Glasener Et Al.: Nitrogen-15 Labeled Legume Nitrogen Sources

L are used commonly in agricultural systems Legume mulches are important sources of N for cereal crop producas a source of N for subsequent crops and for maintion, particularly for organic and resource-poor producers. A field taining soil N levels. This use is particularly important study was conducted using a direct method to determine if the amount in the humid tropics where N fertilizers often are not of N in cereal crops derived from either the shoots or roots of precedeconomically feasible due to poor market and infraing tropical legume cover crops was affected by their chemical compostructure development (Palm and Sanchez, 1991). To sition and mineralization potential. Desmodium ovalifolium Guill. & date, studies attempting to quantify the legume N contriPerr. [ D. adscendens (Sw.) DC. and Pueraria phaseoloides (Roxb.) bution to subsequent crops have been conducted mainly Benth.], were grown in 6.0-m2 microplots and foliar-labeled with 99 in temperate agroecosystems and have dealt primarily atom % 15N urea. A cereal sequence of maize (Zea mays L.)–rice (Oryza sativa L.)–maize followed the legumes. Cereal accumulation with aboveground legume N, ignoring root N because of of legume N from either the shoot (shoot leaf litter) or the rootthe difficulty of harvesting roots and nodules. Moreover, soil sources was evaluated by spatially separating the legume N these assessments of N cycling in cover crop based prosources. This was achieved by interchanging surface applications of duction systems have often relied on indirect methods nonlabeled and 15N-labeled legume shoots with in situ 15N-labeled and that evaluate plant and soil N pools (Ditsch et al., 1993; nonlabeled legume roots. Initially the Desmodium shoot N source Luna-Orea and Wagger, 1996), N release from cover contained 316 kg N ha 1 and roots contained 12.5 kg N ha 1. Pueraria crop residue (Ranells and Wagger, 1991; Luna-Orea et shoots and root N sources initially contained 262 and 14.8 kg N ha 1, al., 1996), and N uptake by a summer crop (Hargrove, respectively. About 90 g kg 1 of the initial N of each legume shoot 1986; Clark et al., 1994). was recovered in the total aboveground tissues from the three cereal crops, while 490 g kg 1 of Desmodium and 280 g kg 1 of Pueraria Nitrogen-15 methodology is useful for resolving N root-soil N sources were recovered. Of the 181 kg N ha 1 accumulated dynamics, whereby 15N-labeled legume cover crops are aboveground by the cereal sequence, the contribution of shoot plus harvested and applied as N sources for subsequent grain root-soil N sources was 200 g kg 1 from Desmodium and 150 g kg 1 crops (Varco et al., 1989; Jordan et al., 1993; Harris et from Pueraria. Cereal N was derived primarily from mineralization al., 1994). Varco et al. (1993) found that 600 g kg 1 of of soil organic matter present before the legumes and possibly from the N was mineralized and subsequently lost from 15NN deposition (precipitation and dry) occurring during the cereal crop labeled hairy vetch (Vicia villosa Roth) residue 30 d sequence. After harvest of the last cereal crop, 13 and 180 g kg 1 of after surface application, yet an average of only 60 g the initial legume N was present as inorganic and organic N fractions, kg 1 was recovered as soil inorganic N for two growing respectively, in the top 75 cm of soil. Even though Pueraria shoots had a lower C:N ratio and concentration of polyphenols than Desmodium seasons. In Australia, Ladd and associates (Ladd et al., shoots, the relative contributions of the shoot N source were similar 1981, 1983; Ladd and Amato, 1986) reported fieldfor both legumes. Decomposition of legume residues, particularly grown wheat (Triticum aestivum L.) recovered between legume shoots, make a meaningful contribution to the N economy of 11 and 280 g kg 1 N from 15N-labeled medic (Medicago cereal crops grown in the tropics. The legume cover crops (root littoralis L.) and an additional 40 g kg 1 recovery by a shoot) contributed nearly 280 g kg 1 of the aboveground N in the second wheat crop. When 15N-labeled red clover (Trifofirst cereal crop and as much as 110 g kg 1 of the N in the third crop lium pratense L.) residue was applied at maize planting, during the 15-mo sequence of cereals. 150 g kg 1 was recovered in the harvested crop and 570 g kg 1 was retained by the soil (Harris et al., 1994). Of K.M. Glasener, American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America, 900 Second Street, these studies, that of Varco et al. (1989) made an indirect NE, Suite 205, Washington, DC 20002; M.G. Wagger and R.J. Volk, estimate of the contribution of legume root N to subseDep. of Soil Science, North Carolina State Univ., Box 7619, Raleigh, quent crops. Only a few direct estimates of N contribuNC 27695-7619; C.T. MacKown, USDA-ARS, Grazinglands Research tion from legume shoot and root residues are available Lab., 7202 W. Cheyenne Street, El Reno, OK 73036. Research supported by the Trop Soils Program and funded in part by Grant no. (Harris and Hesterman, 1990; Russell and Fillery, 1996). DAN-1311-G-00-1049-00 from the USAID. Received 21 June 2001. The objectives of this study were to: (i) quantify the Nov. 2000. *Corresponding author ([email protected]). N contribution from two tropical legume cover crops of differing chemical characteristics (i.e., potentially difPublished in Soil Sci. Soc. Am. J. 66:523–530 (2002). 524 SOIL SCI. SOC. AM. J., VOL. 66, MARCH–APRIL 2002 Fig. 1. Monthly and 11-yr mean (1982–1993) precipitation and temperature at La Jota Experiment Station, Bolivia, 1993 to 1995. Temperature not recorded during March and April 1994. Horizontal bars indicate crop duration of legumes and cereal crops. blocks. Desmodium and Pueraria seed were coated with a ferent rates of decomposition and nutrient release) to sugar-water sticker, inoculated with CIAT 4099 and CIAT a subsequent maize–rice–maize sequence using 15N 3287 Rhizobia, respectively, pelleted with 100 g of CaCO3 methodology, (ii) determine the effect of plant part (powder) per kg of seed, and planted in 50-cm wide rows at (shoot vs. root) on recovery of legume-derived 15N by a seeding rate of 27 kg ha . During an 8-mo growth period, grain crops, and (iii) quantify the legume-derived 15N legumes were maintained weed free by hand and pest free remaining in various soil N fractions at the end of the with applications of malathion (O,O-dimethyl phosphorodithcereal crop sequence. ioate of diethyl mecaptosuccinate). Weeds were allowed to grow in the fallow plots during the 8-mo legume growth period. MATERIALS AND METHODS At termination of legume cover crop growth, weeds in the fallow plot were cut and left on the soil surface. A field study was conducted at La Jota Research Station (16 01 S and 65 25 W; 400 m above sea level), situated in the sub-Andean foothills of eastern Bolivia, on a gently sloping Cover Crop Nitrogen-15 Labeling (0–2%) fine-loamy, mixed, isohyperthermic, Typic DystroIn mid-December (early in legume development to minipepts. Selected soil physical and chemical characteristics of mize root damage), two 6-m microplots containing five lethe surface 20 cm are as follows: 41% sand, 41% silt, 18% gume rows were sectioned off within each plot using galvaclay, bulk density 1.02 Mg m , 13.8 g C kg , 1.8 g N kg , nized steel collars (20 cm high) driven into the soil ≈15 cm to effective CEC 7.0 cmolc kg 1 (1 M NH4OAc, pH 7; 1 M KClreduce lateral flow of water and N. Sixteen microplots were extractable Al), 880 g kg 1 Al saturation, and pH 4.6 (1:1, installed, eight for each species representing two N sources soil:water). Rainfall and temperature data for the site, covand four replications. Within each legume plot, one of the ering the 21-mo study period, are presented in Fig. 1. microplots was randomly selected for foliar treatment with N. Shoots from the N-labeled microplot were exchanged Cover Crop Establishment later with shoots from the adjacent unlabeled microplot to spatially separate N-labeled root from N-labeled shoot From June through October 1993, a 1 ha, 6to 19-yr-old sources of legume N for the following maize–rice–maize sesecondary forest was cut and all aboveground vegetation was quence. carried off the experimental site and burned. Negligible litter Nitrogen-15 was applied to growing legumes beginning in remained on the soil surface, which was essentially undismid-May 1994 when Desmodium shoot dry matter accumulaturbed. The top 75 cm of soil contained 98 kg N ha 1 as tion was 9.7 Mg ha 1 (184 kg N ha 1 ) and Pueraria was 4.2 inorganic N just before the legume cover crops were planted. Mg ha 1 (144 kg N ha 1 ). The soil surface beneath the legume In mid-November, 10by 15-m plots of two tropical legumes canopies was not visible. Desmodium and Pueraria began flow[D. ovalifolium and P. phaseoloides (tropical kudzu)] and fallow check were established and randomized in each of four ering just prior to initiation of N foliar labeling. To prevent GLASENER ET AL.: NITROGEN-15 LABELED LEGUME NITROGEN SOURCES 525 seed set and to minimize N translocation from roots to shoots, containing 60 to 100 g total N was transferred into a 104mL specimen cup, and 0.4 g MgO, 0.2 g of Devarda’s alloy flowers were clipped during the N labeling period, dried, and later combined with the shoot N source. A total of 2.58 g (ground to a powder with a ball-mill), and 7.0 g K2SO4 (to increase solution osmotic potential and reduce H2O vapor N (4.3 kg N ha 1 ) as urea labeled with 99 atom % N was applied to each microplot in four equal foliar applications of pressure) were added. A 6-mm diameter acid-washed filter paper disk, acidified with 10 L of 2.5 M KHSO4, was sus0.645 g N each (18 and 30 May and 6 and 13 June). A 12-d interval occurred between the first and second N applications pended above the extract-reagent mixture to trap volatilized NH3. The cup was capped, gently swirled, and incubated at because of rainy conditions. Each microplot was divided into quadrants to assure uniform distribution of the N solution 40 C for 6 d. Following incubation, the paper disk was removed, dried in a vacuum microcentrifuge, wrapped in a Sn sprayed on the upper plant canopy using a hand-held mister. The solution contained 300 mL H2O (volume necessary to capsule and analyzed for N. The soil organic N fraction was determined by subtracting the inorganic N from total N. maximize foliar coverage while minimizing drip, potential N loss, and soil contamination), a wetting agent to maximize absorption, and the N-labeled urea. Immediately following Cereal Cropping Sequence each foliar application, microplots were covered with a transparent plastic shelter for 2 d to prevent rainfall from washing On 24 June 1994, the harvested unlabeled and N-labeled N off the leaves. legume shoots and litter were used to establish microplots for the cereal cropping sequence. Within each replicate, Nlabeled shoots and litter from the original foliar N-labeled Sampling and Analysis of Nitrogenmicroplot were removed and surface-applied to an adjacent 15-Labeled Microplots microplot containing only unlabeled roots, that is, legumes in Ten days after the final N foliar split-application, legume this new microplot had not received the foliar N application. shoots were cut at the soil surface and removed and then leaf Then, an equal weight of shoots plus litter (means for Desmodlitter was collected. A subsample of shoots was weighed, airium and Pueraria were 19.2 and 8.6 Mg ha , respectively) dried, ground to 1 mm, and analyzed for polyphenolic and from this unlabeled legume microplot were surface-applied lignin concentrations using the procedure described by Palm to the microplot containing only the N-labeled root-soil N and Sanchez (1991) and Van Soest (1963), respectively. The source. After redistributing surface residue, the microplots remaining shoots and leaf litter were weighed, subsampled were left undisturbed for 7 d before annual crops were planted. for moisture determination, oven dried at 65 C, reweighed, Legumes outside the microplots were cut and left on the soil and ground to 1 mm. Obtaining representative subsamples is surface as mulch. frequently a problem with N field-tracer studies; thus the Maize (‘Across 8136’), the first cereal crop of the sequence, oven-dried and ground plant subsamples were thoroughly was sowed 30 June 1994 (≈30 700 plants ha 1 ) using a jab mixed in a twin-shell blender, and 5-g subsamples were ground planter and a 65by 50-cm grid pattern within the microplots, to a fine powder in a dental amalgam ball mill. Finally, plant the border area between microplots, and the fallow check subsamples (≈5 mg) containing at least 100 g N were weighed plots. Post-emergence weed control and Desmodium regrowth into Sn capsules for total N and N determinations with an proved challenging throughout the first maize crop, necessitatautomated flash-combustion analyzer coupled to an isotope ing bi-weekly applications of 1 kg a.i. ha 1 of paraquat [1, 1’ratio mass spectrometer (RoboPrep and TracerMass, Europa dimethyl-4 4’-bipyridinium ion (dichloride salt)] in combinaScientific, Crewe, Cheshire, UK). tion with hand weeding. Pests, particularly a severe Spodoptera The present study was designed to leave the labeled roots frugiperda (J.E. Smith) infestation triggered by a 10-yr record undisturbed and to follow the recovery of N from the comdrought, were controlled with weekly sprayings of methamibined root-soil system by subsequent crops. Immediately foldophos [O,S-dimethyl phosphoramidothiate] and malathion. lowing removal of legume shoots, four soil cores were taken Approximately 6 wk after planting the first maize crop, all at five depths (0–10, 10–25, 25–40, 40–60, and 60–75 cm), microplots were fertilized by hand with 29, 111, 21, and 16 kg air-dried, and sieved (2mm). Obtaining representative soil ha 1 of P, K, Ca, and Mg, respectively. Fertilizer trials had subsamples is often a problem with N field-tracer studies, not been conducted in this region, so fertilizer tests for macromore so than with plants due to the inherent variability of nutrients were made and applications were conservatively calsoil N. Thus the air-dried, sieved root-soil samples were thorculated as twice the nutrient removal from an average local oughly mixed in a twin-shell blender and 5-g subsamples were maize and rice yield of 2.3 and 2.9 Mg ha , respectively. ground to a fine powder in a dental amalgam ball mill. Finally, Aboveground parts of mature maize plants were harvested subsamples of root-soil (≈40 mg) containing ≈25 to 80 g N on 7 Dec. 1994 from each microplot. Grain and stover were were weighed into Sn capsules for total N and N determinaweighed, and subsampled for moisture, total N, and N detertions with an automated flash-combustion analyzer coupled minations. Grain yield was adjusted to a 155 g kg 1 moisture to an isotope ratio mass spectrometer. basis. After removal of the maize plants an equal weight of unlabeled maize stover was surface-applied to the microplots. Inorganic Soil Nitrogen and Nitrogen-15 Analysis Upland rice (‘Bluebell’) followed maize in the cereal sequence and was planted 15 Dec. 1994 (≈100 000 plants ha 1 ) Air-dried and sieved soil samples were shaken with 2 M on a 40by 25-cm planting grid with a jab planter. Before rice KCl (20 g soil 100 mL 1 ) for 1 h, filtered, and analyzed for emergence, the area was sprayed with paraquat (1 kg a.i. total inorganic N (NH4 NO3 ) with an automated flow injecha 1 ) to kill existing vegetation. As with the maize crop, posttion ion analyzer (QuickChem IV, Lachat Instruments, Milemergent weed control and Desmodium regrowth proved waukee, WI). Inorganic N was isolated for N analysis by the challenging, requiring five hand cultivations. Additionally, method of Brooks et al. (1989) as follows. An extract volume weekly spray applications of malathion were necessary to control various insect stresses. Fertilization rates for all nutrients 1 The use of trade names in this publication does not imply endorsewere doubled for the rice crop as visual P deficiencies were ment by the North Carolina Agricultural Research Service or USDAobserved in ≈10% of the maize plants. Aboveground parts of ARS of the product named, nor criticism of similar ones not mentioned. mature rice plants were harvested 7 Apr. 1995 from each 526 SOIL SCI. SOC. AM. J., VOL. 66, MARCH–APRIL 2002 Table 2. Recoveries of foliar-applied 15N-urea in the shoot and microplot, rice grain and straw were weighed, and subsampled root-soil components of microplots grown with Desmodium for moisture, total N, and N determinations. The labeled rice and Pueraria legume cover crops. straw was then returned to the respective microplots, while all grain was removed from the field and oven-dried at 65 C Root-soil (0–75 cm) Legume Microplot for 24 h. Rice yields were not adjusted to a specific moisture cover crop Shoot† Inorganic Organic Total total basis due to logistical constraints, though local experience indicates oven drying at 65 C for 24 h yields an approximate % 15N recovery moisture basis of 80 g kg . Desmodium 73.8a‡ 0.3b 0.9b 1.2b 75.0a Pueraria 75.8a 0.6a 3.4a 4.0a 79.8a Maize followed rice as the last crop in the cereal sequence CV, % 10 22 34 28 10 and was planted 3 May 1995. Cultural practices were the same as previously described for the first maize crop, with the excep† Includes leaf litter plus flowers representing recoveries of 3.9 and 7.7% tion of doubling the original fertilization rate of the first maize of foliar applied 15N-urea to Desmodium and Pueraria, respectively. ‡ Means within a column followed by the same letter are not significantly crop as detailed above. At maturity the aboveground parts different (P 0.05). of the last maize crop were harvested on 15 Sept. 1995 and processed as previously described for the first crop of maize in the cereal sequence. Consequently, the C:N ratio of Desmodium was ≈2-fold Soil sampling was conducted at the end of each cereal crop greater than that of Pueraria. Lignin concentrations were season as previously described. Additionally, tissue, grain, and similar for the two legumes, while the polyphenol consoil samples were analyzed for total N and N enrichment as centration of Desmodium shoots exceeded that of Puerpreviously described. Total plant C was determined with a aria by ≈1.5-fold (Table 1). CHN elemental analyzer (Model 2400, Perkin Elmer, Norwalk, CT). Recovery of Foliar-Applied Nitrogen-15 Recovery Calculations and Statistical Analyses About 750 g kg 1 of the 15N-urea applied to the canopy Recoveries of foliar-applied N-urea were calculated using of the legume cover crops was recovered in the the sum of atom % excess N in the shoot plus soil (0 to 75 aboveground tissues (shoots, flowers, and leaf litter) cm). Recoveries of legume sources of N in the following and the top 75 cm of soil in the 15N-labeled microplots. cereal crops were determined after each cereal harvest by Less than 40 g kg 1 of the foliar applied 15N-urea was measuring the atom % excess N in aboveground tissues and the top 75 cm of the soil profile. For the legume root-soil N found in the root-soil N pool, with the organic fraction source, it was assumed that the N label was contained entirely of the root-soil N pool containing more of the foliarin the root, even though a portion of the root-soil N source applied 15N than the inorganic fraction (Table 2). Recovwas present as inorganic N before planting the first cereal eries of 15N as inorganic, organic, and total N fractions crop. The amount of N contained in the legume roots of each in the soil were greater from Pueraria than Desmodium N-labeled microplot was assumed to be proportionally equal microplots. Previous study of 15N labeling of legume to that measured following excavation of legume roots in an canopies produced similar results; the effectiveness of identical N labeling experiment (Glasener et al., 1998) that 15N foliar labeling of Desmodium and Pueraria meawas conducted simultaneously. Consequently, the recoveries sured an average recovery of 780 g kg 1 in shoots and from this source would represent an upper limit for N derived entirely from legume roots. only 41 g kg 1 in the excavated roots and soil (Glasener The experiment was a split plot design with four replicaet al., 1998). Vasilas et al. (1980) also found the leaves tions. Each legume main plot had two randomized subplots and stems of soybean [Glycine max (L.) Merr.] to be consisting of the legume root and shoot N sources contained the dominant sinks for foliar-applied 15N, and no more in microplots. Nitrogen-15 recovery data was analyzed by lethan 16 g kg 1 of the applied N was translocated to the gume species, N source, and the interaction of these factors roots. Morris and Weaver (1983) applied 15N-labeled using PROC GLM (SAS, 1991). Pre-planned contrasts were imurea to foliage of soybean and recovered an average of plemented to distinguish significant differences among treat680 g kg 1 in the shoots and 17 g kg 1 in the roots after ment effects. three sampling dates. We observed some spray drift beyond the perimeter of the 15N-labeled microplots that RESULTS AND DISCUSSION would account partially for the unrecovered foliar apLegume Cover Crop Traits plied 15N in our study. Additional losses may have occurred as urea was absorbed into the leaf (Wittwer et In mid-June 1994, when 15N-labeled cover crops were al., 1963), hydrolyzed, and then volatilized as NH3. harvested, Desmodium had produced ≈2.9-fold more shoot dry matter (17.0 Mg ha 1 ) than Pueraria (5.9 Mg Within 15N-labeled microplots, total N in the top 75 ha 1 ), and had a N concentration that was only 16 g cm of the soil profile was equivalent (averaging 8250 kg 1 compared with 33 g kg 1 for Pueraria (Table 1). kg ha 1 ) for the two legume cover crops (Table 3). Rather than disturb the soil to excavate legume roots Table 1. Aboveground dry matter and selected chemical traits of to measure their N content, the contribution of legume Desmodium and Pueraria legume cover crops labeled with 15N. root N to the soil total N pool of the microplots was Legume Dry matter Total N Lignin Polyphenols C:N ratio calculated using data from our previous research (GlaMg ha 1 g kg 1 sener et al., 1998) with identical experimental conditions Desmodium 17.0a† 15.9b 125a 15.3a 25a (i.e., same time period, identical growing conditions, Pueraria 5.9b 32.8a 104a 10.6b 13b same legumes) in which shoot and root dry matter and CV, % 5 3 6 1.8 5 N content were measured. Estimated root N was 12.5 † Means within a column followed by the same letter are not significantly different (P 0.05). kg ha 1 for Desmodium and 14.8 kg ha 1 for Pueraria GLASENER ET AL.: NITROGEN-15 LABELED LEGUME NITROGEN SOURCES 527 Table 4. Total inorganic N (NO3 NH4 ) from 1993 to 1995 in Table 3. Total N and 15N enrichment of microplot N sources supplied by labeled legume shoot and root-soil (0to 75-cm depth) the top 75 cm of soil in microplots planted with a legume cover crop followed by a sequence of three cereal crops. components and the calculated N content of roots in the 15Nlabeled microplots. Cereal crop sequence