This study was carried out on a private farm owned by Mr. Tsutomu Nakamura with his cooperation. The farm is located in the city of Suzano, São Paulo, Brazil, which is located in a hilly area of Ultisols. The field had deteriorated after over 40 years of conventional farming. The application of high C:N ratio organic material was initiated in July 2008, and by 2010 was being used on the entire 2-ha farm. The specific method was as follows: (1) the same crop was planted without a break after harvest, so that a crop was always growing; (2) approximately 15 - 20 t∙ha⁻¹∙crop⁻¹ of fresh waste mushroom bed (C:N ratio, 39; moisture, 61.80%; total carbon (T-C), 19.10%; and total nitrogen (T-N), 0.49%) was added to surface soil to a depth of approximately 10 cm using a rotary tiller; (3) no other materials (nitrogen, phosphorus, or potassium fertilizer, minerals, microelements, growth promoters, pH control chemicals, or agricultural chemicals) were used during the study period; (4) commercially available seedlings and seeds were used; (5) weeds were cut with a brush cutter when they began to compete with crops and were left on the fields; and (6) irrigation of crops was not conducted, except under severe drought conditions, during which irrigation was conducted the day before seeding or planting of seedlings and over the following 2 days. The fresh waste mushroom bed was transported directly from a nearby mushroom farm (Sitio TKM, Suzano) and applied immediately. The NO₃-N concentration in the both 0 - 10 cm and 70 - 80 cm of soil were 5.6 mg∙kg⁻¹ soil (determined on Dec 12, 2010). That was the same as that of the topsoil (0 - 10 cm) of the neighboring forest. Despite the low nitrogen concentration, no nitrogen deficiency or pests were observed (Figure 1). The field had no surface runoff under a rainfall rate of 50 mm per hour. Irrigation was not needed even when the drought period exceeded 65 days before soil survey (Nov 19, 2012).

We summed the number of each vegetable harvested from 2010 to 2012 from sales records and multiplied each number by the item’s standard weight. Records of each crop’s area were not available; therefore, the conventional yield of the farm was calculated by weight using the top five items produced on the study field: lettuce (46%), cabbage (23%), napa cabbage (7%), radish (5%), and cauliflower (4%). We converted the weight percentage to an area percentage, then multiplied this by the average yields. The average annual yields per ha for these crops are 21, 32, 32, 29, and 14 t, respectively [14] . For example, when the area of lettuce was 46%, the area of cabbage was 15% (23% × 21/32). The total conventional yield was 17.6 t (46% × 21 + 15% × 32 + 4% × 32 + 4% × 29 + 6% × 14). The total area of these top five crops was 74%, so the conventional yield per ha was 23.7 t (17.6 × 100/74). This is the standard conventional yield value for Japan. The average unit yield of all vegetables in Japan is 23.3 t∙ha⁻¹ and that of Brazil is 12.8 t∙ha⁻¹ [15] . Thus, the conventional yield was estimated to be 13.0 t (= 23.7 × 12.8/23.3).

We selected a top point of the study field (SF) for the soil survey, which had no water or soil inflow from the surrounding area. In the soil survey plot, lettuce (12 crops) and cabbage (2 crops) were planted from July 2008, and butter cabbages planted on April 9, 2012 were growing at the time of survey (Nov 19, 2012). We chose a neighboring farmer’s field as the control field (CF). The CF was located in the same topography and was developed at the same time as the Japanese farmers’ colony and implemented almost the same agricultural methods for approximately 40 years. In the CF, cassava was harvested in January 2012, and corn was grown after the addition of waste mushroom bed without fertilizer and harvested in July 2012. The CF was then kept fallow.

Soil profiles were described according to the Handbook of Soil Survey (Japanese Society of Pedology, 1997). We collected wet soil samples from each soil horizon and sieved them to 2 mm. We took two sets of 20-mL samples: one was used for the determination of free adenosine triphosphate (ATP) and the other for the determination of moisture, and carbon and nitrogen content.

We decided to evaluate microbial activity rather than microbial biomass; we therefore measured free ATP, which is a measure of microbial activity. Soil samples selected by soil profiling were analyzed within 30 min. We placed samples in cups, added 50 mL of water, and stirred for 1 min with vibration (Power Masher, Nippi Inc, Tokyo, Japan). We then added 6 mL of the surface water to a sample tube and centrifuged it at 6500 rpm (2200 × g) for 1 min. We then placed 100 μL of the solution using an autopipette into ATP Water Test Devices (Aquasnap AQ100F, Hygiena International, Camarillo, CA, USA). We measured the ATP with a luminometer (SystemSURE Plus, Hygiena International) 20 s after mixing in luciferase. The amount of free ATP was calculated using the weight and soil moisture of a paired sample.

Paired soil samples were air dried. T-N and T-C concentrations of the soil samples were determined with a NC analyzer (SUMI-GRAPH NC 200F, Sumitomo Chemical, Tokyo, Japan) using the dry combustion method. The removal of nitrogen and carbon by harvested products was estimated by the average data for unit area [16] . The values per m² were 28.4 g C and 3.03 g N for lettuce and 171.6 g C and 14.21 g N for cabbage (butter cabbage was considered cabbage).

The T-C and T-N balance was estimated for lower and upper limits. The lower limit was calculated from the difference between the SF and CF over the entire soil profile; this was used as the lower limit because the CF had already received one input of waste mushroom bed, and the fallow period restores soil fertility [17] . The upper limit was calculated assuming that the top three soil horizons at the SF were the same as the Bw1 horizon. The net balance was estimated by subtracting the input of components in the applied waste mushroom bed and adding removed components in harvested products to the above output.

Thirty-three crop items were sold in the last 2 years, including leafy, fruit, and root vegetables. Lettuce and cabbage accounted for 46% and 23% of the total weight, respectively. The total annual average yield from 2010 to 2012 was 56.5 t∙ha⁻¹, which was higher than the estimated conventional yield (13.0 t∙ha⁻¹).

The structure of the soil showed that aggregates of up to 29 cm formed in soil at the SF (Table 1, Figure 2). These aggregates were not earthworm feces but subangular blocky structures that formed around plant roots. In the SF, the pores of the A horizon were large, and the bulk densities of horizons Ap1 (0.68) and Ap2 (0.84) were far smaller than that of the Bw1 horizon (1.03). In the CF, the thickness of the aggregate was 22 cm and the bulk densities of horizons Ap1 and Ap2 were 0.90 and 0.95, respectively, which were similar to that of the Bw1 horizon (0.99).

The above differences between the two fields are related to differences in the soil carbon and nitrogen concentrations. In horizons Ap1 and Ap2, the concentrations of T-C were 76.6 (mg∙g⁻¹ soil) and 54.1 in the SF, and

SF: study field. CF: control field. Bw1: Bw1 horizon of study field. The SF was provided with approximately 15 - 20 t∙ha⁻¹∙crop⁻¹ of waste mushroom bed over 15 applications from Jul 2008 to Nov 2012. The CF was left fallow after corn harvest in July 2012. The differences in soil structure were mainly observed between horizons 1 and 2.

27.2 and 27.1 in the CF, respectively. The concentrations of T-N of horizons Ap1 and Ap2 were 4.66 and 3.52 in the SF, and 2.05 and 1.99 in the CF, respectively. In both fields, concentrations of T-C and T-N in horizon 3 were 21.0 - 22.1 and 1.42 - 1.51, respectively, which are much lower than the concentrations in the upper horizons. The C:N ratios of horizons 1 to 3 were 16.4, 15.4, and 14.8 at the SF and 13.3, 13.6, and 14.7 at the CF, respectively. The soil C:N ratio at the CF increased monotonically from the top to the bottom horizons. However, the C:N ratio at the SF dropped at horizon 3, and the C:N ratio of the remaining horizons were the same as those of corresponding horizons at the CF (Table 2). The gradation of the soil C:N ratio at the CF mainly reflected the decrease in T-N. The irregularity observed in horizons 1 (Ap1) and 2 (Ap2) at the SF was caused by the high soil T-C concentration.

Overall, the bulk densities of soil horizons 3 and below were similar, while the differences in soil structure were mainly observed between horizons 1 and 2.

The microbial activity of soil at the SF was one order of magnitude higher than that of soil at the CF. At the SF,

CF: control field. Bw1: Bw1 horizon of study field. The T-C and T-N balance was estimated for lower limit (CF base) and upper limits (Bw1 base). “SF Present” denotes the present value from the study field. “Output” denotes the difference between the SF present levels and the original levels. “Input” denotes the total amount of waste mushroom bed. The output/input ratio of soil total nitrogen and carbon were estimated to be 2.68 - 6.00 and 1.30 - 2.35, respectively. Results indicate that this agricultural method allows for the maintenance of soil carbon and nitrogen.

ATP was high in the topsoil from horizons 1 to 4 (0.346, 0.125, 0.012, and 0.015 nmol∙g∙soil⁻¹, respectively) and the total thickness of those horizons was 54 cm (Table 2 and Table 3). At the CF, ATP was also high in the topsoil from horizons 1 to 3 (0.029, 0.052, and 0.017 nmol∙g∙soil⁻¹, respectively) and the thickness was 34 cm (Table 2, Table 3). The total ATP in SF was 54.2 mg∙m⁻², which was approximately five times greater than that of CF (Table 3). At the SF, 92% of ATP was present in horizons 1 and 2 and 65% was present only in horizon 1.

The T-C concentration at the SF was 24,135 g∙C∙m⁻², which was 6520 g higher than that at the CF over the entire soil profile (Table 2). At the SF, the T-C concentration in the top three soil horizons was 16,525 g∙C∙m⁻², which was 9418 g higher than that of the top three horizons at the CF (Table 2). The T-C concentration in the top three horizons at the SF was 11,769 g∙C∙m⁻² higher than the original concentration, assuming that the top three soil horizons of the SF were the same as the Bw1 horizon. In addition, the T-N content at the SF was 1422 g∙N∙m⁻², which was 345 g higher than that at the CF over the entire soil profile (Table 2). In the top three soil horizons, the T-N at the SF was 1047 g∙N∙m⁻², which was 461 g higher than that at the CF (Table 2). At the SF, the increase in T-N in the top three soil horizons was 722 g, assuming that the top three soil horizons of the SF were the same as the Bw1 horizon. The balance was as follows: the input of T-C for 15 crops in 4.5 years from the waste mushroom bed was 5014 g∙m⁻², and 856 g C was removed by harvested products. The input of T-N over the same period was 129 g∙m⁻², and 79 g N was removed. The net output/input (O:I) ratio of carbon was estimated to be 2.05 - 2.52 and that of nitrogen to be 4.20 - 6.62; the O:I ratios exclude harvested products were 1.88 - 2.35 and 3.58 - 6.00, respectively.

The productivity of the SF was four times that of the national average; nevertheless, organic farming rarely exceeds conventional farming in yield levels [18] . The NO₃-N concentration in soil was 5.6 mg∙kg⁻¹ soil, which is lower than the lower limit concentration in soil for conventional or general organic farming (20 mg NO₃-N kg⁻¹) [9] , [10] . The large root system and root zone are required for the absorption of a sufficient amount of nutrition under low nitrogen concentration conditions.

The defining characteristic of the soil A horizon is typically the value of the bulk density. The bulk density and C:N ratio of soil horizons 3 and below were similar. Differences in soil structure were mainly observed between

SF: study field. CF: control field. Horizon: refer to Table 1. Amount of free ATP was calculated by multiplying the concentration of each soil horizon by the soil weight. The total ATP in SF was approximately five times greater than that of CF.

horizons 1 and 2. In general, agricultural activities influence carbon stock to a depth of 0.3 m in the profile in savanna systems [19] ; we therefore focused on the top three horizons in the following discussion. At least 7 cm of soil A horizon was formed in 4.5 years. This is considered to be the result of an increase volume of pores be- cause the weights of soil A horizon at the SF and CF were almost the same (332 and 321 kg∙m⁻², respectively). The increase in the ratio of the thickness of the SF soil A horizon to the CF soil A horizon was 20.6%, while that of the weight was 3.4%. We calculated thickness by multiplying the soil weight of SF by the bulk density of CF, when the increase in T-C contributed by carbohydrates is subtracted. The calculated thickness was 8.4 cm smaller than present SF thickness. This volume of soil pores can absorb approximately 80 mm of rainfall. This excellent soil physical property makes farming without irrigation possible [20] . It is reasonable that no surface runoff occurred under a rainfall rate of 50 mm per hour, or crops could grow without irrigation under the 65-day drought.

The rate of formation of the soil A horizon described above is extraordinarily fast. The increase in soil was estimated to range from 9418 to 11,769 g∙C∙m⁻² in response to the application of 5014 g C by waste mushroom bed over 4.5 years. The average annual increase was 2093 - 2615 g∙C∙m⁻²∙year⁻¹. Soil carbon was shown to increase in natural forests by approximately 0.2 - 12.0 g∙C∙m⁻²∙year⁻¹ globally and by 2.3 - 2.5 g∙C∙m⁻²∙year⁻¹ in a tropical forest [3] . Conversion of cropland to fallow land is the most effective way to increase soil organic carbon; in a meta-analysis of 39 different tropical countries, soil organic carbon increased by 6.0 - 11.8 g∙C∙m⁻² in fallow lands within 7 years of conversion [6] . In the present study, the accumulation of soil carbon was drastically enhanced by the application of high C:N ratio organic material.

The increase in soil nitrogen was estimated to range from 461 to 772 g∙N∙m⁻² in response to the application of 129 g N by waste mushroom bed over 4.5 years. Including 79 g N that was removed from the system in the form of harvested material, the soil nitrogen increase was 91 - 160 g∙N∙m⁻²∙g∙year⁻¹. We topographically surveyed the top point of the field, which had no water or soil inflow from the surrounding area. Therefore, the increase in N is considered to be due to biological nitrogen fixation. This value was one order of magnitude higher than the average increase in soil nitrogen of legume crops (5 - 33 g∙N∙m⁻²∙year⁻¹) [21] . We found that inputting high C:N ratio organic material without additional nitrogen drastically increased the soil nitrogen content.

With respect to sustainability, the SF required an input of approximately 1000 g T-C m⁻²∙year⁻¹; however, the high productivity of SF provides considerable carbon to the soil. The current O:I ratio of carbon is greater than 1, meaning that the SF crops are sustainable without external carbon inputs. The current O:I ratio of nitrogen is also greater than 1.

As indicated by the results of this study, productivity increased four-fold, the thickness of the soil A horizon increase by 7 cm, and soil carbon and nitrogen accumulation were 2093 - 2615 and 91 - 160 g∙N∙m⁻²∙g∙year⁻¹, respectively, over the 4.5-year study period; these are rarely observed phenomena in commercial based tropical agriculture. A similar phenomenon was observed in a method using chopped forest biomass led to carbon accumulation (1160 g∙C∙m⁻²∙yr⁻¹) over a 1.5-year period following deforestation, but the effect was no longer detectable after 3 years [22] . The results were different in later years. SF was continually applied carbon, but the forest provided carbon only one time. We believe that a new concept is needed to accurately interpret the above results. For example, Hitoshi Mine, a farmer in Sao Paulo, has been by adding only organic material with a high C:N ratio (approximately 40) for 30 years in his vegetable fields; he adopted this method after observing forest ecosystems [23] . He aimed to provide carbon to microorganisms as their energy source; this is an example of indirect crop management via microorganisms. There are many types of biological nitrogen fixation systems [24] . In any case, if nitrogen fixation increases by one order of magnitude, an equivalent increase in microbial activity is required. Following the provision of carbon, the microbial activity of soil at the SF was one order of magnitude higher than that of soil at the CF. Biota is still a black box; even if we consider only fungi, the number of species globally is estimated to be 1.5 - 3 million, whereas human beings have only catalogued approximately 0.05 million species [25] [26] . Therefore, providing carbon to microorganisms as an energy source is considered a practical way to enhance the level of their activities such as nitrogen fixation. Further studies are needed to clarify the mechanisms responsible for the results observed in response to the application of organic material with a high C:N ratio.