186 B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200providing adequate habitat and protection for soilorganisms, supplying oxygen to roots, and prevent-ing soil erosion (Denef et al., 2001; Franzluebbers,2002a,b). Aggregate size distribution is considered animportant factor for germination and emergence ofseedlings and for the development of roots (Braunackand Dexter, 1989; Freitas et al., 1999). Furthermore,soil aggregate stability has been recognized as a rele-vant factor in the control of water erosion of acid soilsof the tropics (Roth et al., 1986; Castro Filho et al.,1991) because erodibility of soils is directly relatedto aggregate stability (Kemper and Rosenau, 1986).The continued existence of large pores in the soil thatfavor high inﬁltration rates and aeration depends onthe stability of larger aggregates. Soil aggregation isalso one of the principal processes responsible forcarbon sequestration in soils (Lal et al., 1997) andin turn, structural degradation provokes soil organicmatter loss (Six et al., 1999).The favorable effect of reduced and conservationtillage systems on soil aggregation has been reportedfor different soil types and climates (Alvarenga et al.,1986; Schjønning and Rasmussen, 1989; Oyedeleet al., 1999). Soil management systems that leavemore plant residues on the soil surface generally allowimprovements in soil aggregation and aggregate sta-bility (Carpenedo and Mielniczuk, 1990). Kouakouaet al. (1999) and Dutartre et al. (1993) found a strongcorrelation between aggregate stability in water andthe carbon content of bulk clayey Ferralsols fromAfrica. A medium to high correlation was found be-tween the mean geometric diameter, the meanweightdiameter, the amount of aggregates >2 mm, and to-tal organic carbon content of Latosols from Brazil(Carpenedo and Mielniczuk, 1990; Castro Filho et al.,1998).The effect of crop rotations involving cover cropson soil structural stability is still less understood thanthat of tillage. Gijsman and Thomas (1995) reportedthat the addition of legumes to pastures did not changethe aggregate size distribution but increased the stabil-ity of aggregates against slaking in eastern Colombiansavannas. Castro Filho et al. (1998, 2002) observed noeffect of crop rotations on aggregate stability indicesin a Latosol (Rhodic Ferralsol) from southern Brazil.High root density is frequently associated with betteraggregation (e.g. Silva and Mielniczuk, 1997, 1998).In a study conducted on Alﬁsols from New Zealand,Haynes and Francis (1993) suggested that the largerroot mass and length of grass species indirectly im-prove soil aggregate stability by maintaining a higheramount of microbial biomass that resulted in largerproduction of carbohydrates.No-tillage is a widely accepted conservation prac-tice by farmers, both by smallholders (<50 ha) andlarge scale intensive production farmers in Brazil(Machado and Silva, 2001). The objective of thisstudy was to evaluate the effect of no tillage andconventional tillage under different crop rotations onthe aggregation properties and aggregate stability of aRhodic Ferralsol at various depths. The same soil typeunder secondary forest vegetation was included in thestudy as a reference. Correlation between soil aggre-gation indices and total organic carbon concentrationof the aggregate size classes were determined to eval-uate the relationship between macroaggregation andorganic carbon accumulation in this Ferralsol.2. Materials and methods2.1. Site description and soilSoil samples were collected from a long-term ﬁeldexperiment that was established at Embrapa Soybean,in Londrina (23◦23 S; 51◦11 W), State of Paraná,Brazil, in 1989. The climate is Cfa (subtropical cli-mate; Koeppen, 1931) with a yearly average tempera-ture of 20.7 ◦C (with monthly means of 11 ◦C in Julyand 31 ◦C in February) and a mean annual rainfallof 1622 mm. Two thirds of this rain falls from Octo-ber to March. The soil type was a Rhodic Ferralsol(WRB) (Latossolo Vermelho eutroférrico, BrazilianSoil Classiﬁcation System; Rhodic Eutrudox, USDASoil Taxonomy) formed on basalt and covers approx-imately 10 Mha in Brazil (Embrapa, 1981). It hashigh clay content at 0–30 cm (726–800 g kg−1) andhigh drainage capacity. The dominant component ofthe clay-size fraction (<2 m) is kaolinite (Pavanet al., 1985). Characteristic chemical and physicalproperties are reported in Table 1. Before setting upthe experiment, the area had been cultivated with asoybean (Glycine max L.)–wheat (Triticum aestivumL.) succession for 10 years under conventional tillage(disc plough followed by harrowing twice with lightdiscs).
B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200 187Table 1Chemical, physical, and mineralogical properties of a Rhodic Ferralsol, Londrina, BrazilTreatment Chemical properties Particle size analysispHH2O Al(cmolc dm−3)H + Al(cmolc dm−3)Ca(cmolc dm−3)Mg(cmolc dm−3)K(cmolc dm−3)P(mg kg−1)Sand(g kg−1)Silt(g kg−1)Clay(g kg−1)Forest 6.3 0.1 4.5 14.9 2.6 0.58 30 63 211 726NT rot 6.2 0.1 5.1 7.6 2.6 0.56 800 42 192 766NT suc 5.5 0.2 6.7 4.2 1.5 0.48 350 39 184 777CT rot 6.8 0.1 4.4 5.9 3.6 0.23 310 38 162 800CT suc 5.9 0.1 5.3 5.6 2.2 0.33 260 40 174 786Predominant clay minerals in the clay fraction: kaolinite; other clay minerals in small amounts: gibbsite, hematite and vermiculite withinterlayer hydroxi-Al; NT: no-tillage; CT: conventional tillage (disc plough followed by harrowing twice with light discs); rot: crop rotation;suc: crop succession.The tillage treatments were conventional tillage(disc plough at 18 cm depth followed by harrow-ing twice with light discs at 10 cm depth, CT) andno-tillage (no disturbance to the soil other thanthe sowing operation, NT). Crop succession (suc)included wheat/soybean and the 4-year crop ro-tation (rot) included white lupine (Lupinus albusL.)/maize (Zea mays L.)–black oat (Avena strigosaSchieb.)/soybean–wheat/soybean–wheat/soybean. InBrazil, a crop rotation by deﬁnition always includesa cover crop to form mulch, or green manure com-bined with cash crop(s). Thus, soybean and maizewere summer cash crops while lupine and oat werewinter cover crops. Wheat was the only winter cashcrop in the system. The tillage-rotation plots (CT suc;NT suc; CT rot; NT rot) were 7.5 m × 30 m arrangedin a complete randomized block experimental de-sign with three replications. Additional soil sampleswere collected for reference purposes in the adjacent(800 m from the experiment) secondary sub-montane,semi-deciduous forest, which predominantly con-sisted of Trichillia clausenii, Euterpe edulis andAspidosperma polyneuron (Oliveira-Filho and Ratter,1995).2.2. Soil samplingSoil samples for aggregate stability analysis werecollected at ﬁeld capacity in January 2001 (soil wasunder soybean at V2 stage) at 0–5, 5–10, 10–20 and20–30 cm depths. The ﬁeld sampling procedure is ofgreat importance in the evaluation of soil aggrega-tion (Yoder, 1936). In Brazil, the general procedurefor soil sampling includes the use of a shovel and fur-ther crumbling by hand to pass a 4.76 mm sieve be-fore wet sieving (Embrapa, 1997). Castro Filho et al.(1998) showed that the use of a 4 mm sieve underes-timates the ability of tillage systems, especially NT,to form larger water stable aggregates. Later, CastroFilho et al. (2002) suggested the use of a 19 mmsieve to homogenize soil samples before wet siev-ing. In the present study all samples were taken in apit of 1 m2 dug to 0.4 m. In order to minimize com-pression and to obtain a representative sample forthe aggregation state of the soil, samples were takenusing a bricklayer’s trowel inserted into the soil atthe lower level of each sampling depth. Each soilsample was passed through a 19 mm sieve, at thesite of sampling, by gently breaking apart the soil.Clods and aggregates larger than 19 mm diameter werediscarded. Soil samples were air dried for 24 h inshade (humidity 15 ± 2%). Very dry aggregates canlead to a falsely high resistance to breakdown andresult in higher stability indices (Castro Filho et al.,2002). Air-dried soil samples were placed in plasticbags and cardboard boxes and stored at ambient tem-perature until analysis at Embrapa Soils in Rio deJaneiro.2.3. Fractionation of aggregate size classesAggregate separation methods are considered to beuseful when examining certain scale issues (Younget al., 2001), but there is no standardized procedure forsampling and wet sieving (D´ıaz-Zorita et al., 2002).In this study aggregate size classes were separated by
188 B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200wet sieving following a procedure based on Haynes(2000) and Castro Filho et al. (1998, 2002). Air-driedsoil samples were used in the fractionation of aggre-gate size classes after Haynes (2000), who showed thatthe largest separation values by wet sieving for soilsof contrasting cropping histories were obtained withair-dried samples. The soils were wet sieved througha series of eight sieves (8, 4, 2, 1, 0.5, 0.25, 0.125 and0.053 mm). The line between macro- and microaggre-gates is commonly drawn at 0.25 mm (Edwards andBremner, 1967; Oades and Waters, 1991). In our study,the sieves were selected based on the studies of Yoder(1936), Gijsman (1996), and Salako et al. (1999). Theuse of the 8 mm sieve in the wet sieving was basedon previous studies undertaken on a Rhodic Ferralsolunder different tillage systems from the same regionby Sidiras et al. (1982).A 30 g subsample was moistened by capillarityfor 5 min by placing it on a ﬁlter paper on the topof the 8 mm sieve. The water volume was raised in-side the water tank to wet the ﬁlter paper and thesoil. The ﬁlter paper was then removed and the wetsieving procedure was conducted, replicated twicefor each sample. Aggregate separation was achievedby agitating the sieve series up and down 3.5 cmwith 30 repetitions per minute during a period of15 min. At the end of the 15 min cycle, the stableaggregates from each sieve were gently back-washedoff the sieve into 100 mL glass beakers. Aggre-gates were oven dried (105 ◦C), weighed, and storedin plastic ﬂasks at room temperature for carbonanalysis.2.4. Determination of the size distribution ofaggregatesParameters expressing the size distribution of ag-gregates (aggregation indices) were determined as fol-lows:The Meanweight Diameter (MWD) of aggregates(Kemper and Rosenau, 1986):MWD =ni=1xiwiwhere wi is the proportion of each aggregate classin relation to the whole, xi the mean diameter of theclasses (mm).The Mean Geometric Diameter (MGD) of aggre-gates (Kemper and Rosenau, 1986):MGD = expni=1wi log xini=1wiwhere wi is the weight of aggregates (g) in a size classwith an average diameter xi.The Aggregate Stability Index (AS) of soils (CastroFilho et al., 1998):AS =weight of the dry sample − wp25 −sandweight of the dry sample −sand× 100where wp25 is the weight of aggregates <0.25 mm(g), sand is the weight of particles between 2.0 and0.053 mm (g)2.5. Organic carbon analysisTotal organic carbon (TOC) concentration of wholesoil samples and of the aggregate size classes wasdetermined on ﬁnely ground samples by dry com-bustion with a Perkin-Elmer CHNS/O Analyser2400. Coefﬁcient of variation of the method was3%.2.6. Statistical analysisData were analyzed using the SAS statistical pack-age (SAS Institute, 1990) for analyses of variance(F-test). Signiﬁcant differences in TOC, MWD, MGDand AS between tillage and crop rotation systems weredetermined using Tukey’s Studentized Range (HSD)Test for each depth. Bivariate correlation (Pearson’s)was made between TOC of aggregate size classesand the aggregation indices (AS, MWD, MGD). Thebest-ﬁt power function to aggregate size distributionswas determined by nonlinear least squares, and thedegree of dispersion of data points to the adjustedfunction was measured by the root mean square er-ror (RMSE). All results were based on three replica-tions in the ﬁeld. Comparisons between cultivated andforest sites must be made with care because the for-est site was not part of the experimental design andsampling was performed differently (soil proﬁles 50 mapart from one another in straight line). For this rea-son the results obtained for the forest samples werenot included in the statistical analyses. Comparison of
B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200 189means from the forest samples with samples of the ex-periment was done by using standard deviation (S.D.)values.3. Results and discussion3.1. Aggregate size distributionThe examined soil was well aggregated andshowed high structural stability. In the forest at0–5 cm, 95% of the total soil mass was found inmacroaggregates (Table 2). The amount of soil inmacroaggregates slightly decreased at depth (approx-imately 5% decrease at 20–30 cm compared to the0–5 cm layer). Similarly, in a Ferralsol (pH 4.7 and380 g clay kg−1 soil) from the eastern Llanos (sa-vanna) in Colombia, Gijsman and Thomas (1995)reported that about 90% of soil was in macroaggre-gates.In forest soil, 76% of the soil mass occurred inthe largest aggregate size class (8–19 mm) (Table 2).The consistently higher values for macroaggregatesat depth under the forest indicate the inﬂuence ofdeep rooting combined with no soil disturbance. Discploughing followed by harrowing with light discsled to a decrease in the amount of soil in aggregatescompared to the forest soil, with the most dramaticeffect for the 8–19 mm aggregate size class (70 and83% decrease at 0–5 and 20–30 cm respectively)(Table 2). The effects of crop rotation and the interac-tion of tillage and crop rotation on the distribution ofaggregate size classes in all depths were not signiﬁ-cant at P ≤ 0.05. In the surface layer (0–5 cm), soilunder NT had a greater proportion of the 8–19 mmaggregate size class and a lower proportion of the2–4, 1–2, 0.5–1, 0.25–0.5 mm and microaggregateclasses than under CT. The most pronounced ef-fect was in the 8–19 mm aggregates. At 5–10 and10–20 cm the effect of tillage systems was observedonly in the 4–8 and 2–4 mm aggregate size classesrespectively, with NT having higher proportion thanCT (Table 2). Sidiras et al. (1982) reported a similarresult from another tillage experiment on the samesoil type at Londrina. At the end of 4 years, theyfound that, at 0–10 cm, NT had a greater proportion ofthe largest (5.66–9.52 mm) aggregate size class thanCT.All aggregation indices under secondary forest werehigher than those under agricultural use (NT and CT)(Table 3). Roth et al. (1991), in a study of physical andchemical factors related to the aggregation of a RhodicFerralsol from Londrina, found higher parameter val-ues at 0–10 cm depth in a soil under virgin forest thanin soils under disc or chisel plough. Forest soils, whichare protected under dense vegetation and undisturbedby tillage, possess a surface structure sufﬁciently sta-ble to allow rapid water inﬁltration, to prevent crust-ing, and consequently protect the soil against watererosion.Among the main effects, tillage alone had the onlysigniﬁcant impact on aggregation indices and this wasrestricted to the top 5 cm. Conventional tillage hadsigniﬁcantly lower aggregation indices compared toNT, but only in the 0–5 cm layer (Table 3). The valuesof AS, MWD and MGD of the soil under NT werehigher than under CT and the contrasts were greaterthan those reported by Castro Filho et al. (1998) andRoth et al. (1991) for the same soil type in the samegeographical region, but possibly because sampleswere collected at 0–10 cm depth and separated witha smaller maximum sieve size. Silva and Ribeiro(1992), Oyedele et al. (1999) and Franzluebbers et al.(1999) concluded for different soil types under dif-ferent conditions that soil mechanical disturbancereduced soil structural stability. These results in-dicate that, compared to CT, NT promotes macro-aggregation.Aggregation indices were not different betweencrop rotation (rot—with cover crop) and crop succes-sion (suc—no cover crop or green manure) (Table 3).These ﬁndings are similar to those of Castro Filhoet al. (1998, 2002) who also reported no signiﬁ-cant difference in aggregation indices between croprotations in the same soil type and from the sameregion.Log-normal, fractal and Rosin–Rammler functionscan be used to describe aggregate size distributionto evaluate tillage implement performance (Gardner,1956; Perfect et al., 1993). In this work a best-ﬁt func-tion (in this case power function: y = 1.76x2.33, R2 =0.99) was used to characterize the aggregate size dis-tribution of the forest soil. The degree of dispersionfrom the best-ﬁt function for the forest soil standardwas lower under NT (RMSE = 100) than under CT(RMSE = 205) (Fig. 1).
190B.Madarietal./Soil&TillageResearch80(2005)185–200Table 2Effect of tillage and crop rotation on the distribution of aggregate size classes (g kg−1 soil) of a Rhodic Ferralsol, Londrina, BrazilTreatment Aggregate size classes Macroaggregates Microaggregates8–19 mm 4–8 mm 2–4 mm 1–2 mm 0.5–1 mm 0.25–0.5 mm 0.125–0.25 mm 0.053–0.125 mm 0.25–19 mm 0.053–0.25 mm0–5 cmForesta 765 (115) 106 (45) 42 (21) 20 (11) 11 (8) 5 (3) 2 (0) 1 (0) 949 (30) 3 (1)NT 517a∗∗ 91 58b∗ 71b∗ 89b∗ 59b∗ 28b∗ 12b∗ 885a∗ 40b∗CT 232b∗∗ 90 86a∗ 119a∗ 135a∗ 106a∗ 61a∗ 30a∗ 768b∗ 91a∗rot 394 94 67 84 97 80 45 24 816 69suc 355 86 77 106 127 85 43 18 836 615–10 cmForest 647 (71) 150 (54) 65 (31) 30 (14) 18 (13) 7 (5) 3 (1) 2 (0) 917 (30) 5 (0)NT 418 117a∗∗ 79 119 83 45 20 8 861 28CT 347 77b∗∗ 80 103 120 96 52 20 823 72rot 367 99 80 102 106 81 47 18 835 65suc 399 95 79 120 97 60 25 10 850 3510–20 cmForest 576 (121) 147 (46) 70 (39) 47 (25) 45 (25) 19 (16) 6 (5) 3 (2) 904 (57) 9 (7)NT 307 95 115a∗ 137 110 74 40 15 838 55CT 380 69 63b∗ 107 108 84 38 17 811 55rot 377 68 73 110 106 90 51 20 824 71suc 309 97 106 135 112 68 28 11 827 3920–30 cmForest 561 (180) 128 (46) 93 (60) 55 (42) 38 (23) 25 (17) 18 (18) 4 (2) 900 (24) 22 (20)NT 163 69 99 140 153 138 78 31 762 109CT 94 90 98 132 156 136 89 39 706 128rot 117 73 93 129 156 125 75 30 693 105suc 90 85 104 143 152 149 92 40 723 132NT: no-tillage; CT: conventional tillage (disc plough followed by harrowing twice with light discs); rot: crop rotation; suc: crop succession.Comparison of the means was done by Tukey’s Studentized Range (HSD) Test. Values followed by different letters in the same column and depth, within the same type oftreatment (tillage or crop rotation), are signiﬁcantly different by the F-test in the analysis of variance.a Values in parentheses are standard deviation (n = 3).∗ P ≤ 0.05.∗∗ P ≤ 0.01.
B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200 191Fig. 1. Distribution of aggregate size classes of a Rhodic Ferralsol, Londrina, Brazil. The best-ﬁt power function to aggregate sizedistributions was determined by nonlinear least squares. The degree of dispersion of data points to the adjusted function was measured bythe root mean square error (RMSE). NT: no-tillage; CT: conventional tillage (disc plough followed by harrowing twice with light discs).3.2. Organic carbon in aggregatesSoil under secondary forest had the highest TOCconcentration at all depths (data not shown). Totalorganic carbon was also signiﬁcantly greater underNT (25 g kg−1) than under CT (18 g kg−1) at a depthof 0–5 cm only. The TOC concentration of the forestsoil at the same layer was 44 ± 3 g kg−1. These re-sults are consistent with those of Freixo et al. (2002)who reported greater amounts of TOC at 0–5 cm inforest soils (45 g kg−1) than in NT (26 g kg−1) andCT soils (17 g kg−1) in Passo Fundo, Rio Grandedo Sul State, Brazil. Similar to what was observedfor aggregate distribution and stability, differencesin TOC concentration between the crop rotationand the wheat/soybean crop succession were notsigniﬁcant.The carbon concentration of each aggregate sizeclass (g TOC in aggregate fraction per soil in ag-gregate fraction) of the forest soil was higher thanthat of the soils from the experiment, at least inthe surface 10 cm (Table 4). However at 10–20 cm,larger aggregates (8–19 and 4–8 mm) under cultiva-tion (both CT and NT) had a tendency for higherorganic carbon concentration than under the sec-ondary forest, indicating C incorporation to lowersoil layers under cultivation. Incorporation of organiccarbon at depth, especially through disc plough-ing has also been reported for Rhodic Ferralsols insouthern Brazil by Freixo et al. (2002) and Machadoet al. (2003). Under the forest the highest organicC concentration occurred in the smallest macroag-gregate class (0.25–0.5 mm) and in the microag-gregate classes (0.125–0.250 and 0.053–0.125 mm).
192 B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200Table3Effectoftillageandcroprotationonaggregationindices(AS,MWD,MGD)ofaRhodicFerralsol,Londrina,BrazilTreatmentAS(%)aMWD(mm)bMGD(mm)c0–5cm5–10cm10–20cm20–30cm0–5cm5–10cm10–20cm20–30cm0–5cm5–10cm10–20cm20–30cmForestd100(0)100(0)99(1)98(2)11.1(1.2)9.9(0.5)9.0(1.3)8.7(1.9)1.3(0.0)1.3(0)1.2(0.0)1.2(0.0)NT96a∗∗9794897.9a∗∗188.8.131.52.2a∗∗184.108.40.206CT89b∗∗9294874.3b∗∗5.76.02.51.1b∗∗220.127.116.11rot929393896.36.16.03.18.104.22.168.0suc939696822.214.171.124.126.96.36.199.0NT:no-tillage;CT:conventionaltillage(discploughfollowedbyharrowingtwicewithlightdiscs);rot:croprotation;suc:cropsuccession.ComparisonofthemeanswasdonebyTukey’sStudentizedRange(HSD)Test.Valuesfollowedbydifferentlettersinthesamecolumnanddepth,withinthesametypeoftreatment(tillageorcroprotation),aresigniﬁcantlydifferentbytheF-testofvariance.aAggregatestabilityindex.bMeanweightdiameter.cMeangeometricdiameter.dValuesinparenthesesarestandarddeviation(n=3).∗∗P≤0.01.Distribution of organic carbon among aggregate sizeclasses was more even under cultivation than underforest regardless of the tillage system. Organic carbonconcentration under NT, however, was greater thanunder CT in all aggregate classes at 0–5 cm (Table 4).The positive inﬂuence of NT on organic carbon accu-mulation in the aggregate size classes in the top soillayer was probably due to the absence of soil distur-bance, and mulching that favored macroaggregationand incorporation of plant material to aggregatesthrough a slower macroaggregate turnover rate. In thedeeper sampling layers a similar signiﬁcant tillage ef-fect was observed only in the 2–4 mm size at 5–10 cm.These results show the relevance of shallow samplingfor studies on soil aggregation and aggregate stabilitydue to the stratiﬁcation of soil aggregation propertiesand TOC accumulation under NT. The stratiﬁcation ofsoil properties is an important effect of conservationtillage systems (Mrabet, 2002) that could potentiallybe used as an indicator of soil quality (Franzluebbers,2002a). The impact of crop rotation on organic carbonin aggregate classes was not signiﬁcant at P ≤ 0.05.The distribution of TOC among aggregate sizeclasses (g TOC in aggregate fraction per unit wholesoil) had a pattern similar to that of the soil mass inaggregate size classes (Table 5). The only signiﬁcantimpact on the distribution of TOC among aggregatesize classes was caused by tillage (P ≤ 0.01) fol-lowed by the interaction of tillage and crop rotation(P ≤ 0.05). Tillage systems affected the distribu-tion of TOC among aggregate size classes in the0–5 cm layer, the distribution pattern of TOC un-der NT being closer to that under the secondaryforest.The amount of organic carbon was highest in themacroaggregate classes in both forest and cultivatedsoils at all depths and there was a consistent decreasein organic carbon in all aggregate size classes withdepth (Table 5). Roth et al. (1991) and Castro Filhoet al. (2002), analyzing soil aggregation on the samesoil type from southern Brazil, also reported greatestvalues of organic carbon in macroaggregates (>2 mm).Castro Filho et al. (2002) suggested that larger quan-tities of carbon were associated with increases in ag-gregate sizes. Investigating an area of native savannaon an Inceptic Haplustox and other areas under var-ious rotations of grass or grass-legume pasture withrice in the Eastern Plains of Colombia, Gijsman (1996)
B.Madarietal./Soil&TillageResearch80(2005)185–200193Table 4The effect of tillage and crop rotation on total organic carbon (g TOC in aggregate fraction kg−1 soil in aggregate fraction) in aggregate size classes of a Rhodic Ferralsol,Londrina, BrazilTreatment Aggregate size classes Macroaggregates Microaggregates8–19 mm 4–8 mm 2–4 mm 1–2 mm 0.5–1 mm 0.25–0.5 mm 0.125–0.25 mm 0.053–0.125 mm 0.25–19 (mm) 0.053–0.25 (mm)0–5 cmForesta 39 (7) 47 (4) 51 (3) 55 (5) 60 (7) 73 (16) 71 (11) 61 (11) 41 68NT 25a∗∗ 25a∗∗ 25a∗∗ 24a∗∗ 24a∗∗ 25a∗∗ 27a∗∗ 28a∗∗ 25a∗∗ 27a∗∗CT 19b∗∗ 18b∗∗ 18b∗∗ 18b∗∗ 18b∗∗ 18b∗∗ 18b∗∗ 18b∗∗ 18b∗∗ 18b∗∗rot 22 22 22 22 21 22 23 23 22 23suc 21 21 21 21 20 21 22 23 21 225–10 cmForest 29 (6) 28 (4) 31 (3) 32 (5) 34 (7) 37 (8) 40 (13) 39 (18) 29 40NT 20 20 20a∗ 20 19 19 19 21 20 20CT 19 18 18b∗ 19 17 18 18 17 18 18rot 20 19 19 19 18 19 19 20 19 19suc 19 19 19 19 18 18 19 19 19 1910–20 cmForest 16 (0) 17 (2) 19 (2) 19 (3) 20 (3) 21 (4) 23 (5) 23 (7) 17 23NT 18 18 18 17 19 17 17 17 18 17CT 18 18 17 17 17 17 18 18 18 18rot 18 18 18 18 19 18 18 18 18 18suc 19 18 17 17 17 16 17 18 18 17NT: no-tillage; CT: conventional tillage (disc plough followed by harrowing twice with light discs); rot: crop rotation; suc: crop succession.Comparison of the means was done by Tukey’s Studentized Range (HSD) Test. Values followed by different letters in the same column and depth, within the same type oftreatment (tillage or crop rotation), are signiﬁcantly different by the F-test in the analysis of variance.a Values in parentheses are standard deviation (n = 3).∗ P ≤ 0.05.∗∗ P ≤ 0.01.
194B.Madarietal./Soil&TillageResearch80(2005)185–200Table 5The effect of tillage and crop rotation on total organic carbon (g TOC in aggregate fraction kg−1 whole soil) in aggregate size classes of a Rhodic Ferralsol, Londrina, BrazilTreatment Aggregate size classes Macroaggregates Microaggregates8–19 mm 4–8 mm 2–4 mm 1–2 mm 0.5–1 mm 0.25–0.5 mm 0.125–0.25 mm 0.053–0.125 mm 0.25–19 (mm) 0.053–0.25 (mm)0–5 cmForesta 30 (3) 5 (2) 2 (1) 1 (1) 1 (0) 0 (0) 0 (0) 0 (0) 39 (5) 0 (0)NT 13a∗∗ 2 1 2 2 2 1 0 22a∗∗ 1CT 4b∗∗ 2 1 2 2 2 1 0 13b∗∗ 1rot 9 2 1 2 2 2 1 0 18 1suc 8 2 1 2 2 2 1 0 17 15–10 cmForest 18 (4) 4 (1) 2 (1) 1 (1) 1 (1) 0 (0) 0 (0) 0 (0) 26 (6) 0 (0)NT 8 2a∗∗ 2 2 2 1 1 0 17 1CT 6 1b∗∗ 2 2 2 1 0 0 14 0rot 7 2 1 2 2 1 1 0 15 1suc 8 2 1 2 2 1 0 0 16 010–20 cmForest 9 (2) 2 (0) 1 (0) 1 (0) 1 (0) 0 (0) 0 (1) 0 (0) 14 (1) 1 (0)NT 5 2 2a∗ 2 2 1 1 0 14 1CT 7 1 1b∗ 2 2 1 1 0 14 1rot 6 1 1 2 2 2 1 0 14 1suc 6 2 2 2 2 1 0 0 15 0NT: no-tillage; CT: conventional tillage (disc plough followed by harrowing twice with light discs); rot: crop rotation; suc: crop succession. Zero (0) values indicate that TOCwas present in less than 0.5 g kg−1 quantity.Comparison of the means was done by Tukey’s Studentized Range (HSD) Test. Values followed by different letters in the same column and depth, within the same type oftreatment (tillage or crop rotation), are signiﬁcantly different by the F-test in the analysis of variance.a Values in parentheses are standard deviation (n = 3).∗ P ≤ 0.05.∗∗ P ≤ 0.01.
B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200 195also found that the amount of carbon declined withdecreasing aggregate size.The greatest amount of organic carbon was in thelargest aggregate size class (8–19 mm) at all depthsunder forest (Table 5). At 0–5 cm, 77% of the organiccarbon was in the 8–19 mm aggregate size and 99%in macroaggregates.3.3. Relation between organic carbon and soilaggregation indicesIrrespective of treatment, correlation (Pearson’s)between aggregation indices and organic C concen-tration of aggregate fractions (g TOC in aggregatefraction per unit whole soil) was high, but signiﬁcantonly in the 0–5 cm soil layer (Table 6). However,there was strong and signiﬁcant correlation betweenaggregation indices and organic carbon of certain ag-gregate size classes at all depths (Table 7). Positivehigh correlation (r > 0.80) was found at 0–5 cm layerbetween all aggregation indices and the TOC of the8–19 mm aggregate size class. At lower depths, whileAS showed medium-high correlation (between 0.70and 0.80), MWD and MGD continued to have highcorrelation values with the same aggregate size class.No consistent or high correlation was found for the4–8 and 2–4 mm classes and negative correlationswere found for the 1–2 mm class, however not forall depths. Positive correlations were recorded forlarge macroaggregate sizes and negative correlationsTable 6Bivariate correlation (Pearson’s) between the total organic carbonof whole soil (g TOC in aggregate fraction kg−1 whole soil) andsoil aggregation indices (AS, MWD, MGD) of a Rhodic Ferralsol,Londrina, BrazilSampling depth (cm) ASa (%) MWDb (mm) MGDc (mm)0–5d 0.79∗∗ 0.80∗∗ 0.81∗∗5–10d ns ns ns10–20d ns ns ns20–30d 0.65∗∗ ns ns0–30e 0.45∗∗ 0.57∗∗∗ 0.53∗∗∗ns: not signiﬁcant at P ≤ 0.05.a Aggregate stability index.b Meanweight diameter.c Mean geometric diameter.d n = 15 per aggregate size class for all aggregation indices.e n = 45 per aggregate size class for all aggregation indices.∗∗ P ≤ 0.01.∗∗∗ P ≤ 0.0001.for smaller aggregates, especially microaggregates(0.250–0.053 mm), which showed that as organiccarbon increased, the mass of soil in large macroag-gregates also increased. This led to a decrease in theproportion of smaller aggregate size classes, and byextension, higher aggregation indices. Similar resultswere also found by Castro Filho et al. (1998).Correlation between the soil aggregation indices(AS, MWD, MGD) and the TOC of the 8–19 mm ag-gregate class could be best described with a logarith-mic regression model (positive correlation). The bestregression model was selected by the minimal residualsum of squares. The best-ﬁt regression model for the1–2, 0.5–1, 0.25–0.5, 0.125–0.250, 0.053–0.125 mmaggregate size classes was quadratic (negative corre-lation) (Fig. 2).Evaluation of the correlation between soil aggrega-tion indices and TOC of the 2–4 and 1–2 mm aggregatesize classes was also made for results obtained by thestandard wet sieving method suggested by Embrapa(1997) which involves sieves of 2, 1, 0.5 and 0.25 mmdiameter. This was made to validate the correlationvalues recorded in our work. Using the standard wetsieving method the aggregation indices and the TOC ofthe 2–4 and 1–2 mm classes also did not show positivesigniﬁcant or high or even medium correlation (Table8). Therefore, the correlation values recorded usingthe wet sieving procedure used in our work were notthe result of artifacts created by the sieving method.3.4. Soil aggregation and carbon sequestrationpotentialDistribution of all aggregate size classes at 0–5 cm(Table 2) was signiﬁcantly different due to tillage sys-tems (CT and NT). A signiﬁcant effect of tillage sys-tem was also found on the concentration of TOC in allaggregate size classes at the same depth (Table 4) andon the mass of TOC in the 8–19 mm class (Table 5).These results suggest that TOC is important in theformation of large macroaggregates (8–19 mm) in thisRhodic Ferralsol.Oades and Waters (1991) reported that an aggre-gate hierarchy is not evident in soils where oxides arethe dominant aggregate stabilizing agents (e.g. Ferral-sols with high clay content). It is commonly under-stood that in such soils aggregates between 0.02 and
196B.Madarietal./Soil&TillageResearch80(2005)185–200Table 7Bivariate correlation (Pearson’s) between total organic carbon over aggregate size classes (g TOC in aggregate fraction kg−1 whole soil) and soil aggregation indices (AS,MWD, MGD) of a Rhodic Ferralsol, Londrina, BrazilAggregationindicesAggregate size classes Macroaggregates Microaggregates8–19 mm 4–8 mm 2–4 mm 1–2 mm 0.5–1 mm 0.25–0.5 mm 0.125–0.25 mm 0.053–0.125 mm 0.25–19 (mm) 0.053–0.25 (mm)0–5 cmaAsb 0.84∗∗∗ 0.62∗ ns ns −0.53∗ −0.79∗∗ −0.84∗∗∗ −0.94∗∗∗ 0.83∗∗∗ −0.88∗∗∗MWDc 0.92∗∗∗ 0.55∗ ns −0.76∗∗ −0.81∗∗ −0.89∗∗∗ −0.85∗∗∗ −0.83∗∗∗ 0.83∗∗∗ −0.85∗∗∗MGDd 0.92∗∗∗ 0.59∗ ns −0.71∗∗ −0.77∗∗ −0.90∗∗∗ −0.87∗∗∗ −0.87∗∗∗ 0.85∗∗∗ −0.88∗∗∗5–10 cmaAsb 0.73∗∗ ns ns ns −0.78∗∗ −0.92∗∗∗ −0.99∗∗∗ −0.96∗∗∗ 0.58∗ −0.99∗∗∗MWDc 0.89∗∗∗ ns ns ns −0.90∗∗∗ −0.91∗∗∗ −0.88∗∗∗ −0.83∗∗∗ 0.67∗∗ −0.88∗∗∗MGDd 0.88∗∗∗ 0.52∗ ns ns −0.91∗∗∗ −0.96∗∗∗ −0.95∗∗∗ −0.89∗∗∗ 0.68∗∗ −0.94∗∗∗10–20 cmaAsb 0.77∗∗ ns ns −0.75∗∗ −0.79∗∗ −0.96∗∗∗ −0.99∗∗∗ −0.98∗∗∗ ns −0.99∗∗∗MWDc 0.97∗∗∗ ns −0.60∗ −0.94∗∗∗ −0.87∗∗∗ −0.86∗∗∗ −0.78∗∗ −0.80∗∗ ns −0.79∗∗MGDd 0.96∗∗∗ ns −0.57∗ −0.94∗∗∗ −0.90∗∗∗ −0.92∗∗∗ −0.85∗∗∗ −0.87∗∗∗ ns −0.86∗∗∗0–20 cmeAsb 0.65∗∗∗ 0.46∗∗ ns −0.47∗∗ −0.70∗∗∗ −0.89∗∗∗ −0.94∗∗∗ −0.94∗∗∗ 0.50∗∗ −0.95∗∗∗MWDc 0.84∗∗∗ 0.43∗∗ ns −0.69∗∗∗ −0.83∗∗∗ −0.85∗∗∗ −0.80∗∗∗ −0.76∗∗∗ 0.63∗∗∗ −0.80∗∗∗MGDd 0.81∗∗∗ 0.45∗∗ ns −0.67∗∗∗ −0.84∗∗∗ −0.90∗∗∗ −0.87∗∗∗ −0.84∗∗∗ 0.60∗∗∗ −0.87∗∗∗ns: not signiﬁcant at P ≤ 0.05.a n = 15 per aggregate size class for all aggregation indices.b Aggregate stability index (%).c Meanweight diameter (mm).d Mean geometric diameter (mm).e n = 45 per aggregate size class for all aggregation indices.∗ P ≤ 0.05.∗∗ P ≤ 0.01.∗∗∗ P ≤ 0.0001.
B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200 197Fig. 2. The correlation between organic carbon (g TOC in aggregate fraction kg−1 whole soil) of aggregate size classes and the meanweightdiameter (MWD, n = 45) of a Rhodic Ferralsol at 0–5 cm, Londrina, Brazil. Aggregate size classes: (a) 8–19 mm; (b) 4–8 mm; (c) 2–4 mm;(d) 1–2 mm; (e) 0.5–1 mm; (f) 0.50–0.25 mm; (g) 0.250–0.125 mm; (h) 0.125–0.053 mm. Superscript 1 indicates residual sum of squares(RSSQ). **P ≤ 0.01, ***P ≤ 0.001.
198 B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200Table 8Bivariate correlation (Pearson’s) between total organic carbon overaggregate size classes (g TOC in aggregate fraction kg−1 wholesoil) and soil aggregation indices (AS, MWD, MGD) of a RhodicFerralsol, Londrina, BrazilAggregation indicesa Aggregate size classes2–4 mm 1–2 mm0–5 cmbAsc ns nsMWDd ns 0.53∗MGDe ns ns5–10 cmbAsc ns nsMWDd ns 0.44∗MGDe ns ns10–20 cmfAsc ns nsMWDd 0.38∗ nsMGDe ns nsns: not signiﬁcant at P ≤ 0.05.a Values obtained by EMBRAPA method (Embrapa, 1997).b n = 15 per aggregate size class for all aggregation indices.c Aggregate stability index.d Meanweight diameter.e Mean geometric diameter.f n = 30 per aggregate size class for all aggregation indices.∗ P ≤ 0.05.0.25 mm are highly stable and are not destroyedby agricultural practices (Tisdall and Oades, 1982).Castro Filho et al. (1998) found no effect of tillage onthe distribution of the smaller macroaggregate classes(0.25–0.5 and 0.5–1 mm) at 0–10 cm layer, which isin accordance with the above statement.Recent studies have begun to elucidate the role ofNT on carbon sequestration in soils. Six et al. (1999,2000) described the mechanism of carbon sequestra-tion under NT and CT systems through soil aggre-gation. In loamy soils dominated by 2:1 type clayminerals with well developed aggregates, the capacityof NT for C sequestration was higher due to slowermacroaggregate turnover rate under NT than underCT. They also pointed out that the intra-particulateorganic matter drives this process. Organic matterfractions were not determined in the present study.However, in the Rhodic Ferralsol examined herethere was a strong relationship between the formationof large size macroaggregates (8–19 mm) and TOC.Consequently, the formation of large water-stableaggregates is relevant for carbon sequestration. Thisis supported by the highly positive correlation be-tween aggregation indices, especially MWD, MGD,and TOC of the 8–19 mm aggregate size class. Tillagehad a signiﬁcant effect on the turnover rate of thisaggregate size class, and thus on the capacity of NTto sequester carbon in the soil.4. ConclusionsThe use of the 19 mm sieve at sampling on site forthe homogenization of the samples and the use of the8 mm sieve in the wet sieving procedure enabled usto identify a large (8–19 mm) water stable aggregatesize class. The advantage of this procedure was thatthe capacity of NT to form large stable aggregates wasnot underestimated.Similar to the forest site, NT with crop rotationsigniﬁcantly improved soil aggregation, and accumu-lated organic carbon within aggregates of a RhodicFerralsol at the surface layer (0–5 cm), which weredestroyed by CT. The formation of large macroaggre-gates (8–19 mm) under the conservation tillage system(NT) was important in the accumulation of TOC inthis Rhodic Ferralsol.Effects on soil aggregates were strongly stratiﬁed inthe soil proﬁle due to NT, with the most pronouncedeffect at 0–5 cm. Crop rotation did not signiﬁcantlyaffect soil aggregation.The study of organic carbon in different aggregatesize classes, compared to whole soil organic carbonwas more revealing in describing soil aggregation re-lationships.AcknowledgementsThe scientiﬁc support of Fundação Carlos ChagasFilho de Amparo à Pesquisa do Estado do Rio deJaneiro (The Scientiﬁc Foundation of Rio de JaneiroState, Brazil, FAPERJ Grant No. E-26/150.916/2000)and Embrapa Solos (Rio de Janeiro, Brazil, SEPProject No. 01.1999.203.04) are gratefully acknowl-edged. Special thanks are extended to Robert Boddey(Embrapa Agrobiologia) and Alan Franzluebbers(Joint Editor-in-Chief of Soil & Tillage Research) forthe critical review of this manuscript. The technical
B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200 199and professional support given by Celso Vainer Man-zatto (Embrapa Solos), Carlos Alberto Silva (FederalUniversity of Lavras, UFLA, Brazil), and Celso deCastro Filho (Institute of Agronomy of Paraná State,IAPAR) is highly appreciated by the authors as wellas the advice on statistical analysis by Alberto C. deCampos Bernardi (Embrapa Solos). All the hours ofwork to maintain the analysers by Tatiana F. Lafayettede Sá, Alexandre V. Grillo, Ednaldo da Silva Araújo,Felipe S. dos Santos, Ricardo Soares, and Lucas Fer-nandes de Souza (undergraduate students of the Pon-tiﬁcate Catholic University, PUC-Rio, Rio de Janeiro,the Federal Rural University of Rio de Janeiro State,UFRRJ, Seropédica, and of the Fluminense FederalUniversity, UFF, Niterói, Brazil) are greatly appreci-ated.ReferencesAlvarenga, R.C., Fernandes, B., Silva, T.C.A., Resende, M., 1986.Estabilidade de agregados de um Latossolo Roxo sob diferentesmétodo de preparo do solo e de manejo da palhada do milho.R. Bras. Ci. Solo 10, 273–277.Braunack, M.V., Dexter, A.R., 1989. Soil aggregation in theseedbed: a review. II. Effect of aggregate sizes on plant growth.Soil Till. Res. 14, 281–298.Carpenedo, V., Mielniczuk, J., 1990. Estado de agregação equalidade de agregados de Latossolos Roxos, submetidos adiferentes sistemas de manejo. R. Bras. Ci. Solo 14, 99–105.Castro Filho, C., Lourenço, A., Guimarãesde, M.F., Fonseca,I.C.B., 2002. Aggregate stability under different managementsystems in a red Latosol in the State of Paraná, Brasil. SoilTill. Res. 65, 45–51.Castro Filho, C., Muzilli, O., Podanoschi, A.L., 1998. Estabilidadede agregados e sua relação com o teor de carbono orgˆanico numLatossolo Roxo distróﬁco, em função de sistemas de plantio,rotações de culturas e métodos de preparo das amostras. R.Bras. Ci. Solo 22, 527–538.Castro Filho, C., Vieira, M.J., Casão Jr., R., 1991. Tillage methodsand soil and water conservation in southern Brazil. Soil Till.Res. 20, 271–283.Denef, K., Six, J., Bossuyt, H., Frey, S.D., Elliott, E.T., Merckx,R., Paustian, K., 2001. Inﬂuence of dry–wet cycles on theinterrelationship between aggregate, particulate organic matter,and microbial community dynamics. Soil Biol. Biochem. 33,1599–1611.D´ıaz-Zorita, M., Perfect, E., Grove, J.H., 2002. Disruptive methodsfor assessing soil structure. Soil Till. Res. 64, 3–22.Dutartre, Ph., Bartoli, F., Andreux, F., Portal, J.M., Ange, A.,1993. Inﬂuence of content and nature of organic matter on thestructure of some sandy soils from West Africa. In: Brussaard,L., Kooistra, M.J. (Eds.), Proceedings of the InternationalWorkshop on Methods of Research on Soil Structure/Soil BiotaInterrelationships. Geoderma 56, 459–478.Edwards, A.P., Bremner, J.M., 1967. Microaggregates in soils. J.Soil Sci. 18, 64–73.Embrapa (Centro Nacional de Pesquisa de Solos), 1997. Manualde métodos de análise de solo, 2nd ed. Centro Nacional dePesquisa de Solos, Rio de Janeiro, Brazil. pp. 47–49.Embrapa (Centro Nacional de Pesquisa de Solos), 1981. Mapa desolos do Brasil. 1:5.000.000. Serviço Nacional de Levantamentoe Conservação de Solos, Rio de Janeiro, Brazil.Franzluebbers, A.J., 2002a. Soil organic matter stratiﬁcationratio as an indicator of soil quality. Soil Till. Res. 66, 95–106.Franzluebbers, A.J., 2002b. Water inﬁltration and soil structurerelated to organic matter an its stratiﬁcation with depth. SoilTill. Res. 66, 197–205.Franzluebbers, A.J., Langdale, G.W., Schomberg, H.H., 1999. Soilcarbon, nitrogen, and aggregation in response to type andfrequency of tillage. Soil Sci. Soc. Am. J. 63, 349–355.Freitas, P.L., Zobel, R.W., Snyder, V.A., 1999. Corn root growthin soil columns with artiﬁcially constructed aggregates. CropSci. 39 (3), 725–730.Freixo, A.A., Machado, P.L.O.A., Santos, H.P., Silva, C.A.,Fadigas, F.S., 2002. Soil organic carbon and fractions ofa Rhodic Ferralsol under the inﬂuence of tillage and croprotation systems in southern Brazil. Soil Till. Res. 64, 221–230.Gardner, W.R., 1956. Representation of soil aggregate sizedistribution by a logarithmic-normal distribution. Soil Sci. Soc.Am. Proc. 20, 151–153.Gijsman, A.J., 1996. Soil aggregate stability and soil organicmatter fractions under agropastoral systems established in nativesavanna. Aust. J. Soil Res. 34, 891–907.Gijsman, A.J., Thomas, R.J., 1995. Aggregate size distributionand stability of an Ferralsol under legume-based and pure grasspastures in the Eastern Colombian Savannas. Aust. J. Soil Res.33, 153–165.Haynes, R.J., 2000. Interactions between soil organic matterstatus, cropping history, method of quantiﬁcation and samplepretreatment and their effects on measured aggregate stability.Biol. Fert. Soils 30, 270–275.Haynes, R.J., Francis, G.S., 1993. Changes in microbial biomass C,soil carbohydrate composition and aggregate stability by growthof selected crop and forage species under ﬁeld conditions. J.Soil Sci. 44, 665–675.Kemper, W.D., Rosenau, R.C., 1986. Aggregate stability and sizedistribution. In: Klute, A. (Ed.), Methods of Soil Analysis. PartI. Physical and Mineralogical Methods. Agronomy MonographNo. 9. American Society of Agronomy, Soil Science Societyof America, Madison, WI, pp. 425–442.Koeppen, W., 1931. Grundriss der Klimakunde. Gruyter Verlag,Berlin/Leipzig, p. 388.Kouakoua, E., Larré-Larrouy, M.-C., Barthès, B., Freitas, P.L.,Neves, C., Sala, G.-H., Feller, C., 1999. Relations entre stabilitéde l’agrégation et matiére organiqe totale et soluble à l’eauchaude dans des sols ferrallitiques argileux (Congo, Brésil).Can. J. Soil Sci. 79 (4), 561–569.
200 B. Madari et al. / Soil & Tillage Research 80 (2005) 185–200Lal, R., 2000. Physical management of the soils of the tropics:priorities for the 21st century. Soil Sci. 165 (3), 191–207.Lal, R., Kimble, J., Follett, R.F., 1997. Pedospheric processes andthe carbon cycle. In: Lal, R., Blum, W.H., Valentine, C., Stewart,B.A. (Eds.), Methods for Assessment of Soil Degradation. CRCPress, Boca Raton, pp. 1–8.Machado, P.L.O.A., Silva, C.A., 2001. Soil management underno-tillage systems in the tropics with special reference to Brazil.Nutr. Cycling Agroecosyst. 61, 119–130.Machado, P.L.O.A., Sohi, S.P., Gaunt, J.L., 2003. Effect ofno-tillage on turnover of organic matter in a Rhodic Ferralsol.Soil Use Manage. 19, 250–256.Mrabet, R., 2002. Stratiﬁcation of soil aggregation and organicmatter under conservation tillage systems in Africa. Soil Till.Res. 66, 119–128.Oades, J.M., Waters, A.G., 1991. Aggregate hierarchy in soils.Aust. J. Soil Res. 29, 815–828.Oliveira-Filho, A.T., Ratter, J.A., 1995. A study of the originof central Brazilian forests by the analysis of plant speciesdistribution patterns. Edinb. J. Bot. 52, 141–194.Oyedele, D.J., Schjønning, P., Sibbesen, E., Debosz, K., 1999.Aggregation and organic matter fractions of three Nigeriansoils as affected by soil disturbance and incorporation of plantmaterial. Soil Till. Res. 50, 105–114.Pavan, M.A., Bingham, F.T., Pratt, P.F., 1985. Chemical andmineralogical characteristics of selected acid soils of the stateof Paraná, Brazil. Turrialba 35 (2), 131–139.Perfect, E., Kay, B.D., Ferguson, J.A., Da Silva, A.P., Denholm,K.A., 1993. Comparison of functions characterizing the dryaggregate size distribution of tilled soil. Soil Till. Res. 28, 123–139.Roth, C.H., Castro Filho, C., de Mediros, G.B., 1991. Análisede fatores f´ısicos e qu´ımicos relacionados com a agregaçãode um Latossolo Roxo distróﬁco. R. Bras. Ci. Solo 15, 241–248.Roth, C.H., Pavan, M.A., Chaves, J.C.D., Meyer, B., Frede, H.-G.,1986. Efeitos das aplicações de calcário e gesso sobre aestabilidade de agregados e inﬁltração de água em um LatossoloRoxo cultivado com cafeeiros. R. Bras. Ci. Solo 10, 163–166.Salako, F.K., Babalola, O., Hauser, S., Kang, B.T., 1999. Soilmacroaggregate stability under different fallow managementsystems and cropping intensities in southwestern Nigeria.Geoderma 91, 103–123.SAS Institute, 1990. SAS User’s Guide: Statistics, Version 6thEdition. SAS Institute Inc., Cary, NC.Schjønning, P., Rasmussen, K.J., 1989. Long-term reducedcultivation. I. Soil strength and stability. Soil Till. Res. 15,79–90.Sidiras, N., Henklain, J.C., Derpsch, R., 1982. Vergleich von dreiBodenbearbeitungsverfahren im Bezug auf einige physikalischeEigenschaften, Boden- und Wasserkonservierung und Erträgevon Soja und Weizen auf einem Ferralsol. Z. Acker- undPﬂanzenbau (J. Agron. Crop Sci.) 151, 137–148.Silva, I.F., Mielniczuk, J., 1998. Sistemas de cultivo ecaracter´ısticas do solo afetando a estabilidade de agregados. R.Bras. Ci. Solo 22, 311–317.Silva, I.F., Mielniczuk, J., 1997. Ação do sistema radicular deplantas na formação e estabilização de agregados. R. Bras. Ci.Solo 21, 113–117.Silva, M.S.L.da., Ribeiro, M.R., 1992. Inﬂuˆencia do cultivocont´ınuo da cana-de-açúcar em propriedades morfológicas ef´ısicas de solos argilosos de Tabuleiro no Estado de Alagoas.R. Bras. Ci. Solo 16, 397–402.Six, J., Elliott, E.T., Paustian, K., 1999. Aggregate and soil organicmatter dynamics under conventional and no-tillage systems.Soil Sci. Soc. Am. J. 63, 1350–1358.Six, J., Elliott, E.T., Paustian, K., 2000. Soil macroaggregateturnover and microaggregate formation: a mechanism for Csequestration under no-tillage agriculture. Soil Biol. Biochem.32, 2099–2103.Tisdall, J.M., Oades, J.M., 1982. Organic matter and water stableaggregates in soils. J. Soil Sci. 33, 141–163.Yoder, R.E., 1936. A direct method of aggregate analysis of soilsand a study of the physical nature of erosion losses. J. Am.Soc. Agron. 28 (5), 337–351.Young, I.M., Crawford, J.W., Rappoldt, C., 2001. New methodsand models for characterising structural heterogeneity of soil.Soil Till. Res. 61, 33–45.