Nitrogen Saturation in TemperateForest Ecosystems

Hypotheses revisited

John Aber, William McDowell, Knute Nadelhoffer, Alison Magill, Glenn Berntson, MarkKamakea, Steven McNulty, William Currie, Lindsey Rustad, and Ivan Fernandez

N itrogen emissions to the at-mosphere due to human ac-e tivity remain elevated in in-

dustrialized regions of the world andare accelerating in many developingregions (Galloway 1995). Althoughthe deposition of sulfur has beenreduced over much of the UnitedStates and Europe by aggressive en-vironmental protection policies, cur-rent nitrogen deposition reductiontargets in the US are modest. Nitro-gen deposition remains relativelyconstant in the northeastern UnitedStates and is increasing in the South-east and the West (Fenn et al. in press).

The US acid deposition effects

John Aber is a professor, Alison Magi11 i sa laboratory supervisor, Glenn Berntson isa research assistant professor, and MarkKamakea is a graduate student at theComplex Systems Research Center, Uni-versity of New Hampshire, Durham, NH03824. William McDowell is a professorin the Department of Natural Resources,University of New Hampshire, Durham,NH 03824. Knute Nadelhoffer is a seniorscientist at the Ecosystem Center , MarineBiological Laboratory, Woods Hole, MA02543. Steven McNulty is director of theSouthern Global Change program, USDAForest Service, Raleigh, NC 27606. Will-iam Currie is an assistant professor at theAppalachian Laboratory, Center for Envi-ronmental Science, University of Mary-land, Frostburg, MD 21532. LindseyRustad is a research forester at the North-eastern Research Station, USDA ForestService, Durham, NH 03824. IvanFernandez is a professor in the Depart-ment of Applied Ecology and Environ-mental Sciences, University of Maine,Orono, ME 04469. 0 1998 AmericanInst i tute of Biological Sciences .

November 1998

Recent research raisesquestions on the

processes of N retentionin soils, and how much

protection theseprocesses offer to forestand stream ecosystems

program in the 1980s (the NationalAtmospheric Precipitation Assess-ment Program, or NAPAP) fundedresearch mostly on the effects of sul-fur deposition rather than on thoseassociated with nitrogen. Since thecompletion of this program and thepassage of the 1990 Clean Air ActAmendments, the US regulatory com-munity has not supported substan-tial additional research on acidicdeposition. Consequently, the po-tential for nitrogen deposit ion tocontribute to soil and surface wateracidification remains unresolved.Unlike the European Community,which has pursued an active andwell-coordinated international pro-gram on nitrogen deposition effects(the NITREX program; Wright andvan Breeman 1995, Wright andRasmussen 1998) and on criticalloads for nitrogen (Nilsson andGrennfelt 1988, Henricksen et al.1992, Warfvinge and Sverdrup 1992,Wright and Rasmussen 1998), re-search on this topic in the UnitedStates remains scattered and piece-

meal. Policy and regulatory activityhave also, until very recently, beenvirtually nonexistent.

Despite this lack of support anddirection, the US scientific commu-nity has continued to express con-cern over the long-term effects ofnitrogen deposition on forests, grass-lands, streams, and estuaries (e.g.,Aber et al. 1995, Wedin and Tilman1996, Asner et al. 1997, Vitousek etal. 1997, Fenn et al. in press). Theseconcerns cover the deposition notonly of oxides of nitrogen (NO, andespecially nitrate, NO,-), which origi-nate mainly from fossil fuel combus-tion, but also of ammonium (NH.,+),which originates from the produc-tion and use of fertilizers in agricul-ture. In parts of Europe and much ofAsia, deposition of ammonium ex-ceeds that of nitrate, sometimes byseveralfold (Galloway 1995).

In 1989, we published a review inBioScience of the known effects ofnitrogen deposit ion on temperateforest ecosystems in which we setforth a series of hypotheses aboutthe long-term consequences of con-tinuously elevated nitrogen inputs(Aber et al. 1989). We stressed thepotential for nitrogen additions tolead to nitrate and aluminum mobilityin soils, causing soil and stream acidi-fication, nutrient:nitrogen imbalancesin trees, and forest decline. The inte-grated set of hypotheses (Figure 1)suggested that responses to nitrogendeposition would not be linear andwould therefore not be captured insimple dose-response funct ions.Rather, we expected them to behighly nonlinear, with critical thresh-

9 2 1

Figure 1. Initial set ofhypotheses describingthe integrated responseof ni t rogen- l imited tem-perate forests to chronicnitrogen addit ions. Thetwo graphs show pre-dicted responses forsoils (top) and plants(bottom). The stages ofresponse are drawnfrom Smith (1974) andBormann (1982), assummarized in Aber etal. (1989). Stage 0 ispretreatment and as-sumes strong nitrogenlimitations on growth.Stage 1 is characterizedby high nitrogen reten-tion and a fertilizer ef-fect of added nitrogenon tree growth. Stage 2includes the inductionof nitrification andsome nitrate leaching,al though growth is s t i l lhigh. In Stage 3, treegrowth declines, and ni-trification and nitrateloss continue to in-crease, as the fractionof mineralized ammo-nium that nitrifies in-creases. Figure redrawnfromAberetaL(1989).

old points. Centralto most responseswas the induction of,or increase in, net ni-trification.

A d d i t i o n s Soturotion DehlineBegin

Stage 0 I 2 3

Faliar N Concentration

Fine Root Mass

Stage 0

,

A d d i t i o n sBeain

1 8

S a t u r a t i o n D e c l i n e

We used these hypotheses to framea series of field experiments, similarto the NITREX studies in Europe,and one extensive survey in forestecosystems across New York andNew England. These recent studiesallow a more complete summary ofthe process that has come to be callednitrogen saturation. Although thisterm has been defined in many ways(Agren and Bossata 1988, Aber et al.1989, Stoddard 1994), all definitionsinclude increases in nitrogen avail-ability over time that result in thealleviation of nitrogen limitations onrates of biological function and in-creases in nitrate mobility in soils.

gen saturation. Finally, we examinethe evidence for three different gen-eral mechanisms by which added ni-trogen may be retained in forest eco-systems and offer a novel hypothesisas to how this retention may occur.

Theory behind theoriginal hypotheses

In this article, we review the theorybehind the initial 1989 hypotheses.We then combine results derived fromtesting these hypotheses with resultsfrom other experiments to presentour current understanding of nitro-

Our initial review of the literature(see Aber et al. 1989 for references)pinpointed the production and mo-bility of nitrate as key processes con-trolling ecosystem responses to ni-trogen deposition. Predicted changesin stream and sail chemistry, as wellas plant responses, all followed fromnitrate dynamics. From previouslypublished studies, we derived thefollowing general understanding ofprocesses affecting nitrogen cyclingand nitrate production.

Under background or low nitro-

From this general pattern of car-bon and nitrogen limitations on plantand microbial function, we hypoth-esized that the addition of inorganicnitrogen to temperate forests in min-eral form would result, at least ini-tially, in increased nitrogen uptakeand growth by plants, and that therewould be little change in the immo-bilization of nitrogen by microbesdue to preexisting energetic limita-tions (Figure 2a). Rather, most of thenitrogen added to these systemswould be taken up by the nitrogen-limited organisms-the plants-andrelatively little by soil microbes. Theprocess of nitrogen saturation, bywhich forest ecosystems show de-creased retention efficiency of addednitrogen, should then be controlledmainl? by plant uptake, rather thanby sol1 microbial uptake, with thesecondary process of microbial nitri-fication increasing as nitrogen avail-ability begins to saturate plant de-mand. This logic led directly to theinitial overarching hypothesis (Figure1) that plant processes would changefirst in response to nitrogen additions,

922 BioScience Vol. 48 No. 11

2 3

.

gen deposition conditions, plantgrowth-in temperate forests is-mostfrequently limited by nitrogen avail-ability, while the metabolism of free-living soil microbes is most generallylimited by carbon (or energy). Theseessential constraints on the bio-geochemistry of forested ecosystemsare self-perpetuating in that plantsincorporate available nitrogen re-leased by microbial decompositioninto carbon compounds that are en-ergetically and structurally difficultto decompose. The partial decompo-sition of fresh plant tissues shed aslitter results in both a further con-centration of nitrogen in the result-ing organic matter (humus) and afurther degradation in the decom-posability of this residue. The rate-limiting step in the nitrogen cycle oftemperate forest ecosystems is thusthe humus decay process, and thelargest pool of nitrogen in these sys-tems is in soil organic matter. Inaddition, published studies suggestedthat in acid forest soils, nitrificationis controlled mainly by competitionbetween plants and nitrifying mi-crobes such that nitrification tendsto occur only where the supply ofammonium is high relative to plantdemand.

‘, ..

Figure 2. Evolution of hypotheses regarding the main pathway ofnitrogen assimilation and retention in forest ecosystems exposed

a

to chronically elevated nitrogen inputs: The width of the arrowsexpresses the relative amount of added nitrogen following eachpathway. (a) The initial hypothesis was that plant uptake wouldbe the main sink, resulting in increased photosynthesis and treegrowth, with recycling of nitrogen through litter and humus to theavailable pool (Aber et al. 1989). This mechanism should saturatequickly, resulting in early increases in nitrate leaching. (b) The

? very large rates of ni trogen retent ion in soi ls led us to quest ion theassumption that free-living soil microbes were carbon or energylimited. We hypothesized that pools of available carbon in soilswere driving increased microbial growth in competition with

6 plant uptake (Aber 1992). A requirement of this hypothesis waslarge increases in soil respiration of CO,. (c) Field measurementssuggest significant immobilization of nitrogen in soils withoutincreased CO, production. Mycorrhizal assimilation using plantcarbon derived directly from photosynthate is a process that most Humusclosely matches the limitations on carbon and nitrogen cyclingimposed by those measurements. DOC, dissolved organic carbon. DOC

with changes in soil microbial pro-cesses such as nitrogen mineraliza-tion and nitrification changing sec-ondarily in response to alterations inlitter input and quality, and to thesaturation of plant demand for inor-ganic nitrogen. Do our experimentsand those of others support or refutethis central hypothesis?

ExperimentsTo test the hypotheses contained inFigure 1, we established experimen-tal manipulations in four site/standcombinations in New England, andalso examined C:N and otherelement:nitrogen ratios in foliageand forest floor materials from aseries of 161 spruce-fir forests alongan nitrogen deposition gradientacross the New York/New Englandregion (Figure 3). Intensive site-levelexperiments can be questioned as totheir generality or applicability overa larger region, whereas regional sur-veys can be questioned due to thesimultaneous change in,several envi-ronmental parameters (e.g., climate,geology, soils) along any spatialtransect. Thus, a combination of space-for-time substitution in the regionalsurvey, and controlled experiments atthe intensive sites, provides a morerobust set of conclusions.

Manipulated sites included twostands (red pine plantation and na-tive mixed hardwoods) at theHarvard Forest in Petersham, Mas-sachusetts, where a chronic nitrogenamendment experiment is one of thecore activities of the National Sci-

November 1998

ence Foundation’sLong Term Ecologi-cal Research (LTER)program (Aber et al.1993, Magi11 et al.1997). Started in1988 and still con-tinuing, this experi-ment involves addi-tions of concentrateddissolved ammo-nium nitrate in sixmonthly doses fromMay to October ofeach year. Total ni-trogen additions are0, 5, and 15 g.m-2*yr-1. A similar

l b

1 h ’ / Mycorrhizal /

I \DOC

plot-level experi-ment was conducted :

in a second-growthN co2 - - - -D

northern hardwoodADDED

stand as part of theBear Brook Water-shed manipulation

t

program in Maine,which is funded by -the Environmental

I

Protection AgencyM y c o r r h i z a l /

Roots(Kahl et al. 1993,Rustad et al. 1993,Magi11 et al. 1996).

Free-LivingMicrobes

This experiment, \liwhich occurred from Humus1988 to 1991, in-cluded nitrogen ad- DOCditions of 0,2.8, and5.6 g*m-2*yr-1 as ni-tric acid. The fourth site, Mt. spruce-fir survey described below.Ascutney in Vermont, was selected Nitrogen additions, which rangedas a mature, healthy spruce-fir sys- from 0 to 3.1 g*m-2*yr-1 (McNulty andtern to complement the regional Aber 1993, McNulty et al. 1996),

923

Figure 3. Loca-tion of study sites. 1Harvard Forest,Bear Brook, andMt. Ascutney aresites of intensivemanipulation ex-periments. Thetransect sites area set of 161spruce-fir forestsused for an exten-sive survey ofchanges in forestfloor and foliarchemistry along anitrogen deposi-tion gradient.

Field Sites

included a combination of ammo-nium and nitrate added as a concen-

Ecosystem responses and

trated solution three times per year,intersite comparisons

due to the shorter growing season.These additions are continuing. The161 spruce-fir stands in the regionalsurvey (Figure 3) were located in 11areas from New York to Maine(McNulty et al. 1990, 1991) thatreceive from 0.4 to 1.3 g*m-2.yr-1 inestimated current nitrogen deposi-tion (Ollinger et al. 1993).

At each of the intensive manipula-tion sites, continuous measurementswere made of key components of thenitrogen cycle, including net annualnitrogen mineralization and nitrifica-tion, foliar nitrogen concentration andMg:N and Ca:Al ratios, and eitherwood plus foliar net primary produc-tivity (NPP) or basal area increment(net increase in the total cross-sec-tional stem area of all trees on aplot). Nitrogen leaching losses weremeasured at the Harvard Forest andBear Brook Watershed sites with ten-sion lysimeters placed below the root-ing zone. Because tension lysimeterscould not be buried in the shallow,stony soil at Mt. Ascutney, resin bagswere used to assess relative rates ofcation and anion movement below therooting zone. Foliar nitrogen, magne-sium, calcium, and aluminum concen-trations were measured on samplescollected in midsummer from all sites.At the regional survey sites, only foliarand forest-floor samples were col-lected, and these were analyzed fornitrogen and element:nitrogen ratios,as well as for potential net nitrogenmineralization and nitrification (asmeasured by a one-time, laboratoryincubation of forest-floor samples).

Results from the experiments de-scribed above (e.g., McNulty et al.1991,1996, Aber et al. 1995, Magi11et al. 1996, 1997) can be combinedwith those from similar studies inEurope and elsewhere in the UnitedStates (e.g., Christ et al. 1995, Gilliamet al. 1996) to generate a more com-plete picture of the nitrogen satura-tion process. In particular, the Euro-pean NITREX program (Wright andRasmussen 1998) carried out a se-ries of nitrogen additions to nitro-gen-poor sites (by fertilizer additions)and nitrogen removals from nitro-gen-rich sites ( by the construction ofunder-canopy roofs). There have alsobeen experimental additions of ni-trogen at the watershed scale in theUnited States (Adams et al. 1993,Kahl et al. 1993) and observationalstudies of nitrogen cycling in nitro-gen-saturated systems (Van Miegroetand Cole 1985, Johnson et al. 1991).

Nitrogen mineralization, nitrifica-tion, and nitrate loss. Our originalhypothesis predicted that chronicnitrogen additions would result inlong-term increases in nitrogen min-eralization as added nitrogen becameincorporated into soil organic mat-ter, reducing C:N ratios and increas-ing nitrogen release during decom-position. All sites and studies didshow an initial increase in net nitro-gen mineralization, as did the standsin the midpoint of the spruce-firtransect relative to the lowest nitro-gen deposition stands (Figure 4a).Maximum measured rates of net ni-

.

trogen mineralization ranged from1.2-to 2.4 times control (or low ni-trogen deposition) values. However,longer-term responses in all but onestand, the Harvard Forest hard-woods, show actual decreases in netnitrogen mineralization from earlypeak rates, dropping near or belowcontrol values (Figure 4a).

Reports of reduced rates of nitro-gen mineralization under nitrogen-enriched conditions can be found inboth recent and older studies (e.g.,Baath et al. 1981, Soderstrom et al.1983, Fog 1988). In the NITREXstudies in Europe, Tietema (1998),using 15N pool dilution, found in-creases in both gross mineralizationand gross immobilization rates acrossa nitrogen deposition gradient, butnet nitrogen mineralization peakedat intermediate nitrogen deposition.Gundersen et al. (1998) reported in-creases in field-measured net nitro-gen mineralization rates with nitro-gen addition only at nitrogen-limitedsites, whereas those with high initialrates of nitrogen cycling showed de-creases in net nitrogen mineraliza-tion with further nitrogen additions.

Two extant hypotheses could ex-plain this decline in nitrogen miner-alization rates. The first is that theaddition of nitrogen randomizeschemical bond structures in soil or-ganic matter, reducing efficiencies ofextracellular catabolic enzymes anddecreasing decay rates (e.g., Berg1986). Thus, nitrogen applicationsresult in higher nitrogen concentra-tion in soil organic matter, whichreduces decomposability and net ni-trogen mineralization despite a nar-rower C:N ratio. The second hy-pothesis states that the productionof humus-degrading enzymes by soilmicrobes (especially fungi) is sup-pressed in the presence of elevatedconcentrations of mineral nitrogen,thereby reducing nitrogen mineral-ization (Keyser et al. 1978, Fog 1988,Tien and Myer 1990). The relativeimportance of these two mechanismsunder field conditions cannot beevaluated with available data, al-though the inhibition of enzyme pro-duction may be especially relevantwhere experimental manipulationsresult in very high rates of nitrogenaddition.

We also predicted that chronic ni-trogen additions would increase net

924 Bioscience Vol. 48 No. 11

nitrification, or induce it where pre-viously absent, and that nitrate mo-bility and leaching losses would in-crease. Again, with one exception,this prediction was borne out. Boththe Mt. Ascutney and regional tran-sect data yielded a linear and highlysignificant relationship between for-est-floor nitrogen concentration andthe fraction of mineralized nitrogennitrified (McNulty et al. 1996). Thisrelationship is nearly identical to thatderived from the NITREX data sets(Tietema and Beier 1995). The onlyexperimental stand that showed nei-ther a decrease in net nitrogen min-eralization nor an increase in netnitrification was the Harvard Forestmixed hardwood site, the most nitro-gen limited site as discussed below.

However, although increases innitrate cycling and loss were detectedin most stands, these increases weresmall relative to the additions, show-ing that nitrogen retention efficiencywas relatively high in all sites. TheNITREX study sites also showedrapid reductions in nitrate losses instands to which nitrogen inputs wereexperimentally reduced, but small tono increases in nitrate losses withnitrogen additions (Bredemeier et al.1998), suggesting efficient retentionof added nitrogen in nitrogen-lim-ited systems. In a broader survey ofnitrate leaching in response to nitro-gen deposition in over 100 stands inEurope, Dise and Wright (1995)showed that nitrate leaching lossesaverage only approximately 30% ofinorganic nitrogen deposition. Simi-lar results have been compiled forNorth American sites by Peterjohnet al. (1996; see also Christ et al.1995), although individual sites mayapproach input:output ratios of 1.0(e.g., Johnson et al. 1991) or maygreatly exceed this ratio in cases inwhich nitrogen fixation by root sym-bionts is a major pathway for nitro-gen addition (Van Miegroet and Cole1985).

Foliar nitrogen. As we hypothesized,all sites showed increases in foliarnitrogen concentration, with the larg-est proportional changes occurringin the Harvard Forest pine stand andat the Mt. Ascutney site. These in-creases resulted in decreased Mg:Nratios in all sites (Figure 4b). Inter-estingly, foliar Ca:Al ratios also de-

Figure 4. Summary of changesover time in several ecosystemprocesses in response to long-term,chronic nitrogen additions. (a)Fractional change in nitrogenmin-eralization (“Maximum” is thehighest value obtained in the highnitrogen deposition stands beforedeclining to the “High nitrogen”value at the end of the experimentor transect). (b) Element ratios infoliage in the highest nitrogendeposition treatments expressedas a fraction of control values(e.g., a value of 0.8 under C:Nmeans that high nitrogen treat-ment foliar C:N ratios are 80% ofcontrol values). (c) Change inwood and total aboveground NPPin the highest nitrogen depositiontreatment expressed as a fractionof control-site values. Data fromAber et al. (1995).

Ii

creased in all but one site.The most probable explana-tion for these decreases isthat increased nitrate mobil-ity in soils increases cationleaching losses and soil acidi-fication. These results arealso consistent with thosefrom a watershed-level ex-periment in North America(Gilliam et al. 1996) and fromthe NITREX experiments(Dise and Wright 1995, Gun-dersen et al, 1998). Becauseforest decline has been linkedto Ca:AI imbalances (e.g.,Shortle and Smith 1988), thefinding that this condition canbe induced by experimentalincreases in nitrogen avail- ‘-

ability alone may provide a unifyingprinciple in forest-decline research,

Increased foliar nitrogen concen-tration can also affect ecosystem car-bon balance through increased ratesof net photosynthesis (Field andMooney 1986, Reich et al. 1995).This increase can increase above-ground biomass production (the fer-tilizer effect) and may also provideadditional carbon (and energy) tofuel belowground processes.

Net primary production. Althoughwe had hypothesized, based on re-search in Europe (e.g., Schulze 1989),that nitrogen saturation could leadto forest decline, we did not expectto see any response of this type withinthe lifetime of the intensive experi-

ments. However, both of the experi-mentally manipulated evergreenstands (Harvard Forest pine and Mt.Ascutney) and the regional surveyhave shown either declining treegrowth (Figure 4c) or increased treemortality (McNulty et al. 1996). Thisevidence of decline occurred despitethe increases in foliar nitrogen, whichmight be expected to increase netphotosynthesis (Reich et al. 1995).Increased mortality at Mt. Ascutneyis significantly correlated with bothnitrogen addition rate and foliarCa:Al ratio (McNulty et al. 1996)and occurred within the first 6 yearsof fertilization. Declining pinegrowth rates at the Harvard Foresthave not yet led to increased mortal-ity, but surveys of mortality along

November 1998 925

.

HFHW-> HFP- I I I I I I I)B B W - >

Ascutney - - - - - - - - qTransect- - - - - - - - - +

0 1 2 3

StageFigure 5. Revised set of hypotheses on the response of temperate forest ecosystems tolong-term, chronic nitrogen additions. Changes from initial hypotheses (Figure 1)include the reduction in nitrogen mineralization in stage 3 and the addition of foliarCa:Al and Mg:N ratios. Abbreviations at the top of the figure denote relative degree ofnitrogen saturat ion in each of the study si tes and the transect stands prior to increaseddeposit ion. The posit ion of the arrowheads summarize how far toward saturation eachsite moved, as summarized by the data presented in this art icle. We hypothesize thatprevious land-use history determines the initial position and that deciduous standsmove more slowly toward saturation than evergreen stands. HFP, Harvard Forest pinestand; HFHW, Harvard Forest hardwood; BBW, Bear Brook Watershed.

the spruce-fir transect (Rock et al.1986, McNulty et al. 1991) showconsistent trends of increasing mor-tality with increasing nitrogen depo-sition. These findings suggest thatthe significant decline in high-eleva-tion spruce-fir forests in the last threedecades in western New England andeastern New York (Siccama et al.1982, Hornbeck and Smith 1985)may have been due in part to in-creased nitrogen deposition. In addi-tion, Boxman et al. (1998) reporti n c r e a s e d t r e e g r o w t h i n t w oNITREX sites with high nitrogencycling following the reduction ofni t rogen inputs through under-canopy roof construction.

addit ions, al though root ni trogenconcentration nearly doubled (Magi11et al. 1997). In the NITREX studies,by contrast, fine-root biomass in-creased significantly following ni-trogen exclusion from the site withthe highest nitrogen availabili ty(Boxman et al. 1998, Gundersen etal. 1998).

Hypotheses revisited

Fine roots. Because it is difficult tomeasure fine-root biomass, fewerdata are available on responses offine roots to nitrogen deposition. Atboth Harvard Forest s tands, nochanges in fine-root biomass weremeasured through year 4 of nitrogen

Information gained through nearly adecade of intensive research can besynthesized into a set of summaryrelationships that define the responseof temperate-zone forests to accu-mulated nitrogen deposition (Figure5). The principal differences betweenFigure 5 and our initial hypothesesare the downturn in nitrogen miner-alization as saturation is approachedand the addition of the foliar Mg:Nand Ca:Al trends. This descriptionof nitrogen saturation as a progres-sive “syndrome” of concurrent re-

Nor do all stands move towardsaturation at the same rate. The ar-rows in Figure 5 suggest how fareach site has moved toward satura-tion and possible decline during thetreatment period. The deciduousstands did not move as rapidly to-ward saturation as did the evergreenstands. The Harvard Forest hard-wood s tand, which was the mostnitrogen limited initially, receivednitrogen in the amount of nearly 90g/m’ over the first 6 years withoutthe initiation of significant nitrifica-tion or nitrate leaching. In contrast,total nitrogen inputs of less than 15g/m* were sufficient to induce sig-nificant decreases in foliar Mg:N andCa:AI ratios and increases in treemortality in spruce-fir forests on Mt.Ascutney. Despite the presence ofsome nitrification and nitrogen leach-ing in control plots, the Bear Brookwatershed deciduous forest also ex-hibited only moderate increases innitrate loss (Magi11 et al. 1996),whereas those in the red pine standat the Harvard Forest increased dra-matically. Across all experiments,growth declined or mortality in-creased in evergreen stands but notin deciduous stands.

926 Biokience Vol. 48 No. 11

sponses to long-term, chronic nitro-gen deposition is in agreement withthe use of a generalized indicator of“nitrogen status” derived from prin-ciple components analysis for theNITREX sites by Gundersen et al.(1998).

It is apparent that not all of thecontrol or pretreatment stands fallon the same part of the x-axis inFigure 5. Measured values for netnitrogen mineralization, nitrifica-tion, nitrate leaching, foliar nitrogenconcentration, and NPP all suggestthat the Harvard Forest hardwoodcontrol stand is extremely nitrogenlimited and that it has not movedvery far toward nitrogen saturation.In contrast, the Bear Brooks hard-wood control stand shows some ni-trification and nitrate leaching, sug-gesting that this stand started at adifferent point along the x-axis. Ini-tial differences between the HarvardForest pine stand and the spruce-firstands can also be expressed as dif-ferent initial positions along this x-axis, with those locations being de-termined by measured rates ofnitrogen cycling and loss.

.Do these findings suggest that Figure 6. One-year rp--

j greater cumulative inputs -of nitro-I gen are required to induce nitrogen

saturation in deciduous forests? Dothey further suggest that nitrogen

/ saturation does not pose a problemfor forest health in deciduous for-ests, or only that higher nitrogen

A retention rates will allow a delay inthe onset of decline? Only the con-tinuation of these long-term experi-

iments will provide answers to thesequestions.

Does land-use historydetermine ecosystem response?Why do stands appear to start atdifferent points within the nitrogen-saturation continuum? Before begin-ning our experiments, we thoughtthat the Bear Brook Watershed hard-wood stand in Maine would have ahigher nitrogen retention capacitythan the Harvard Forest hardwoodsbecause of higher cumulative nitro-gen deposition at the Harvard Forestsite. The opposite is true. We nowhypothesize (Aber and Driscolll997,Foster et al. 1997) that prior land-use history, reaching back even lOO-200 years, can play a significant rolein preconditioning forest responseto nitrogen deposition. Essentially,the greater the previous extractionof nitrogen from a site by agricul-tural conversion, fires, or harvest-ing, the greater the nitrogen limita-tion on net photosynthesis and forestgrowth and the larger the amount ofnitrogen deposition required to movetoward saturation. Previous land-usehistory appears to be more impor-tant than either current or total ac-cumulated nitrogen deposition in de-termining current nitrate leachinglosses across the landscape in thenortheastern United Sta,tes. The samehypothesis can be applied to Euro-pean sites, where forest-floor C:Nratio, reflective of both past land useand current nitrogen status, is sig-nificantly correlated with nitrogenleaching losses (Tietema and Beier1995, Gundersen et al. 1998). Inhigh-elevation sites in the southeast-ern United States, both the nitrogendeposition rate (which increases withelevation) and previous land use havebeen shown to affect nitrate concen-trations in streams (Silsbee and Larson1982, 1983, Nodvin et al. 1995).

November 1998

change in nitrate con-centration in soil solu-tion collected below therooting zone in trenchedplots within a red pineplantat ion and a nativehardwood stand at theHarvard Forest. In con-t ras t to s tands showinglong-term retention ofadded nitrogen, standswith disrupted canopy-root connections beginto leach ni t rogen within1 year (from Aber et al.

- H a r d w o o d s. . . . . fqnes

1983).L

This view implies that under un-disturbed, mature, or steady-stateconditions, all forests should be nearthe middle of stage 2 (Figure 5),where nitrogen losses are sufficientto balance nitrogen inputs in deposi-tion (see discussion below). Standsto the left of this stage in Figure 5represent disturbed conditions, inwhich the removal of nitrogen fromthe site has resulted in a slower ni-trogen cycle. For example, theHarvard Forest hardwood site waspart of a working farm for over acentury, and although it was neverplowed, it was adjacent to a plowedfield and was undoubtedly harvestedmany times. There was also a groundfire through this stand in the 19.50s.In contrast, the Bear Brook stand isin a remote area of eastern Maine,and only one commercial harvest canbe documented. It is thus closer tothe “natural,” or stage 2, condition.

A paradox: Nitrogen retentionin carbon-limited soilsFrom our initial literature review,we had hypothesized that addingnitrogen in the amount of 15g*m-2*yr-1 to the Harvard Forest siteswould move those systems rapidly tonitrogen saturation due to limita-tions on the increase in photosynthe-sis that could be achieved and wouldbe needed to assimilate additionalnitrogen (Figure 2a). This assump-tion was based in part on the resultsfrom an older trenching experimentconducted in similar stand types atthe Harvard Forest, in which 1 x 2 msoil monoliths were isolated fromaboveground plant processes by in-stallation of a 1 m deep trench sur-

Month

J

rounding the plot. This experimentresulted in highly elevated nitrateconcentrations in soil solutionswithin a year of treatment (Figure 6;Aber et al. 1983). Other stands ex-amined in a comparative studyshowed nearly immediate nitratelosses (Vitousek et al. 1979). Trench-ing effectively cuts off plant uptakeof mineral nitrogen. In a system thatcycles nitrogen in the amount of 7.5-8.0 g-m-“+yr-’ (Magi11 et al. 1997),trenching would be expected to re-sult in fairly large nitrogen lossesfrom the system in a short period oftime. We thought the 1 year delaywas quite a long response time, andwe expected the delay before theinitiation of nitrate losses from thenitrogen amended plots to be simi-lar, that is, l-2 years.

Thus, the most striking and unex-pected result from this set of studieswas the very high nitrogen retentionefficiency in all treatments over longperiods of time (Table 1). Even thehighest nitrogen addition plot in thepine stand at the Harvard Forest, inwhich nitrogen leaching losses rosedramatically after the first 2 years,still maintained an overall retentionefficiency of 85% over the initial 6years of treatment. The hardwoodstand at the Harvard Forest retainednearly 90 g/m2 of nitrogen in the first6 years with no loss of nitrate andextremely low soil-extractable ni-trate. This long-term net nitrogenaccumulation is a major factor re-ducing the negative impacts of nitro-gen deposition on forests and sur-face waters (e.g., nitrate andaluminum mobility, cation depletion,soil and water acidification, and for-est decline; Schulze 1989, Herrick

927

.

Table 1. Mean annual inorganic nitrogen retention efficiency for several experimental plots at the Harvard Forest and BearBrook Watershed.”

Location Time period Stand TreatmentNitrogen inputs Nitrogen outputs Retention

(gWb Wm2)b efficiencyc

Harvard forest 1988-1993 Pine ControlPine Low nitrogenPine High nitrogenHardwood ControlHardwood Low nitrogenHardwood High nitrogen

Bear Brook 1988-1991 Hardwood ControlHardwood Low nitrogenHardwood High nitrogen

4.832.487.4

4.832.487.4

2.011.9 0.324.3 1.8

0.20.62.70.10.60.50.1

.96

.98

.85

.98

.98

.99

.95

.97

.93

“Data from Magill et al. (1996, 1997).bNitrogen inputs include both background deposition (0.8 g.m-2.yr-‘) and fertilizer additions.‘Retention efficiency calculated as (l- (nitrogen outputs/nitrogen inputs)).

and Friedland et al. 1991, Richter etal. 1992, Kahl et al. 1993, McNultyand Aber 1993, Boxman et al. 1995,Emmett et al. 1995).

The difference between the effectsof trenching and nitrogen additionson nitrate losses suggests that sever-ing the connection between plantsand the root and soil systems, de-spite the potential creation of a siz-able pool of decomposition substratefor nitrogen immobilization by mi-crobes due to root death, causes amuch larger and more immediatedecrease in nitrogen retention effi-ciency than even very large additionsof mineral nitrogen. This hypothesisin turn suggests that the presence offunctional roots in the soil matrixmay have an effect on nitrogen cy-cling and retention beyond the pro-cess of nitrogen uptake for plant useand that a functional root system,with associated mycorrhizal sym-bionts, may be more important thanthe free-living microbial communityor soil detrital pools in determiningnitrogen retention.

Peterjohn et al. 1996) similar to thosein the Harvard Forest pine stand(Magi!! et al. 1997), DIN losses donot generally approach input levelsin the absence of forest decline ordisturbance (but see Johnson et al.1991).

The high nitrogen retention ratesreported here are not unusual. Theyare generally consistent with a largesurvey of European data (Dise andWright 1995), in which average re-tention of dissolved inorganic nitro-gen (DIN) was approximately 70%,and did not appear to decline signifi-cantly or consistently even at veryhigh nitrogen deposition rates. Re-tention was somewhat higher in sys-tems in which ammonium inputsdominated over nitrate. A smallerset of US data shows similar efficien-cies (Peterjohn et al. 1996). Althougha few sites have shown increases inDIN loss over long time periods(Murdochandstoddard 1991,1993,

DIN exports are not the only ni-trogen losses from forested systems.Dissolved organic nitrogen (DON)can be the dominant form of nitro-gen exported from systems in whichDIN losses are low (e.g., McDowelland Asbury 1994, Hedinet al. 1995).At moderate rates of DIN deposi-tion, inclusion of DON losses in standnitrogen balances can reduce calcu-lated nitrogen retention efficienciessignificantly (Table 2). Because DONexports appear to cover a much nar-rower range than DIN losses and donot appear to increase significantlywith high nitrogen addition rates(Table 2; Currie et al. 1996), totalnitrogen retention efficiencies, in-cluding DON, may actually increaserather than decrease at higher levelsof nitrogen addition (Table 2). Otherpathways for ecosystem nitrogenloss, such as denitrification and am-monia volatilization, appear to betoo small to be important in mostsites with high nitrogen inputs (butsee Tietema and Verstraten 1992).

et al. 1997), and on soil nitrogenpartitioning (e.g., Matschonat andMatzner 1995) indicate that two-thirds or more of added nitrogenresides in soil organic matter. More-over, most of this nitrogen did notpass through aboveground plant tis-sues (see also Johnson 1992). Theseresults may at first appear to contra-dict our hypothesis that plant pro-cesses should both respond first tonitrogen deposition and controlprogress toward nitrogen saturation.However, only the final, long-termsink for nitrogen is in soils. We donot know the mechanisms (plant,microbial, or soil-chemical) throughwhich this stabilization occurred.Strong correlations between nitratemobility and foliar nitrogen concen-tration in several studies (Tietemaand Beier 1995, Magi!! et al. 1997)suggest that plant nitrogen status andnitrogen processing might still be acritical part of the nitrogen satura-tion process.

What mechanism(s) might accountfor this very high rate of nitrogenretention? Studies on altered C:Nratios in forest-floor organic matterunder coniferous forests (McNultyet al. 1991, McNulty and Aber 1993,Tietema and Beier 1995), on long-term distribution of added lSN trac-ers (Buchman et al. 1995, Nadel-hoffer et al. 1995, Koopmans et al.in press), on ecosystem-level nitro-gen budgets (Aber et al. 1993, Magi!!

The importance of soil organicmatter as the final sink for addednitrogen caused us first to rethinkour initial assumption that soil mi-crobes are energy and carbon limited(Aber 1992). What if the soils at theHarvard Forest were so nitrogendepleted that microbes were limitedby availability of nitrogen? If so,then there would have to be a pool ofunused but bioavailable carbon inthe soils that soil microbes used toimmobilize the added nitrogen. Thenitrogen additions should then re-sult in a pulse of CO, from soils as thisavailable carbon pool was tapped. Wethen hypothesized that dissolved or-ganic carbon (DOC) might be thisavailable soil pool, and should there-fore be converted to CO, and de-

9 2 8 BioScience Vol. 48 No. 11

-.

Table2, Nitrogen retention efficiency in different Harvard Forest stands under different nitrogen treatments, with and withoutthe inclusion of dissolved organic nitrogen (DON).”

Parameter ControlPine

Low nitrogen High nitrogen ControlHardwood

Low nitrogen High nitrogen

DIN deposition 0.8b 0.8 0.8 0.8 0.8 0.8DON deposition .06 .06 .06 .06 .06 .06Nitrogen additions 0 5.0 15.0 0 5.0 15.0DIN leaching 0.03 0.1 2.1 0.02 0.1 0.1

I D O N leaching .54 .44 .36 .32 .46 .35

t

Without DONTotal inputs :::3 5.8 15.8 0.8 5.8 15.8Total outputs 0.1 2.1 0.02 0.1 0.1Nitrogen retention

efficiency (%) 96 98 87 97 98 99

With DONTotal inputs .86 5.86 15.86 0.86 5.86 15.86Total outputs .57 0 .54 2 .46 0 .34 0.56 0 .36Nitrogen retention

efficiency (%) 34 91 85 60 90 98

“Data are for the period October 1992-September 1993 and come from the chronic nitrogen addition experiment at the Harvard Forest(Currie et al. 1996, Magill et al. 1997). All values are given in g=m-2.yr-1. Measured nitrogen trace gas fluxes were found to be near zero.

pleted by the microbial response tonitrogen additions (Figure 2b).

None of these occurred. Differ-ences in soil CO, efflux across treat-ments were undetectable, although asimple calculation of the CO, pro-duction that should result from im-mobilization of added nitrogenthrough microbial biomass produc-tion suggested that fluxes should bevery large and easily detected (Micks1994). Similar results have beenfound in other studies (Roberge 1976,Foster et al. 1980, Flanagan and VanCleve 1983, Soderstrom et al. 1983,Mattson 1995). In addition, DOCflux below the forest floor was notaltered significantly, and DON fluxesincreased only slightly through a shiftin the DOC:DON ratio (Currie et al.1996). There were also no signifi-cant reductions in the bioavailabilityof DOC in leachate from the forestfloor (Yano et al. in press).

l Neither alter concentrations ofDOC or the bioavailability of DOCin soil solution nor increase CO,efflux from soils (therefore, it mustnot draw on previously untappedpools of labile carbon in soils);

l Use only additional carbon madeavailable through increased rates ofnet photosynthesis resulting fromhigher foliar nitrogen concentrations;

sured changes in soil CO, efflux (i.e.,undetectable; Micks 1994). The firstcomparison places an upper limit onthe amount of additional carbon thatmight be available through detritusproduction or root exudation. Thesecond comparison bounds the in-crease in microbial respiration thatcould result from an increase in bio-mass production.

l Require the continued integrityof the plant-root-soil system.

Three classes of soil processesmight account for the measured con-versions of mineral nitrogen to soilorganic nitrogen. In order from mostto least conventional, these are:

l Immobilization by free-living mi-crobes through biomass production;

l Abiotic incorporation into ex-isting soil organic matter;

Possible mechanisms fornitrogen incorporationMultiple sets of observations from ourfield experiments and those of othersallow us to provide criteria that anyprocess must meet to be considered an

I

Iimportant mechanism in the retention

Iof added nitrogen in the Harvard For-est sites. Such a process must:

.

l Conversion to organic nitrogenforms by mycorrhizae, without bio-mass production.

l Incorporate nitrogen into soilorganic matter without first incor-porating it into aboveground plantbiomass;

We have calculated the amount ofcarbon required to immobilize thenitrogen added in the high nitrogentreatments at the Harvard Forest andthe increase in soil CO, efflux thatwould result for each of these mecha-nisms (Table 3). These estimatedfluxes can be compared with the es-timated increase in net photosynthe-sis predicted to occur through in-creased foliar nitrogen concentration(reduced by carbon going to increasedwood NPP; Table 4), and with mea-

Microbial immobilization. Microbialimmobilization is frequently men-tioned as the primary process forincorporation of mineral nitrogeninto soil organic matter. Short-termlSN pool dilution experiments havedemonstrated that gross nitrogenmineralization and immobilizationrates measured in soils may be manytimes higher than net rates, such thatlarge quantities of nitrogen could beadded to soil organic matter by thismechanism (Davidson et al. 1992, Hartet al. 1994). Stark and Hart (1997)proposed that gross fluxes of nitrateare also high in forest soils, evenwhen no evidence for net nitrification,and no extractable nitrate, can befound. These authors hypothesizedthat rapid microbial immobilizationcould be the primary mechanism in-volved in the retention of added DIN.In contrast, Tietema (1998) foundno evidence for gross nitrate immo-bilization in several forest soils ofdiffering nitrogen availability.

If microbial immobilization is theprimary process incorporating min-eral nitrogen into soil organic mat-

November 1998 929

Table 3. Calculated demand for carbon and production of CO, for the three hypothesized pathways for incorporation of thenitrogen (15 g.m-2-yr-1) added to the chronic nitrogen stands at the Harvard Forest.

Nitrogen X C:N ratio X C a r b o n - u s e = Total carbon Total carbonPathway incorporation of product’ efficiency requirementb respired’

Nitrogen incorporation 15 g.m-**yr-’ 8 3 g C required per 360 g*m-z.yr-’ 240 g.m-*.yr-’through production by g C producedfree-living microbes

Abiotic incorporation 15 g.m-‘ayr-’ Not known 0 g C required perg C produced

0 g.m-z*yr’ 0 g*m-Z.yr-L

Nitrogen incorporationby mycorrhizalassimilation

15 g.m+yr-’ 6.25 1 g C required per0.8 g C produced

117.2 gem-*.yr-’ 23.4 gm-‘*yr-’

‘C:N ratio of product in the case of microbial production is the assumed C:N ratio of microbial biomass. For mycorrhizal assimilation, theC:N ratio of product is the C:N ratio in proteins-the assumed product-not biomass.bTotal carbon requirement is obtained by multiplying nitrogen incorporation by C:N ratio of product by carbon-use efficiency.‘Respired carbon is the fraction of total carbon demand that is not incorporated into product (e.g., two-thirds of 360 g.m-Z*yr-’ for microbialgrowth; 20% of 117.2 g.m-“syr-’ for mycorrhizal assimilation).

ter, then what is the source of thelabile carbon required to drive thisprocess? Nitrogen immobilizationthrough biomass production by free-living soil microbes carries high car-bon requirements resulting from theC:N ratio in biomass produced andthe relatively low efficiency of car-bon utilization, which results fromthe need to degrade soil organic com-pounds to obtain carbon and energy.Microbial carbon-use efficiency isboth very difficult to measure andvaries widely between and withinstudies (e.g., Schimel 1988). Soilfungi (e.g., mycorrhizae) may be moreefficient users of carbon than free-living prokaryotes (Zak et al. 1996),but this efficiency is offset by therelatively high C:N ratio of fungi ascompared with bacteria (Tietema1998). Low carbon-use efficiency hasthe additional effect of driving rela-tively large CO, fluxes from soils. Ifsoil microbial activity under fieldconditions is energy or carbon lim-ited, not nitrogen limited (e.g., Fos-ter et al. 1980, Baath et al. 1981,Flanagan and Van Cleve 1983), thenadditions of DIN should not stimu-late increased metabolic rates, norshould soils contain an untapped poolof labile carbon available to driveincreased immobilization of addedDIN. Thus, even if background ratesof DIN cycling are high in untreatedstands, an additional source of car-bon would be required to immobi-lize the additional DIN received.

timated fluxes. Multiplying 15 g ni-trogen added by a C:N ratio in mi-crobial biomass of 8.0, and dividingby an assumed carbon-use efficiencyof 33%, yields a total net increase incarbon demand of 360 g.m-2.yr-1.This calculated value greatly exceedsthe estimated increases in net photo-synthesis (less the increased carbonallocation to wood; Table 4). The33% carbon-use efficiency meansthat two-thirds of the carbon used isrespired, so that soil CO, effluxshould increase by 240 g*m-2*yr-1.An increase of this magnitude at theHarvard Forest would have beendetectable with the methods used,but no increase was measured thereor in several other locations. Assum-ing that most soil microbial biomassis fungal rather than bacterial andapplying Tietema’s (1998) C:N ra-tios (25:l) and carbon-use efficien-cies (60%) results in an even highertotal carbon demand of 626 g-m-2-yr-‘,with respiration about the same (2.50g-m-2-yr-1).

Abiotic incorporation of DIN intosoil organic matter. Significant ratesof direct chemical incorporation ofammonium into soil organic matterhave been measured by mass balanceusing sterilized soil samples (Nom-mik 1970, Axelsson and Berg 1988,Schimel and Firestone 1989, Sen andChalk 1995). Measured rates of di-rect chemical incorporation ofNH,-N per kg of carbon range from0.0036 mg/day (Nommik 1970) to112 mg/day (Sen and Chalk 1995). Itis difficult to.interpret these resultsin terms of field nitrogen retentionrates because the methods used (e.g.,incubation times, ammonium concen-trations, extraction procedures, andsoil sterilization techniques) vary be-tween studies. In particular, all meth-ods of soil sterilization (i.e., gaseousfumigation, solutions of heavy metalsalts, gamma irradiation, and auto-claving) may influence the incorpo-ration of ammonium by alterationof the substrate. However, Foster etal. (1985a, 1985b) have shown thatsignificant immobilization can oc-cur through reactions between am-monia and organic matter in forestsoils treated with urea, which cre-ates both high pH and high ammo-nia concentrations-conditions thatfoster this reaction.

Our calculations of both carbonrequired and carbon respired forimmobilization through microbial bio-mass production (Table 3) exceed es-

These calculations indicate thatthere is neither enough of an increasein carbon fixation, nor enough of anincrease in soil CO, efflux, to sug-gest that microbial immobilizationthrough biomass production is theprimary mechanism involved in theimmobilization of added nitrogen atthe Harvard Forest. Moreover, thisprocess would also be highly depen-dent on existing soil detrital pools,and not affected by continuing car-bon inputs from plants, and so is notin agreement with the comparison oftrenched plot and nitrogen amend-ment results.

Abiotic incorporation of DIN intosoils has also been estimated by therapid (15-30 min) disappearance oflSN tracer in pool dilution experi-ments (Davidson et al. 1991). Thisrapid incorporation was not stronglyinfluenced by autoclaving of soilcores. Both the speed of immobiliza-tion and the insensitivity to auto-

930 Biokience Vol. 48 No. 11

.

clavingsuggest that a biotic processeswere involved.

The incorporation of mineral ni-trogen into soil organic matterthrough chemical fixation reactionshas been verified by 15N NMR spec-trometry. Laboratory studies haveshown that ammonium (Thorn andMikita 1992) and other nitrogencompounds, such as glycine (Benzing-Purdie et al. 1983), undergo reac-tions with soil organic matter thatresult mainly in heterocyclic formsof nitrogen (e.g., indoles and pyr-roles) and some amides (Thorn andMikita 1992). Biotic nitrogen incor-poration in soils transforms nitro-gen primarily into amides (Almen-dros et al. 1991, Knicker et al. 1993,Clinton et al. 1995).

However, 15N NMR spectra fromsoil organic matter and compost resi-dues in which biotic processes werenot excluded suggest that directchemical fixation explains 15% orless of the total nitrogen retention.Natural abundance spectra of soilorganic matter (Knicker et al. 1993),spectra from labeled organic matteramendments (Almendros et al. 1991),and spectra from labeled mineralnitrogen amendments (Clinton et al.1995) show that approximately 85 %of the nitrogen retained occurs in theamide form.

Abiotic nitrogen incorporationrequires only the existence of suit-able organic substrates and environ-mental conditions. It requires noadditional carbon inputs and gener-ates no additional CO, efflux (Table3), so it is compatible with the car-bon balance restrictions already dis-cussed. However, estimates of nitro-gen immobilization rates derivedfrom both rapid short-term loss of15N tracers and from 15N NMR spec-trometry are too small to account forthe magnitude of nitrogen retentionobserved in our experimental sites innortheastern North America andthroughout the NITREX sites inEurope.

Mycorrhizal assimilation. A third,and less frequently discussed processfor conversion of DIN to soil organicnitrogen is through mycorrhizal as-similation and exudation. Mycor-rhizae are ubiquitous in forest soils(Fahey and Hughes 1994) and canconstitute a large fraction of the to-

Table 4. Estimated net increase in carbon available to drive nitrogen immobilization.a

Pine Pine Deciduous DeciduousControl (low nitrogen) (high nitrogen) (low nitrogen) (high nitrogen)

0 99 249b 49 75

*Net increase in carbon availability is calculated as the estimated increase in total netphotosynthesis due to increased foliar nitrogen concentration (from a model of canopyphotosynthesis validated at the Harvard Forest, PnET-Day [Aber et al. 19961) minus carbonallocated to increased aboveground production. Values in g*m-z.yr-‘.bThis number may be too high. Actual growth rates of trees have declined in this stand, as haveboth Mg:N and Ca:AI ratios in foliage, suggesting that net photosynthesis may be declining.

tal fungal biomass, which, in turn,can be a large fraction of the totalmicrobial biomass (Killham 1994).This fraction may be especially largeunder nitrogen-limiting conditionsand may decline with nitrogen satu-ration (Tietema 1998). Mycorrhizaeare longer lived than free-living mi-crobes, act as an extension of rootsystems, and are generally thoughtof as a mechanism by which plantsincrease the efficiency of nutrientuptake (Rygiewicz et al. 1984, Yanai1994). Still, there is some uncertaintyabout which partner, the plant or thefungus, primarily controls this inter-action (Rygiewicz and Andersen 1994).

If mycorrhizae were to assimilatemineral nitrogen, convert the nitro-gen into extracellular enzymes, andexude these enzymes back into thesoil, where they could be stabilizedby well-known condensation reac-tions between proteins and humiccompounds, the estimated carbon re-quirements and CO, productionwould be much lower than for free-living microbial immobilization(Table 3) and well within the rangeof estimated increases in net carbonavailability (Table 4). Mycorrhizalnitrogen assimilation reduces the es-timated carbon cost of nitrogen in-corporation by using photosynthatecarbon directly rather than requir-ing that the carbon be first convertedto biomass by plants and then shedas litter and decomposed by soil mi-crobes. The respirational efficiencyof conversion of photosynthate tostructural polymers is estimated tobe approximately 80% (growth res-piration equals 25% of total carbonrequired for growth; Penning de Vriesetal. 1974,SprugelandBenecke 1991),in contrast to the 33% efficiency wehave used for free-living microbialimmobilization. Thus, carbon re-quirements are one-third as large andCO, effluxes one-tenth as large for

the hypothesized mycorrhizal path-way (Table 3), in comparison withmicrobial biomass production.

Because of the direct reliance ofmycorrhizae on continued allocationof photosynthate from crowns toroots, this mechanism would be themost sensitive of the three to disrup-tion of the plant-root-soil linkage.If rapid uptake by mycorrhizae isresponsiblefor the high rates of grossnitrogen immobilization in soils mea-sured by 15N pool dilution, then thequestion of the carbon source to drivethis process may also be answered-it is plant photosynthate and notdetrita.1 or soil organic carbon.

A new hypothesisSeveral sets of studies described inthis article have established bound-ary conditions on the mechanism bywhich the majority of nitrogen addedto forests in nitrogen addition ex-periments becomes incorporated intosoil organic matter. Of the three gen-eral classes of processes discussed inthis article, only the third-mycor-rhizal assimilation and exudation,using photosynthate from the hostplant as the carbon source-fitswithin the limits set by field observa-tions. We hypothesize (Figure 2c)that mycorrhizal assimilation andexudation is the dominant processinvolved in immobilization of addednitrogen. Tietema (1998) has sug-gested that during nitrogen satura-tion, soil microbial communitiesmove from being fungal, and prob-ably mycorrhizal, dominated to be-ing bacterial dominated. This loss ofmycorrhizal function could be a keyprocess leading to increased nitrifi-cation and nitrate mobility.

Unfortunately, no methods arecurrently available to determine theassimilation-exudation balances ofmycorrhizae. Still, the hypothesis can

November 1998 931

be tested inferentially. We predictthat severing the connections be-tween plants and mycorrhizal roots,through a simple set of trenched plotexperiments, will lead to immedi-ate reductions in the rate of grossnitrogen immobilization (as mea-sured by lSN pool dilution) and toincreased nitrate mobility. If ourhypothesis is correct, gross immo-bilization of “N would decreasewithin days to weeks of trenchingas the carbon source fueling immo-bilization, plant photosynthate, isexcluded. If our hypothesis is in-correct, then the pulse of detritalmaterial created by root death as aresult of trenching should lead, alsoin days to weeks, to increased grossnitrogen immobilization.

References cited 1: 359-369.

Aber JD. 1992. Nitrogen cycling and nitro-gen saturation in temperate forest ecosys-tems. Trends in Ecology & Evolution 7:220-223.

Bormann FH. 1982. The effects of air pollu-tion on the New England Landscape.Ambio 11: 338-346.

Aber JD, Driscoll CT. 1997. Effects of landuse, climate variation and N depositionon N cycling and C storage in northernhardwood forests. Global Biogeochemi-cal Cycles 11: 639-648.

Boxman AW, van Dam D, van Dyck HFG,Hogervorst RF, Koopmans CJ. 1995. Eco-system responses to reduced nitrogen andsulphur inputs into two coniferous foreststands in the Netherlands. Forest Ecoloavand Management 71: 7-30.

.,,

Aber JD, Melillo JD, McClaugherty CA,Eshleman KN. 1983. Potential sinks formineralized nitrogen following distur-bance in forest ecosystems. Pages I79-192 in Hallberg, R, ed. EnvironmentalBiogeochemistry. Ecological Bulletin.Stockholm (Sweden): Publishing House/F R N .

Boxman AW, Blanck K, Brandrud TE, EmmetBA, Gundersen P, Hogervorst RF, KjonaasOJ, Persson H, Timmermann V. 1998.Vegetation and soil biota response to ex-perimentally-changed nitrogen inputs inconiferous forest ecosystems of theNITREX project. Forest Ecology andManagement 101: 65-80.

Aber JD, Nadelhoffer KJ, Steudler P, MelilloTM. 1989. Nitrogen saturation in notth-;rn forest ecosystims. Bioscience 39: 378-386 .

Bredemeier M, Blanck K, Xu XJ, Tietema A,Boxman AW, Emmet B, Moldan F,Gundersen P, Schleppi P, Wright RF. 1998.Input-output budgets at theNITREX sites.Forest Ecology and Management 101: 57-64 .

At the beginning of this article, weset up a dichotomy between plantand microbial control over the reten-tion of nitrogen added to forest eco-systems and the timing of nitrogensaturation. Our new hypothesis sug-gests that this dichotomy was anoversimplification of the problem,and recognizes the fundamental im-portance of mycorrhizal symbiosesin mediating the carbon and nitro-gen cycles of temperate forest eco-systems. Although the processes in-volved are microbial, and would bemeasured as such by “N pool dilu-tion methods, gross rates of nitrogencycling in soils are driven by, andtied directly to, current rates of plantphotosynthesis. This view reconcilesthe high rates of gross nitrogen im-mobilization reported for many sys-tems with the strongly carbon andenergy limited nature of the soilecosystem. One of the key implica-tions of this hypothesis is that in-creased nitrate mobility requires “satu-ration” of plant nitrogen demand.Consequently, accurately determin-ing plant nitrogen status-for ex-ample, by foliar analysis-could pro-vide the best indicator of ecosystemnitrogen status and degree of nitro-gen saturation.

Aber JD, Magi11 A, Boone R, Melillo JM,Steudler P, Bowden R. 1993. Plant andsoil responses to chronic nitrogen addi-tions at the Harvard Forest, Massachu-setts. Ecological Applications 3: 156-166.

Aber JD, Magi11 A, McNulty SG, Boone RD,Nadelhoffer KJ, Downs M, Hallett R.1995. Forest biogeochemistry and pri-mary production altered by nitrogen satu-ration. Water Air and Soil Pollution 85:1665-1670.

Buchmann N, Schulze ED, Gebauer G. 1995.iSN-ammonium and *jN-nitrate uptake ofa 15 year old Picea dies plantation.Oecologia 102: 361-370.

Christ M, Zhang Y, Likens GE, Driscoll CT.1995. Nitrogen retention capacity of anorthern hardwood forest soil under am-monium sulfate additions. EcologicalApplications 5: 802-812.

Aber JD, Reich PB, Goulden ML. 1996. Ex-trapolating leaf CO, exchange to thecanopy: A generalized model of forestphotosynthesis validated by eddy correla-tion. Oecologia 106: 257-265.

Adams MB, Edwards PJ, Wood F, andKochenderfer JN. 1993. Artificial water-shed acidification on the Fernow Experi-mental Forest, USA. Journal of Hydrol-ogy 150: SOS-S19.

Clinton PW, Newman RH, Allen RB. 1995.Immobilization of lSN in forest litter stud-ied by iSN CPMAS NMR spectroscopy.European Journal of Soil Science 46: SSl-556 .

Currie WS, Abet JD, McDowell WH, BooneRD, Magill AH. 1996. Vertical transportof dissolved organic C and N under long-term N amendments in pine and hardwoodforests. Biogeochemistry 35: 471-505.

Davidson EA, Hart SC, Firestone MK. 1991.Agren GI, Bosatta E. 1988. Nitrogen satura-

tion of terrestrial ecosystems. Environ-mental Pollution 54: 185-197.

Almendros G, Frund R, Gonzales-Vila FJ,Haider KM, Knicker H, Ludemann HD.1991. Analysis of i3C and iSN CPMASNMR-spectra of soil organic matter andcomposts. FEBS Letters‘i82: 119-121.

Asner GP. Seastedt TR. Townsend AR. 1997.

Measuring gross nitrogen mineralization,immobilization. and nitrification bv i5Nisotope dilution in intact soil cores. Jour-nal of Soil Science 42: 335-349.

-. 1992. Internal cycling of nitrate insoils of a mature coniferous forest. Ecol-ogy 73: 1148-1156.

The decoupling of terrestrial carbon andnitrogen cycles. Bioscience 47: 226-234.

Axelsson G, Berg B. 1988. Fixation of ammo-nia (i5N) to Pinus silvestris needle litter indifferent stages of decomposition. Scandi-navian Journal of Forest Research 3: 273-279 .

Dise NB, Wright RF. 1995. Nitrogen leach-ing in European forests in relation to ni-trogen deposition. Forest Ecology andManagement 71: 153-162.

Emmett BA, Brittain A, Hughes S, Gorres J,Kennedy V, Norris D, Rafarel R, ReynoldsB, Stevens PA. 1995. Nitrogen additions(NaNO, and NH,NO,) at Aber forest,Wales: I. Resoonse of throuehfall and soil

Baath E, Lundgren B, Soderstrom B. 1981.Effects of nitrogen fertilization on theactivity and biomass of fungi and bacteriain a podzolic soil. Zentralblat fur Bak-teriologie, Mikrobiologie und Hygiene IAbt Originale C 2: 90-98.

1 v

water chemistry. Forest Ecology and Man-agement 71: 45-60.

AcknowledgmentsFwnding for the research summarizedhere was provided by the NationalScience Foundation’s Long TermEcological Research Program (LTER)and the US Environmental Protec-tion Agency.

Benzing-Purdie L, Ripmeester JA, PrestonCM. 1983. Elucidation of the nitrogenforms in melanoidins and humic acid bynitrogen-15 cross polarization-magic anglespinning nuclear magnetic resonance spec-troscopy. Journal of Agricultural and FoodChemistry 31: 913-915.

Berg B. 1986. Nutrient release from litter andhumus in coniferous soils-a mini review.Scandinavian Journal of Forest Research

Fahey TJ, Hughes JW. 1994. Fine root dy-namics in a northern hardwood forestecosystem: Hubbard Brook ExperiementalForest, NH. The lournal of Ecology 82:533-548. -

-.

Fenn ME, Poth MA, Aber JD, Baron JS,Bormann BT. lohnson DW. Lemlv AD,McNulty SG,‘dyan DF, Stottlemyer R. Inpress. Nitrogen excess in North Americanecosystems: A review of predisposing fac-tors, geographic extent, ecosystem re-sponses and management strategies. ECO-logical Applications.

Field CJ, Mooney HA. 1986. The photosyn-

932 Biokience Vol. 48 No. 1 I

organic matter quality in taiga ecosys-

thesis-nitrogen relationship in wild plants,

tems. Canadian lournal of Forest Research

Pages 25-55 in Givinish TJ, ed. On the

13: 795-817. ”

Economy of Plant Form and Function.New York: Cambridge University Press.

Flanagan PW, Van Cleve K. 1983. Nutrientcycling in relation to decomposition and

Fog K. 1988. The effect of added nitrogen on-the rate of decomposition of organic mat-ter. Biological Review 63: 433-462.

Foster DR, Aber JD, Melillo JM, BowdenRD, Bazazz FA. 1997. Forest response todisturbance and anthropogenic stress.Bioscience 47: 437-445.

Foster NW, Beauchamp EG, Corke CT. 1980.Microbial activity in Pinus banksiunaLamb. forest floor amended with nitrogenand carbon. Canadian Journal of Soil Sci-ence 60: 199-209.

-. 1985a. Immobilization of i5N-labeledurea in a jack pine forest floor. Soil Sci-ence Society of American lournal49: 448-4 5 2 .

- - 1985b. Reactions of iSN-labeled ureawith jack pine forest-floor material. SoilBiology a<d Biochemistry. 17: 699-703

Galloway JN. 1995. Acid deposition: Per-spectives in time and space. Water, Airand Soil Pollution 85: 15-24.

Gilliam FS, Adams MB, Yurish BM. 1996.Ecosystem nutrient responses to chronicnitrogen inputs at Fernow ExperimentalForest, West Virginia. Canadian Journalof Forest Research 26: 196-205.

Gundersen P, Emmet BA, Kjonaas OJ,Koopmans CJ, Tietema A. 1998. Impactof nitrogen deposition on nitrogen cyclingin forests: A synthesis of NITREX data.Forest Ecology and Management 101: 37-56 .

Hart SC, Nason GE, Myrold DD, Perry DA.1994. Dynamics of gross nitrogen trans-formations in an old growth forest: Thecarbonconnection. Ecology75: 880-891.

Hedin LO. Armesto Il. lohnson AH. 1995.Patterns of nutrien;loss from unpolluted,old-growth temperate forests: Evaluationof biogeochemical theory. Ecology 76:493-509.

Henriksen A, Kamari J, Posch M, WilanderA. 1992. Critical loads to surface watersin the Nordic countries. Ambio 21: 356-363 .

Herrick GT, Friedland Al. 1991. Winterdessication and injury of subalpine redsoruce. Tree Phvsioloav 8: 23-36.

Hornbeck JW, Smith RB.“i985. Documenta-tion of red spruce growth decline. Cana-dian Journal of Forest Research. 15: 1199-1201

Johnson DW. 1992. Nitrogen retention inforest soils. Journal of EnvironmentalQuality 21: 1-12.

Johnson DW, Van Miegroet H, Lindberg SE,Todd DE, Harrison RB. 1991. Nutrientcycling in red spruce forests of the GreatSmoky Mountains. Canadian Journal ofForest Research 21: 769-787.

Kahl JS, Norton SA, Fernandez IJ, NadelhofferKJ, Driscoll CT, Aber JD. 1993. Experi-mental inducement of nitrogen saturationat the watershed scale. EnvironmentalScience and Technology 27: 565-568.

Keyser P, Kirk TK, Zeikus JG. 1978. Lig-

ninolytic enzyme systems of Phanerochaetechrysosporium: Synthesized in the absenceof lignin in response to nitrogen starvation.

Knicker H, Frund R, Ludemann HD. 1993.The chemical nature of nitrogen in native

lournal of Bacterioloav 135: 790-797.

soil organic matter. Naturwissenschaften

Kiliham K. 1994. Soil &ology. New York:

80: 219-221.

Cambridge University Press.

Koopmans C, Tietema A, Boxman AW. Inpress. The fate of iSN enriched throughfallin two coniferous stands at different ni-trogen deposition levels. Biogeochemistry.

Magill AH, Downs MR, Nadelhoffer KJ,Hallett RA, Aber JD. 1996. Forest ecosys-tem response to four years of chronicnitrate -and sulfate additions at LeadMountain, Maine. Forest Ecoloev andManagement 84: 29-37. “’

Magill AH, Aber JD, Hendricks JJ, BowdenRD, Melillo JM, Steudler P. 1997. Bio-geochemical response of forest ecosys-tems to simulated chronic nitrogen deposi-tion. Ecological Applications 7-402-415.

Matschonat G. Matzner E. 1995. Ouantifica-tion of ammonium sorption in acid forestsoils by sorption isotherms. Plant and Soil169: 95-101.

Mattson KG. 1995. CO, efflux from conifer-ous forest soils: Comparison of measure-ment methods and effects of added nitro-gen. Pages 329-341 in Lai R, Kimble J,Levine E, Stewart BA, eds. Soils and Glo-bal Change. Boca Raton (FL): Lewis Pub-lishers.

McDowell WH, Asbury CE. 1994. Export ofcarbon, nitrogen and major ions fromthree tropical montane watersheds. Lim-nology and Oceanography 39: 111-125.

McNulty SG, Aber JD. 1993. Effects ofchronic nitrogen additions on nitrogencycling in a high-elevation spruce-firstand. Canadian Journal of Forest Re-search 23: 1252-1263.

McNulty SG, Aber JD, McLellan TM, KattSM. 1990. Nitrogen cycling in high eleva-tion forests of the northeastern US inrelation to nitrogen deposition. Ambio19: 38-40.

McNulty SG, Aber JD, Boone RD. 1991.Spatial changes in forest floor and foliarchemistry of spruce-fir forests across NewEngland. Biogeochemistry 14: 13-29.

McNulty SG, Aber JD, Newman SD. 1996.Nitrogen saturation in a high elevationspruce-fir stand. Forest Ecology and Man-agement 84: 109-121.

Micks P. 1994. Soil respiration and nitrogenapplication in two forest stands: Short-and long-term responses. Master’s thesis.University of New Hampshire, Durham,N H .

Murdoch PS, Stoddard JL. 1991. The role ofnitrate in stream acidification in theCatskill Mountains, New York. WaterResources Research 28: 2707-2720.

-. 1993. Chemical characteristics andtemporal trends in eight streams of theCatskill Mountains. New York. Water.Air and Soil Pollution 67: 367-396. ’

Nadelhoffer KJ, Downs M, Fry B, Aber JD,Magill AH, Melillo JM. 1995. The fate ofi5N-labeled nitrate additions to a north-ern hardwood forest in eastern Maine,USA. Oecologia 103: 292-301.

Nilsson J, Grennfelt P. 1988. Critical loadsfor sulphur and nitrogen. Report from aworkshop held at Skokioster, S w e d e n , 19-24 March. 1988. Cooenhaeen: NordicCouncil ofMinisters. * -

Nodvin SC, Van Miegroet H, Lindberg SE,Nicholas NS, Johnson DW. 1995. Acidicdeposition, ecosystem processes and ni-trogen saturation in a high elevation south-ern Appalachian watershed. Water, Airand Soil Pollution 85: 1647-1652.

Nommik H. 1970. Non-exchangeable bind-ing of ammonium and amino nitrogen byNorway spruce raw humus. Plant and Soil33: 581-595.

Ollinger SV, Aber JD, Lovett GM, MillhamSE, Lathrop RG. 1993. A spatial model ofatmospheric deposition for the northeast-ern US. Ecological Applications 3: 459-472 .

Penning de Vries FWT, Brunsting AHM, VanLaar HM. 1974. Products, requirementsand efficiency of biosynthesis: A quantita-tive approach. Journal of Theoretical Bi-ology 45: 358-377.

Peterjohn WT, Adams MB, Gilliam FS. 1996.Symptoms of nitrogen saturation in twocentral Appalachian hardwood ecosys-tems. Biogeochemistry 35: 507-522.

Reich PB, Kloeppel B, Ellsworth DS, WaltersMB. 1995. Different photosynthesis-ni-trogen relations in deciduous hardwoodand evergreen coniferous tree species.Oecologia 104: 24-30.

Richter DD, Johnson DW, Dai KH. 1992.Cation exchange and Al mobilization insoils. Pages 341-377 in Johnson DW.Lindberg SE, eds. Atmospheric Deposi-tion and Forest Nutrient Cycling. NewYork: Springer-Verlag.

Roberge MR. 1976. Respiration rates fordetermining the effects of urea on the soil-surface organic horizon of a black sprucestand. Canadian Journal of Microbiology22: 1328-1335.

Rock BN, Vogelmann JE, Williams DL, Vogel-mann AF, Hoshizaki T. 1986. Remotedetection of forest damage. Bioscience36: 439-445.

Rustad LE, Fernandez IJ, Fuller RD, DavidMB, Nodvin SC, Halteman WA. 1993.Soil solution response to acidic depositionin a northern hardwood forest. Agricul-ture, Ecosystems and Environment 47:117-134.

Rygiewicz PT, Andersen CP. 1994. Mycor-rhizae alter quality and quantity of car-bon allocated below ground. Nature 369:58-60.

Rygiewicz PT, Bledsoe CS, Zasowski RJ.1984. Effect of ectomycorrhizae and solu-tion pH on iSN-nitrate uptake by conifer-ous seedlings. Canadian Journal of ForestResearch 14: 893-899.

Schimel DS. 1988. Calculation of microbialgrowth efficiency from i5N immobiliza-tion. Biogeochemistry 6: 239-243.

Schimel JP, Firestone MK. 1989. Inorganic Nincorporation by coniferous forest floormaterial. Soil Biology and Biochemistry21: 41-46.

Schulze ED. 1989. Air pollution and forestdecline in a spruce (Picea abies) forest.Science 244: 776-783.

Sen S, Chalk PM. 1995. Biological interac-tions between soil nitrogen and alkaline-

November 1998 933

hydrolysing nitrogen fertilizers. Biologyand Fertility of Soils 20: 41-48.

Shortle WC, Smith KT. 1988. Aluminum-induced calcium deficiency syndrome indeclining red spruce. Science 240: 239-240.

Siccama TG, Bliss M, Vogelmann HW. 1982.Decline of red spruce in the Green Moun-tains of Vermont. Bulletin of the TorreyBotanical Club 109: 162-168.

S&bee DG, Larson GL. 1982. Water qualityof streams in the Great Smoky MountainNational Park. Hydrobiologia 89: 97-115 .

-. 1983. Comparison of streams in loggedand unlogged areas of Great Smoky Moun-tain National Park. Hydrobiologia 102:99-111.

Smith WH. 1974. Air pollution effects on thestructure and function of temperate forestecosystems. Environmental Pollution 6:111-129.

Soderstrom B, Baath E, Lundgren B. 1983.Decrease in soil microbial activity andbiomass owing to nitrogen amendments.Canadian Journal of Microbiology 29:1500-1506.

Sprugel DG, Benecke U. 1991. Measuringwoody-tissue respiration in young A&esamabilis trees. Tree 4: 88-98.

Stark JM, Hart SC. 1997. High rates of nitrifi-cation and nitrate turnover in undisturbedconiferous forests. Nature 385: 61-64.

Stoddard JL. 1994. Long-term changes inwatershed retention of nitrogen: Its causesand aquatic consequences. Pages 223-284in Baker LA, ed. Environmental Chemis-

e

Total Environment 1131 67-87.

try of Lakes and Reservoirs. Washington

Tien M. Mver SB. 1990. Selection and char-

(DC): American Chemical Society.Thorn KA, Mikita MA. 1992. Ammonium

fixation by humic substances: A nitrogen-15 and carbon-13 study. Science of the

acterizaiion of mutants of Phanerochaetechrysosporum exhibiting ligninolytic ac-tivity under nutrient rich conditions. Ap-olied and Environmental Microbiologyj6: 2540-2544.

-.

Tietema A. 1998. Microbial carbon and ni-trogen dynamics in coniferous forest floormaterial collected along a European ni-trogen deposition gradient. Forest Ecol-ogy and Management 101: 29-36.

Warfvinge P, Sverdrup H. 1992. Calculat-

GE, Matson PA, Schindler DW, Schlesinger

ine critical loads of acid deoosition withPROFILE-a steady-state soil chemistry

WH, Tilman D. 1997. Human alteration

model. Water, Air and Soil Pollution 63:119-143.

of the global nitrogen cycle: Sources andconsequences. Ecological Applications7: 737-750.

Tietema A, Beier C. 1995. A correlative evaiu-ation of nitrogen cycling in the forestecosystems of the EC projects NITREXand EXMAN. Forest Ecology and Man-agement 71: 143-152.

Tietema A, Verstraten JM. 1992. Nitrogencycling in an acid forest ecosystem in theNetherlands under increased atmosphericnitrogen input. Biogeochemistry 15: 2146.

Van Miegroet H, Cole DW. 1985. Acidifica-tion sources in red alder and Douglas-firsoils: Importance of nitrification. Soil Sci-ence Society of America Journal 49: 1274-1279.

Vitousek PM, Gosz JR, Grier CC, MelilloJM, Reiners WA, Todd RL. 1979. Nitratelosses from disturbed ecosystems. Science204: 469-474.

Vitousek PM, Aber JD, Howarth RW, Likens

Wedin DA, Tilman D. 1996. Influence ofnitrogen loading and species composi-tion on the carbon balance of grass-lands. Science 274: 1720-1723.

Wright RF, Rasmussen L. 1998. Introduc-tion to the NITREX and EXMAN projects.Forest Ecology and Management 101: l-8 .

Wright RF, van Breeman N. 1995. TheNITREX project: An introduction. For-est Ecology and Management 71: l-6.

Yanai RD. 1994. A steadv state model ofnutrient uptake accounting for newlygrown roots. Soil Science Society ofAmerica Journal 58: 1562-1571.

Yano Y, McDowell WH, Kinner NE. Inpress. Quantification of biodegradabledissolved organic carbon in soil solutionwith flow-through bioreactors. Soil Sci-ence Society of America Journal.

Zak DR, Ringelberg DB, Pregitzer KS,Randlett DL, White DC, Curtis PS. 1996.Soil microbial communities beneathPopulus grandidentata grown under el-evated atmospheric CO,. Ecological Ap-plications 6: 257-262.

934 BioScience Vol. 48 No. I1