Animal density and the percent of livestock farms have

Transcription

1 TECHNICAL REPORTS: WASTE MANAGEMENT Evaluating Slurry Broadcasting and Injection to Ley for Phosphorus Losses and Fecal Microorganisms in Surface Runoff Jaana Uusi-Kämppä* MTT Agrifood Research Finland Helvi Heinonen-Tanski University of Kuopio The recent growth in the size of dairy cattle farms and the concentration of farms into smaller areas in Finland may increase local water pollution due to increased manure production and slurry application to grass. Therefore, a field study was conducted to monitor losses of total phosphorus (TP), dissolved reactive phosphorus (DRP), and fecal microorganisms in surface runoff from a perennial ley. Cattle slurry was added once a year in June (Study I) and biannually in June and October (Study II). The slurry was surface broadcast or injected into the clay soil. The field had a slope of 0.9 to 1.7%. Mineral fertilizer was applied on control plots. Biannual slurry broadcasting increased DRP (p < 0.001) and TP losses (p < 0.001) and numbers of fecal microorganisms in surface runoff waters. The highest losses of TP (2.7 kg ha 1 yr 1 ) and DRP (2.2 kg ha 1 yr 1 ) and the highest numbers of fecal coliforms (880 colony-forming units [CFU] per 100 ml) and somatic coliphages (2700 plaque-forming units [PFU] per 100 ml) were measured after broadcasting slurry to wet soil followed by rainfall in fall Injection reduced the TP and DRP losses in surface runoff by 79 and 86%, respectively, compared with broadcasting (17 Oct Oct. 1999). Corresponding numbers for fecal coliforms were 350 CFU (100 ml) 1 and for somatic coliphages were 110 PFU (100 ml) 1 in surface runoff after injection in October Slurry injection should be favored when spreading slurry amendments to grassland to avoid losses of P and fecal microorganisms in runoff to surface waters. Copyright 2008 by the American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Published in J. Environ. Qual. 37: (2008). doi: /jeq Received 13 Aug *Corresponding author (jaana.uusi-kamppa@mtt.fi). ASA, CSSA, SSSA 677 S. Segoe Rd., Madison, WI USA Animal density and the percent of livestock farms have increased in certain areas in recent decades, increasing problems with slurry management. The number of dairy farms declined by half, but the number of dairy cows declined by only one fifth between 1995 and 2004 (Information Centre of the Ministry of Agriculture and Forestry, 1995; 2005). On several enlarged Finnish dairy farms, fields are under silage grass production, and slurry is applied to grasslands after the first cutting in summer instead of earlier application methods, such as before plowing in autumn or before sowing in spring to cereal fields. Due to the large percentage of grass fields, cattle farms have difficulties finding enough fields for slurry applications. There are no previous reports of measurements in Finland of P losses from slurry injection vs. surface application to grass lands. Although manure and slurry are important sources of phosphorus (P) and nitrogen (N) in cereal and grass production, they can pollute surface waters significantly. The contribution of agricultural P to surface waters is estimated to be 2600 t yr 1, representing 63% of the total anthropogenic P load in Finland (Nyroos et al., 2006). For the past few decades, industry, municipalities, and fish farming have decreased their release of P to the environment, leaving agriculture as the largest single source of human-derived P in surface water. Although the P accumulation in Finnish soils is not as high as for certain North American or Western European areas, the situation is getting worse in intensive animal production areas. The new EU member states are likely to continue the same development enlargement of farms and concentration of cattle farms in certain areas and thus increase the pollution of the Baltic Sea. In Finland, there are several policy measures that have been taken, such as two national water protection target programs, the EU agri-environmental scheme, and the implementation of Nitrates Directive, all aimed at decreasing N and P losses from agricultural sources (Mitikka et al., 2005) (e.g., Finnish farms have storage capacity of slurry and manure for the 12 mo, but sometimes the spreading of slurry during growing season is impossible due to wet weather conditions). The maximum amount of fall-applied cattle slurry is 20 t ha 1, and surface application onto grass is allowed J. Uusi-Kämppä, MTT Agrifood Research Finland, Plant Production Research, E-talo, FI Jokioinen, Finland. H. Heinonen-Tanski, Univ. of Kuopio, PL 1627, FI Kuopio, Finland. Abbreviations: CFU, colony-forming units; DRP, dissolved reactive phosphorus; P Ac, concentration of ammonium acetate extractable P in soil; PFU, plaque-forming units; PP, particulate phosphorus; TP, total phosphorus. 2339

2 until 15 September by the Government Decree on the restriction of discharge of agricultural nitrates into water (931/2000; Finlex, 2000b). However, injection of slurry in the grassland is allowed in fall. The EU Agri-Environmental Support scheme is even tighter; it forbids manure application onto grass after 31 August unless slurry is injected or the grassland is plowed immediately after surface application. Many studies (e.g., Turtola and Yli-Halla, 1999; Smith et al., 2001; Withers et al., 2001; Kleinman et al., 2002) have shown that concentrations of dissolved reactive phosphorus (DRP) and total phosphorus (TP) may be extremely high in surface runoff from fields where fertilizer, manure, slurry, or other amendments are applied to the soil surface instead of incorporated into the soil. Incorporation of P amendments has effectively decreased runoff P losses from fields compared with surface P applications (Dunigan and Dick, 1980; Withers et al., 2001; Kleinman et al., 2002). Gessel et al. (2004) pointed that there are short-term and long-term effects of manure application on the quantity and quality of runoff. In their study, swine slurry applied and incorporated into the soil reduced runoff volume and sediment loss during the growing season compared with soil with no applied manure. There was also no difference in runoff TP load compared with plots without applied manure. However, the DRP loss increased in spring runoff as manure application rates increased. If manure and mineral fertilizer are continually applied at levels exceeding crop uptake, P accumulates on the soil surface to such an extent that the loss of P to surface runoff becomes a priority management concern (Sharpley, 1995). Sharpley et al. (1986) also showed that there is a significant linear relationship between the soil P content of the top 1 cm of surface soil and DRP concentrations in runoff from cropped and grassy watersheds. Kleinman et al. (2002) pointed out that runoff DRP concentrations correlated with the water-soluble P concentration in surface-applied manure. High P content in surface soil and high water-soluble concentrations of amendments may increase runoff DRP concentrations. Indeed, slurry applications to wet soil and after rainfall produce the highest risk of runoff P losses (Misselbrook et al., 1995). In studies from Europe, North America, and New Zealand, manure and slurry applications, especially in fall and winter, have been shown to cause large losses of DRP and TP (Niinioja, 1993; Turtola and Kemppainen, 1998; Smith et al., 2001; Withers et al., 2001) and fecal microbes (Nunez-Delgado et al., 2002; Shima et al., 2002; Tyrrel and Quinton, 2003) to downstream waters. Slurry application, to frozen soils in particular, was found to increase nutrient runoff to waters after snow melt in Finland (Niinioja, 1993; Turtola and Kemppainen, 1998). The practice is now forbidden in Finland by a Government Decree (931/2000; Finlex, 2000b). Turtola and Kemppainen (1998) reported the cumulative load of TP was 15 and 53 kg ha 1 after fall and winter spreading of cattle slurry to perennial grassland on a fine sand over a 2-yr period. In addition to nutrients, animal fecal wastes contain high numbers of fecal microorganisms. Some of them are useful for degrading fecal components; however, some can cause deterioration of the grass or silage, and some can be pathogenic. Due to the fertilization effect, manure or sewage continues to be generally used, but silage deterioration must be avoided (Heinonen-Tanski et al., 1998). There are reports of spreading problems with pathogens and epidemics caused by enteric microorganisms common to cattle and humans, such as the bacteria Salmonella (Placha et al., 2001), Mycobacterium bovis (Scanlon and Quinn, 2000), and hemorrhagic Escherichia coli O157:H7 (McGee et al., 2001) and protozoa such as Cryptosporidium (Bodley-Tickell et al., 2002). The fecal coliforms, sulfite-reducing clostridia and enterococci are used as general bacterial indicators of fecal contamination (American Public Health Association, 2005) and warnings about the risks for human and animal health. The bacterial viruses, coliphages, and especially RNA-coliphages survive better than enteric bacteria in different inhibiting environments (Koivunen and Heinonen-Tanski, 2005). If coliphages survive in runoff waters, then enteric cattle viruses would also survive, and if pathogenic viruses are present, their spread would be possible. This paper reports data on the transfer of DRP, TP, and fecal microorganisms in surface and near-surface runoff (0 30 cm) during a 5-yr study, where cattle slurry was applied by surface broadcasting or injection into a perennial ley only in June after grass cutting ( , Study I) and in June and October after grass cuttings ( , Study II). Subsequently, we monitored the residual effect of slurry application on easily soluble P concentrations in soil and surface runoff from barley fields in 2001 and from pasture in (Study III). Mineral fertilizer was surface applied onto the control plots in spring and summer. The amounts of easily soluble P in different soil layers were analyzed in samples taken before fertilization or slurry application in spring and fall. Phosphorus in the harvested yield of grass ( ) and barley (2001) was measured and used in calculations of nutrient balances. Materials and Methods The Experimental Field The surface runoff experimental field (0.34 ha) was located at MTT Agrifood Research Finland, at Jokioinen, southwestern Finland, and consisted of eight plots (6 70 m), each hydrologically isolated with plastic film to a depth of 0.6 m (Fig. 1). The soil was classified as a Typic Cryaquept (Soil Survey Staff, 1996) containing 61% clay (particle size <0.002 mm) in the plow layer. The plots had a fairly even slope (<0.9%), with a short steeper slope ( %) at the lower end. In the uppermost soil layer (0 0.1 m), the initial concentration of ammonium acetate extractable P (P Ac ) ranged from 8.3 to 12 mg L 1, corresponding to a satisfactory P status for Finnish cultivated soils (Viljavuuspalvelu, 2000). Surface and near-surface runoff (referred to hereafter as surface runoff) to a depth of 30 cm was collected in a modified collector trench planned by Puustinen (1994) at the lower end of each plot and fed by pipes into plastic tanks (2.0 m 3 ) 2340 Journal of Environmental Quality Volume 37 November December 2008

3 buried into the soil. The water collected was protected against further contamination caused by vegetation and animals. Water volume was measured by flow meters (Oy Tekno-Monta Ab; JOT-company, 1992) when the tanks were being emptied, and representative subsamples were taken for laboratory analyses. The volume of drainage water could not be measured, although the field area had been subsurface drained. Initially, the slurry used in the experiment was from bulls raised on an experimental farm of MTT and fed with barleybased concentrate and grass silage. Under the barn, there was a collecting tank from which the slurry was occasionally pumped to outdoor slurry tanks. The slurry was mixed before transfer from the collecting tank to the outdoor slurry tanks and before applications to the field. Since October 1999, the slurry used was from a nearby commercial dairy farm, where the slurry flowed directly from a gutter to outdoor slurry tanks. The slurry was mixed before each application. On both of the farms, the slurry tanks were emptied twice a year, in spring and in autumn. The analyses of slurries are presented in Table 1. Fig. 1. Schematic diagram of the experimental field. Slurry and mineral fertilizer was applied to a central area (shown in gray) of each plot: 3 50 m in Study I (annual slurry application) and 5 50 m in Study II (biannual slurry application). Surface runoff water was collected from the whole plot area (6 70 m) and fed by pipes into plastic tanks buried in the soil. A 10-m-wide buffer zone at the lower edge was left without fertilization under permanent grass cover between the grass field and water collectors. The field was surrounded by an open ditch. The arrow shows the direction of the slope on the field. (C, control; I, injection; B, broadcasting). Experimental Design A mixture of timothy (Phleum pratense) and meadow fescue (Festuca pratensis) was grown on the plots. After cutting the grass, slurry was applied annually in June ( ) in Study I and in June and October ( ) in Study II. Annual application of slurry after the first grass cut in summer is the typical method on Finnish farms. In Study II, a further autumn application of slurry was examined, which farms may sometimes do to empty the slurry tanks before winter (e.g., after a rainy summer season). During Study III (which concerns residual effects of slurry), the grass field was plowed on 26 Oct. 2000, and barley with hay was sown the following spring. Residual effects of the slurry applications on barley production in 2001 and plant-available P in soil and surface runoff from barley (2001) and pasture ( ) were measured in Study III. Phosphorus inputs for different studies are presented in Table 2. A 10-m-wide vegetated buffer zone at the bottom of the slope, which was not manured or fertilized, was left to prevent surface runoff of P and fecal microbes from the slurry amended fields to surface water. The experimental treatments were as follows: (1) Surface broadcasting of cattle slurry onto the ley (three replicates), (2) shallow injection of cattle slurry into the ley soil (depth of m; three replicates), and (3) control NPK fertilization onto the ley (two replicates). In Study I, cattle slurry was applied after the first cut in June by a Vogelsang spreader, which had 10 tines with 0.3 m spacing (Kapuinen, 1998), with a working width of 3.0 m (amended area 3 50 m). Surface broadcasting was performed with a band spreading unit equipped with a small splash plate under the hose. On the injection plots, slurry was shallow injected to a depth of 0.06 to 0.1 m. During Study II, slurry broadcasting was performed by a Teho-Lotina spreader with 0.47 m tine spacing and disc coulters holding the applicator up and equipped with a small splash plate under the each drill. Slurry injection was performed using the same applicator, the injection depth being 0.05 to 0.1 m. When the working width of the spreader was 2.5 m, the slurry-applied area was 5 m wide with two applications (amended area 5 50 m). In Study III, three heifers under 2 yr old (5.8 au ha 1 ) were grazed on the experimental field area where slurry had been earlier applied and on four buffer strips below the field for 11 d in summer 2002 and 40 d in summer 2003 and summer Water Analyses The samples for microbial analyses were taken from fresh runoff water, which was usually stored for 1 to 3 d in the tanks. The samples were collected in new, sterile, 1000-mL polyethylene bottles. The water samples were sent in cool Table 1. Concentrations of TN, NH 4 N, TP, K, Ca, and Mg, dry matter (DM) content and ph in the cattle slurries. Date TN NH 4 N TP K Ca Mg DM ph g kg 1 (wet wt.) % Annual slurry application (Study I) 14 May NA NA NA 8.0 NA June NA NA NA 6.9 NA Biannual slurry application (Study II) 29 June NA Oct NA 30 June Oct June Oct NA NA NA Beef cattle slurry was applied 14 May June 1999, and dairy cattle slurry was applied 27 Oct Oct NA, not available. Uusi-Kämppä and Heinonen-Tanski: P & Microbe Runoff from Grass 2341

4 Table 2. Application dates, total phosphorus (P) applications in slurry (s) and mineral fertilizer (m), the amended plot area and amount of slurry used during experimental years Dates Area Amount of slurry Broadcasting Injection Control m 2 t ha 1 (wet wt.) P, kg ha 1 Annual slurry application (Study I) June s 36s 23 m June s 38s 16 m Annual P application Biannual slurry application (Study II) 29 June s 44s 18 m 16 Oct s 35s 11 May m 30 June s 51s 20 m 27 Oct s 20s 8 May m June s 29s 20 m 23 Oct s 28s Annual P application Residual effects of slurry (Study III) 11 May m 10 m 10 m 6 May m 6 m 6 m Annual P application ( ) Estimated mean annual P applications to the whole plot area. The P from the dung of grazing cattle was not estimated. boxes to the laboratory within 20 h and thus were analyzed 1 d after sampling. The storage tanks were washed with cold tap water once a year (in summer). Microbial content of runoff water samples was determined by measuring levels of somatic and RNA (MS2) coliphages (E. coli ATCC and as hosts) according to the single-layer method of Grabow and Coubrough (1986) modified by Rajala- Mustonen and Heinonen-Tanski (1992) corresponding to American standard methods (2005) except that RNase was not confirmed for RNA coliphages. Total coliforms were cultivated on m-endo agarles (Difco ; SFS 3016, 2001), fecal coliforms on mfc agar (Difco, ; SFS 4088, 2001) and confirmed, since the fall of 1998, by an oxidase test. Enterococci were cultivated on KF streptococcus agar (Oxoid CM701) and confirmed with 3% H 2 O 2 and on bile-aesculin-azid agar (Difco ; SFS 3014, 1983). The Finnish SFS standards follow the corresponding American methods (American Public Health Association, 2005). Sulfite-reducing clostridia were determined according to the European Norm on a self-made media (SFS- EN , 1993) and incubated in an Oxoid anaerobic jar. For counting geometric means or log-transformed numbers for statistical significances (Tukey test), half of the detection limit of 0.5 colony-forming units (CFU) or plaque-forming units (PFU) per 100 ml was used when the count was 0 per 100 ml. Finnish and American standards are based on the analysis volume of 100 ml, and therefore this volume was used. Representative water samples for nutrient analyses were collected in 500-mL polyethylene bottles when the collection tanks were emptied. The nutrient samples were taken more often than the samples for microbial analyses. The time interval between water sampling during peak runoff periods varied between a few hours and 2 wk, depending on rains. The water samples were stored in the dark (4 C) from a few days to several weeks before nutrient determination. The storage time did not have a significant impact on the phosphorus concentrations (Turtola, 1989). Water sample analyses included measuring concentrations of total solids, dissolved reactive phosphorus (DRP), and TP. Total solids were measured to estimate erosion losses from the field. Dissolved reactive phosphorus was determined by a molybdate blue method, using ascorbic acid as the reducing agent (Murphy and Riley, 1962) after filtration (nuclepore polycarbonate; pore size 0.2 μm). Total phosphorus was analyzed by the same method after peroxodisulfate digestion. Dissolved reactive phosphorus and TP were determined using the FIAstar autoanalyzer according to the Finnish standards SFS 3025 (1986) and SFS 3026 (1986), respectively. Particulate P (PP) was calculated as the difference between TP and DRP. For statistical analyses, log-transformation of the DRP and PP values were used. The data from the three studies were analyzed together using a mixed model whereby study, treatment, and their interactions were used as fixed effects, whereas block, block treatment, and block study were used as random effects. Each block included two or three adjacent plots with different treatments. Analyses were performed using SAS/MIXED procedure. Soil and Plant Analyses Initially, soil was sampled in each plot to depths of 0 to 10, 10 to 20, and 40 to 60 cm with a drill. Since the spring of 1997, surface soil was sampled from depths of 0 to 2, 2 to 5, and 5 to 10 cm (four cores for each depth level were pooled). Surface samples were dug up using a shovel, and soil layers were separated using a trowel. Initially, in plots with slurry injection, soil samples were taken randomly in the application area. In October 1999, 2342 Journal of Environmental Quality Volume 37 November December 2008

5 Table 3. Monthly precipitation and mean of surface runoff (number of sampling dates in parentheses), and mean losses of total phosphorus (TP), dissolved reactive phosphorus (DRP), and particle-bound phosphorus (PP) from plots where slurry was broadcast (B) or injected (I) into soil and mineral fertilized plots. Surface runoff TP DRP PP Treatment Precipitation B I C B I C B I C B I C mm kg ha 1 Annual slurry application (Study I) 1 Jan. 18 June (12) June 31 Dec (7) Jan. 27 June (14) June 31 Dec (9) Total Biannual slurry application (Study II) 1 Jan. 29 June (17) June 16 Oct (4) Oct. 31 Dec (8) Jan. 30 June (14) July 27 Oct (0) Oct. 31 Dec (3) Jan. 22 June (13) June 20 Oct (1) Total Residual effects of slurry (Study III) 21 Oct. 31 Dec (9) Jan. 8 May (6) May 16 Sept (1) Sept. 31 Dec (3) Total C, control. April 2000, and October 2000, samples were taken separately from the slits where slurry was injected and from the area between the slits. Cameron et al. (1996) have shown that the injected dairy pond sludge is confined to a relatively narrow zone close to the point of injection. Thus, when concentrations of plant-available phosphorus (P Ac ) of the soil were calculated in our study, 20% of the application area was estimated to represent the slit area where slurry was concentrated, whereas the remaining 80% was assumed to represent the lessconcentrated area between the slits. Plant-available phosphorus was determined by extracting dried, ground, and sieved soil samples with acid ammonium acetate at ph 4.65 (Vuorinen and Mäkitie, 1955). Total phosphorus content of plants was analyzed from digested samples by inductively coupled plasma emission spectroscopy (Huang and Schulte, 1985). Plant-available phosphorus was analyzed statistically using a mixed model, whereby depth, treatment, and their interactions were used as fixed effects and block, block treatment, and block depth were used as random effects. Analyses were performed using SAS/MIXED procedure. Results Phosphorus in Runoff Annual Slurry Application (Study I, ) Beef cattle slurry was applied to only a small area (3 50 m) after the first grass cuts in June 1996 and Over the 2-yr study, cumulative surface runoff was 120 to 140 mm, the cumulative losses of TP and DRP were small ( kg ha 1 and 0.3 kg ha 1, respectively), and the differences among treatments were nonsignificant (Table 3). Most of the runoff and P losses occurred in spring 1996 before slurry applications. The TP concentration was generally below 1.0 mg L 1 (Fig. 2). In all treatments, over half of TP (53 66%) was bound to soil particles. Biannual Slurry Application (Study II, ) Annual P losses in surface runoff from plots where slurry was biannually broadcast over 3 yr were clearly greater than previously (Table 3). On the plots where slurry was broadcast, the cumulative losses of TP and DRP were 3.7 kg ha 1 and 2.8 kg ha 1, respectively (Table 3). The TP and DRP losses were significantly greater (p < 0.001) from broadcast plots compared with injection and control plots. Injection decreased the loads of TP and DRP by 65 and 75%, respectively. The loads of TP and DRP from the plots where slurry was injected were similar to the control (Table 3). In contrast to Study I, most of the P was now lost as DRP, representing 75, 55, and 72% of TP in the runoff water from surface broadcasting, injection, and control plots, respectively. The PP load was also significantly greater (p < 0.01) from the broadcast plots compared with other plots. In June 1998, there were heavy rains after the grass harvest (58 mm during a period of 5 d), but the slurry had not yet been applied. The runoff varied from 3 to 8 mm from the plots applied with slurry in the previous summer, whereas the mean cumulative load of TP over the first 5 d was negligible at 0.02 kg ha 1. The concentration of TP in surface runoff was low (Fig. 2). Indeed, in October 1998, slurry amendments (32 35 kg P ha 1 ) were made to wet soil and were followed by Uusi-Kämppä and Heinonen-Tanski: P & Microbe Runoff from Grass 2343

6 Fig. 2. Surface runoff total phosphorus (TP) concentrations after slurry application from 1996 to Slurry applications are marked by arrows. frequent rainfall events (38 mm during a period of 5 d). During these 5 d, the surface runoff from the field was from 8 to 9 mm. The runoff was twice the amount it had been in June 1998, and the cumulative load of TP was now 18- to 24-fold greater for plots treated by surface broadcasting of slurry than it had been in June. In a 5-d period in October 1998, incidental TP and DRP transfers were equivalent to 0.9 to 1.2 kg ha 1 and 0.6 to 0.9 kg ha 1, respectively, from the plots where slurry was broadcast. Injection decreased TP and DRP losses by 88 and 92%, respectively. The mean incidental TP concentrations in runoff water were also high during the first 3 d after the slurry application (16 mg L 1 from broadcasting and 2.2 mg L 1 from injection; Fig. 2), most of the TP being in the soluble form. The control, fertilized in June, peaked at only 0.8 mg L 1. During the following 6 mo, the mean runoff DRP and TP losses were 2.2 and 2.7 kg ha 1, respectively, from the plots treated by surface broadcasting (Fig. 3). Most of the TP and DRP leached out just a few days after application and during snow melt the following April. Injection of slurry into the ground decreased the loads of DRP and TP by an average of 86 and 79%, respectively, in the fall of 1998 and the following spring. During May through October 1999, the surface runoff and phosphorus losses were almost negligible. Fig. 3. Cumulative surface runoff dissolved reactive phosphorus (DRP) loss from grass with slurry broadcasting, slurry injection, and mineral fertilization from June 1998 to July Slurry applications in June 1998 (50 t ha 1 ), October 1998 (40 t ha 1 ), and June 1999 (60 t ha 1 ) are marked by arrows. The error bar shows SD. Residual Effects (Study III, ) After plowing the field on 26 Oct. 2000, the erosion, TP, and PP losses were greater for all treatments in surface runoff; PP represented approximately 90% of TP (Table 3). The concentrations of total solids were normally below 0.5 g L 1, whereas after plowing the values varied between 1.0 g L 1 and 3.0 g L 1 (data not shown). Because runoff volume was low (20 30 mm) in spring 2001, erosion was only 60 to 90 kg ha 1 in this period. During the first year after plowing, there were significant differences in DRP losses (p = 0.01) among all treatments and tendency in TP losses (p = 0.07) between broadcasting and injection. The loads of DRP and TP in surface runoff from the whole experimental field under barley and grass were small, being 0.03 kg ha 1 and 0.1 kg ha 1, respectively, from May 2001 to December 2001 (Table 3). Mean surface runoff (180 mm), TP (2.0 kg ha 1 ), and DRP (1.3 kg ha 1 ) from the pasture over the 3-yr study period was half of that from the silage grass with biannual slurry broadcasting in Study II. If the runoff from the pasture had been as high as in Study II, the DRP and TP losses might have been higher and possibly even the same as from the surface broadcasting plots. Microbial Content of Waters At the beginning of Study II, the numbers of coliphages, sulfite-reducing clostridia, and fecal coliforms were low, often less than the detection limit (1 CFU or PFU 100 ml 1 ), at least in some of the replicates. Similarly, the numbers of total coliforms and enterococci were low (<500 CFU 100 ml 1 ). In two cases, the runoff waters collected from plots with surface broadcasting slurry contained more total coliforms and enterococci than the control plots or the plots injected with slurry (Table 4). On 21 June 1998, the situation changed when water samples were collected 360 d after the last slurry application and after many days of heavy rain. A clear microbial contamination of the abundant runoff water was found. The same situation was found again in the fall of 1998 after the new slurry applications (Table 4). In all these cases, slurry broadcasting caused the most serious microbial contamination. Generally, if the slurry application was done in the fall, the enteric microbial numbers of runoff waters were high, especially if the slurry had been broadcast, but also if the slurry had been injected (Table 4). The runoff waters in spring were also still polluted, although less than in the fall. During grazing, the levels of fecal coliforms and enterococci were as high as during slurry applications. In November 2003, the levels of fecal coliforms and enterococci were significantly lower (p < 0.001) in runoff from the pasture with 10-m-wide nongrazed buffer zones than from the pasture with grazed buffers (Table 4). Phosphorus in Soil Annual Slurry Application (Study I, ) At the start in May 1996, the mean P Ac (9.5 mg L 1 ) in the soil surface (0 10 cm) corresponded to a satisfactory phos Journal of Environmental Quality Volume 37 November December 2008

7 Table 4. The geometric means of the numbers of somatic coliphages (SCP), RNA-coliphages (RCP), sulfite-reducing clostridia (SRC), fecal coliforms (FC), total coliforms (TC), and enterococci (EC) in surface runoff waters. Sampling date (+ number of days since the last slurry application) Treatment SCP RCP SRC FC TC EC PFU per 100 ml CFU per 100 ml Biannual slurry application (Study II) 11 May 1998, +319 broadcasting <1 <1 < * 270 injection <1 <1 < control <1 <1 < June 1998, after very heavy rains, +360 broadcasting injection ,000 10,000 10,000 control ,000 1 Sept. 1998, +64 broadcasting 1.1 <1 < injection 0.6 <1 < control <1 <1 < Oct. 1998, after very rainy days, +5 broadcasting 2700*** 130*** 1500*** , injection 110*** 1 69* control 0.7 <1 < Apr. 1999, after winter, +180 broadcasting 2.5 <1 < injection 22*** 0.9 < control <1 <1 < Dec. 1999, +35 broadcasting injection 100** 74** * control 1.5 < Oct. 2000, +8 broadcasting 4500** 210** 130*** *** 10,000*** injection 3000** 120** ,000*** control < Grazing (Study III) 5 Nov. 2003, 2 mo after grazing grazed BS <1 <1 <1 250*** NA 70*** nongrazed BS <1 <1 <1 33 NA July 2004, during grazing grazed BS NA NA <1 11,000 NA 6400 nongrazed BS NA NA < NA Sept. 2004, 11 d after grazing grazed BS 9 12 <1 30,000 NA 22,000 nongrazed BS <1 16,000 NA 5900 * Statistical significance from control at p < ** Statistical significance from control at p < *** Statistical significance from control at p < The statistical significance found between plots where slurry was broadcast or injected. BS, buffer strip; NA, not available. In Study III, statistical significance between the grazed and non-grazed BS. phorus status for cultivated soil (Viljavuuspalvelu, 2000). In May 1997, 11 mo after the first slurry application, there were no clear differences in P Ac between treatments (p = 0.11). Biannual Slurry Application (Study II, ) During Study II, the concentration of P Ac seemed to increase in the 2-cm-layer of surface soil in the slurry broadcasting plots. A significant interaction of treatment by soil depth occurred for P Ac (p < 0.01; Fig. 4) at each sampling. In September 1998, 11 wk after the third slurry application, mean P Ac at a depth of 0 to 2 cm was found to be highest in the slurry broadcasting plots (20.6 mg L 1 ; p < 0.001). Plantavailable phosphorus was also slightly elevated (16.4 mg L 1 ; p < 0.01) in the injection plots at a depth of 5 to 10 cm. This time there were no differences in the other depths. In October 1999, the highest P Ac, 65 mg L 1 (p < 0.001; Fig. 4), corresponding to an excessive P status (Viljavuuspalvelu, 2000), was measured from 2-cm-thick soil surface layers in the plots with slurry broadcasting after a very dry summer and 16 wk after the fifth slurry application. Differences in P Ac were also observed in the depths of 5 to 10 and 20 to 40 cm (p < 0.01). In April 2000 and October 2000, there were differences in the P Ac in the 0- to 2-cm (p < 0.001), the 5- to 10-cm (p < 0.01), and the 10- to 20-cm (p = 0.04) soil depths. There seemed to be clear differences in P Ac between the three injection plots. The reason was the uneven distribution of the slurry by the drills. The P Ac seemed always to be higher in injection slits, at depths of 5 to 10 and 10 to 20 cm, compared with the area between slits. The P Ac of mineral fertilized grass plots did not change, partly due to an even, lower P input. Only in the last samples (October 2000) P Ac values of the control were relatively high, at 20.5 mg L 1 (Fig. 4). Residual Effects (Study III, ) After plowing the field, there were no statistically significant differences in P Ac between treatments at a depth of 0 to 20 cm (p = 0.29). In May 2001, the mean values were 7.2 to 8.5 mg L 1, being slightly higher in injection plots. At a depth of 40 to 60 cm, the P Ac was low (<1 mg L 1 ), corresponding to a rather poor P status. During grazing, the mean P Ac of surface soil (0 2 cm) on broadcasting, injection, and control plots decreased slightly (8.3, 9.7, and 6.6 mg L 1, respectively) at the end of the 3-yr pasture period. Uusi-Kämppä and Heinonen-Tanski: P & Microbe Runoff from Grass 2345

8 Fig. 4. The concentration of easily available phosphorus (P Ac ) in different soil layers when slurry was broadcast (n = 3) or injected (n = 3) or when mineral fertilizer (n = 2) was used. Sampling dates were 15 May 1997 (46 wk after the first application), 15 Sept (11 wk after the third application), 20 Oct (16 wk after the fifth application), 26 Apr (26 wk after the sixth application), and 19 Oct (17 wk after the seventh application). Table 5. Total phosphorus balance during the three studies. TP balance Broadcasting Injection Control kg ha 1 Annual slurry application (Study I) 1996 output input output balance output input output balance balance Biannual slurry application (Study II) 1998 output input output input balance input output input input input balance input output input output input balance balance Residual effects of slurry (Study III) 2001 input output balance balance Output, harvested P in grass/cereal yields; inputs, added P in mineral fertilizer/slurry. Phosphorus Uptake by Plants The P uptake of grass was the highest (29 40 kg ha 1 ) in the first experimental year (Table 5). Subsequently, P uptake decreased, being lowest (11 13 kg P ha 1 ) in the dry year 1999, when only one grass yield could be harvested. The P uptake was normally the highest in the mineral fertilized plots, without any meaningful differences between the slurry application treatments. In the first residual year (2001), the P uptake of barley was slightly higher in the plots with injected slurry (27 kg P ha 1 ) than in the plots with surface broadcasting or controls (24 or 23 kg P ha 1 ), reflecting differences in the yield of above-ground biomass. Phosphorus Balances Phosphorus balances are presented in Table 5. During the five grass years, the P surplus in soil was 161 and 182 kg ha 1 in slurry broadcasting and injection plots, respectively, whereas it was only 16 kg ha 1 in the control. In Study I, the field P balance was negative ( 26 kg ha 1 ) in the control plots, whereas the P surplus was 18 and 25 kg ha 1 in the plots where slurry was broadcast and injected, respectively. Discussion Total P and DRP losses to surface runoff water were the greatest from the grassland with surface broadcasting of slurry. Injection below the soil surface reduced the TP load to the same level as in the control (mineral fertilizer was broadcast onto the soil surface), although the amount of fertilizer P applied to control plots was only half of that added in slurry. Kleinman et al. (2002) also reported that mixing of manure and slurry into the soil significantly decreased P losses relative to surface application, such that DRP losses from amended, mixed soils were not significantly different from those of unamended soil. Turtola and Kemppainen (1998) found that TP and DRP concentrations and losses in runoff did not increase after an autumn slurry application if plowing was done immediately, whereas the losses were drastically increased after fall and winter slurry application without plowing. There were a few reasons for the greater TP and DRP losses in Study II compared with Study I. Owing to the larger application area (5 50 m) in Study II than in Study I (3 50 m), and two slurry applications rather than the single application of Study I, more slurry and phosphorus (29 33 kg P ha 1 ) was annually added to the slurry plots in The other reason for the 2346 Journal of Environmental Quality Volume 37 November December 2008

9 greater P losses in Study II was the better uptake of P by grass in summer compared with the P uptake from the slurry applied in October. The incidental losses were also great in fall Concentrations of eroded material, TP, and PP in surface runoff increased after plowing in fall Presumably the erosion and PP loads were somewhat reduced by the 10-mwide buffer strips and by the late plowing. Incidental Losses Extremely high surface runoff TP and DRP losses and high values of fecal microbes were observed in October 1998 after the broadcasting of slurry onto wet soil followed by heavy rains. The respective losses from controls were small because no P fertilizer had been applied after June. Injection of slurry into soil greatly decreased TP and DRP losses and fecal microorganisms in surface runoff waters. In spite of high surface runoff, measured in June 1998, the runoff TP and DRP losses were negligible because P sources were not yet applied. These results are similar to those of Preedy et al. (2001), who stated that incidental losses of TP and DRP can occur when slurry applications are coincident with the onset of rainfall. They highlighted the risk associated with incidental P transfers and proposed that short-term decision-making was perhaps the most immediately viable method for mitigating P loss. In our study, TP and DRP losses peaked soon after slurry application in October 1998 and thereafter during snow melting the following spring. Results show that when rainfall immediately follows application of a P source, the potential for P loss peaks and then declines over time, verifying the results of Gascho et al. (1998) and Sharpley and Syers (1979). In the dry summer and fall 1999, the TP and DRP losses were low after the autumn application. Less P was also applied in 1999 because the dairy cattle slurry used contained only 0.5 g P kg 1, whereas the beef cattle slurry used earlier contained 0.8 g P kg 1. The phosphorus application rate in manure is known to affect the concentrations of DRP and TP in runoff from soils with manure broadcasting (Kleinman and Sharpley, 2003). Readily Available Phosphorus in Soil and Its Effects on Phosphorus Surface Runoff Losses The mean P Ac was initially 9.5 mg L 1 in the top 10-cm soil layer of all plots. Because there were no changes in P Ac at the depth of plowing, we started to sample the soil surface. In the topsoil (0 2 cm) of P-amended plots, a clear increase in P Ac was recognized when slurry was added biannually. On the slurry broadcast plots, the mean P Ac doubled or even increased to as much as sixfold, probably partly because the slurry had not infiltrated the soil during the dry summer. In fall 1998, the mean P Ac was elevated in the 2-cm-deep surface soil layer in the broadcasting plots (20.6 mg L 1 ) compared with the injection and control (11.4 mg L 1 ) plots before the autumn applications. Thus, the broadcasting of slurry onto soil with a good P status and coincidental rainfall gave rather high DRP and TP losses (2.2 and 2.7 kg ha 1 yr 1, respectively). This is in agreement with Sharpley et al. (1986, 2001), who found a link between high soil P and DRP losses. The role of the recently applied P as a key source of P in runoff was pointed out by Kleinman et al. (2002). The TP and DRP losses from the fall of 1998 were one third of the total of all losses over the 3-yr study period. A rather high P Ac (65 mg L 1 ) was measured in October 1999, but the surface runoff and the losses of TP and DRP were small because there were no heavy rains. Because the experimental field was quite flat, the risk of surface runoff was smaller than that from the steeper fields. The surface runoff DRP loss was greatest from the plots where slurry was broadcast. Turtola and Yli-Halla (1999) also reported that on a low-p soil surface, application of slurry may increase the P status at the soil surface in a few years and multiply the P loading to surface runoff. In our study, when slurry was injected into the soil, the P Ac increased most in the 2- to 5-, 5- to 10-, and 10- to 20-cm soil depths. On the plots with mineral fertilization, the P Ac in the topsoil (0 2 cm) varied from 9.6 to 20.5 mg L 1 in Studies I and II. Phosphorus Balances The mean surplus of P was extremely high (32 36 kg ha 1 yr 1 ) on slurry-amended plot areas, whereas the average P surplus in Finnish agricultural land was estimated to be 13 kg ha 1 yr 1 in the period 1995 through 1999 (Antikainen et al., 2005). Moreover, the farm gate balance for P has been found to be 12 kg ha 1 on dairy farms (Virtanen and Nousiainen, 2005; Uusi-Kämppä et al., unpublished data). The amounts of P sources applied were high (52 79 kg TP ha 1 yr 1 ) during slurry years. For practical farming, the maximum amounts of plantavailable P and N allowed for silage grass on any soil P status were 30 and 180 kg ha 1, respectively. In P application calculations, 75% of the TP in slurry was normally estimated to be plant available. According to the current agri-environmental scheme ( ), 85% of TP in livestock dung is considered as plant available. In Study II, the maximum plant-available P was exceeded annually by 8 to 29 kg ha 1. The applied P amounts can sometimes be greater than allowed for in a single year, and the 4-yr cumulative amount of P applied may be rotated for 4 yr. In our study, the input of P was rotated for years and Thus, the study is relevant to the situation of P application amounts on Finnish dairy farms. On the other hand, during Study II, the amount of TN was considerably greater than 170 kg ha 1, which in Finland is the maximum amount allowed by Government Decree (931/2000; Finlex, 2000b). The maximum amount of TN in slurry spread to soil was already exceeded during the summer application by 0 to 50 kg ha 1 (Uusi-Kämppä and Mattila, unpublished). The autumn slurry applications in Study II were not allowed by Government Decree (931/2000; Finlex, 2000b). The estimated animal unit per hectare was also high (3 5.8 au ha 1 ) on the slurry application areas because the mean animal unit per hectare is supposed to be 1.5 au ha 1 (Environmental Protection Decree 169/2000; Finlex, 2000a). Uusi-Kämppä and Heinonen-Tanski: P & Microbe Runoff from Grass 2347

10 Microbial Content of Surface Runoff Water Injection reduced the occurrence of enteric microorganims in runoff waters, but this treatment cannot guarantee water that would be free of enteric microorganisms. The fecal microorganisms could have been controlled best if the application of slurry had been done in spring or early summer so that solar UV irradiation and other factors could destroy a proportion of the enteric microorganisms (Caslake et al., 2005). The enteric microorganisms of slurry may survive more than 1 yr in Finnish winters, as shown in the present work, and many soil bacteria survive in tundra permafrost (Khlebnikova et al., 1990). Most slurry-contaminated runoff waters would not meet quality requirements set for bathing waters (The European Parliament and the Council of the European Union, 2006), where the numbers of E. coli (fecal coliforms include this species mainly) should be less than 500 CFU per 100 ml and the numbers of enterococci less than 200 CFU per 100 ml in excellent quality inland waters. These surface waters, if used for crop irrigation, would risk the spread of contaminating microorganisms to food chains (Beuchat, 2006). These waters would also not have met the limits for irrigation in Alberta, Canada, where water should contain less than 100 CFU per 100 ml of fecal coliforms and less than 1000 CFU per 100 ml total coliforms (Alberta Environment, 1999) or in California, where the treated wastewater for unrestricted irrigation should contain less than 2.2 CFU per 100 ml total coliforms measured by the most probable numbers technique (California Regional Water Quality Control Board, 1991). The total coliforms were used as official index organisms in the former European bathing water directive (The Council of the European Communities, 1976), which was rejected in 2008; therefore, this group was analyzed although it was known to indicate also nonfecal contamination (American Public Health Association, 2005). Our results show that the surface runoff water always contains relatively high numbers of total coliforms, including E. coli, although the numbers of coliphages or sulfite reducing clostridia could be less than the detection numbers, indicating different survival and spreading abilities between different enteric microorganisms. All analyzed enteric microorganisms were able to be found in cases when manure contamination was evident. Some enteric microorganisms may also have been transported from slurrytreated plots to non-slurry treated control plots by water movement, wind, wild animals, and other factors as well as from grazed buffer zones to nongrazed buffers. Drainage Water and Phosphorus Leaching Levels of P in drainage waters could not be measured for this study. Therefore, we tried to estimate P leaching by taking soil profiles for P analyses in spring and fall. According to the P Ac in soil profiles, leaching of P did not seem to be a great problem from grassland. The lack of drainage water was not thought to be a serious problem on the clay soil because in many other Nordic studies (e.g., Uhlen, 1978; Turtola and Kemppainen, 1998) the P losses in drainage water have been quite small after slurry applications. Sharpley (1995) observed that sorption of P by P-deficient subsoils generally resulted in lower concentrations of DRP in subsurface than in surface runoff. Additionally, Cameron et al. (1996) reported that leaching of orthophosphate and its concentration in drainage water were uniformly negligible after surface application and subsurface injection of dairy pond sludge to pasture on fine sandy loam soil. However, there is a risk of P leaching into drainage water from heavily manured (Kleinman and Sharpley, 2003; Hountin et al., 2000) or grazed fields (Sharpley and Syers, 1979). Conclusions The risk of surface runoff becoming contaminated with different forms of P and fecal microbes can be minimized by avoiding application of slurry to excessively wet soils, during rainy periods, and when heavy rainfalls are forecasted. If slurry is surface broadcast to wet soil and heavy rains (e.g., thunderstorms in summer) occur shortly thereafter, the runoff TP, DRP, and fecal microbe losses from slurry-amended soils can be extremely high. Shallow injection of slurry below the soil surface significantly decreased P loss in surface runoff relative to broadcasting by decreasing the content of easily soluble P at the soil surface (0 2 cm). The surface runoff TP and DRP losses from injected grassland were not significantly different from those of the mineral-treated grassland (control plots). Injection only partially prevented microbial losses of runoff waters compared with broadcasting. In the future, the injection of slurry into grassland would be favored where it is technically feasible. To adjust P fertilization, calculations of field and farm P balances, especially on livestock farms, are recommended. Best management practices, such as avoiding application of slurry to wet soil or outside of the growing season, application of optimum amounts of slurry, injection of slurry into soil, and buffering areas at the field edges, will reduce water pollution from the application of slurry. The P losses in surface runoff were low in dry years and when slurry was applied only in summer. The risk for incidental losses of TP and DRP may increase in the future if climate change increases rainfall and melting waters during winter. The continuous growth of cattle farms and the concentration of livestock farms in certain agricultural areas may also increase P pollution and may increase the risks of enteric pathogen transmission to the environment. Therefore, environmental aspects must be taken into account when choosing any agricultural practice. Acknowledgments We are grateful to Mr. Risto Tanni, Mr. Ari Seppänen, Mr. Aaro Närvänen, and Petri Kapuinen, Lic.Sc. (Agr.Eng.) for their technical assistance during the experiment. We thank biometrician Mr. Lauri Jauhiainen for his statistical expertise. The evaluation of the manuscript by Professor Eila Turtola is gratefully acknowledged. Financial support for this study was provided by the Ministry of Agriculture and Forestry Journal of Environmental Quality Volume 37 November December 2008