.42A2 9 to 15 2.5 3.88B21h 15 to 18 12.6 3.82B2ir 18 to 80 7.1 4.63Cx > 80 0.9 5.16

tics (Table 1). In this study, the A and Bhorizons were extensively sampled todetermine the abundances and distribu-tions of biological agents (nonwoodyroots, bacteria, and fungi) and geochemi-cal agents (iron and aluminum sesquiox-ides) that typically govern phosphorusmovements and solubilities in acid soils(7). The Cx horizon, a discontinuousfragipan, is impermeable to deep seepageand root growth, and, as such, does notdirectly influence phosphorus biogeo-chemistry.We surveyed the populations of roots,

bacteria, and fungi during the 1977 grow-ing season by first extracting intact soilcores from specified depths within theprofile. Fine (5 0.6 mm in diameter)nonwoody roots were carefully removedfrom cores, and their lengths, diameters,numbers of intact root tips, and ovendry weights were measured (7). Bacteriaand fungi were inventoried by direct mi-croscopic counts pf soil homogenates (7,8).Peak concentrations of fine roots oc-

curred in the surface 2 cm of the forestfloor. Abundances declined with depthin a near exponential fashion with aslight discontinuity in the A2 and B21hhorizons (Fig. 1). We calculate that thereare, during the growing season, a totalof 24.7 ± 3.2 km, 24.6 ± 3.2 mi2, and707 ± 103 g of fine root length, surfacearea, and biomass, respectively, beneatheach square meter of soil surface. Roottips total (3.8 ± 0.6) x 106 per squaremeter. Forty to 50 percent of these totalsoccur in the top 10 cm of the profile.

Soil microbes were similarly distribut-ed with peak concentrations of (10.4 +

2.2) x 109 bacteria per cubic centimeterof soil and 402 ± 299 m offungal hyphaeper cubic centimeter occurring in the AOhorizon. Bacterial abundance declined to(1.1 ± 0.4) x 109 per cubic centimeter inthe A2, rose to (3.0 ± 1.0) x 109 in theB21h, and then fell with depth throughthe B2ir where densities averaged(0.8 ± 0.3) x 109. Fungal hyphae fol-lowed a similar decline with depth:32 ± 19 m cm-3 in the A2, 100 ± 31 inthe B21h, and 40 ± 26 in the B2ir. Ap-proximately 60 percent of the soil bacte-ria and fungi occurred in the upper 10 cmof the profile.Many factors, including moisture and

oxygen availability, may shape distribu-tions of roots and microbes, but peaks inthe AO and B21h suggest the importanceof annual litterfall and soil organic mat-ter as sources of carbon and nutrientsfor growth (Table 1). Codistributions ofroots, microbes, and organic matter, inturn, suggest localized transfers of nutri-ents between living and dead biomass, a

391

on

Oct

ober

4, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

4, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

4, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

20

- 40E

600

100

/

0 50 5 10 15Root length (cm/cm 3)

20

Fig. 1. The distributions of fine roots (< 0.6mm in diameter) in Hubbard Brook soils.Mean root densities are expressed as rootlength (in centimeters) per cubic centimeter ofsoil. Profiles for fine root surface area, bio-mass, and root tip density were similar.

key process for biological regulation ofphosphorus cycling and retention.We examined chemical regulation of

phosphorus retention through a series ofequilibrium experiments designed tomeasure (i) the capacities of soils takenfrom designated horizons to sorb andrelease dissolved inorganic phosphoruswhen shaken for 24 hours with KH2PO4solutions ranging in concentration from 0to 2 mM (7, 9) and (ii) the equilibriumphosphorus concentrations maintainedwhen sorption and release processescame into balance (10). These analysesshowed that AO and A2 horizon soilswere relatively inert with respect tophosphorus geochemistry. Composedprimarily of organic matter and siliceousresidues of heavily weathered minerals,these surface soils chemically removed

gram of soil. AO horizon soils releasedsignificant amounts of phosphorus to dis-tilled water, and they tended to weaklybuffer solution phosphorus concentra-tions at 2000 to 7000 ,ug of DIP per liter.A2 horizon soils tended to release littlephosphorus to solution and weakly main-tained equilibrium phosphorus concen-trations at 109 ± 42 jig of DIP per liter.

In contrast, soil from B21h and B2irhorizons chemically fixed large amountsof dissolved phosphorus (their sorptioncapacities for phosphorus exceeded 1.5mg g-') and released only trace amountsof phosphorus when equilibrated withdistilled water. As such, they strictlybuffered equilibrium phosphorus con-centrations at 6 ± 2 and 1 ± 0 ,ug of DIPper liter, respectively. The latter result isimportant. B2ir horizon soils regulatesolution phosphorus concentrations atlevels also found in first-order streamsdraining the forest ecosystem.

In acid soils, iron and aluminum ses-quioxides typically dominate phospho-rus fixation through replacement of hy-droxyl groups by phosphate (3). Thismechanism appears active in HubbardBrook spodosols. Phosphorus sorptioncapacities of ten soil samples from eachof the five horizons were highly correlat-ed (r2 = 0.95; P s 0.001) with concen-trations offree, surface-reactive iron andaluminum measured by HCI extraction(7, 11). Correlations with concentrationsof extractable calcium, magnesium, andmanganese, other metals that can chemi-cally fix phosphate, were insignificant.

0

000 Blo. -0-

000000000 00000 00000 0000000000000 00oa0o0o00

*40 00 0000000000000 )oOOoOo

a 0000 000E 0

60 J00000000g 000 Geo.60 0

0 0 000080 ogogog0ogg

0 50Peak concentration (%)

100

Fig. 3. Comparison of the relative distribu-tions of biological agents (fine roots, bacteria,and fungi) versus geochemical agents (sur-face-reactive iron and aluminum sesquiox-ides) that contribute to phosphorus retentionin Hubbard Brook soils.

2). Concentrations of free iron peaked inthe B21h and then diminished through theremainder of the profile. Free aluminumwas more deeply distributed, with con-centrations peaking in the upper B2ir.

Soils in the B horizons can also sorblarge amounts of organic phosphorusthrough coprecipitation of dissolved or-ganic matter with iron (12). Equilibrationexperiments showed B2ir soils to be ca-pable of sorbing more than 0.3 mg ofdissolved organic carbon (DOC) pergram of soil and of maintaining equilibri-um DOC concentrations at 2.0 to 3.0 mgliter-' (13). Phosphorus associated withthis organic carbon is, of course, fixed intandem.

only small amounts of phosphorus from Three representative profiles showed The above surveys indicate that Hub-solution. Their sorption capacities aver- low concentrations of free iron and alu- bard Brook soils are highly stratifiedaged less than 0.05 mg of phosphorus per minum in the AG and A2 horizons (Fig. with respect to phosphorus biogeochem-

istry (Fig. 3). The forest floor, rich inroots, bacteria, and fungi but poor in free

HCI-extractable Fe and Al (mg/g) iron and aluminum, is a zone of poten-0 10 20 0 10 20 0 10 20' tially high biological control but minimal

o a, I I ' a I I II I geochemical control over phosphoruspO Fe l % _o_ o Fo. ,o FeO | movements. Underlying B horizons,

20L --°E3% / I L --° I L ---, Iwith high concentrations of free metals20 -no | O $ |and relatively low densities of roots and

00,' IlAl microorganisms, are just the opposite.

o oo 0 oo We used this stratification to distin-40 / / Al A l|- guish the relative roles of biological and

E% 0 ole I,' I0 chemical processes in shaping phospho-m' co en rus movements within and through the

s60 o/ soil system. Soil solutions were collectedcl ~~~~~~00a

a during two growing seasons from 12 ly-IDt O simeters placed in the lower A2 horizon

80 _ ( and from six lysimeters placed in themiddle of the B2ir. Solutions were ana-

lyzed for their phosphorus contents, and100 6 results were used to examine phospho-

a rus flux through the soil profile. Solu-11%\ tions collected from the lower A2 hori-

120 zon contained only 24 18 ,ug of totalFig. 2. Distributions of HCl-extractable iron and aluminum in three soil profiles at Hubbard phosphorus per liter and 1 ± 1 ,ug of DIPBrook. per liter. These concentrations were far

392 SCIENCE, VOL. 223

below those chemically maintained by Ahorizon soils in laboratory equilibra-tions, an indication of regulation throughbiological uptake. Using computer-basedestimates of hydrologic flow across theA horizon-B horizon interface (7, 14)and assuming that the phosphorus con-tents of these solutions did not varyseasonally, we calculated that 0.23 kgper hectare of total phosphorus leachfrom the A horizons to the B horizonsannually. This amount is only 3 percentof the 8.35 kg ha- year-l estimated tobe released within the A horizons (15),and it indicates strict biological retentionof phosphorus within surface soils. Solu-tions collected from lysimeters placed inthe middle of the B2ir invariably con-tained < 1 p.g of total phosphorus perliter, substantiating the hypothesis ofgeochemical control of dissolved inor-ganic and organic phosphorus in subsoilsand eventually stream waters.Our results indicate dual and stratified

regulation of phosphorus cycling and re-tention at Hubbard Brook. Phosphorusis biologically conserved within this for-est community by close coupling of bio-logical decomposition and uptake pro-cesses in the surface soils. Underlyingiron- and aluminum-rich B horizons, inturn, function as massive geochemicalbuffers that regulate the constant andlow-level losses of dissolved phosphoruswith stream water.These findings are signifi

plaining changes in phosphcchemistry after forest distuiperiments at Hubbard Brookonstrated massive increaserwater exports of calcium,and nitrate-nitrogen after clLosses were attributed to dbiological retention mechanias to acidification of caticsites in the profile (6, 16). Stlosses of dissolved inorganicdid not increase after clear-epresumably because chemicEin the B horizons continuedWith disruption of biologicalsurface horizons, however,that large amounts of phosph(ed from the forest floor toB horizons. Recovery ofstored in those subsoils may r

important aspect of forest rejdisturbance.

Native Plants, Inc.,Salt Lake City, Utah 84108

F. HG

School of Forestry and Envi,Studies, Yale University,New Haven, Connecticut 06.

27 JANUARY 1984

References and Notes1. G. E. Likens et al., Biogeochemistry of a For-

ested Ecosystem (Springer-Verlag, New York,1977).

2. N. Stark and C. F. Jordan, Ecology 59, 434(1978).

3. R. L. Parfitt and R. J. Atkinson, Nature (Lon-don) 264, 740 (1976); L. J. Evans and G. W.Smillie, Ir. J. Agric. Res. 15, 65 (1976).

4. Values for the phosphorus fluxes are as pub-lished (1), with modifications based on revisedestimates of root turnover and of the amounts oforganic and inorganic paosphorus in HubbardBrook soils (7).

5. J. L. Meyer and G. E. Likens, Ecology 60, 1255(1979).

6. Soil Survey Staff, Soil Taxonomy (AgricultureHandbook No. 436, Government Printing Of-fice, Washington, D.C., 1975).

7. T. E. Wood, thesis, Yale University (1980).8. J. E. Hobbie, R. J. Daley, S. Jasper, Appl.

Environ. Microbiol. 33, 1225 (1977); E. A. Pauland R. L. Johnson, ibid. 34, 263 (1977).

9. S. R. Olsen and F. S. Watanabe, Soil Sci. Soc.Am. Proc. 21, 144 (1957).

10. A. W. Taylor and H. M. Kunishi, J. Agric. FoodChem. 19, 824 (1971).

11. Soils were extracted for 1 hour with 12N HCI.

We analyzed extracts for iron and aluminum,using methods in M. L. Jackson, [Soil Chemi-cal Analyses (Prentice-Hall, Englewood Cliffs,N.J., 1958)] and in W. K. Dougan and A. L.Wilson [Analyst 99, 413 (1974)].

12. M. Schnitzer, Soil Sci. Soc. Am. Proc. 33, 75(1969). '

13. W. H. McDowell and T. E. Wood, Soil Sci., inpress.

14. C. A. Federer and D. Lash, Research Report 19(Water Resources Research Center, Durham,N.H., 1978).

15. The estimate of annual phosphorus release with-in the AO plus A2 horizons (8.35 kg ha-' year')was calculated as the sum of phosphorus inputsto these strata with precipitation (0.04), through-fall and stemflow (0.56), aboveground and be-lowground litter (7.02), root exudates (0.10), andmineral weathering (0.90), minus the net annualaccretion of phosphorus in the A horizons withaccumulating organic matter (0.27). Completedocumentation is given by Wood (7).

16. G. E. Likens, F. H. Bormann, N. M. Johnson,D. W. Fisher, R. S. Pierce, Ecol. Monogr. 40, 23(1970).

17. J. E. Hobbie and G. E. Likens, Limnol.Oceanog. 13, 734 (1973).

8 August 1983; accepted I November 1983

Microwave Measurements of Carbon Monoxide on Titan

Abstract. The ratio oftheflux density of Titan was measured in two 200-megahertzbands, one centered on the (1-0) rotation line ofcarbon monoxide at 115.3 gigahertzand the other 2600 megahertz lower. The measurements were made with a complex-correlation technique on the new millimeter-wavelength interferometer at the OwensValley Radio Observatory, Big Pine, California. The excess flux in the carbonmonoxide band is interpreted as a strong detection ofcarbon monoxide and a mixingratio, assumed constant, of 6 x 10-5. The brightness temperature of Titan at 112.6gigahertz is 69 ± 10 kelvins, consistent with atmospheric emission from just belowthe tropopause.

Until recently, the known constituentsicant in ex- of the Titan atmosphere were CH4,)rus biogeo- C2H6, C2H2, and CH3D. Experiments on

rbance. Ex- Voyager during its flyby of Titan estab-c have dem- lished that N2 was the primary constitu-s in stream ent, with no more than 3 percent CH4potassium, (1). Numerous other species were de-

lear-cutting. tected, all of which could be products ofLisruption of the parent compounds N2 and CH4. Weisms as well began a series of measurements in early)n-exchange 1982 to detect the first rotational transi-tream water tion of CO at 115.271 GHz, since thatphosphorus line is very strong in the terrestrial plan-cutting (17), ets and the detection of an oxygen com-

al processes pound in Titan would be extremely im-to function. portant (2). Concurrently, Samuelson etcontrols in al. (3) interpreted an emission feature init is likely the Infrared Interferometer Spectrome-orus migrat- ter (IRIS) spectrum at 667 cm-' as C02,sinks in the and recently prepared a paper in whichphosphorus they predicted a constant CO mixingrepresent an ratio of 1.1 x 10-4 based on their photo-growth after chemical model and a CO2 mixing ratio

of 1.5 x 10-9. In their model the CO/TIM WOOD CO2 ratio is very large because of the

low abundance of H20, which is an

important source of OH, a critical mole-. BORMANN cule in the conversion of CO into CO2.. K. VOIGT Lutz et al. (2) detected CO in Titan withronmental Earth-based measurements in the (3-0)

rotational-vibration bands near 6350511 cm-' and interpreted their measurement

with a constant mixing ratio of 6 x 10-5with an uncertainty of a factor of 3. Thisuncertainty arises from a multitude ofcauses, including the possible presenceof aerosol, cloud, and haze layers inTitan's atmosphere, which would seri-ously influence their interpretations. Themicrowave lines offer a more directmeans of measuring the abundance,since the emission is only affected bygaseous absorption and confusion fromother chemical species is very unlikely.Any liquid droplets in clouds wouldprobably be nonpolar and, consequently,would not be significant absorbers. Theprimary problem is one of signal-to-noiseratio.

Titan was observed on 7 and 8 May1983 by use of the millimeter-wave-length, two-element interferometer atthe Owens Valley Radio Observatory,Big Pine, California. The interferometerconsisted of two 10.4-m telescopes (4)positioned to give a baseline of 50 meast-west. At a nominal wavelength of2.6 mm the fringe spacings on the sky inthe direction toward the Saturn systemranged from about 10.6 to 21.4 arc sec-onds. The interferometer fringes werecontinuously positioned on Titan withcomputer-controlled phase tracking, us-ing a precise Titan ephemeris (5). During

393