Effects of cattle treading and soil moisture on phosphorus and sediment losses in surface runoff from pasture

Abstract

Cattle treading is known to influence surface runoff contaminants such as phosphorus (P) and suspended sediment (SS). The objective of this work was to determine how simulated cattle treading affects losses of P and SS in surface runoff from four soils (Brown, Melanic, Pallic and Recent Gley) with contrasting resistance to treading damage and drainage properties. Soils were excavated, placed in boxes (0.8 m long and 0.20 m wide) and one of three moisture treatments were imposed (10, 50 and 90% of available water holding capacity) before surface runoff was induced via simulated rainfall; P fraction and SS were determined in the resulting runoff. The Pallic and Recent Gley soils lost the most P and SS compared with the better structured Brown and Melanic soils. Greater losses of particulate phosphorous (PP) and SS were observed at higher soil moisture contents. In contrast, dissolved P in runoff was greater at lower soil moisture contents, especially dissolved reactive phosphorus (DRP), which accounted for up to 90% of total P losses at 10% soil moisture in the Recent Gley soil. Measurement of soil water soluble P, a surrogate for DRP at a range of soil moisture contents, suggests this could be due to lysis of soil microbial biomass upon drying. Although concentrations and loads of DRP in surface runoff were greatest at 10% moisture, the risk of surface runoff is less than the risk posed by PP and SS losses with treading on wet soils, where surface runoff is more likely. This was particularly true for the Pallic and Recent Gley soils. It is therefore recommended that, to minimise SS and P losses to surface water, care be taken when grazing these soils under wet winter–spring conditions. Although collected under controlled conditions, these data help to explain the wide range of potential SS and P losses that may occur in a field due to hydrologic and soil variation.

Keywords:

• compaction
• pugging
• particulate phosphorus
• dissolved phosphorus
• water soluble phosphorus
• microbial biomass phosphorus

Introduction

Agricultural land use in New Zealand has intensified since the 1990s with greater concentrations of animal stock on pastures (Parfitt et al. 2008). Coupled with intensification, land is now being grazed that was previously thought to be marginal or uneconomic for use, including soils that have poor structural stability or experience low rainfall (Hewitt & Shepherd 1997). Losses of suspended sediment (SS) and phosphorus (P) from these soils have been shown to increase with grazing intensity and treading damage (e.g. McDowell et al. 2003a). However, additional factors such as soil moisture, grazing duration, vegetative cover and soil physical properties may be equal, or more important, in influencing how treading affects P and SS losses (Climo & Richardson 1984; McDowell et al. 2008;). Given the sensitivity of many of New Zealand's surface waters to P enrichment (McDowell et al. 2009) and SS inputs, it is important to examine how some of these factors may influence losses, and reinforce best management practices, to protect surface water quality.

It is accepted that compaction occurs on medium to wet soils and pugging can result from grazing saturated soils (Bilotta et al. 2007; Drewry 2006; Scholefield et al. 1985). Treading damage decreases soil infiltration rates and increases ponding and surface runoff of P and SS (Greenwood 1999; McDowell et al. 2008; Taboada & Lavado 1993). Further, the crushing and burial of pasture (a major source of P loss (McDowell et al. 2007)) caused by hoof action will negate any benefit pasture provides as a protective cover against soil erosion (McDowell et al. 2008; Nash & Halliwell 1999). However, it is unclear whether or not P and SS losses increase with increasing soil moisture on trodden soils and, if so, which P fractions are affected: dissolved or particulate phosphorus (PP). For instance, at one site in South Otago, McDowell et al. (2003a ,b) found low losses over winter associated with the sedimentation of P-containing particles in deep hoof imprints, whereas McDowell & Houlbrooke (2009) found higher losses over winter at a dry site receiving irrigation in North Otago.

In addition to soil moisture, soils with poor structural stability and drainage tend to be more affected by treading than well structured soils (Russell et al. 2001). With decreasing infiltration rates, the potential for saturation can be greater, due to fewer macropores (McDowell et al. 2003b; Tian et al. 2007). Furthermore, under ponded and waterlogged conditions, the soil may begin to experience oxygen deficiencies (Greenwood & McNamara 1992) that could promote P release and loss as ferric minerals may become reduced in winter (Magid et al. 1996). Previous work has documented P and SS losses from Pallic soils (Monaghan et al. 2007), but given the wet winter conditions that exist in the New Zealand provinces of Otago and Southland, the likelihood of surface runoff is probably high irrespective of soil type.

The objective of this study was to determine the impact of treading and moisture status on the loads and concentrations of dissolved and PP and SS in surface runoff from four different pasture soils.

Materials and methods

Experimental design

Four soil types were investigated, representing those commonly used or currently being developed for pastoral grazing by cattle in southern and eastern Otago and Southland, New Zealand: a Recent Gley (Momona silt; Aquent) and Pallic soil (Warepa silt; Fragiochrept) of moderate to high structural vulnerability (SV) and a Brown (Cargill silt loam; Dystrochrept) and Melanic (Oamaru silt loam; Rendoll) soil of moderate to low SV according to Hewitt and Shepherd (1997) (soil name and USDA classification given in parentheses) (Table 1). The Brown and Melanic soils are silt loams and the Pallic and Recent Gley are silts. All soils had been in pasture for >10 years and were sampled at similar available water holding capacity (10–25%) in autumn 2008 from paddocks chosen to have the same slope (∼5%) and southerly aspect.

Table 1  Mean chemical and physical properties of the four soils used. The F-statistic is given for comparison of means between soil types when appropriate.

Parameter	Brown (Cargill silt loam)	Melanic (Oamaru silt loam)	Pallic (Warepa silt)	Recent Gley (Momona silt)	F-statistic
WSP (mg l^−1)^a	1.02	1.12	2.11	3.56	<0.001
Olsen P (mg kg^−1)	19	24	21	41	<0.001
Total P (mg kg^−1)	747	601	209	680	<0.001
Organic C (g kg^−1)^b	70	170	30	50	–
P retention (%)	56	26	17	37	–
Bulk density (g cm^−3)	0.83	0.61	1.1	0.82	<0.001
Sand (g kg^−1)^b	175	235	105	55	–
Silt (g kg^−1)^b	620	640	840	900	–
Clay (g kg^−1)^b	210	130	55	40	–
K sat (mm h^−1)^c	462	7055	461	135	<0.05
Macroporosity (% v:v)	21	34	11	17	<0.001
^aWater soluble phosphorus.
^bDetermined on bulked soil samples. Particle size distribution presented according to USDA classification.
^cSaturated hydraulic conductivity (Ksat) was calculated using the log100 scale for normality.

Twelve soil samples (0–7.5 cm depth) were taken at each site to determine background chemical properties; additional samples at 0–5 cm depth were taken for soil physical analyses (see below). Soils to be placed under a rainfall simulator to generate surface runoff were excavated using a purpose-built cutting blade modified from the work McDowell et al. (2007). Briefly, this involved hammering the 0.8 m long by 0.2 m wide blade to 0.15 m depth and carefully excavating topsoil. Soils were then trimmed to 12.5 cm. Pasture (all >90% ground cover) was trimmed to 5 cm height and turfs were placed into boxes. When in plastic lined boxes, one of three treatments was imposed: 90% of available water holding capacity (AWHC); 50% of AWHC; or 10% of AWHC.

AWHC was calculated using the bulk density and volumetric soil moistures calculated from saturated and air-dry soil cores collected at the same time as soils for boxes. These moisture treatments are hereafter referred to as 10%, 50% and 90%. There were four replicates of each treatment for each soil type, except for the 90% treatment, which had 16 replicates.

After treatments had been established for at least 7 days, each boxed soil was trodden on by an artificial cow hoof five times (20 imprints m⁻²) to simulate treading during a 24 h grazing event (McDowell et al. 2003a). The artificial cow hoof was modelled on a 2 year old Friesian cow and delivered 250 kPa of pressure over a 90 cm² area (Di et al. 2001). Soils were then left outdoors under shelter for 24 h.

Surface runoff collection and analyses

Soils were placed under a rainfall simulator (tap water; P less than the detection limit of 0.001 mg P l⁻¹) and rainfall applied to boxed-turfs at a rate 30–35 mm h⁻¹ to create surface runoff. Each box was inclined at a 5% slope as in the field. The rainfall simulator used one TeeJet 1/4HH-SS30WSQ nozzle (Spraying Systems Co., Wheaton, IL) positioned approximately 250 cm above the soil surface to produce rainfall with size, velocity and impact angle approximating natural rainfall (Shelton et al. 1985). The nozzle, plumbing, in-line filter and pressure gauge were fitted onto a 305 cm high by 305 cm wide by 305 cm deep aluminium frame with tarpaulins on each side to provide a wind screen. The 30–35 mm h⁻¹ rainfall intensity has a return frequency of approximately two–three times a year for a 10 min event (NIWA 2010). After initiation of surface runoff, it was collected for 30 min. Runoff was collected and measured in a measuring cylinder and a 200 ml sub-sample was taken to the laboratory for further analysis.

Samples of surface runoff were filtered (0.45 µm) immediately after collection and analysed for dissolved reactive phosphorus (DRP) within 24 h. Filtered and unfiltered samples were refrigerated at 5°C for digestion within 7 days using persulphate (Eisenreich et al. 1975), which, after colorimetric analysis, determined total dissolved phosphorus (TDP) and total phosphorus (TP). Fractions defined as dissolved unreactive phosphorus (DUP) and PP were defined as the difference between TDP and DRP and TP and TDP, respectively. SS was analysed by weighing the oven-dried residue on a GF/A glass fibre filter paper (0.7 µm pore size; GF75-MFS Advantec Inc., Pleasanton, CA) before and after filtration of a known volume of sample.

Soil analyses

Soils (0–7.5 cm depth) for chemical analyses were air-dried, crushed and passed through a 2 mm sieve before bicarbonate-extractable P (Olsen P (Olsen et al. 1954)), water soluble P (WSP (McDowell & Condron 2004)) and total P, after digestion in aqua regia (4:1 v:v concentrated HCl:HNO₃ (Crosland et al. 1995)), were determined. All P analyses were measured in duplicate and used the colorimetric method of Watanabe & Olsen (1965).

Soil physical measurements included macroporosity (volumetric percentage of pores >30 µm), bulk density and saturated hydraulic conductivity (K _sat) as outlined by Drewry & Paton (2005).

The effect of soil moisture treatments on the potential for DRP release and the influence of microbial biomass (before and after chloroform fumigation) were further investigated by determining soil WSP for each soil type at a range of soil moistures (10, 20, 50, 75 and 100% of AWHC). Field moist samples (∼500 g at 20–30% of AWHC at sampling) were sieved to 2 mm. Triplicate sub-samples (each 100 g air-dried equivalent) were then either fumigated with chloroform (Jenkinson & Powlson 1976) or not, and left at room temperature to dry to the equivalent moisture (determined using earlier estimates of AWHC, bulk density and daily assessment of weight). If the soil was not wet enough, sufficient sterile water was added and mixed into the soil using a sterile spatula until the required moisture content was reached (by weight). Soils were left to equilibrate for 1 week, with daily correction for evaporation if needed, before extraction of WSP.

Statistical analyses

Mean concentrations and loads of P fractions and SS in surface runoff were tested for normality and transformed if necessary before being subjected to an ANOVA, fitting terms for soil type, treatment and the factorial interaction of soil type and treatment. The F-statistic or the least significant difference at P<0.05 (LSD₀₅) is presented to compare means between soils and treatments, and their interaction.

Results and discussion

Surface runoff

Soil physical analyses showed that macroporosity was greatest for the Melanic soil followed by the Brown, Recent Gley and Pallic soils, while the opposite was true for bulk density (Table 1). Previous work has shown an inverse relationship between macroporosity and surface runoff volumes, due to the decreased water holding capacity (fewer large pores) and time to ponding caused by pore blockage (e.g. McDowell et al. 2003a). Greater compaction (lower macroporosity and higher bulk density) was also observed under irrigated practices when compared with dryland treatments for two out of three years on a Pallic silt loam (Houlbrooke et al. 2009). In the present study, macroporosity in the Pallic soil was the least (11%) and the volume of surface runoff was significantly (P<0.05) greater (2.17 l) than that of the other soils (Melanic, Brown and Recent Gley producing 1.67, 1.65 and 1.23 l, respectively (see Table 2)) for the 90% soil moisture treatment.

Table 2  Mean load of P fractions (dissolved reactive P (DRP), dissolved unreactive P (DUP), particulate P (PP) and total P (TP)), suspended sediment (SS) and volume lost in surface runoff for each soil type and treatment during the rainfall simulation. The F-statistic (using log transformed data) is presented for treatment comparisons.

		Mean load (kg P ha^−1)
Soil type	Moisture (%)	DRP	DUP	PP	TP	SS (kg ha^−1)	Volume (l)
Brown	10	0.021	0.003	0.001	0.025	0.08	1.05
	50	0.004	0.001	0.006	0.011	0.17	1.59
	90	0.001	0.002	0.011	0.014	0.52	1.65
Melanic	10	0.013	0.002	0.002	0.017	0.06	1.09
	50	0.003	0.001	0.006	0.010	0.10	1.14
	90	0.002	0.002	0.014	0.018	0.78	1.67
Pallic	10	0.024	0.001	0.001	0.026	0.07	1.09
	50	0.009	0.001	0.005	0.015	0.39	1.14
	90	0.013	0.005	0.031	0.049	2.32	2.17
Recent Gley	10	0.090	0.002	0.005	0.097	0.08	1.43
	50	0.016	0.003	0.004	0.023	0.16	1.12
	90	0.038	0.003	0.013	0.054	0.92	1.23
F-statistic soil		<0.001	0.498	0.095	<0.001	0.006	0.046
F-statistic moisture		<0.001	0.508	<0.001	0.079	<0.001	0.021
F-statistic soil×moisture		0.003	0.110	0.002	0.056	0.014	0.094

Unsurprisingly, the 90% soil moisture treatment produced significantly more surface runoff than the 50% and 10% soil moisture treatments except for the Recent Gley (Table 2). The likely explanation for this is that most pores were already filled with water, thereby decreasing the time to produce surface runoff. For example, it took nearly twice as long to produce surface runoff in the Melanic soil at 10% soil moisture compared with the 90% soil moisture treatment (data not shown). This was not the case for the Recent Gley, which produced a greater volume of runoff from the 10% than the 90% soil moisture treatment. Due to restricted drainage and infiltration rates much greater than rainfall intensity (Table 1), saturation excess is the most likely cause of surface runoff. Data for the Recent Gley also suggest that a proportion of surface runoff may have been caused by infiltration excess conditions, possibly due to surface smearing. However, overall no difference was found in surface runoff volumes for the interaction of soil type and moisture.

Phosphorus and SS losses

Soil type

Fig. 1 and Table 2 present the mean concentrations and loads of P fractions and SS lost in surface runoff for each soil type and moisture regime. The greatest mean TP load occurred from the Recent Gley soil (0.059 kg P ha⁻¹), followed by the Pallic, Brown and Melanic soils (0.03, 0.017 and 0.015 kg P ha⁻¹, respectively). Although most surface runoff was collected from the Pallic soil at 90% soil moisture, the P lost appeared to be more influenced by the Olsen P concentration of the Recent Gley (41 mg kg⁻¹) compared with the other soils (19–24 mg kg⁻¹). This is consistent with the findings of Pote et al. (1996). However, it has also been established that the concentration of DRP in surface runoff is a function of the quotient of both P retention and Olsen P (McDowell & Condron 2004), which can be approximated by WSP. Among the soils tested here, the Recent Gley also exhibited the greatest WSP and DRP concentration (see Table 1 and 2, Fig. 1). Given that 40–81% of the TP load in surface runoff was as DRP, this indicates that soil P concentration (Table 1) outweighed the influence of greater surface runoff in the Pallic soil (Table 2). Although saturation-excess surface runoff has been found to be the more common pathway for P losses in Southland and Otago compared with infiltration-excess surface runoff, soils in the field are deeper than 12.5 cm and may be subjected to a wider range of hydrological processes (e.g. subsurface seepage) than those simulated here. The PP and DUP accounted for much less, on an individual basis, of the TP load in surface runoff (12–49% for PP and 5–12% for DUP; Table 2).

Fig. 1  Mean concentrations of P fractions and SS lost in surface runoff for each soil and moisture content. The LSD₀₅ is given for the interaction between soil type and moisture content.

The Pallic soil yielded the greatest SS concentration and load in surface runoff (Table 2, Fig. 1). McDowell (2006) reported losses of 1499 kg SS ha⁻¹ from a Fragic Pallic soil within a grazed dairy catchment in South Otago. The large erosion losses in this study compared with the New Zealand average for dairy of 299 kg SS ha⁻¹ (McDowell & Wilcock 2008) were ascribed to the poor aggregate stability of the Pallic soil (Rousseva 1988) as well as livestock grazing stream banks. Hewitt & Shepherd (1997) incorporated factors that influence aggregate stability within their measure of a soil's SV to cope with stress. Among the soils tested in the present study, the Pallic soil exhibited the greatest SV (0.69), primarily due to the low amounts of Al and Fe oxy-hydroxides and low organic carbon concentration (Table 1). Although classed as having ‘moderate structural vulnerability’ (SV of 0.5–0.6), the Recent Gley soil also exhibits poor drainage (Hewitt & Shepherd 1997) and experienced large SS losses in surface runoff at 90% soil moisture (Table 2). The greater organic C concentration and drainage (as indicated by macroporosity values) for the Melanic and Brown soils (Table 1) may have contributed to decreased SS losses in surface runoff for 90% soil moisture (Table 2, Fig. 1).

Influence of soil moisture

Among soil moisture treatments, TP loads varied from an average of 0.010 to 0.097 kg P ha⁻¹ (Table 2). Although differences in TP load across treatments were not significant, differences among DRP and PP fractions and SS losses were. On average, the greatest SS loads were lost from the 90% soil moisture treatment followed by the 50% and 10% treatments (1.134, 0.205 and 0.075 kg ha⁻¹, respectively (Table 2)). This is most likely due to increases in erosion power with increasing runoff volume (e.g. Quinton et al. 2001). Since PP losses usually arise via similar erosive processes (Ballantine et al. 2009), increased surface runoff also explains the greater PP load and concentration with 90% soil moisture content compared with the runoff produced under 50% and 10% soil moisture treatments (Table 2, Fig. 1).

In addition to increased surface runoff, the treading impact increased with soil moisture (Climo & Richardson 1984; Drewry 2006). Under wet conditions, the friction between soil particles is lessened and particles can move along points of weakness, decreasing soil strength and altering its structure (Platto et al. 1978). Visual inspection of the hoof imprints indicated that while some compaction was probable on soils at 50% soil moisture content, raised edges around the imprint (Bilotta et al. 2007), indicative of pugging, were probable on the soils at 90% soil moisture. A past study of a Pallic soil in South Otago showed that sediment and P loss in surface runoff was restricted by ponding and sedimentation in deep imprints (McDowell et al. 2003a ,b). However, the sedimentation potential of the imprint will be dependent upon depth, whereby sedimentation is probably negated in shallow imprints by turbulent surface runoff, much like turbid waters of a shallow lake on a windy day. The imprints in the study of McDowell et al. (2003b) were up to 20 cm deep, while those in the present study were only up to 12 cm deep, particularly for 90% moisture contents. Although imprint depth may have been restricted by the depth of the soil boxes (12.5 cm), the salient point is that PP and SS loss increased with soil moisture content to the point that PP losses accounted for >60% of TP losses in soils at 90% soil moisture. Possible reasons for the increase in TP and SS, in addition to surface runoff volumes, include the crushing and smearing of aggregates at the soil surface (Platto et al. 1978), an increase in soil bulk density beneath the imprint (O'Connor 1956) and an increase, via imprints, in the surface area of soil interacting with surface runoff.

In contrast to PP and SS, DRP concentrations and loads decreased with increasing soil moisture; within the 10%, 50% and 90% soils, the load of DRP lost accounted for 80%, 53% and 40% of TP losses, respectively. Clearly, compared with SS and PP, different processes for DRP were affected by soil moisture.

Water soluble P and soil moisture

Water soluble P extractable from soil (WSP-soil) can be used as a surrogate for DRP in surface runoff (McDowell & Condron 2004). Hence, WSP was used to further investigate the changes in DRP with soil type and moisture. The principal hypothesis was that desiccation and lysis of the soil microbial biomass at 10% soil moisture increased availability and losses of DRP in surface runoff (Sparling et al. 1987). If true, this may partly explain why concentrations of DRP in summer surface runoff events tend to be enriched for certain soil textures (Magid and Nielsen 1992; Magid et al. 1996). An alternative hypothesis is that DRP concentration becomes enriched and builds up simply because of reduced soil moisture in drier periods. Or, in other words, DRP concentration in surface runoff under saturated excess conditions from wetter soils is lower due to dilution. Turner & Haygarth (2001) also reported that total amounts of WSP were minimal in moist soils (those drained from saturation for 48 h), but increased by 185–1900% after drying.

Fumigation with chloroform, followed by extraction of fumigated and unfumigated soils, has been used as a tool to give a measurement of microbial biomass P (Brookes et al. 1982). Although usually extracted with bicarbonate, this work sought to extract the P fraction likely available to surface runoff and hence water was used (WSP-microbial biomass). Precedents where extraction with water has been used to establish WSP-microbial biomass exist for soils (Morel et al. 1996; Oehl et al. 1998) and stream sediments (McDowell 2003). The data in Fig. 2 indicate that WSP-soil and WSP-microbial biomass were much greater, at most soil moisture contents, in the Recent Gley soil than other soils, which mirrored DRP in surface runoff (Fig. 1). Among the soils, the relative proportion of WSP-microbial biomass compared with WSP-soil tended to be lower at 10% moisture than either 50% or 100% moisture (Fig. 2). This was especially so for the Pallic soil, which had very little WSP-microbial biomass for any given soil moisture. The exact reasons for this difference are unclear. The increase in the proportion of WSP-microbial biomass in the Melanic soil, up to 50% moisture, suggests that lysis played a limited role in supplying DRP for surface runoff compared with the other soils where WSP-soil increased while WSP-microbial biomass decreased. Although DRP enrichment at 10% soil moisture for the Melanic soil could be related to dilution/concentration, Haynes & Swift (1985) provided an alternative hypothesis. When looking at the effect of air-drying on soil chemical properties on four acid soils in New Zealand, they showed that air-drying increased the extraction of native soil P compared with moist soils. The increase in P extractability was thought to be associated with the decomposition of organic matter complexes upon drying. Although speculative, the enriched organic C concentration of the Melanic soil (170 g kg⁻¹, Table 1), an attribute for low SV (Hewitt & Shepherd 1997), makes this alternative hypothesis possible. Another reason could be that the microbial population was more resilient to desiccation, perhaps imparted by high organic matter concentrations (Jenkinson & Powlson 1976).

Fig. 2  Mean concentration of water soluble P extractable from soil (WSP-soil) and from microbial biomass for each soil type and moisture content. The LSD₀₅ is given for the interaction between soil type and moisture for WSP-soil and for WSP-microbial biomass.

In contrast to the other soils, DRP loads in surface runoff from the Recent Gley soil at 90% soil moisture were more than double the DRP loads observed at 50%, despite a small but not significant difference in surface runoff volumes (Table 2). Magid et al. (1996) suggested that ferric minerals may become reduced via anaerobic conditions during the winter. Greenwood & McNamara (1992) noted that treading can cause a soil to become waterlogged and oxygen deficient. The data in Fig. 2 illustrate that, for the Recent Gley soil, WSP-soil was greatest at either wet or dry extremes and, for 100% soil moisture, reduction via anaerobic conditions is very likely. In addition to the reduction of ferric-P, anaerobic conditions can contribute P via the death of plant roots (Gradwell 1968) and micro and macro fauna. At 10% soil moisture, it is likely that WSP-soil peaked due to the lysis of soil microbial biomass because WSP-microbial biomass was least at 10% soil moisture. Although the experimental conditions of this work may have led to a re-inoculation of microbes during the week where soil was left to equilibrate open to the air, it is highly unlikely that this would match the size and diversity of the original population. The fact that differences between WSP-soil and WSP-microbial biomass were large also indicates that re-inoculation (which would minimise this difference) was not a significant factor.

These data suggest that the majority of DRP loss occurs from dry soils, while greater PP and SS losses occur from wetter soils; this implies that the periods of greater risk after a 24 h grazing event by cattle (with associated treading) are summer and autumn for DRP, and winter and spring for PP and SS. However, runoff in Otago and Southland is mostly generated by saturation-excess conditions (Srinivasan et al. 2007), implying that the risk of surface runoff associated with summer and autumn is less than in winter and spring. As shown with the Recent Gley soil, saturated soils or soils at field capacity for prolonged periods also have the potential to experience reducing conditions (Gradwell 1968), which can influence DRP losses. The current experiment did not allow for the fragipan within the Pallic subsoil; this severely limits drainage at 25 cm depth. This limitation, coupled with the low P retention, suggests that the Pallic soil has excessive PP and SS loss via saturation-excess surface runoff. In contrast, the Brown and Melanic soils had good structural stability and were unlikely to experience such reducing conditions. Best management practices (BMPs), such as restricted grazing and a decrease in stocking frequency or density should be considered during wet spring–winter conditions in Otago and Southland. Such BMPs could decrease damage caused by the hooves to topsoil and the removal of pasture cover, and thus protect against erosion and losses of sediment P (Nash & Halliwell 1999).

Conclusions

The equivalent of 24 h treading by a mechanical hoof on soils of contrasting soil moistures caused differences in the concentrations and loads of P fractions and SS lost in surface runoff from four soil types. The Pallic and Recent Gley soils were shown to be more susceptible to SS and P losses than the Brown and Melanic soils. With increasing soil moisture, the combination of physical disturbance and crushing and burial of pasture (depleting the protective cover against soil erosion) caused an increase in sediment and PP losses. In contrast, increases in DRP losses from drier soils were attributed to the release of soil microbial biomass P, except in the Melanic soil where the effect was either absent or much decreased. The concentration of WSP in the Recent Gley soil was greater for 100% moisture than that at 50% moisture and this was attributed to increased P release under reducing conditions.

The data indicate that the risk to surface water quality from runoff, when soil moisture is low, is high due to the coincidence of enriched bioavailable P (DRP) in the runoff and warm temperatures and high light levels that promote algal growth. However, the risk posed by losses of PP and SS with treading is also great, because surface runoff is more likely in wetter periods. This winter-derived loss is particularly important for the Pallic and Recent Gley soils studied, and suggests that care should be taken when grazing these soils under wet spring–winter conditions in Otago and Southland. Although collected under controlled conditions, these data will be useful in explaining the wide range of potential SS and P losses that may occur in the field due to hydrologic and soil variation.

Acknowledgements

This work was conducted for Pastoral 21 Environment programme (contract C10X0603), funded by FRST, Dairy NZ, Fonterra and Meat and Wool New Zealand.