Fertilization Effects on Soil Organic Matter Turnover and Corn Residue C StorageE. G. Gregorich,* B. H. Ellert, C. F. Drury, and B. C. Liang

ABSTRACTSoil organic matter turnover is influenced by N; thus long-term

fertilization of corn (Zea mays L.) may significantly affect soil organicmatter levels. Effects of fertilization on soil organic matter turnoverand storage of residue C under continuous corn were evaluated usingsoils from a long-term field experiment in Ontario. Total organic Cand natural I3C abundance measurements indicated that fertilized soilshad more organic C than unfertilized soils, the difference accounted forby more C4-derived C in the fertilized soils. About 22 to 30% of thesoil C in the plow layer had turned over and was derived from cornin the fertilized soils; in unfertilized soils only 15 to 20% was derivedfrom corn. Assuming that organic matter turnover follows first-orderkinetics, the half-life of Cj-derived C in the surface 10 cm of bothsoils was the same, about 19 yr. Natural I3C abundance measurementsand estimates from a soil organic matter model indicate that 10 to20% of the added residue C was retained in the soil. Fertilized soilshad more light fraction (LF) C than unfertilized soils. More than 70%of the C in the LF of fertilized soils was derived from corn; inunfertilized soils only 41% was derived from corn. The half-life ofC.i-derived C in the LF was shorter than 10 yr. These results indicatethat adequate fertilization increases crop yields, in turn leading togreater C storage, and that fertilization does not significantly alterthe rate of turnover of native soil organic matter.

THE AMOUNT OF ORGANIC MATTER in soil is a functionof the amount of plant residues returned to the soil

and the rate at which those residues decompose. It isoften reported that organic residue addition is one of themost important factors influencing organic matter levels.Larson et al. (1972) found that changes in soil organicC were linearly related to the amount of residue appliedto soil under continuous corn. Rasmussen et al. (1980)made similar observations and also noted that the changeswere independent of the type of residue applied.

Many soils have received applications of inorganicamendments for several decades, and it is recognizedthat the addition of fertilizer on a regular basis leads toan increase in soil organic matter (Campbell and Zentner,1993; GlendiningandPowlson, 1991). The rate of changein soil organic matter is dependent on a number of factors,including the initial level of organic matter (Campbell etal., 1976) and texture (Bauer and Black, 1981). Liangand Mackenzie (1992) observed that soil C content in-creased by 18% after 6 yr of continuous corn fertilizedat relatively high rates.

The LF is a transitory pool of organic matter betweenfresh plant residues and humified soil organic matter. TheLF concentration in soil is highly variable and dependson the amount and characteristics of C inputs and soilenvironmental factors that affect rates of decomposition

E.G. Gregorich and B.C. Liang, Agriculture and Agri-Food Canada,Centre for Land and Biological Resources Research, Ottawa, ON, K1AOC6 Canada; B.H. Ellert, Agriculture and Agri-Food Canada, ResearchCentre, Lethbridge, AB, T1J 4B1 Canada; and C.F. Drury, Agricultureand Agri-Food Canada, Research Centre, Harrow, ON, NOR 1GO Canada.Received 15 Feb. 1995. "Corresponding author.

Published in Soil Sci. Soc. Am. J. 60:472-476 (1996).

(Gregorich and Janzen, 1995). The LF has been sug-gested as a sensitive indicator of changes of soil organicmatter because of its responsiveness to management prac-tices (Gregorich et al., 1994). Field studies have sup-ported this concept and showed that soil organic matteraccumulation is linked to accumulation of LF (Wanderet al., 1994). Studies in western Canada have shownthat application of N fertilizer significantly increases LFC in continuous cropping systems (Janzen et al., 1992;Biederbeck et al., 1994).

Data from field experiments with 14C-labeled plantmaterial have been used to develop models that describethe decomposition and turnover of soil organic matter(Jenkinson, 1977; Voroney et al., 1989). These modelsusually partition the incoming residue into two compart-ments, each decomposing by a first-order process, butone much more quickly than the other.

The introduction of C4 plants to soil previously devel-oped under Cs vegetation results in the soil organic mattercontaining two isotopically different sources of C andprovides a means of partitioning soil organic matter asto origin. The natural 13C abundance method has beenused to estimate soil organic matter turnover both intropical (Martin et al., 1990) and temperate soils (Bales-dent et al., 1987; Gregorich et al., 1995).

The objective of this study was to use the I3C isotopicmethod to determine the long-term effects of fertilizationon the turnover of soil organic matter and storage of Cderived from corn residue in a medium-textured soil insouthwest Ontario.

METHODS AND MATERIALSThe soil used in this experiment is a Brookston clay loam,

a poorly drained soil (clayey, mixed, mesic Typic Haplaquoll)located at the Eugene F. Whelan experimental farm (Agricul-ture and Agri-Food Canada, Woodslee, Ontario). The averageannual temperature at the experimental site is 8.7°C; theaverage maximum growing season (May-September) tempera-ture is 24°C and the average minimum temperature is 13°C.The average annual total precipitation is 876 mm, with rainfallaccounting for 769 mm. The average maximum evapotranspira-tion rate is 654 mm. The average textural analysis for thissoil is 280 g kg-' sand, 350 g kg~' silt, and 370 g kg"1 clay.Although complete records of agricultural management of theexperimental site prior to 1954 are not available, it is knownthat alfalfa (Medicago saliva L.) and red clover (Trifoliumpratense L.) were grown for several years between 1940 and1954. The land was summer fallowed in 1954; tile drains(100-mm diam.) were installed in 1955 at a depth of 71 cmand a spacing of 12.2 m. The plots, 76.2 m long by 12.2 mwide, were centered longitudinally above the tile drains (Boltonet al., 1970). Corn was grown on all plots from 1956 to 1958to reduce residual effects and obtain uniform data. In 1959,12 plots, consisting of six cropping treatment plots with fertil-izer and six without, were implemented. A continuous corntreatment with fertilizer and one without fertilizer as wellas the fertilized continuous bluegrass (Poa pratensis L.) sodtreatments were used in this study. The fertilized treatmentsreceived 16.8 kg N ha~', 67.2 kg P ha"1, and 33.0 kg K

472

GREGORICH ET AL.: FERTILIZATION EFFECTS ON SOIL ORGANIC MATTER 473

ha~' at planting and 112 kg N ha~' of NH4NO3 sidedressed(incorporated).

Corn was planted in 1.0-m rows at a density of 37 000 to50 000 seeds ha~' from 1956 to 1982 and at 55 000 seeds ha~'from 1982 to 1990. Herbicides were applied as required tocontrol weeds. Ten rows of corn (33 m in length) were har-vested from each treatment. Corn residues were incorporatedinto the soil by moldboard plow each autumn whereas thebluegrass was left on the soil surface.

Prior to seeding in the spring of 1991, four replicate soilsamples were obtained from the plow layer (0-26 cm) of eachtreatment. One sample was also taken from the 15-cm layerbelow the plow layer in each treatment for a total samplingdepth of approximately 40 cm. Soil samples were mixed, airdried, and sieved through a 2-mm sieve to remove visibleplant fragments. The LF organic matter from the surface 10cm of each soil was separated by flotation on a solution ofNal with the density adjusted to 1.8 g cm~3 (Gregorich andEllert, 1993). The organic C content of samples was determinedusing a Carlo Erba T1500 elemental analyzer (Carlo ErbaStrumentazione, Milan, Italy). Because the soils were free ofcarbonates to a depth of 60 cm, total soil C is equivalent tosoil organic C. Bulk density of the plow layer was determinedusing soil cores; bulk density values for depths below the plowlayer were obtained from published data measured at the samesite (McKeague et al., 1987).

The 8I3C values were determined by combustion of 3 mgC mixed with CuO (1:50) in the vacuum-combustion systemdescribed by Swerhone et al. (1991). Carbon dioxide generatedin the combustion tubes was separated by cryogenic distillation,collected in breakseals, and analyzed on a VG-SIRA 12 isotoperatio mass spectrometer (Isotech, Middlewich, England). Thenatural abundance of heavy isotopes is expressed as parts perthousand (%o), relative to the international PDB standard usingdelta'units (8). The 813C value is calculated from the measuredC isotope ratios of the sample and standard gases as

513C(%0) = [(^sample ~ /?standard)//?standanl]103 [1]

where R is the I3C/12C ratio of the sample or standard gas.The soil under the sod treatment was taken as the reference,

and the proportion of soil C derived from corn residues sincethe experiment was initiated (X) was calculated as

X=(6-5s)/(6cr-6s)100 [2]where 8 = 8I3C value of sample from corn soil, 8cr = —13. l%o,the mean 813C value of corn residues measured on four repli-cates of each of leaves, cobs and grain, stalks and roots; and8S = 8'3C value of the sod soil. The mean 813C value for thesod profile was — 26.4%o; in a nearby forest soil, the meanvalue for a profile with an equivalent depth was —26.9%o.

The grain yields for the 32-yr period showed considerableyear-to-year variability, which was presumably due to climaticfactors (Drury and Tan, 1995). After 1960, the fertilizedtreatment consistently had greater corn yields than the unfertil-ized treatment. The average annual grain yield (dry weight)on fertilized plots was 5.09 Mg ha~'; on the unfertilized plotsit was 1.45 Mg ha~' for the 32-yr period. The actual amountof corn straw returned to the soil during the experimentalperiod was not measured. The amount of C from corn residuesthat was returned to the soil was estimated using the grainyields, assuming a C content of 45% in plant tissues and aroot/shoot ratio of 0.15 (Geisler and Krutzfeldt, 1984), whichis similar to the value estimated by Balesdent and Balabane(1992) using natural I3C abundance in a study of maize root-derived soil organic C. The harvest index (ratio of grain drymatter/total aboveground dry matter) was assumed to be 45 %

for the fertilized corn and 30% for the unfertilized corn (Geislerand Krutzfeldt, 1984). A minimum value of 3 Mg ha"1 y r~ 'of corn residue was assumed for the unfertilized corn whenthe crop failed (no grain yield), as occurred in 1983 and 1985,and when the grain yield was <1 Mg ha"' yr"'. The totalamount of corn residue returned to the soil was estimated tobe 286 Mg ha~' for the fertilized system and 159 Mg ha"1

for the unfertilized system (Fig. 1).The amount of C remaining in the soil was estimated using

the model used by Woodruff (1949):

dC/dt = A- FkFC - (1 - F)ksC [3]which, integrated, gives

C, = FCoe-m + (1 - F)C0e-*5' + FA/kF(l - e^')+ (1 - F)A/ks(l - e-k5') [4]

whereC, = corn residue-derived C in the soil at time t, Mg

ha"1

Co = corn residue-derived C in the soil at time 0 = 0A = rate of corn-residue C addition, Mg ha"1 yr~ 'F = fraction of corn-residue C in the fast fraction =

0.70kf = first-order decay coefficient for the fast fraction,

yr-1 = 2.77ks = first-order decay coefficient for the slow fraction,

yr-1 = 0.0866The fraction of corn-residue C in the fast fraction and thedecay coefficients for the fast and slow fraction were obtainedfrom Jenkinson's (1977) field study of the decomposition of'4C-labeled ryegrass (Lolium multiflorum Lam.). Results ofthat study indicated that 70% of the incoming plant materialdecomposes with a half-life of 0.25 yr, and the remainderwith a half-life of 8 yr. Voroney et al. (1989) obtained similarparameters from decomposition studies on soils from Saskatch-ewan, Canada. Other work by Jenkinson and Ayanaba (1977)indicated that the rates and the pattern of decomposition ofmaize leaves and ryegrass were similar.

1u

200 -

150 -

100

50 -

Fertilized cornUnfertilized com

1960 1975 1980

Year

Fig. 1. Estimated cumulative corn residue returned to the soil (stoverand roots) under fertilized and unfertilized treatments with thefollowing assumptions: (i) harvest index was assumed to be 45%for fertilized and 35% for unfertilized corn, and (ii) a minimumvalue of 3 Mg ha~' of corn residue was assumed to return to thesoil annually.

474 SOIL SCI. SOC. AM. J., VOL. 60, MARCH-APRIL 1996

Table 1. Natural abundance of I3C(5I3C), total organic C, C4-derived C, Cj-derived C, and C4-derived C/organic C in fertilized andunfertilized corn soils.

Crop

Fertilized corn

Unfertilized corn

Sod

Depth

cm0-10

10-2626-420-10

10-2626-410-10

10-2222-40

813C

%o-22.7 ± 0.3t-23.1 ± 0.4

-25.7-24.0 ± 0.3-24.1 + 0.3

-25.2-26.7 + 0.3-26.0 ± 0.1

-26.5

Total C

21.8 + 1.721.7 + 0.9

6.519.5 ± 0.519.8 + 0.6

7.451.3 + 3.230.1 ± 3.2

11.1

C4-derived C

v lip-'6.44.80.43.82.90.7-—-

Cj-derived C

15.416.96.1

15.716.96.7

51.330.111.1

C4/organic C

%30226

201510-—-

t Mean + standard deviation.

RESULTS AND DISCUSSIONTurnover and Storage of Carbon

The incorporation of crop residues with different Cisotope signatures resulted in an in situ 13C labeling ofsoil organic matter since 1959, when the experiment wasinitiated. This allowed us to distinguish between the Cs-and C4-C in these soils. The 8 C values of the soilorganic matter sampled from the sod soil ranged from-26.0 to -26.7%o. Organic C from the corn soil plowlayers was enriched in 13C compared with the B horizonsbecause of the addition of C4-C material (Table 1), whichis consistent with the results of Gregorich et al. (1995),who found that 88% of the C derived from corn residuewas found within the plow layer. The similar 813C valueswithin the corn plow layers reflect the homogenizationof the surface soil resulting from annual mixing byplowing. After 32 yr of fertilization, the 8I3C value inthe surface 10 cm increased from — 26.7%o under grassto — 22.7%o under corn. The plow layer of the fertilizedcorn soil was more enriched in 13C than that of theunfertilized corn. From 22 to 30% of the organic C inthe plow layer of the fertilized corn soil had turned overand was derived from corn, whereas only 15 to 20% ofthe organic C was derived from corn in the unfertilizedsoil (Table 1). The concentration of C3-C in the plowlayer the soils (15-17 g kg"1 soil) was unaffected byfertilizer. The corn-derived C found below the plowlayer was small (6-10%) and probably derived fromcorn roots and soluble corn exudates.

To account for changes in soil density that may haveoccurred during the experiment, the data in Table 1 wereexpressed on a mass basis (Table 2), and the stocks of

organic C compared. Since the start of the experimentthe amount of organic C in both corn soils decreasedby about 32 to 38%, assuming no change has occurredunder the bluegrass sod. This agrees with the findingsof Mattel and Deschenes (1976), who reported that lossesof organic C with annual cropping in eastern Canadaranged from 30 to 40% in soils where erosion was notsignificant.

As found by others (Campbell and Zentner, 1993;Glendining and Powlson, 1991), the unfertilized soil hadless organic C than the fertilized soil (Table 2). Thisdifference was accounted for by the larger amount ofcorn-derived C present in the fertilized soil (19.5 Mgha"1) than in the unfertilized soil (12.4 Mg ha"1). Theamount of C3-C was the same in both soils, approxi-mately 70 Mg ha"1 (Table 2). Assuming that the initialorganic C (A0) decayed exponentially with time (t) (i.e.,A, — A0Q~kr), the mean decay constant (k) of the Cs-derived C in the surface 10 cm of both soils would be0.036 yr"1, which is equivalent to a half-life of 19 yr.Gregorich et al. (1995) studied organic matter turnoverin similar Ontario soils that had been under continuouscorn for 25 yr and estimated a half-life of 13 yr forCs-C in the 0- to 10-cm layer of soil.

The sod soil had a larger amount of LF organic matterthan both corn soils and the fertilized corn soil hadmore than the unfertilized corn soil (Table 3). The LFaccounted for 10% of the total soil C in the surface 10cm of the sod soil, but only 5 and 2% in the fertilizedand unfertilized soils respectively. The LF of the ferti-lized soil was enriched in 13C (a 813C value of - 17.2%o),which indicated that 72% of the C was derived from

Table 2. Amounts of total organic C, C4-derived C, and C3-derived C in soils under fertilized and unfertilized corn after 32 yr.

Crop

Fertilized corn

Unfertilized corn

Sod

Horizon

AplAp2

B

AplAp2

B

AlA2B

Depth

cm0-10

10-2626-42Total0-10

10-2626-41Total0-10

10-2222-40Total

Organic C

28.5 + 1.3t46.2 + 2.0

14.689.3

25.4 + 2.239.2 + 2.1

16.781.3

62.7 + 4.844.0 + 4.9

25.0131.7

C4-derived C

Mgha-'8.4 + 0.5

10.2 + 1.00.9

19.55.0 + 0.45.7 + 1.0

1.712.4-——-

C3-derived C

20.1 ± 1.536.0 + 2.7

13.769.7

20.4 + 2.233.5 + 5.6

15.068.962.744.025.0

131.7

t Mean ± standard deviation.

GREGORICH ET AL.: FERTILIZATION EFFECTS ON SOIL ORGANIC MATTER 475

Table 3. Amounts of total organic C, C4-C, and C,-C in the lightfraction organic matter in the surface 10 cm of fertilized andunfertilized corn soils after 32 yr.

Light fraction

Crop 8"C C,-CC4-derived Cj-derived

Total C C C

%o - g C k g - ' soilFertilized

corn -17.2 ± 0.2t 72 1.11 ± 0.08 0.80 0.31Unfertilized

corn -21.7 ±0.5 41 0.48 ± 0.08 0.20 0.28Sod_______-27.7 ± 0.6 - 4.91 ± 2.74 _____4.91t Mean + standard deviation.

corn. Approximately 41% of the LF C in soil underunfertilized corn was derived from corn. If the referencesod LF changed with time or decay, then these estimatedproportions may not be precise. For example if the sodLF were l%o less negative, then the amount of C derivedfrom corn in the fertilized LF would be 70%; if it werel%o more negative the proportion would be 74%. Therelatively greater amount of LF C in the fertilized cornsoil indicates that this pool of organic matter is susceptibleto changes in management practices (Gregorich and Jan-zen, 1995). This partially decomposed organic matteralso accounted for a greater proportion of the C4-C inthe fertilized corn soil than in the unfertilized corn soil.The LF accounted for >12% of the C4-C in the surface10 cm of soil under fertilized corn, but only about 5%of the C4-C under unfertilized corn. The larger amountsof corn residue returned to the soil (discussed below)apparently have a direct effect on the amount of LF C.

The C3-C content of the LF of both fertilized andunfertilized soils was the same, 0.3 g kg"1 soil. Thelabile nature of the LF is shown by the fact that 94%of the original €3 -C had turned over since the start ofthe experiment. The estimated half-life of the LF Cs-Cin both corn soils was about 8 yr, the same as thatestimated by Gregorich et al. (1995) in a soil that wasunder continuous corn for 25 yr. This half-life is similarto the half-lives reported in studies using 14C-labeledplant residues (Jenkinson, 1977; Voroney et al., 1989).Angers et al. (1995), who used natural 13C abundanceto estimate the half-life for macroorganic matter in Que-bec soils that had been under continuous corn for 11 yr,obtained a value of 10 yr.

Estimates of Storage from Yield DataThe amount of C remaining in the fertilized and unfer-

tilized soils was estimated from assumed crop residueinput using a double exponential model (Woodruff, 1949;Bartholomew and Kirkham, 1960). The model assumesthat soil organic matter is heterogeneous and composedof two components that decompose at different rates(Jenkinson, 1977). The estimated amount of C stored insoil, derived from the annual additions of corn residue,including stover and roots, was 14.1 Mg ha~' for the

O Fertilized comn Unfertilized corn

1965 1985 1990

YearFig. 2. Soil organic C derived from corn residue under fertilized and

unfertilized treatments, assuming that the decomposition of thecorn residue in soil follows a double exponential model (Woodrow,1949). The C content of the corn residue was assumed to be 45%.

fertilized soils and 7.8 Mg ha"1 for the unfertilized soils(Fig. 2).

Fertilized soil had larger amounts of C derived fromcorn (22-30% of the total C) than unfertilized soil (15-20% of the total C). In the fertilized soil, the estimateby the above-cited model of the amount of C4-C re-maining after 32 yr (14.1 Mg ha"1) was slightly lowerthan the amount measured by the natural abundance of13C (19.5 Mg ha"1). In the unfertilized soil, the amountof C4-C measured by the natural abundance of 13C (12.4Mg ha"1) was larger than that estimated by annual inputsof corn residue. These data indicate that after 32 yr ofcropping to corn, between 10 and 20% of corn residue-Cinputs was retained as total soil C, similar to the findingsof Balesdent et al. (1990).

The smaller amount of corn-derived C estimated forthe unfertilized soil by the decomposition model mayhave been due to the underestimation of the root/shootratio. Geisler and Kriitzfeldt (1984) reported that theratio of shoot to root for dry matter increased withincreasing N concentration in the soil. In addition, ac-cording to Jenkinson (1977), a two-compartment modelis an oversimplification of the real processes of decayin soil. It takes no account of the formation and decayof biomass, nor of the formation and decay of the inertmaterial that radiocarbon dating has shown is present.

CONCLUSIONSSoils under continuous corn, fertilized for >30 yr,

had greater amounts of soil C than systems that wereunfertilized. The difference between the soils was ac-counted for by estimating the C4-derived C using thenatural 13C abundance method. We estimated that infertilized soils, from 22 to 30% of soil C in the plowlayer had turned over and was derived from corn resi-dues, whereas in unfertilized soils, only 15 to 20% wasderived from corn residue. Only a small portion (10-20%) of the corn residue C inputs remained in thesesoils after 32 yr. Estimated half-lives of the surface soil

476 SOIL SCI. SOC. AM. J., VOL. 60, MARCH-APRIL 1996

C indicated that fertilization did not significantly alterthe decomposition rate of C3-derived C.

The LF organic material in the surface 10 cm of bothsoils was labile; between 40 and 70% of the LF C wasderived from corn residues. However, the LF C contentof fertilized soils was more than two times that of theunfertilized soils, and the difference between the soilswas related to the amount of C4-derived C. The estimatedhalf-life of C3 -C in the LF was shorter than 10 yr, whichis consistent with findings of other studies conducted ontemperate soils.

ACKNOWLEDGMENTSB.C. Liang acknowledges the Natural Sciences and Engi-

neering Research Council of Canada for providing a post-doctoral scholarship. This research was conducted as part ofthe Evaluating Changes in Soil Organic Matter study of theNational Soil Conservation Program. Funding for this researchwas provided by Research Branch, Agriculture and Agri-FoodCanada, and the Program of Energy Research and Development(Natural Resources Canada). The "C determinations wereperformed at the Stable Isotope Lab., Dep. of Soil Science,Univ. of Saskatchewan, Saskatoon, Saskatchewan.