Post on 18-Feb-2017




2 download

Embed Size (px)


<ul><li><p>KINETICS OF SOIL NITROGENMINERALIZATION FROM UNDISTURBED</p><p>AND DISTURBED SOIL</p><p>Ariel Ringuelet* and Omar AntonioBachmeier</p><p>Catedra de Edafologa, Facultad de Ciencias Agropecuarias,</p><p>Av. Valparaso s/n. C.C. 509, Cordoba 5000, Argentina</p><p>ABSTRACT</p><p>Knowledge of soil nitrogen (N) mineralization processes is</p><p>essential for modeling soil processes in agriculture. Many authors</p><p>have found discrepancies in N mineralization between disturbed</p><p>and undisturbed samples. Nevertheless, most simulation models</p><p>use a first-order kinetic model (exponential model) for all the</p><p>layers under study, devised from studies using disturbed and</p><p>superficial samples. The goal of the present study was to establish</p><p>the best kinetic model to explain and predict N mineralization as</p><p>affected by sample disturbance and soil depth in two soils of the</p><p>semiarid region of Argentina. Disturbed (D: sieved ,2 mm andquartz mixed) and undisturbed (UD) samples from two</p><p>Haplustolls were subject to successive incubations and extractions</p><p>to assess N mineralization rates. The amount of N mineralized in</p><p>3703</p><p>DOI: 10.1081/CSS-120015916 0010-3624 (Print); 1532-2416 (Online)</p><p>Copyright q 2002 by Marcel Dekker, Inc.</p><p>*Corresponding author. E-mail:</p><p>COMMUNICATIONS IN SOIL SCIENCE AND PLANT ANALYSIS</p><p>Vol. 33, Nos. 19 &amp; 20, pp. 37033721, 2002</p><p>2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.</p><p>MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016</p></li><li><p>disturbed samples was up to 200% greater than undisturbed.</p><p>Replicate D samples from the Torriorthentic had significantly</p><p>higher variability p , 0:05 than samples from the Torrifluventic.In the latter, UD samples had a kinetic heterogeneity that was not</p><p>apparent in the D samples. These results suggest that the</p><p>incubation-technique for D samples is ineffective for these soils.</p><p>Key Words: Nitrogen; Mineralization; Undisturbed samples;</p><p>Kinetic models</p><p>INTRODUCTION</p><p>Knowledge of the quantity of N supplied to a growing crop from</p><p>mineralization of soil organic matter is important to improve the efficiency of</p><p>N fertilizer and reduce the risks of polluting water resources and atmosphere.</p><p>There exists a wide variety of chemical and biological methods to assess N</p><p>mineralization in the laboratory and in situ.[1] The question arises, whether</p><p>the mineralization on disturbed samples can adequately predict mineralization</p><p>for in situ structure soil conditions (undisturbed samples).</p><p>Stanford and Smith[2] developed an incubation method using disturbed,</p><p>dried and rewetted soil samples at 358C. They suggested that nitrogenmineralization follows first-order (exponential) kinetics for a wide variety of</p><p>soils, where the first-order constant k was found not to differ significantly</p><p>between soils, whereas the initial pool of potentially nitrogen (N0) varied widely.</p><p>Since then, many authors have used this method, and the concept involved in it, to</p><p>study mineralization processes.</p><p>Alternatives to the exponential model have been proposed: a double</p><p>exponential model[3] with two components of potentially mineralizable nitrogen,</p><p>each representing organic pools that differ in their resistance to decomposition</p><p>(i.e., different rate constants). Bonde and Rosswall[4] modified this model by</p><p>replacing the resistant pool with a linear term that accounted for an apparently</p><p>unlimited organic pool (zero order kinetics). While exponential models can best</p><p>explain N mineralization from plant residues,[5] linear models seem to be</p><p>adequate to represent nitrogen mineralization from soil humus,[6] a more resistant</p><p>organic fraction. Simard and Ndayegamiye[7] found that the cumulative N</p><p>mineralization curves were best described by the Gompertz equation, derived</p><p>from the assumptions that the mineralization rate increases in the early stages and</p><p>the efficiency of the release process will decrease with time because of the slower</p><p>activity of the mineralizing flora or the exhaustion of mineralizable N.</p><p>RINGUELET AND BACHMEIER3704</p><p>2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.</p><p>MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016</p></li><li><p>Although many N mineralization kinetic models have been proposed, most</p><p>crop yield simulation models use first-order (exponential) kinetics for different</p><p>organic matter pools, namely litter, manure, humus, stable, fresh organic N,</p><p>active organic N, biomass, and soil organic pools.[8] This over-simplification</p><p>masks dynamic changes in N mineralization, originating from root activity,[9]</p><p>from type, placement and timing of residue input,[5] by differences in</p><p>management histories[10] and by seasonal fluctuations.[11]</p><p>Most N mineralization studies have been conducted on disturbed samples.</p><p>Some investigators have found discrepancies in N mineralization between</p><p>disturbed and undisturbed samples. Nordmeyer and Richter[12] observed that in</p><p>undisturbed samples N mineralization increases nearly linearly with time,</p><p>whereas disturbed samples show clearly a mineralization flush during the first 20</p><p>incubation days, showing that any disturbance introduced by soil preparation has</p><p>a strong influence on subsequent N mineralization. Cabrera and Kissel[13]</p><p>obtained considerable N mineralization overprediction using disturbed samples,</p><p>possibly explained by the pretreatment of soil samples prior to the incubation.</p><p>Drying and rewetting the soil is known to induce a flush of nitrogen</p><p>mineralization.[14] Mineralization rates decrease with succesive incubation</p><p>periods in undisturbed samples,[15] suggesting a mineralizationimmobilization</p><p>process in soil microsites.</p><p>The purposes of the present study were to find N mineralization models that</p><p>would properly describe the data obtained with disturbed and undisturbed</p><p>samples and to determine whether the models obtained for disturbed samples</p><p>could in any way be used to predict the pool of N mineralized under undisturbed</p><p>conditions.</p><p>MATERIALS AND METHODS</p><p>Soils</p><p>The work was undertaken in soils of the Semiarid Chaco Region in</p><p>Argentina, a vast phytogeographical region where soil N is limiting and there is</p><p>scant knowledge of its dynamics.</p><p>Two soils of fluvial origin, representative of the area surrounding the Cruz</p><p>del Eje River valley, in the Province of Cordoba, Argentina, were used. One was</p><p>a coarse loam mixed thermic Torrifluventic Haplustoll, of average fertility, and</p><p>the other was a sandy-loam mixed thermic Torriorthentic Haplustoll, of low</p><p>fertility (Table 1).</p><p>The selected plots in these two soils were in bare fallow at the time of</p><p>sampling (May of 1994). A furrow-irrigated squash crop (Cucurbita pepo ) was</p><p>harvested two months earlier, without incorporating residues. Before that, there</p><p>KINETICS OF SOIL N MINERALIZATION 3705</p><p>2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.</p><p>MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016</p></li><li><p>had been alfalfa for three years, and, previously, five years of alternate</p><p>horticultural crops (tomato and squash).</p><p>Soil Sampling</p><p>Three sites for each soil were randomly chosen (Fig. 1): two cropped (CrA</p><p>and CrB), and a noncropped site (NCr). The latter was included in order to</p><p>understand the changes in N dynamics that result from soil alteration, since this</p><p>site had not been under cultivation disturbance for the last 30 years. The three</p><p>sampled sites were in a straight line perpendicular to the direction of the furrows.</p><p>The distances between the samples sites were approximately 100 m randomly</p><p>selected.</p><p>In each site, a 0:70 m 0:70 m area was defined, where samples were takenat two depths: in and below the tillage layer (1518 cm and 3033 cm, respecti-</p><p>vely). The tillage layer depth was chosen in order that the first depth was selected</p><p>just below the tillage disturbance layer, in order to get undisturbed soil cores. The</p><p>3033 cm sampling depth corresponds to the central layer of the AC horizon.</p><p>The effect of soil physical disturbance was assessed by taking intact cores</p><p>which were used in the incubation studies (UD) with similar soils which were</p><p>sieved to create a disturbance effect (D). There were three repetitions for each</p><p>type of sample (Fig. 1).</p><p>Subsamples in the laboratory were composite samples taken from each</p><p>depth within each sampling area. Replicates for UD were three cores taken 40</p><p>(^5) cm apart, within the sampling area.</p><p>UD samples were taken with a steel cylinder (7.5 cm diameter), and</p><p>transferred to a PVC cylinder of the same inner dimension with the help of a</p><p>Table 1. Selected Characteristics of the Soils Used in the Study</p><p>Soil Torrifluventic Haplustoll Torriorthentic Haplustoll</p><p>Depth (cm) 1020 2535 1020 2535</p><p>Organic carbon (g kg21) 17.2 8.4 9.5 5.3</p><p>Total N (g kg21) 1.8 1.2 1.2 0.9</p><p>Phosphorus (mg kg21) 19.5 20.9 7.2 1.3</p><p>pH 7.5 8.0 7.3 8.2</p><p>Clay (g kg21) 181 161 106 101</p><p>Silt (g kg21) 384 299 190 190</p><p>Sand (g kg21) 435 539 704 709</p><p>Texture class Loam Sandy loam Sandy loam Sandy loam</p><p>Bulk density (Mg m23) 1.17 1.16 1.21 1.19</p><p>RINGUELET AND BACHMEIER3706</p><p>2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.</p><p>MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016</p></li><li><p>plunger. A fiberglass disc, a styrofoam lid, and a plastic screen were placed at the</p><p>end of each soil column in order to keep it unaltered during transport, incubation</p><p>and leaching. Samples were covered with waterproof, oxygen-permeable low</p><p>density polyethylene film (30mm thick) to prevent moisture loss. The UDsamples from 3033 cm were taken exactly below the 1518 cm samples. After</p><p>removal, samples were kept at 48C until the beginning of the incubation, no morethan 72 hours later.</p><p>Disturbed Samples Conditioning</p><p>Field-moist soil samples from each depth were gently passed through a</p><p>2-mm sieve (9 mesh) and mixed with ashed, acid-washed quartz[2] sieved through</p><p>1-mm screen (18 mesh) in a 1:1 ratio (80 g soil 80 g quartz). The soil sampleswere not dried to avoid an initial mineral N-flush, commonly evidenced after the</p><p>Figure 1. Soil sampling scheme. Above: the three sample sites within each plot for each</p><p>soil. Below: sample site 0:70 m 0:70 m:</p><p>KINETICS OF SOIL N MINERALIZATION 3707</p><p>2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.</p><p>MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016</p></li><li><p>drying process, in the first stages of the process.[14] The addition of quartz was</p><p>meant to maintain an adequate drainage after re-packing the soil columns. These</p><p>samples were incubated in cylinders similar to those used for the UD samples</p><p>(310 mL).</p><p>Sample Incubation</p><p>The soil samples, contained in the PVC cylinders, were initially leached</p><p>and then incubated, at a soil moisture close to field capacity up to 27 weeks.</p><p>Mineralized N from the Torrifluventic Haplustoll soil was measured on days 6,</p><p>20, 38, 65, 91, 131, and 187. In the Torriorthentic Haplustoll soil the measures</p><p>were taken on days 14, 28, 54, 94, 134, and 174.</p><p>After incubation, mineral N extractions were carried out using the nutritive</p><p>solution of Ref. [2] as extractant: aliquots of 50 mL of extraction medium</p><p>(0.5 mM of each K2SO4, MgSO47H2O and CaSO42H2O, and 1.2 mM</p><p>Ca(PO4H2)2, up to a total volume of 400 mL, were used to leach all mineralized</p><p>N. A vacuum of 20.033 MPa was applied to the UD samples to ensure full perco-lation of the leaching solution within 24 h of initiating the extraction process.</p><p>Net mineralized N was calculated as the sum of NO3-N plus NH4-N in the</p><p>percolated solution. The concentration of NO3-N was analyzed using a specific</p><p>electrode (ORION 93-07) and the output was recorded with an ORION 901</p><p>Ionalizer.[16] The concentration of NH4-N was spectrophotometrically analyzed,</p><p>using the indophenol blue method.[16]</p><p>Soil moisture was measured periodically by weighing the soil cores plus</p><p>their containers. When necessary, enough extractive solution was added and the</p><p>cylinders were covered with low density, highly oxygen porous polyethylene, to</p><p>maintain the soil moisture close to field capacity. The incubation chamber</p><p>temperature was maintained at 308C (^18C).</p><p>Chemical and Physical Analyses</p><p>To characterize the soils, two composite samples of each soil were taken</p><p>from the three sites: one at 1320 cm and another at 2835 cm.</p><p>The following analyses were carried out: organic C by the Walkley and</p><p>Black method,[17] total N content by micro-Kjeldahl,[18] soil pH potentiome-</p><p>trically measured at a soil:water ratio of 1:1, particle-size using the pipette</p><p>method,[19] and extractable phosphorus by the NaHCO3 method.[20]</p><p>RINGUELET AND BACHMEIER3708</p><p>2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.</p><p>MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016</p></li><li><p>Statistical Methods</p><p>Mineralization potentials (N0) and rate constants (k,p,h) were estimated</p><p>using a nonlinear least squares method as described by Smith et al.[21] Data were</p><p>fit to various kinetic models (Table 2). Parameters were estimated by numeric</p><p>iteration using the GaussNewton algorithm (NLIN procedures of SAS).</p><p>Goodness of fit was tested by means of studentized residual versus predicted</p><p>values, by the relation sum of regression squares/total sum of the squares</p><p>(SRS/TSS), and through the asymmetric standard error of parameter estimators.</p><p>Models which did not fit any of the three individual replications were discarded.</p><p>Correlations, regressions and T-tests were also performed with SAS.[25]</p><p>RESULTS AND DISCUSSION</p><p>Quantities of Nitrogen Mineralized</p><p>Comparing D and UD treatments, the amount of N mineralized in disturbed</p><p>samples was larger than that in undisturbed samples, such as was found by other</p><p>authors.[13,15,26] However, in our study, this extra-mineralization was much</p><p>lower. In eight treatments, the D samples mineralized between 0 and 52%, based</p><p>on N mineralized by UD samples (Tables 3 and 4), and in the other three</p><p>treatments (at 3033 cm depth samples from Torrifluventic soil), the D samples</p><p>Table 2. Kinetics Models Fitted to Data</p><p>Model Equation References</p><p>Zero-order kinetics</p><p>(lineal model)</p><p>Nm b0 pt Addiscott[22]</p><p>First-order kinetics</p><p>(exponential model)</p><p>Nm N0 1 2 exp 2kt Stanford and Smith[2]</p><p>Exponential linealmodel</p><p>Nm N0 1 2 exp 2kt pt Wedin and Pastor[23]</p><p>Gompertz model Nm N01 exp 2h exp 2kt2N02 exp 2h</p><p>Simard and</p><p>Ndayegamiye[7]</p><p>Double first-order kinetics</p><p>(double exponential model)</p><p>Nm N01 1 2 exp 2ktN02 1 2 exp 2kt</p><p>Deans et al.[24]</p><p>Nm accumulated mineralized N (mg kg21) at time t; b0 intercept; p zero-ordermineralization constant; N0 potentially mineralizable N (mg kg21); k first-ordermineralization constant (days21); h proportion...</p></li></ul>


View more >