The field experiments with different types of management were established in 1975 at Yamaguchi Prefecture Agricultural Experimental Station, Japan (34˚14'N, 131˚4'E). Yamaguchi is most west Prefecture of Honshu island. Yamaguchi City has a humid subtropical climate: mean annual temperature is ca. 15˚C and mean annual precipitation ca. 1800 mm. The soil at this field site was classified as Gray Lowland soil (Fluvisols). Each plot of the field experiment was 200 m², and the treatment of plot was carried out without replications. For the comparison of potential role of compost application, three plots were selected: (i) plot F, only chemical fertilizers containing N, P, and K were applied; (ii) plot F + LC, chemical fertilizers plus a low level of compost were applied; (iii) plot F + HC, chemical fertilizers plus a high level of compost were applied. The same plots were used as paddy fields for Japanese rice (e.g., Koshihikari) in summer and as upland fields for barley in winter until June 2001. The application rate of N, P₂O₅, and K₂O for each crop was 100 kg∙ha⁻¹, respectively (N: 200 kg∙ha⁻¹∙y⁻¹; P₂O₅: 200 kg∙ha⁻¹∙y⁻¹; K₂O: 200 kg∙ha⁻¹∙y⁻¹). After harvest (June and November) of each crop, rice straw-cow dung compost (the mixture of rice straw and cow dung in the ratio of 7 to 3 was stacked in a composting room for 6 months with mixing intermittently) was applied at the rates of 5 Mg∙ha⁻¹ (10 Mg∙ha⁻¹∙y⁻¹) for the low level and 15 Mg∙ha⁻¹ (30 Mg∙ha⁻¹∙y⁻¹) for the high level. However, since June 2001 (26 years after the start of field experiment), these plots were used only as paddy fields and consequently the amounts of chemical fertilizers and compost applied were reduced by half. Water irrigation and weed control were performed conventionally.

To obtain an average soil sample in each plot, soils were taken from the plow layer (0 - 15 cm) of five sites across each of the three plots and mixed well. In many experiments, the soils were air-dried, gently crushed, passed through a 2-mm mesh sieve, and then employed for analytical determinations.

As described later, the long-term application of chemical fertilizer and compost into the plots increased organic C (OC) and total N (TN) contents of whole soil. The OC and TN contents (g∙kg⁻¹) increased from 13.1 and 1.20 before the start of the field experiment to 17.6 and 1.56 for plot F, 22.1 and 2.18 for plot F + LC, and 30.5 and 2.84 for F + HC, respectively, in April 2007 (32 years after the start of the field experiment) [12] .

Soils were taken from plots F, F + LC, and F + HC in October 1994 (19 years after the start of field experiment) [6] . Moist soil samples (<2 mm, sieved without air-drying) were used for the determination of enzyme activities and microbial biomass N. The activities of organic C-decomposing enzymes (α-glucosidase, β-glucosidase, α-galactosidase, and β-galactosidase), organic N-decomposing enzymes (protease, β-acetylglucosaminidase, and adenosine deaminase), and organic P-decomposing enzymes (phosphomonoesterase [pH 6.5], phosphomonoesterase [pH 11], and phosphodiesterase) were determined according to the methods described previously [13] - [17] . On the other hand, OC, TN, and cation exchange capacity (CEC) were analyzed according to the method described elsewhere [18] , using air-dried soil sample (<2 mm). The relationships between the enzyme activities and the chemical properties of soils were examined by statistics analysis (correlation coefficient and t-test).

The amount of microbial biomass N in the moist soils described above was determined according to the chloroform fumigation-extraction method reported by Brookes et al. [19] .

The amount of phosphate-extractable organic N (PEON) in the soils, which were taken from plots F and F + HC in June 2000 (25 years after the start of field experiment), was determined [8] . This extraction method, using neutral 1/15 mol∙L⁻¹ phosphate buffer as an extractant, was proposed as a useful index for estimating the amount of available N in Japanese paddy soils [20] .

The amount of hydrolyzable amino acid-N in the soils, which were taken from plots F and F + LC, and F + HC in October 2008 (33 years after the start of field experiment), was determined as described elsewhere (e.g., [9] ).

Soils were taken from plots F, F + LC, and F + HC in November 1996 (21 years after the start of field experiment) [7] . Part of moist soil was used to determine the hyphal length after sieving with a 2-mm mesh sieve. The remaining part was air-dried, gently crushed, and then sieved with a 4-mm mesh sieve, followed by a 2-mm mesh sieve. These sieved samples (2-4 mm in diameter) were used for the determination of soil aggregation and other analytical determinations.

In this study [7] , the aggregate stability index was represented by the degree of aggregation. The soil samples were sieved by the wet sieving method [21] [22] , using the apparatus designed by Daiki Rika Kogyo Co. Ltd., Tokyo, Japan. The details of the sieving procedure were described previously by Ibrahim et al. [23] . The degree of aggregation is an expression of the degree to which small particles are aggregated. The aggregation degree was estimated from the differences of mass weights of aggregate fractions obtained before and after the digestion of OM with H₂O₂ [23] , Furthermore, relationships between the degree of macroaggregation (>0.25 mm) and the contents of OC, TN, hydrolyzable carbohydrates, and dithionite-citrate-bicarbonate (DCB)-soluble Al and Fe, or hyphal length were examined by statistics analysis (correlation coefficient and t-test).

Soil enzymes are mainly microbial origins, and they play an important role in the transformation and decomposition of organic components involving C, N, or P in soils (e.g., [15] ). A number of studies have shown that the application of organic amendments can increase the activities of enzymes in soils. For example, according to Kanazawa [5] [26] continuous application of manure or compost into various paddy soils in Japan increased the enzyme activities such as β-glucosidase, cellulase, protease, β-glucosaminidase, L-glutaminase, and L-aspara- ginase. Sato and Omura [27] found that continuous compost application increased the activities of protease, adenosine deaminase, and β-acetylglucosaminidase in the soil of an Andosol paddy field. On the other hand, Omura et al. [28] reported that continuous compost application into a greenhouse field increased the activities of hydrolytic enzymes such as protease, L-glutaminase, and L-asparaginase. Chang et al. [29] described that compost application increased enzyme activities, such as protease and β-glucosidase, in greenhouse cultivation soils, although the degrees of increase were influenced by the quantity of compost applied. Shindo [30] found that continuous compost application increased the activities of protease, β-acetylglucosaminidase, and adenosine de- aminase in the upland soils differing in parent materials and that protease and β-acetylglucosaminidase activities were correlated significantly with the amount of N mineralized during the incubation of their soils. However, enzyme activities in double cropping soils have received little attention. Thus, Shindo and Shojaku [6] investigated the effect of compost application on the activities of organic C-, N-, and P-decomposing enzymes of soil in a field subjected to double cropping (19 years after the start of field experiment). Furthermore, the relationships between the enzyme activities and the chemical properties or microbial biomass of soil were examined.

The activities of organic C-decomposing enzymes (α-glucosidase, β-glucosidase, α-galactosidase, and β-galacto- sidase) in the soils of plots F, F + LC, and F + HC ranged from 2.4 to 14.9 U (μmol∙min⁻¹∙kg⁻¹ dried soil) (Table 1). On the other hand, the activities of organic N-decomposing enzymes (protease, β-acetylglucosaminidase, and adenosine deaminase) and organic P-decomposing enzymes (phosphomonoesterase [pH 6.5], phosphomonoesterase [pH 11], and phosphodiesterase) varied from 3.4 to 21.7 U and from 4.6 to 61.8 U, respectively (Table 1). All the activities of organic C-, N-, and P-decomposing enzymes increased remarkably in the order: plots F < F + LC < F + HC. Furthermore, in most enzymes, their activities were correlated significantly (r ³ 0.950, p < 0.05) with the contents of OC, TN, and microbial biomass N or CEC value (Table 2). These findings unambiguously indicate that the degrees of increase in those enzyme activities are influenced by the amount of compost applied as well as the enzyme types. Also, it appears that the compost application contributes to the improvement of soil qualities in the double cropping field.

Chang et al. [29] reported that compost application increased microbial (bacteria, actinomycetes, and fungi) populations and enzyme activities, such as protease and β-glucosidase, in soils of greenhouse cultivation, although the populations and activities were influenced by the amount of compost applied. As described above, the compost application at the high level (plot F + HC) increased largely the activities of organic C-, N-, and P-decomposing enzymes, contents of OC, TN, and microbial biomass N, and the CEC value. The reasons are proposed as follows: (i) the growth of paddy rice and barley is accelerated to a larger extent in plot F + HC than in plots F and F + LC, due to the larger amount of compost application; (ii) as a result, larger amounts of plant remains such as root and stubble are left in plot F + HC than in the other plots; (iii) those remains and compost serve as energy and nutrient of soil microorganisms and, simultaneously, contribute to the maintenance and improvement of soil qualities, thereby increasing the amount, population, and species of microorganisms, and thus (iv) larger amounts of organic C-, N-, and P-decomposing enzymes are produced and released in plot F + HC, compared with the other plots.

a. Plot F: Only chemical fertilizers containing N, P, and K were applied; b. Plot F + LC: Chemical fertilizers plus compost at low level (10 Mg∙ha⁻¹∙y⁻¹) were applied; c. Plot F + HC: Chemical fertilizers plus compost at high level (30 Mg∙ha⁻¹∙y⁻¹) were applied. *Standard deviation (n = 4).

Nitrogen is an essential element for plants. In the surface layer of most soils, over 90% of N occurs in organic forms. It is well known that hydrolyzable N is the major source of mineralizable N in soils, and this hydrolyzable N contains protein-like N compounds (e.g., [31] ). Continuous manure application could increase significantly the amount of amino acid-N in the hydrolyzable fractions of soils (e.g., [32] ). Similar results were obtained for the double cropping soils (33 years after the start of field experiment) studied (Table 3) [9] . The amount of PEON, containing protein-like N compounds, is employed as a useful index for estimating the amount of available N in Japanese paddy soils [20] . According to Shindo et al. [8] , in the soils taken from plots F and F + HC (25 years after the start of field experiment), the amounts of PEON and TN in plot F + HC were 1.2 and 1.7 times larger than those in plot F, respectively (Table 3).

It is accepted that microorganisms are a transformation agent of OM as well as a labile reservoir of nutrients such as N, P, S, and C in soils (e.g., [33] [34] ). Microbial bodies are formed and then died may accumulate in soils. Predation upon soil microorganisms may release most of the immobilized C, N and P, which would otherwise be unavailable to crop plants. Matsumoto et al. [35] suggested that protein-like N compounds in the PEON fraction, derived from the remains of microorganisms, were adsorbed on surfaces of soil colloids and became a source of mineralizable N in soils. Mineralization of organic N in soils proceeds with the mediation of organic N-decomposing enzymes. As described earlier, compost application increased the amount of microbial biomass N as well as the activities of organic N-decomposing enzymes in the soils of plot F + LC and, especially, plot F + HC (Table 1 and Table 3). Based on these findings, it is concluded that the compost application, especially at the high level, is effective for the maintenance and enhancement of N fertility in the double cropping soils studied.

a. Correlation coefficients of 0.950 or higher indicate p < 0.05.

a. See Table 1; b. Data from Shindo et al. [8] ; c. Data from Shindo and Shojaku [6] ; d. Data from Nguyen and Shindo [9] .

Water-stable aggregates are essential for the maintenance of a good structure for plant growth. These aggregates are divided into microaggregates (<0.25 mm) and macroaggregates (>0.25 mm). Microaggregates show a relatively high stability against physical disruption [36] , whereas macroaggregates are sensitive to soil management [37] .

The long-term incorporation of organic amendments such as straw increased the amount of 1 - 20 mm aggregates as well as the OM content in a loamy sandy soil [38] . Angers and N’Dayegamiye [39] showed that long- term application of manure increased the contents of C, N, and total carbohydrates of the whole soil and of the particle size fractions. Furthermore, it was reported that OC, TN, carbohydrates, DCB-soluble Al, and hyphae affect the stabilization of aggregates mainly based on the correlation with soil aggregation [23] [40] -[43] . In most of these studies, however, the role of only a few aggregate-forming factors was compared at the same time. Furthermore, the effect of continuous compost application on soil aggregation and its related factors in the fields subjected to double cropping remains to be studied. Thus, the influence of compost application on soil macroaggregation and the factors involved in the macroaggregation were investigated using the soil samples taken from plots F, F + LC, and F + HC (21 years after the start of field experiment) [7] .

The average recovery of water-stable aggregates (>2.0 mm, 1.0 - 2.0 mm, 0.5 - 1.0 mm, 0.25 - 0.5 mm, 0.1 - 0.25 mm, and <0.1 mm) by duplicate wet sieving ranged from 99 to 100% in all the soils of plots F, F + LC, and F + HC. Thus, the degree of aggregation in five size fractions (>2.0 mm, >1.0 mm, >0.5 mm, >0.25 mm, and >0.1 mm) was determined. In all the size fractions, the degree of aggregation increased in the order: plots F < F + LC < F + HC. The degree was 1.8 to 13 times higher in plot F + HC, compared with plot F. The degree of macroaggregation (>0.25 mm) was 35% for plot F, 43% for plot F + LC, and 62% for F + HC, indicating that the compost application at the high level promoted largely macroaggregation.

It is assumed that a complex mechanism may be involved in the formation as well as the increase of the amount of water-stable macroaggregates. To gain a fundamental understanding about the mechanism in the double cropping soils, the relationships between the degrees of macroaggregation and several properties of soils were examined. The degree of macroaggregation showed significant correlations (r ³ 0.950, p < 0.05) with the contents of OC, TN, hydrolyzable carbohydrates, and DCB-soluble Al (Table 4). Furthermore, these contents were significantly intercorrelated (r ³ 0.950, p < 0.05) among themselves. As described later, continuous compost application largely increased the amounts of fulvic acids (FAs) and, especially, humic acids (HAs) in the soils of experimental plots. The increase in the amounts of negatively charged organic constituents such as hydrolyzable carbohydrates (involving polysaccharides), HAs, and FAs and of positively charged DCB-soluble (active) Al which are capable of binding soil particles into secondary organomineral complexes, i.e., water-sta- ble aggregates, may promote macroaggregation. On the other hand, Oades [44] reported that a higher degree of macroaggregation was provided by fungal hyphae through the physical enmeshment of soil particles. N’Dayegamiye and Angers [45] observed that the application of manure to soil increased the fungal population. Ibrahim and Shindo [46] found that when the soil amended with rice straw was incubated at 30℃ under the moist conditions, the

a. Correlation coefficients of 0.950 or higher indicate p < 0.05.

amount of the >2.0 mm macroaggregates remarkably increased, accompanying with the increases of hyphal length and of the amount of microbial biomass C. Furthermore, the compost application at the high level in our studies (plot F + HC) markedly increased the hyphal length (264 km kg⁻¹ dried soil), compared with plots F (197 km) and F + LC (192 km) [7] . Based on the results obtained, in the fields subjected to long-term double cropping under study, it appears that non-humic (polysaccharides) and humic substances (HAs and FAs), active Al, and hyphae contribute together to the formation as well as the increase of the amount of water-stable macroaggregates. Continuous application of compost, especially at the high level, enhanced the role of those factors in the improvement of the soil structure.

Humus is an integral part of soil and affects various soil properties as well as global C cycle. It is well known that continuous application of organic amendments can affect the humus composition of paddy and upland soils (e.g., [24] [47] ). However, the effect of compost application on the humus composition of double cropping soils remains to be studied.

It is easy to assume that organic amendments such as compost, manure, and plant remains incorporated into soils become small in size and the amounts of mineral particles adhering and/or being associated with them increase, with the progression of transformation and degradation by abiotic and biotic agents. According to Roppongi and Mundie [48] , OM of recent plant origin is considered to be preferentially recovered in the sand-size fraction, whereas more microbially processed material can be found in the silt- and clay-size fractions.

Thus, we investigated the qualitative and quantitative changes of humus in the double cropping soils as influenced by continuous compost application, using whole soils taken from plots F, F + LC, and F + HC and their particle size fractions [8] [10] [11] .

Long-term compost application (21 years after the start of the field experiment) increased remarkably the contents of TOM, HA, and FA in the whole soils in the order: plots F < F + LC < F + HC (Table 5) [10] . Humic acid content exceeded largely FA content in all the plots and, especially, plot F + HC. On the other hand, the degree of darkening of HA increased in the order: F + HC < F + LC < F. In other words, the HA content was the highest in plot F + HC, but the degree of darkening of HA was the lowest. The reasons are proposed as follows: (i) the growth of paddy rice and barley is accelerated to a larger extent in plot F + HC than in the other plots, due to the continuous compost application at the high level; (ii) as a result, larger amounts of plant remains such as root and stubble, which can be transformed into HA and FA, are left in plot F + HC than in the other plots; (iii) the degrees of darkening of HAs in the compost used [10] and plant remains decayed and degraded in soils [24] are very low; and thus (iv) continuous compost application at the high level contribute to the accumulation of larger amounts of TOM, HAs with a low degree of darkening, and FAs, although part of indigenous humus may be decomposed by soil microorganisms, presumably due to priming effect.

The effects of organic amendments on humus composition have also been studied for the soils of upland fields. For example, Roppongi et al. [49] reported that the incorporation of compost into an upland field (Fulvisols) induced an increase in the amounts of TOM, HA, and FA and a decrease in the darkening degree of HA. Similar results were obtained even when cattle manures were applied into upland fields of an Andosol [50] and a Cambisol [51] . Hirahara [52] found that long-term (55 years) compost application into an Andosol upland field decreased the δ13C values of FA and, especially, HA, the reduction being a larger at a higher level of compost application. These findings indicate that the accumulation of plant material-derived HAs with a low degree of darkening in soils is accelerated by continuous application of compost and manure.

a. Data from Shindo and Shimada [10] ; b. See Table 1; c. Δlog K and RF values stand for color coefficient and relative color intensity, respectively.

¹³C-NMR analysis can provide valuable information on the C form in complex macromolecules. Thus, Shindo et al. [8] characterized the C species of HAs and FAs, which were isolated from the soils of plots F and F + HC (25 years after the start of the field experiment), by their ¹³C-NMR analyses (TOSS method) (Figure 1). In the ¹³C-NMR spectra of all the HAs, peaks appeared at 30, 55, 74, 130, 150, and 173 ppm. On the other hand, in the ¹³C-NMR spectra of all the FAs, peaks appeared at approximately 20, 75, 100, and 175 ppm. In this study, individual C types were not assigned from ¹³C-NMR spectra because both HAs and FAs are heterogeneous polymers. The peaks at 55 and 150 ppm in the NMR spectra of the HAs, which show the presence of lignin-like compounds [53] , were more intense for the HA with a lower degree of darkening in plot F + HC, compared with plot F. This finding suggests that part of lignin is transformed into HA with a low degree of darkening. To gain a better understanding of the C species in the HAs and FAs, the contents of individual C species (alkyl-C, O-alkyl-C, aromatic-C, and carbonyl-C) and the quantitative contribution of individual C species relative to total C in the HAs or FAs were estimated from the ¹³C-NMR spectra and C contents of HA and FA (Table 6). Although the compost application increased the contents of all C species in the HAs and FAs, the quantitative contribution of individual C species relative to total C in the HAs or FAs did not differ largely between plots F and F + HC. However, there were distinct differences in the distribution of C species between the HAs and FAs. The percentage contribution of individual C species in the HAs of plots F and F + HC was the highest for aromatic-C (37% and 38%, respectively), generally followed by carbonyl-C (24% and 24%), alkyl-C (22% and 19%), and O-alkyl-C (17% and 19%). On the other hand, in the FAs of plots F and F + HC, the distribution was the highest for O-alkyl-C (48% and 45%, respectively), generally followed by carbonyl-C (27% and 24%), alkyl-C (12 and 17%), and aromatic-C (13% and 14%). The HAs were characterized by presence of large amount of aromatic-C and small amount of O-alkyl-C, and the inverse relationship was found for the FAs. The aromatic-C species may involve the decaying and degradation products of lignin, while O-alkyl-C may include plant and microbial carbohydrate residues.

The soils taken from the three plots (32 years after the start of field experiment) were divided physically into seven fractions as described earlier. In many particle size fractions, the amounts (g C kg⁻¹ whole soil) of TOM, HA, and FA increased in the order: plots F < F + LC < F + HC [11] . This result indicates that the accumulation of TOM, HA, and FA in the particle size fractions of double cropping soils was influenced by the amounts of plant materials supplied, involving compost (as described earlier). Here, percentage distribution of the amounts of TOM, HA, and FA in each fraction relative to those in the whole soil was compared between plots F and F + HC (Figure 2). In both plots, 62% of TOM, 66% - 70% of HA, and 77% - 78% of FA was occupied in SIA (2 - 20 μm) and CLA (<2 μm) fractions. Most of the rest OM existed in CSA- and MSA-DP (212 - 2000 and 53 - 212 μm, respectively) fractions. The distribution values in CSA- and MSA-MP (212 - 2000 and 53 - 212 μm, respectively) and FSA (20 - 53 μm) fractions were low or very low.

a. Data from Shindo et al. [8] ; b. See Table 1.

Aoyama and Kumakura [50] reported that the application of large amounts of manure (up to 320 Mg∙ha⁻¹∙y⁻¹) into Kuriyagawa Andosol (upland field) for 20 years increased significantly the amount of particulate OM (>53 μm), but it did not affect the amount of mineral-associated OM (<53 μm). The increase of particulate OM was influenced largely by the amount of manure applied, and it was in the order of 80 < 160 < 320 Mg∙ha⁻¹∙y⁻¹. A similar result was also obtained for Hirosaki Andosol (upland field) [54] . However, in the double cropping fields described above, where compost was applied at the rates of 10 Mg∙ha⁻¹∙y⁻¹ (plot F + LC) and 30 Mg∙ha⁻¹∙y⁻¹ (plot F + HC) for about 30 years, the amount of TOM was much larger in the SIA and CIA fractions than in the other fractions. In Fujisaka Andosol (upland field) where compost was incorporated continuously at the rates of 11 and 34 Mg∙ha⁻¹∙y⁻¹ for about 40 years [47] , the distribution pattern of total OC in size fractions was similar to that of TOM in our studies. These findings may indicate that in the soils where organic amendments are continuously applied, the accumulation pattern of OM in the particle size fraction is influenced largely by the amounts of amendments applied.

The optical properties (∆ log K and RF values) of HAs in CSA-DP, MSA-DP, SIA, and CLA fractions were examined, since these fractions contained enough amounts of HAs to determine their properties. Kumada [24] proposed to classify soil HAs employing the optical properties. According to his classification system, the lower the ∆ log K value and the higher the RF value, the higher the degree of darkening of HA. In both plots F and F + HC, the degree of darkening of HA was the highest in CLA, followed by SIA, MSA-DP, and CSA-DP fractions (Figure 3). Although the amounts of TOM and HA in the whole soil and SIA and CLA fractions were much larger in plot F + HC than in plot F [11] , the degrees of darkening of HAs in both fractions were much lower in plot F + HC than in plot F (Figure 3). These results indicate that larger amounts of plant materials, which can be transformed into HA with a low degree of darkening, were supplied into plot F + HC than plot F.

In the physical fractionation method used, after CLA fraction was collected by sedimentation, SIA fraction was recovered. This means that free and fine decayed plant materials (a low density, not or hardly connected to mineral particles) may be recovered in CLA fraction. However, in plot F + HC, the amount of HA was much larger in SIA fraction than in CLA fraction (Figure 2), and the darkening degree of HA was much lower in SIA than in CIA (Figure 3). These results may indicate that in plot F + HC, larger amounts of decayed plant materials containing HA with a low degree of darkening were accumulated (presumably due to the complexation with mineral matrices) in SIA fraction than in CLA fraction. On the other hand, in plot F, the amount and dar-

kening degree of HA did not differ largely between SIA and CLA fractions (Figure 2 and Figure 3). This may be because smaller amounts of plant materials were supplied into plot F, compared with plot F + HC. The effects of continuous compost application on the quantitative and qualitative changes of humus were limited in plot F + LC, due to the low level of compost application.