Relationship between nitrogen and phosphate surplus from agricultural production and river water quality in two types of production structure

Abstract

We examined the relationship between nitrogen (N) and phosphate (P₂O₅) surplus derived from agriculture and river water quality. We selected two river basins; one was a paddy farming area (Omoigawa) and the other was an intensive livestock husbandry area (Nakagawa). Nitrogen and P₂O₅ surpluses, defined as the difference between their input and output on regional farmland, from farmland in Omoigawa were twice those of Nakagawa; the surpluses came mainly from chemical fertilizer use. Although N and P₂O₅ surpluses in Nakagawa were lower, Nakagawa had a large amount of non-utilized livestock excreta, twice the N surplus on farmland in Nakagawa. Residual N and P₂O₅ in the river basin, caused by surplus and non-utilized livestock excreta, were approximately 20% and 40%, respectively, higher in Nakagawa than in Omoigawa. Outflows of N and P₂O₅ to river water were higher in Omoigawa than Nakagawa. By excluding domestic sewage N and P₂O₅ in river water, we calculated the loads of non-point source N and P₂O₅. Non-point source N and P₂O₅ were higher in Omoigawa than Nakagawa. This inverse result might be caused by the different source of residual N (i.e. chemical N fertilizer or livestock excreta).

Key words:

• chemical fertilizer
• manure
• nitrogen
• phosphate
• river water

INTRODUCTION

Agricultural practices affect water quality significantly. In Sweden, the proportion of farmland in a basin was positively correlated with N in river water (Kyllmar et al. 2006), and the type of livestock also affected river water quality; that is, extensive cattle farming caused lower N concentrations than intensive pig farming. In Finland, the level of N surplus on farmland affected river N level (Salo and Turtola 2006). Woli et al. (2004) found a significant positive relationship between proportion of farmland and the concentration of N in river water in Hokkaido, Japan. Woli et al. (2004) defined the slope of the regression line as the impact factor and indicated that the impact factor had a significant relationship with N surplus on farmland, chemical fertilizer application and chemical fertilizer plus manure application. Hatano (2005) summarized the relationship between N load by agriculture and river water quality in Hokkaido, Japan.

Environmental damage by N is not limited only to surface water. Kumazawa (1999) summarized ground water quality and stated that the N concentration was higher in the area of upland fields and livestock husbandry, which cause high N surplus, than in paddy farming areas, which have low N surplus. Nishio (2003) calculated the effect of crops on ground water N concentration. The values indicated that excess organic and chemical fertilizer N in crop production causes N contamination of ground water.

Mishima et al. (2002) reviewed problems caused by P₂O₅ fertilizer by examining the trend in chemical P₂O₅ fertilizer use from 1966 to 1999, and then calculating the P₂O₅ balance in Japan in 1997. Although excessive P₂O₅ fertilizer use and accumulation in farmland basically did not cause damage to crops, excess damage was observed on Japanese radish, Garland chrysanthemum (Shungiku, leaf vegetable) and cucumber plants. Because of low crop response by P₂O₅ fertilization and low P₂O₅ fertility in Japanese farmland soil, large amounts of P₂O₅ had been applied and it peaked in mid 1980s. However, available P₂O₅ has increased and a large proportion of upland fields and facility farmland have excessive available P₂O₅ compared with the recommended upper limit of soil P₂O₅ fertility. The estimated P₂O₅ removed by crops from farmland was 16% of the applied P₂O₅, and outflow from farmland soil was only 0.6% of the applied P₂O₅, therefore, most P₂O₅ remained in the farmland soil. This accumulated P₂O₅ raises the risk of eutrophication of surface water by surface runoff, especially in the rainy and typhoon season, or may cause sudden outflow to a lower layer of farmland soil when P₂O₅ accumulated too much. In the Netherlands, the application of P₂O₅ in livestock excreta to farmlands is limited to less than 125 kg P₂O₅ ha⁻¹ to prevent water contamination (Tamminga and Wijnands 1991).

However, there have not been enough studies examining the relationship between river water quality and different types of agricultural production on a regional scale. In the present study, we selected a paddy farming area and a livestock husbandry area in Tochigi prefecture, central Japan, and examined the relationship between river water quality and nutrient surplus caused by agricultural production.

MATERIALS AND METHODS

Site description and data source

We selected the Omoigawa river basin (Omoigawa) and the upper Nakagawa river basin (Nakagawa) (Fig. 1) as study sites. The amount of farmland area in the river basins was approximately 17% at both sites; however, the structure of the agricultural production at each site differed. In Omoigawa, paddy rice fields occupied 70% of the total farmland, which is higher than the Japanese average (55%), and the rest was basically upland fields covered by vegetables (Table 1). In Nakagawa, although the amount of paddy rice fields was close to the Japanese average (58%) (Table 1), Nakagawa had threefold higher cattle density and twofold higher pig density than the Japanese average (Table 2) and second dominant farmland was forage field for cattle husbandry (24%) (Table 2). Therefore, we defined Omoigawa as a paddy farming area and Nakagawa as an intensive livestock farming area.

We determined the shape of the river basins using old district borders that were defined in 1950 because they constituted the smallest scale on which we could obtain publicly available statistical data. Data on the agricultural production and population of these areas were obtained from the Agricultural Census of 2000 (Ministry of Agriculture, Forestry and Fisheries 2001a) and Agricultural Statistics in Tochigi prefecture (Kanto Nosei Bureau Tochigi Statistics Office 2001).

Estimation of N and P₂O₅ flow

Figure 2 shows an N and P₂O₅ flow model. Chemical fertilizer and feed are taken into the river basin and products are taken out as crops and livestock production. Nitrogen and P₂O₅ surplus on farmland as a result of crop production and the disposal of livestock excreta to local land in the basin contribute to the contamination of river water. Sewage from homes and industry in the basin also contributes to the contamination. The model is divided into three parts (livestock husbandry, crop production and residents), and N and P₂O₅ flow associated with the three parts were estimated as follows.!

Livestock husbandry

Nitrogen and P₂O₅ in the feed consumed by livestock were estimated from the Cost of Livestock Production (Ministry of Agriculture, Forestry and Fisheries 1991, 2001b), which provided the consumption of formula feed and other feeds, such as roughage, for each kind of livestock. The Feed Nutrition Standard (Ministry of Agriculture, Forestry and Fisheries 1987) provides the N and P₂O₅ concentration in feed except for formula feed, which is provided by the Formula Feed Course Committee (1980) for each kind of livestock. Feed consumption by livestock was estimated from the head number of each kind of livestock, their consumption of each kind of feed and their N and P₂O₅ concentration. The amount of self-sufficiency roughage (forage, dent corn and sorghum) N and P₂O₅ were estimated from planted areas of roughage, roughage yield per area, and the N and P₂O₅ concentration in these feeds. The roughage produced was assumed to be consumed by cattle in the area because the amount produced was smaller than the amount consumed (Ministry of Agriculture, Forestry and Fisheries 2001b). Feed consumption minus self-sufficiency roughage was defined as purchased feed.

The amounts of N and P₂O₅ in livestock excreta were estimated from the head number of each species of livestock and the excretion of N and P₂O₅ per head (Harada 1999). In Omoigawa, cattle and pig excreta are separated into dung and urine; the dung is composted and the urine is stored before use for crop production or disposal to the environment. In Nakagawa, we used data from the Ministry of Agriculture, Forestry and Fisheries (2000) for the type of composting (type of co-material; dung and urine mixed or divided) of cattle, pig and poultry excreta. Part of the N in excreta volatilizes during composting and storage. The rate of N volatilization from dung and urine was estimated from the

Figure 1  Map of the study areas. Lines on the Tochigi Prefecture map indicate old municipal borders set in 1950. The shapes of the river basins were approximated using this border because this cell is the smallest statistical dataset we could find.

Table 1 Land use in the upper Nakagawa and Omoigawa river basins

	Nakagawa			Omoigawa			National average (%)
	Area (ha)	(Planted area [ha])	Occupation (%)	Area (ha)	(Planted area [ha])	Occupation (%)
Total area	123,090			92,800
Farmland area	23,273			19,030
Planted farmland	21,475	(21,539)	100	16,048	(17,646)	100	100
Paddy field	12,478	(12,921)	58	11,289	(12,887)	70	55
Upland field	3,768	(3,325)	18	4,122	(4,122)	26	25
Orchard	128	(128)	1	432	(432)	3	7
Forage field	5,101	(5,165)	24	205	(205)	1	13
Non-planted farmland	1,602			3,187

Table 2 Livestock husbandry in the upper Nakagawa and Omoigawa river basins

	Nakagawa		Omoigawa		National average
	No.	Density (head ha^−1)	No.	Density (head ha^−1)
Daily cattle	33,960	1.58	5,737	0.36	0.38
Beef cattle	23,969	1.12	14,808	0.92	0.58
Pig	87,673	4.08	37,654	2.35	2.00
Layer	706,250	32.89	1,665	0.10	38.67
Broiler (sold)	83,036	3.87	2,825	0.18	119.08

difference between the N : P₂O₅ ratio of raw excreta and that of the standard content of compost (Tochigi Prefecture 2002) and from the amount of volatilization indicated by Harada (1999). The amount of manure application was estimated from Tochigi Prefecture (1994, 1996, 1997), and the rest is disposed of as non-utilized livestock excreta to local landfill, or released to rivers after sewage treatment, to become the potential of environmental load.

Milk production was estimated from the head number of daily cattle over 2 years old and the milk production per head (Ministry of Agriculture, Forestry and Fisheries 2001c). Egg production was estimated from the head number of layers over 6 months old and egg production per head (Ministry of Agriculture, Forestry and Fisheries 2001c) and N and P₂O₅ concentration (Resources Council, Science and Technology Agency 2000). Beef cattle, pig and broiler production were estimated from feed consumption of each type of livestock minus dung and urine. These were summarized then recorded as livestock production.

Figure 2  Nitrogen and P2O5 flow model. Nitrogen and P2O5 flows were estimated using this model. A, disposal of livestock excreta; B, recycle pathway of forage production consumed by livestock and manure application to farmland.

Crop production

Farmland receives N and P₂O₅ as chemical fertilizer and manure. The amount and types of manure applied came from the Report of Soil Environmental Basics (Tochigi Prefecture 1994, 1996, 1997). Chemical fertilizer application for each crop type came from a database about the production environment prepared by the Ministry of Agriculture, Forestry and Fisheries (1998). Crop N and P₂O₅ production was estimated from the planted area (Ministry of Agriculture, Forestry and Fisheries 2001a), yield (Ministry of Agriculture, Forestry and Fisheries 2001b), water (Ministry of Education, Culture, Sports, Science and Technology (MECSST) 2000), and the N and P₂O₅ concentration (Owa 1996) of each type of crop. Natural N and P₂O₅ input by N₂-fixation (for example, 100 kg N ha⁻¹ for soybean) and irrigation (1.7 mg N kg⁻¹ and 0.08 mg P₂O₅ kg⁻¹, set as 1500 mm) came from Yatazawa (1978) and (Tochigi Prefecture (2002), respectively, and natural N output by denitrification was assumed to be 30% of the chemical fertilizer input into paddy fields (Sekiya 1987). Part of the crop by-product, such as rice straw, was removed and used in the upland fields, for example, as bedding material and roughage for livestock husbandry. The amount of N and P₂O₅ removal of crop by-product was estimated from the Report of Soil Environmental Basics (Tochigi Prefecture 1994, 1996, 1997) and the Feed Nutrition Standard (Ministry of Agriculture, Forestry and Fisheries 1987). The other crop by-products were assumed to have been plowed into the farmland.

Residents

The basic unit of N and P₂O₅ output per capita (12.0 g N day⁻¹, 3.28 g P₂O₅ day⁻¹) came from Kunimatsu and Muraoka (1989). These values include industry as well as domestic wastewater. The number of residents in Nakagawa was 205,125 and in Omoigawa the number was 532,502. The number of residents in the river basin multiplied by these values gives the sewage output N and P₂O₅ by resident. We assumed that the sewage is treated by sewage treatment plants and that the effluent is then discharged into the river. Sewage treatment removes 49% of N and 64% of P₂O₅ (Kunimatsu and Muraoka 1989) as sewage sludge.

Surplus on farmland and residual N and P₂O₅

The sum of the N and P₂O₅ inputs (chemical fertilizer, manure, N₂-fixation, irrigation) minus the sum of the outputs (crop, removal of crop by-products, denitrification) gives the N and P₂O₅ surplus on farmland. The surplus on farmland potentially and indirectly contributes to river water contamination by N and P₂O₅ as a non-point source.

Part of the livestock excreta was not used as manure and was disposed of into the environment (local land or river; Fig. 2, lines “A”). This non-utilized livestock excreta potentially and indirectly contributes to river water contamination by N and P₂O₅.

The surplus plus non-utilized livestock excreta are defined as the residual N and P₂O₅ that could potentially be a source of non-point contamination by crop and livestock production in the river basin.

River water quality and quantities of N and P₂O₅ in the river water

On Omoigawa, N and P₂O₅ flow was estimated from annual water flow in 2000 (973.89 × 10⁶ mg) and geometric average of N and P₂O₅ concentrations from 1989 to 1999 (result of the Public Water Quality Measurement: http://www-gis.nies.go.jp/intro/intro.html [in Japanese]). On Nakagawa, annual N and P₂O₅ flows from 1989 to 1998 were estimated from N and P₂O₅ concentrations in the river water and water flow in each year (River Association Japan 2001), weighted and then averaged.

RESULTS AND DISCUSSION

Nutrient balance in agricultural production

Livestock husbandry

From the point of view of N and P₂O₅ flow, livestock production in Nakagawa (1,870 Mg N, 941 Mg P₂O₅) was more than 2.5-fold the scale of that in Omoigawa (526 Mg N, 373 Mg P₂O₅) (Table 3). Although more feed was produced in Nakagawa (1,364 Mg N, 677 Mg P₂O₅) (Fig. 2; Line B) than in Omoigawa (88 Mg N, 30 Mg P₂O₅), more feed was also bought in Nakagawa (6,440 Mg N, 3,235 Mg P₂O₅) than Omoigawa (2,180 Mg N, 1,213 Mg P₂O₅) (Table 3). As a result, fourfold more dung N and 2.5-fold more urine N was excreted in Nakagawa (dung 2,888 Mg N, urine 3,046 Mg N) than Omoigawa (dung 749 Mg N, urine 775 Mg N). These excreta were processed into manure and stored urine, containing, respectively, approximately fourfold more manure N and 1.6-fold more stored urine N in Nakagawa (manure 2,249 Mg N, stored urine 2,858 Mg N) than Omoigawa (manure 543 Mg N, stored urine 461 Mg N) (Table 3).

In Nakagawa, a large amount of cattle manure was not utilized (420 Mg N, 469 Mg P₂O₅). Most cattle and pig urine and pig and poultry manure was also not utilized (Table 5). Although the rate of utilization of manure was higher (93% in N basis), the quantity of use was lower in Omoigawa (503 Mg N, 629 Mg P₂O₅) than in Nakagawa (1,281 Mg N, 1,431 Mg P₂O₅). In Omoigawa and Nakagawa, the non-utilized manure and urine pose a risk of nutrient leaching into the river

Table 3 Nutrient balance of livestock production in the upper Nakagawa river basin and the Omoigawa river basin

		Purchased feed		Self-stuffed feed		Products		Excreta				Manure		Stored urine		N
		N	P2O5	N	P2O5	N	P2O5	Dung N	Dung PP2O5	Urine N	Urine PP2O5	N	P2O5	N	P2O5	volatilization
Nakagawa	Total	6,440	3,235	1,364	677	1,870	941	2,888	2,740	3,046	232	2,249	2,858	715	114	2,395
	Cattle	4,100	1,790	1,364	677	1,107	543	2,176	1,874	2,180	49	1,701	1,901	499	23	1,582
	Pig	1,529	1,036			390	357	273	497	866	183	359	588	217	91	563
	Poultry	811	409			373	41	438	369			189	369			249
Omoigawa	Total	2,180	1,213	88	30	526	373	749	775	992	95	543	775	461	95	738
	Cattle	1,532	775	88	30	373	231	630	559	617	14	424	559	287	14	536
	Pig	645	437			152	142	118	215	375	80	118	215	174	80	201
	Poultry	2	1			1	0	1	0			0	0			1

after disposal to the farmland, although the N and P₂O₅ pathways to the river water are not known.

Crop production

Paddy fields received the most chemical fertilizer at both sites at the same level (Nakagawa 970 Mg N and 1,353 Mg P₂O₅, Omoigawa 949 Mg N and 1,320 Mg P₂O₅); however, chemical fertilizer application on upland fields in Omoigawa (673 Mg N, 1,045 Mg P₂O₅) was approximately twice that in Nakagawa (377 Mg N, 577 Mg P₂O₅). This difference resulted from the different coverage of upland fields and the different crop types. In Omoigawa, vegetables, which need a large amount of chemical fertilizer, were more dominant than in Nakagawa, where cereals were more dominant. Conversely, chemical fertilizer use for forage was higher in Nakagawa (285 Mg N and 317 Mg P₂O₅) than in Omoigawa (45 Mg N and 38 Mg P₂O₅). In total, chemical fertilizer application was higher in Omoigawa (1,817 Mg N and 2,659 Mg P₂O₅) than Nakagawa (1,686 Mg N and 2,340 Mg P₂O₅) (Table 4). This difference was a result of the difference in planted area (Table 1) and manure application in Nakagawa (1,281 Mg N and 1,431 Mg P₂O₅) was more than twice the chemical fertilizer application per planted area of upland field in Omoigawa (163 kg N ha⁻¹) and Nakagawa (113 kg N ha⁻¹). In Nakagawa, more than half of the manure (796 Mg N and 889 Mg P₂O₅) was used to produce forage. Paddy fields in Nakagawa also received relatively more manure (449 Mg N and 501 Mg P₂O₅) than those in Omoigawa. This might be because of the removal of a large amount of rice straw in Nakagawa (Table 4), used mainly for roughage and bedding material for livestock. Crop production in Nakagawa (2,011 Mg N and 989 Mg P₂O₅) was twice that in Omoigawa (945 Mg N and 440 Mg P₂O₅). Forage accounted for half of the crop production in Nakagawa. The production of other crops was the same in Nakagawa and Omoigawa.

Surplus N in Omoigawa (1,561 Mg N) was twice that in Nakagawa (785 Mg N) (Table 4). More than 60% of the N surplus came from upland fields in Omoigawa (962 Mg N) as a result of the larger inputs of chemical fertilizer and manure that are used for crop (mainly vegetable) production. The low N surplus in Nakagawa came from higher crop production (2,011 Mg N) and the removal of more by-products (328 Mg N) than Omoigawa (crop 945 Mg N, by-product 65 Mg N), although N input by chemical fertilizer and manure was higher in Nakagawa (2,967 Mg N) than in Omoigawa (2,320 Mg N) (Table 4). In contrast, P₂O₅ surplus at both sites was similar (Omoigawa 2,701 Mg P₂O₅, Nakagawa 2,663 Mg P₂O₅) (Table 4). This result came from the high input of manure in Nakagawa (1,431 Mg P₂O₅) and the low crop production (440 Mg P₂O₅) in Omoigawa.

In Nakagawa, forage fields received relatively large amounts of manure (796 Mg N, 889 Mg P₂O₅), which came mainly from cattle excreta among the crop fields and produced and supplied forage for cattle (Fig. 2, line B). This N cycle might reduce the N surplus in Nakagawa by reducing the need to import feed into the district.

Residual N and P₂O₅ in agricultural production

Surplus N and P₂O₅ on farmland (Nakagawa 785 Mg N, 2,663 Mg P₂O₅; Omoigawa 1,561 Mg N, 2,701 Mg P₂O₅) plus non-utilized livestock excreta N and P₂O₅ (Nakagawa 1,684 Mg N, 1,540 Mg P₂O₅; Omoigawa 532 Mg N, 249 Mg P₂O₅) (Table 5) were defined as residual N and P₂O₅ (Nakagawa 2,469 Mg N, 4,203 Mg P₂O₅; Omoigawa 2,093 Mg N, 2,951 Mg P₂O₅) (Table 6). Residual N and P₂O₅ can contaminate river water as non-point source contaminates through leaching and surface run off from farmland and from the other local land where dung and urine were disposed, or directly to surface water.

Sewage discharge from residents

The amount of N and P₂O₅ in sewage in Omoigawa (2,332 Mg N, 637 Mg P₂O₅) was greater than in Nakagawa (898 Mg N, 245 Mg P₂O₅) because of the smaller population in Nakagawa (Table 6). Part of the N and P₂O₅ were removed as sewage sludge in the sewage treatment process, and then discharged to the river (Omoigawa 1,190 Mg N, 229 Mg P₂O₅; Nakagawa 458 Mg N, 88 Mg P₂O₅) (Table 6).

Nitrogen and P₂O₅ in river water and their sources

Water flow and N and P₂O₅ concentrations in river water in Nakagawa were 1,184 × 10⁶ m³ year⁻¹, 1.591 mg N L⁻¹ and 0.096 mg P₂O₅ L⁻¹, respectively, and those in Omoigawa were 974 × 10⁶ m³ year⁻¹, 3.755 mg N L⁻¹ and 0.275 mg P₂O₅ L⁻¹. Annual flows were 1884 Mg N and 114 Mg P₂O₅ in Nakagawa and 3,657 Mg N and 268 Mg P₂O₅ in Omoigawa (Table 6). Although water flow was larger in Nakagawa than in Omoigawa, N and P₂O₅ concentrations were higher in Omoigawa than in Nakagawa. Therefore, the amounts of N and P₂O₅ in river water were approximately twofold and 2.5-fold higher in Omoigawa than in Nakagawa, respectively.

Nitrogen and P₂O₅ in river water minus sewage discharge could be thought of as non-point-source nutrients. Values of non-point-source N and P₂O₅ were positive in both basins, and the value was larger in Omoigawa (2,467 Mg N, 38 Mg P₂O₅) than in Nakagawa (1,426 Mg N, 26 Mg P₂O₅).

Jordan et al. (1997) indicated that one-third of anthropogenic N input leached to surface water. In Omoigawa, N surplus plus N from non-utilized livestock excreta (i.e. residual N caused by agricultural production) was approximately 15% smaller than non-point-source N in river water. This difference might be caused by different fertilizer use in the past and/or the different prevalence rate of the domestic sewage system. The method and place of sewage sludge disposal have the potential to contaminate river water (Nagumo et al. 2004). Although we knew the location of the large sewage disposal plant, we could not follow the fate of the residue using the available reports. Other N sources, such as industrial wastewater, might not affect river water N because Omoigawa did not appear to have very large industries. In contrast, residual N in Nakagawa was as twice N quantity as in river water in the year. This difference might arise from the different forms of residual N. In Omoigawa, most residual N was N surplus on farmland and surplus caused by the large use of chemical fertilizer N, especially on upland fields (Table 4). Most of the rest of the residual N, non-utilized livestock excreta N, was urine, which has high fertilizer efficiency, being easily mineralizable and leached. In contrast, most residual N in Nakagawa was non-utilized livestock N and manure was dominant (Table 6). Yanan et al. (1997) stated that farmyard manure application with chemical fertilizer reduced N leaching. Diez et al. (1997) also found less N leaching from organic fertilizer than from chemical fertilizer by fixation in the soil. Maeda et al. (2003) reported that leaching of N from chemical-fertilizer-treated farmland and pig-manure-treated farmland was the same in the fourth year of their experiment; however, the level of N in pig manure was twice that in chemical fertilizer. These reports suggest that N surplus caused by manure might load less N to the environment than chemical fertilizer derived surplus. Nielsen and Jensen (1990) indicated that although the application of manure affects the level of N leaching, crops were the main factor controlling N leaching. Paddy rice and forage were the main crops in Nakagawa, and paddy rice and upland crops in Omoigawa. This difference in crops might also affect the N load to river water. However, the risk of N load to the environment was higher in Nakagawa than in Omoigawa because of the higher surplus of agricultural production. Nishio (2003) reported that the time lag between N surplus on farmland and an increase of N in groundwater is 1–10+ years. However, the time lag from groundwater to river water is unknown. In Nakagawa, the numbers of dairy cattle, beef cattle and pigs were small in the 1960s, increased in the 1980s, and increased markedly in the 1990s, although layer and broiler numbers were reducing (Fig. 3) (Ministry of Agriculture, Forestry and Fisheries 1966, 1971, 1976, 1981, 1986, 1991, 1996, 2001d). Therefore, the risk of N load to the river might be increasing as a result of increases in cattle and pigs in recent years. On sandy soil, a larger N dose caused more N leaching and more N surplus nearly equal to N leaching, even if N was applied in organic form (Hoffmann and Johnsson 1999; Sexon et al. 1996). Although fine texture soil reduces N leaching, the risk of eutrophication of river water increases because N would be retained in the soil by its buffer effect. Therefore, even if the amount of N in riverwater in Nakagawa is smaller than Omoigawa, we should take note of the residual N and water quality.

Non-point source P₂O₅ in river water accounted for 0.6% of residual P₂O₅ in Nakagawa and 1.3% of residual P₂O₅ in Omoigawa (Table 6), although residual P₂O₅ per planted area in Nakagawa (196 kg P₂O₅ ha⁻¹) was similar to that in Omoigawa (184 kg P₂O₅ ha⁻¹). The reason why outflow from residual P₂O₅ was higher in Omoigawa than Nakagawa is unknown. Under estimation of sewage-derived P₂O₅ might have affected

Figure 3  Livestock numbers in the upper Nakagawa River basin. This graph was derived from agricultural census data from Tochigi prefecture. Intensive dairy and beef cattle production have increased in the past 20 years, and broiler production has declined. Pig production and layer breeding have increased rapidly in recent years.

this result. The low contribution of residual P₂O₅ to river water at both sites might be caused by P₂O₅ fixation by the soil. Takeuchi (1997) reviewed nutrient outflow from farmland and reported that even though many observations reported N outflow from farmland, cases of P₂O₅ were rare. Mishima et al. (2002) indicated that P₂O₅ outflow accounted for 0.6% of P₂O₅ applied to farmland soil in Japan. These results might indicate that surplus P₂O₅ is caught in farmland soil tightly and does not easily flow out to the environment. However, Hechrath et al. (1995) revealed that accumulated P₂O₅ suddenly started to flow out when the concentration reached a soil dependent specific level. Even if the P₂O₅ accumulation level is under the specific level, farmland soil with a higher concentration of P₂O₅ may cause higher environmental damage, such as eutrophication by surface runoff in the rainy season and during typhoons. Therefore, P₂O₅ accumulation raises the risk of environmental damage. Soil in more than half of the farmlands in both basins was Andisol, which has a high P₂O₅ fixation activity, so farmers might need a large amount of P₂O₅ fertilizer. However, conventional P₂O₅ fertilizer application sometimes exceeded the recommended level. For example, conventional chemical P₂O₅ fertilizer application for Japanese radish is 282.4 kg P₂O₅ ha⁻¹ (Ministry of Agriculture, Forestry and Fisheries 1998), although the recommended rate is 200 kg P₂O₅ ha⁻¹ (Tochigi Prefecture 2002). Compliance with recommended levels is needed to reduce the risk, even if obvious environmental impacts do not appear.

These results regarding the relationship between N and P₂O₅ in river water and in residues on agricultural production indicate that residual N, particularly as a result of chemical fertilizer, affects river water quality and residual P₂O₅ affects river water quality to a lesser extent compared with N.

Promotive and substitutional use of livestock excreta instead of chemical fertilizer or introducing sewage treatment for cattle and pig urine would help reduce the risk of damage to river water quality. The distribution of manure over broad areas might be another solution; however, manure is distributed solely within the production area (Biotic Waste Recycle Workshop 1999), and the basins we selected were large enough to not warrant bringing the manure out of the basin. Increasing feed production or limiting the number of livestock might be needed in Nakagawa.

ACKNOWLEDGMENT

We thank Dr Akira Hayakawa for his suggestion to start this work.