Measurement and prediction of nitrogen loss by simulated erosion events on cultivated forest soils of contrasting structure

Soil & Tillage Research 83 (2005) 204–217 www.elsevier.com/locate/still

Measurement and prediction of nitrogen loss by simulated erosion events on cultivated forest soils of contrasting structure P.C. Teixeiraa,b, R.K. Misrac,* a

Cooperative Research Centre for Sustainable Production Forestry, Private Bag 12, Hobart, Tasmania 7001, Australia b Department of Plant Science, University of Tasmania, Private Bag 55, Hobart, Tasmania 7001, Australia c Faculty of Engineering and Surveying, The University of Southern Queensland, Toowoomba, Queensland 4350, Australia Received 15 April 2003; received in revised form 24 June 2004; accepted 14 July 2004

Abstract The objectives of this study were to determine nitrogen (N) loss associated with erosion of forest soils and to explore the role of soil structure and other factors governing N enrichment of sediment to aid prediction of N loss. We measured erosion, size distribution of aggregates in the sediment and N distribution in various aggregate fractions using simulated rainfall on samples of three cultivated forest soils of contrasting structure (repacked in trays) exposed to four erosion conditions. Both sediment loss and N loss increased with slope and kinetic energy of rainfall suggesting greater dependency of N loss on sediment loss than on N concentration in the sediment. Irrespective of erosion treatments and soil type, the bulk sediment and its size fractions were mostly richer in N than those of the uneroded soils. The enrichment ratio (ER) and concentration ratio (CR) of N for sediment cast some doubt on the application of raindrop stripping as a mechanism of N enrichment of sediment for soils of widely differing characteristics. Previously published models of N enrichment of sediment did not predict ER for our soils satisfactorily. However, an empirical model using published data on erosion and N loss agreed well with the results. This predictive method only requires information on sediment loss that can be easily obtained from an erosion model. # 2004 Elsevier B.V. All rights reserved. Keywords: Enrichment ratio; Erosion; Forest soils; Nitrogen loss; Soil structure

1. Introduction Loss of nutrients by erosion has often been attributed to selection by erosion and deposition processes (Stoltenberg and White, 1953; Rose and * Corresponding author. Tel.: +61 7 46312805; fax: +61 7 46312526. E-mail address: [email protected] (R.K. Misra).

Dalal, 1988), leading to sediment that is rich in organic matter and clay (Bedell et al., 1946; Stoltenberg and White, 1953; Barrows and Kilmer, 1963; DeBano and Conrad, 1976; Menzel, 1980; Alberts and Moldenhauer, 1981; Flanagan and Foster, 1989). Organic residues are among the first constituents to be removed by erosion because they are concentrated in the surface soil and are less dense than other soil constituents (DeBano and Conrad, 1976). Even with

0167-1987/$ – see front matter # 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.still.2004.07.014

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

similar density, fine soil fractions are known to erode more easily than coarse fractions (Barrows and Kilmer, 1963; Flanagan and Foster, 1989), because fine fractions take longer to settle in runoff water than coarse fractions. Most soils tend to lose a greater percentage of organic matter and fine material when sediment concentration of runoff water is low than when it is high. This selective loss of soil components rich in nutrients is believed to arise from the type of erosion processes involved. Erosion by rainfall results from the impact of raindrops on soil causing localised shear stresses and breakdown of aggregates. In contrast, shear stress arising from runoff water is distributed over a large area so that sediment concentration in a runoffdriven erosion event can be much greater than that during a rainfall-driven erosion event (Misra and Rose, 1995), despite less aggregate breakdown. Both particle and aggregate size and composition of sediment are determined by the extent of aggregate breakdown (Loch and Donnollan, 1982; Proffitt et al., 1993), and by deposition before the sediment reaches the point of measurement at the downslope end of erosion plot. The extent of aggregate breakdown depends mainly on aggregate strength, the energy of raindrop impact, and the depth of water on the soil surface. Without a significant depth of water protecting the soil surface, the outer layer of some soil aggregates may be peeled off by raindrop impact producing fine soil material, a mechanism described as raindrop stripping by Ghadiri and Rose (1991a). The outer layers of aggregates may be richer in nutrients than inner parts, so that raindrop stripping produces fine material richer in nutrients than material of the same size in the uneroded soil (Ghadiri and Rose, 1991a,b). Most reports of raindrop stripping as a mechanism of chemical enrichment of sediment are for clay soils with stable structure (Ghadiri and Rose, 1991a,b; Wan and El-Swaify, 1998). How nutrient enrichment mechanisms operate in soils of less stable structure and other texture classes has not been established. Losses of nitrogen (N) by erosion are estimated to range from 1 to 100 kg ha
1 per year (White, 1986; Rose and Dalal, 1988), implicating erosion as a major cause of long-term decline in the fertility of agricultural soils. Similar information on N loss is currently limited for soils of forest plantations, where surface soil is usually enriched in N because of woody

205

residues left from harvesting the previous tree crop. Although coarse and heavy residues may minimise erosion, they may break down into fine and partially decomposed material, which is readily removed by runoff. N loss in an erosion event (SN, kg ha
1) is given by: SN ¼ SL  NS

(1)

where NS is the N concentration in the sediment (kg kg
1) and SL is sediment lost per unit area (kg ha
1). The composition of sediment as a potential pollutant of waterways is often described in terms of enrichment ratio (ER). In the context of N losses, ER refers to the ratio of N in the sediment (NS) compared to that in the uneroded soil (NU), i.e., ER ¼

NS NU

(2)

An ER > 1 denotes the sediment is richer in N than the uneroded soil, and an ER < 1 denotes impoverishment of the sediment compared with the uneroded soil. As the sediment loss (SL) can be predicted with erosion models for single events (e.g. WEPP, Lane and Nearing, 1989; GUEST, Misra and Rose, 1996) or for multiple events on an annual basis (e.g. USLE, Wischmeier and Smith, 1978; RUSLE, Renard et al., 1991), N loss may be predicted from the measurement of NS using Eq. (1) or from the knowledge of ER and NU using Eq. (2). Menzel (1980) developed a relationship between ER and sediment loss (SL) by simplifying the equations proposed by Massey and Jackson (1952): lnðER0 Þ ¼ u þ m lnðSL Þ where ER0

(3)

is the predicted value of ER, and u and m are regression parameters. Regression equations are useful for predictive purposes, but they are empirical and therefore do not explain the underlying processes relating to erosion and nutrient loss. Palis et al. (1990) tested the theoretical framework of Rose and Dalal (1988) for differentiating erosion by rainfall only or by the combined effects of rainfall and runoff. This framework assumed that the N concentration of any size class of sediment is similar to that of the same size class of uneroded soil, which is expected to work well in situations where introduction of inflow water from the upslope end of a runoff plot could increase the depth of runoff water to eliminate raindrop stripping. Thus, it is possible to design

206

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

erosion experiments where raindrop stripping could be manipulated independent of slope and slopelength. The objectives of the present study were: (1) to compare N loss for contrasting erosion events with simulated rain on bare soils of varying structure taken from forest plantations, (2) to investigate the effects of soil structure on the N enrichment of sediment, and (3) to evaluate Menzel’s method of predicting N loss resulting from erosion.

proportional to the diameter of raindrops, so two different kinetic energies were obtained with raindrops of 2.60 mm (high KE) and 0.65 mm (low KE). Raindrops of low KE were obtained by intercepting large raindrops by a net of pore size 1.0 m  1.5 mm placed 0.5 m above the erosion tray. 2.2. Erosion treatments Fourerosiontreatments wereapplied toall threesoils, and each combination of erosion treatment and soil was replicated three times. The erosion treatments were:

2. Materials and methods 2.1. Erosion experiments Three soils (Table 1) were collected from the upper 20 cm of recently cleared and mound-cultivated eucalypt plantations in Tasmania. Each soil was airdried, stones and coarse organic debris removed, and clods crushed by hand to reduce the size to 8 mm. Erosion trays 1.0 m long (downslope), 1.0 m wide and 0.1 m depth were filled with each soil and used for the measurements of erosion under simulated rain. Each tray had a buffer area of 1.0 m  0.1 m along both sides of the tray to reduce soil loss by splash. A runoff collector at the downslope end of each tray allowed sampling of all runoff and sediment from an area of 1.0 m  0.8 m. A portable, rotating disc rainfall simulator of the design of Grierson and Oades (1977) was used to simulate rain of a constant rate (116 1.8 mm h
1) using tap water. The kinetic energy (KE) of rain was varied without changing the rainfall rate for some of the erosion treatments. KE of the rain is directly

Treatment 1: low rainfall KE and gentle slope (28), Treatment 2: high rainfall KE and gentle slope (28), Treatment 3: high rainfall KE and steep slope (168) without drying cycle, and Treatment 4: high rainfall KE and steep slope (168) with a drying cycle. For treatment 4, soil was dried under shade for 15 days between two consecutive erosion events to represent a drying cycle. These four erosion treatments permitted the comparison of erosion at varying KE and constant slope, varying slope and constant KE, and the effects of a drying cycle on erosion for the high KE and steep slope condition. 2.3. Soils In this report, we refer to soils used in erosion experiments by their locations (D: Dover, R: Ridgley, and M: Maydena), which are classified as Redoxic Hydrosol, Red Ferrosol and Grey Kurosol, respec-

Table 1 Properties of the soils used in erosion experiments Soil/site properties

Soil D (Dover)

Soil R (Ridgley)

Soil M (Maydena)

Latitude Longitude Soil type Sand % (2.0–0.02 mm)a Silt % (0.02–0.002 mm)a Clay % (<0.002 mm)a Organic carbon (g kg
1) pHb EC (Ds m
1)b

438220 S 1468570 W Redoxic Hydrosol 80.2 11.1 8.7 36.3 4.59 0.107

418100 S 1458500 W Red Ferrosol 22.4 23.8 53.8 92.5 4.77 0.184

428420 S 1468410 W Grey Kurosol 40.5 24.2 35.3 21.6 4.44 0.092

a b

By weight. Soil water ratio (1:5).

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

tively (Isbell, 1996). Properties relevant to erosion experiments for these soils are given in Table 1. The aggregate stability of these soils to wetting was in the order M (very susceptible) > D (intermediate) > R (least susceptible), as quantified by the index of Teixeira and Misra (1997). 2.4. Sampling and measurements Uneroded soil (sampled prior to simulated erosion events), and sediment in runoff water (sampled at 0–5, 12.5–17.5, and 35–40 min after runoff started) were wet-sieved using a sieve shaker for 3 min into six aggregate size-classes (>2000, 2000–1000, 1000– 500, 500–250, 250–53, and <53 mm). Before the start of each replicated erosion event, the aggregate size distribution was measured for two samples of each soil type (n = 24). For each replicated combination of erosion treatment and soil type, the aggregate size distribution of sediment was obtained for each sampling time (n = 12). Total N was measured in sediment sampled in the first and last 5 min of each erosion event (0–5 and 35–40 min) to examine temporal variation of N concentration during erosion. Sub-samples of about 0.3 g were digested in a mixture of sodium-thiosulfate and salicylic acid (Rayment and Higginson, 1992) and analysed colorimetrically in a flow injection analyser (QuikChem800, Lachat Instruments, USA). For each soil type, total N was also measured on the bulk soil (3 soils  4 erosion treatments  3 replicates) and on three replicates of each size class of uneroded soil. For some erosion treatments, the amount of sediment after size fractionation was too small for total N determination. N concentration of each size class could be measured on all three replicates of treatments 2 and 3 at 0–5 min, but for other erosion treatments and at 35–40 min after erosion, N concentration was measured on sediment samples of one replicate only. N concentrations for the whole sediment samples were calculated from the distribution of N concentrations in each size class and the mass per cent of that size class. 2.5. Statistical analysis Analysis of variance (ANOVA) for sediment loss at various times of sampling considered soil type and erosion treatments as independent factors. When a

207

factor was significant (p < 0.05), mean values of the treatments for that factor were compared using the least significant difference (LSD) at p = 0.05. Size distribution of sediment and the uneroded soil were not subjected to ANOVA to minimise the number of comparisons. Instead they were compared graphically using means and standard errors. N concentrations in some size fractions of sediment were not subjected to ANOVA if they were unreplicated because of small sample size. Parameters calculated from N concentration in sediment (e.g. N loss and Enrichment ratio) were also reported as single values. Most of these data were used to predict N loss using regression methods.

3. Results 3.1. Distribution of N with aggregate size in the uneroded soil N concentration in the uneroded soil (NU) was significantly greater (p < 0.01) in soil R (0.004 kg kg
1) than in soil D (0.0014 kg kg
1) and M (0.0014 kg kg
1). N concentration of soil aggregates (Fig. 1) varied the most among aggregate sizes for soil D (coefficient of variation, CV = 57%), followed by soil M (CV = 33%) and soil R (CV = 10%). Overall, soil R had the highest N concentration

Fig. 1. Variation of N concentration with mean size of uneroded soil aggregates for soils D, R, and M. Horizontal dotted lines represent the mean N concentrations for the bulk soil samples without size fractionation. Vertical bars over mean values indicate S.E. of mean (n = 3). Some error bars are not visible, as they are smaller than the size of symbols.

208

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

for most size-classes (Fig. 1) and soil M the lowest N concentration. 3.2. Sediment loss and N concentration in sediment Sediment losses by erosion (SL, mg ha
1) were significantly influenced by soil type (p  0.001) and erosion treatment (p  0.001), as well as their interaction (p  0.001). Although SL at 0–5 min of erosion was significantly greater than at 35–40 min (p = 0.001), interaction between time of sampling and soil and/or erosion treatment was not significant. Values of SL in 5 min of erosion (averaged over 0–5 and 35–40 min) for all soils and erosion treatments are presented in Fig. 2. For all erosion treatments, SL was in the order M > D > R. There was little effect of erosion treatment on SL for soil R. For the other two soils, SL increased with increasing slope and rainfall KE. Size-distribution of aggregates in uneroded soil and sediment at the beginning (0–5 min), middle (12.5– 17.5 min), and end of erosion (35–40 min) for each soil and erosion treatment is presented in Figs. 3–5. For all soils with low slope (treatments 1 and 2), there

were always more aggregates in the finest fraction (<53 mm) of the sediment than in the uneroded soil. The sediment from soil D contained more aggregates in the 500–2000 mm fraction than the uneroded soil. There were also more 1000–2000 mm aggregates in the sediment than in the uneroded soil R in erosion treatment 1 (i.e., low slope and low KE). With high slope (treatments 3 and 4), aggregates <53 mm were more abundant than the uneroded soils R and M. For soil D, the proportion of aggregates 250– 500 mm in size was greater in sediment than the uneroded soil. There was some indication of a greater proportion of 1000–2000 mm size in sediment than the uneroded soil at longer duration of erosion for treatment 4 of soil R. The N concentration in each aggregate size class was usually greater in the eroded sediment than in the uneroded soil (Fig. 6). The difference in N concentration between sediment and soil was usually greater for erosion treatments 1 and 2 (low slope) than for treatments 3 and 4 (high slope). For a given size fraction, the difference in N concentration between sediment and uneroded soil was large for some soils, but not others. 3.3. N loss and enrichment ratio (ER)

Fig. 2. Mean values of sediment loss (SL) measured over five minutes of erosion (0–5 and 35–40 min) for soils D, R and M and the four erosion treatments (T1–T4), where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle. Vertical bars with different letter(s) denote significantly different values of SL (p < 0.05).

With low slope (treatments 1 and 2), N loss (SN, kg ha
1) during 5 min of erosion (the mean N loss for the 0–5 and 35–40 min periods) was in the order D > M > R (Fig. 7). With high slope, N loss was in the order M > D > R, even though the N concentration of sediment (NS) was generally in the order R > D M (Fig. 6). For all soils, N loss was greater at high KE (treatment 2) than at low KE (treatment 1) and on steep slope (treatments 3 and 4) than at low slope (treatments 1 and 2). The variation of SN (Fig. 7) for various soils and erosion treatments was similar to that of SL (Fig. 2). The overall concentration of N in the sediment was greater than that in the uneroded soil, as indicated by values of ER > 1 (Fig. 8). With low slope (treatments 1 and 2) ER was in the order D > R > M, and for steep slope (treatments 3 and 4) it was in the order R > D > M. We calculated values of ER of N for a range of agricultural soils under different crops, cultivation and weather conditions using the data from Massey and

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

209

Fig. 3. Aggregate size distribution of sediment eroded from soil D at three sampling times during a 40 min erosion event and for various erosion treatments (T1–T4), where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle. Aggregate size distribution of the uneroded soil is given for comparison. For this and other figures, data for any particular aggregate size class is represented as an average size (e.g. 500– 1000 mm is represented as 750 mm).

Jackson (1952), Sharpley (1985), Rose and Dalal (1988), Palis et al. (1990), McIsaac et al. (1991), Catt et al. (1994), Hansen and Nielsen (1995) and Sombatpanit et al. (1995). The mean value of ER from these data was 1.86 (S.E. = 0.139, n = 88), which is similar to the mean value of 1.97 for the forest soils obtained from the data in Fig. 8 (S.E. = 0.199, n = 24 from 3 soils  4 erosion treatments  2 sampling times). 3.4. Prediction of enrichment ratio and N loss It is often acknowledged that if ER can be predicted then N loss can be estimated without a need for N analysis of sediment (Menzel, 1980; Sharpley, 1980). In this section, we compare the measured values of ER from our experiments with predicted values of ER (denoted as ER0 ) obtained by the method of Menzel (Eq. (3)) using regression parameter values of u = 2 and m =
0.2, as suggested by Menzel (1980). ER0 using Menzel’s method generally overestimated ER for all soils, except for some data for soil D (Fig. 9).

The values of ER0 shown in Fig. 9 were also used to predict N loss (S0N ) using Eq. (1) with measured values of sediment loss (SL, Mg ha
1) and N concentrations of the uneroded soils. S0N showed a significant linear relationship with SN (Fig. 10, r2 = 0.98, p < 0.001). However, the regression line had a slope of 1.22 ( 0.031), which is significantly different (p > 0.05) from the 1:1 line. These results (Figs. 9 and 10) suggest that inadequacy in the estimation of ER had a small influence on predicted N loss, provided sediment loss was measured. Although the estimation of ER0 with Menzel’s equation allowed rapid estimation of N loss, N concentration of the uneroded soil (NU) was still required. To avoid determination of NU, it would be useful to obtain a regression between SN (in kg ha
1) and SL (in kg ha
1) directly without the consideration of ER0 as follows: S0N ¼ a1 þ a2 SL

(4)

The parameters a1 and a2 of Eq. (4) were obtained by fitting a linear regression to the available data for

210

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

Fig. 4. Aggregate size distribution of sediment eroded from soil R at three sampling times during a 40 min erosion event and for various erosion treatments (T1–T4), where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle.

Fig. 5. Aggregate size distribution of sediment eroded from soil M at three sampling times during a 40 min erosion event and for various erosion treatments (T1–T4), where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle.

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

211

Fig. 8. Enrichment ratio of N (ER) for various soils (D, R and M) and erosion treatments (T1–T4), where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle.

Fig. 6. Variation of N concentration with aggregate size at two sampling times in the four erosion treatments (T1–T4) for soils D, R and M. T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, T4: high rainfall KE and steep slope (168) with a drying cycle, and U: uneroded soil.

Fig. 7. Average N lost during 5 min of erosion from soils D, R and M with the four erosion treatments (T1–T4), where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle.

SN and SL (Table 2). These data (n = 151) were obtained from sites used for a range of agricultural crops (alfalfa, wheat, potatoes, barley, corn, sorghum), natural vegetation (range land) and bare soil (fallow), on a range of slopes in various climates, and from experimental plots ranging in area from 1 m2 to 4.8 ha. The regression parameters were a1  0, and a2 = 0.00161 (S.E. = 4  10
5), with r2 = 0.92 (p < 0.001) (Fig. 11).

Fig. 9. Enrichment ratio of N predicted with Menzel’s equation (ER0 ) vs. measured ER for various soils (D, R and M), erosion treatments and two sampling times.

212

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

(4) may have a more general application in erosion than Menzel’s. To explore the reasons for enrichment of N in sediment further, a concentration ratio (CR) of N for each size class of aggregate was calculated. Concentration ratio is the ratio of the N concentration in sediment to that in the uneroded soil for any given size fraction. The equivalent term for the whole sediment is the enrichment ratio (ER), referred to earlier. For all soils, values of CR were generally greater for low slope (treatments 1 and 2) than high slope (treatments 3 and 4) for a range of aggregate sizes (Fig. 13). Fig. 10. N lost by erosion predicted with Menzel’s equation (S0N ) as a function of measured N loss (SN) for various soils (D, R and M), erosion treatments and sampling times. Solid line represents linear regression fitted to the data and dotted lines the 95% confidence intervals of the fitted regression.

Predicting N loss (S0N ) for the experimental data of this study using the linear regression in Eq. (4) provided the best estimate, because the slope of the line (1.04 0.04, with r2 = 0.97, p < 0.001) did not differ significantly from the 1:1 line (Fig. 12). Compared with the prediction of SN with Menzel’s method (Fig. 10), using constant N concentration for sediment (a2) and Eq. (4), the prediction of S0N was closer to the observed SN, which suggests that this Eq.

4. Discussion 4.1. N concentration in sediments, N loss, and enrichment ratio of N N losses for various soils and erosion treatments (Fig. 7) during 5 min of erosion indicated the least N loss (0.2 kg ha
1) was for soil R with the least sediment loss (Fig. 2). N loss was the largest (>15 kg ha
1) for soil M, which also had the largest sediment loss for all erosion treatments (Fig. 2). Thus, the soil with the highest fertility (as indicated by N concentration in Fig. 1), soil R, was not only less susceptible to erosion but also to loss of fertility when compared with soils D and M.

Table 2 Characteristics of the erosion experiments from which the data on sediment loss (SL) and nitrogen loss by erosion (SN) were used to obtain the regression parameters of Eq. (4) Location

Plot area

Plot slope (%)

Soil texture

Soil cover

n

Type of rain

Source

California (USA) Mississippi (USA) Illinois (USA) UK Denmark

36 m2 100 m2 33 m2 864 m2 66 m2

1–3 5 3–5 9 10

N N S N N

DeBano and Conrad (1976) McDowell and McGregor (1980) McIsaac et al. (1989, 1991) Catt et al. (1994) Hansen and Nielsen (1995)

4.0–4.8 ha 46 m2

1–3 4

6 10

N N

Richardson and King (1995) Sombatpanit et al. (1995)

Texas (USA) Canada

1 m2 100 m2

NM 9–10

Rangeland Agr. crops Agr. crops Agr. crops Agr. crops, fallow Agr. crops Agr. crops, fallow Agr. crops Agr. crops, fallow

4 5 56 12 36

Texas (USA) Thailand

Loam Silt loam Silt loam Sand Loamy sand, sandy loam Loam Loam

4 18

S N

Torbert et al. (1996) Hargrave and Shaykewich (1997)

Clay Clay, sandy loam, sandy clay loam, clay loam

The number of data available (n) is given. N: natural rain, S: simulated rain, NM: not mentioned.

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

213

Fig. 11. Variation of N loss (SN, kg ha
1) with sediment loss (SL, mg ha
1) for a range of erosion conditions. Dotted lines represent 95% confidence intervals of the regression line.

For all sizes of aggregates, the N concentration was greater in sediment than the uneroded soil (Fig. 6). However, these differences in N concentration did not influence N losses for the various soils and erosion treatments (Fig. 7). Pattern of N losses for various soils

Fig. 12. Nitrogen loss predicted by Eq. (4) (S0N ) as a function of measured N loss (SN) for various soils (D, R and M), erosion treatments and sampling times. Solid line indicates linear regression and dotted lines the 95% confidence intervals.

and erosion treatments was similar to the pattern of soil losses (Fig. 2). The lack of sensitivity of N loss to N concentration in soils could be due to two reasons. First, the sediment is dominated by aggregates of a certain aggregate size (Figs. 3–5) that is related to the soils’ susceptibility to wetting and dispersion (Teixeira and Misra, 1997). Second, the N concentration of the dominant sized aggregate in the sediment may be quite different from the average N concentration of the whole sediment. In general, the soil with the most stable structure and highest resistance to dispersion (soil R) showed little erosion and N loss; and the soil with the highest susceptibility to dispersion (soil M) showed the most erosion and N loss (Figs. 2 and 7). Rainfall-driven erosion processes (e.g. rainfall detachment) tend to dominate at gentle slopes and runoff-driven processes (e.g. runoff entrainment) at steep slopes (Misra and Rose, 1995). It is possible that micro-scale shear stresses developed on aggregates at gentle slope is more effective than macro-scale shear stresses (of lower magnitude) at steep slope in raising the values of CR and ER of N in sediment. For soils with a more stable structure, the increased concentration of N-rich aggregates in the sediment probably

214

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

Fig. 13. Variation in N concentration ratio (CR) with size of aggregates in sediment derived from three soils, and with different erosion treatments (T1–T4, where T1: low rainfall KE and gentle slope (28), T2: high rainfall KE and gentle slope (28), T3: high rainfall KE and steep slope (168) without drying cycle, and T4: high rainfall KE and steep slope (168) with a drying cycle) and time of sediment sampling (minutes) during erosion. Horizontal dashed line represents CR = 1.

results from the presence of plant residues and an Nrich outer layer removed from larger aggregates by raindrop stripping (Ghadiri and Rose, 1991a,b). In previous studies, raindrop stripping phenomenon has not been evaluated on soils of a wide range of textures and structures. Indeed, data on the distribution of N in various size fractions of the aggregates in the uneroded soils is limited in the literature. In most studies, nutrient concentration has been shown to decrease with increasing aggregate size in clay soils with relatively stable structure (Ghadiri and Rose,

1991a,b; Wan and El-Swaify, 1998). However, this was not observed in any of the three soils we studied (Figs. 1 and 6). Previous studies have shown that large, waterstable aggregates have a surface coating that is rich in organic matter (and organic forms of nutrients) and an inner core that is poor in nutrients because nutrients diffuse slowly into aggregates from surface towards the centre. This theory of nutrient distribution in aggregates supports that CR of sediment would be expected to decrease with increase in aggregate size with CR < 1 for coarsely aggregated sediment because of N loss from surface through raindrop stripping and CR > 1 for finely aggregated sediment that is generated from the rich surface coating of water stable aggregates. These hypotheses relating to aggregation and N distribution do not take slaking and dispersion of soil into account. Aggregate size distribution and N distribution with aggregate size in dispersive soils are likely to fluctuate under natural wetting and drying cycles. An increase in the CR of sediment with aggregate size observed in these soils (e.g. soil M in Fig. 13) might be due to the presence of plant residues in the sediment. For the other two soils of moderate susceptibility to dispersion (D and R), evidence of raindrop stripping was also limited, again implying the presence of N-rich fragments in sediment consisting of partly decomposed plant residue and particulate organic matter. This agrees with observations that losses of nutrients and organic matter are greater where rainfall erosion dominates runoff erosion (giving low erosion rates) than where runoff predominates and erosion rate is high (Stoltenberg and White, 1953; Rose and Dalal, 1988). 4.2. Prediction of enrichment ratio and N loss Nitrogen and other nutrients lost through erosion have substantial on-site impacts affecting fertility and productivity of soil and off-site impacts through pollution of waterways. Prediction of nutrient loss through erosion is achieved using Eq. (3), for example, with the CREAMS model (Knisel, 1980), which can predict enrichment ratio of both nitrogen and phosphorus by erosion. The regression parameters of Eq. (3), calculated for the data available in this study, were u = 2.6 0.55 and m =
0.13 0.09 (p < 0.05), which were within the ranges, u = 2 1, and m =
0.2 0.1, suggested by Menzel (1980). However,

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

215

Table 3 Parameters u and m of Eq. (3) calculated from the published data (with standard errors in parentheses) Source

u

m

n

r2

McIsaac et al. (1991) Catt et al. (1994) Sombatpanit et al. (1995) Sharpley (1980) Ghadiri and Rose (1991b)a Ghadiri and Rose (1991b)

0.26 (0.091)
0.77 (0.289) 3.69 (0.403) 2.48 0.9 1.8


0.01 (0.012) 0.20 (0.044)
0.29 (0.051)
0.27
0.05
0.21

34 12 10 – – –

0.05 (NS) 0.68 (p = 0.001) 0.81 (p < 0.001) – – –

The size of data set (n) and the coefficient of determination (r2) of the regression equation are indicated. Parameter values for some data sets without statistics are also shown as reported; NS is not significant at p < 0.05. a For experiments dominated by rainfall detachment.

Fig. 9 shows that prediction of ER0 with the parameters u and m of Menzel, did not agree well with the observed ER. The results suggest that small differences in the parameters can produce large differences between estimated and observed values of ER. The parameters of Eq. (3) calculated for the data of other authors in Table 3 differed significantly from the values suggested by Menzel (1980), although some data of Sharpley (1980) and Ghadiri and Rose (1991b) showed these parameters to be in reasonable agreement with Menzel (1980). These conflicting results raise doubts over wide applicability of Menzel’s equation to all erosion situations. Although the prediction of ER by the method of Menzel (1980) was poor in our erosion experiments, predicted values of N loss using the inaccurate estimates of ER agreed reasonably with the observed N loss (Fig. 10). Thus, ER need not be estimated accurately to predict N loss quite well. This is because the variation in sediment loss (SL) is usually much greater than the variation in N concentration of sediment (NS) that is required for the calculation of N loss. As SL has a stronger influence than NS in predicting N loss, the variation of N loss (SN) for various soils and erosion treatments in Fig. 7 was similar to the variation of SL (Fig. 2). The linear relationship between N loss and sediment loss (Eq. (4) and Fig. 11) based on the data for various erosion experiments (Table 2) implies that a N concentration (NS) of 0.0016 kg kg
1 (0.16% N) may be used for a wide range of soils and erosion conditions to obtain a good estimate of N loss. The magnitude of a2 and the type of regression in Eq. (4) were similar to those reported by McIsaac et al. (1991), Gachene et al. (1997) and Hargrave and Shaykewich (1997). Thus, loss of N by erosion could

be reasonably estimated from knowledge of sediment loss alone. The value of 0.16% of N is within the range of N concentration reported for the cultivated layer of most soils by Bremner (1965) (0.08–0.4%) and Foth and Ellis (1988) (0.1–0.2%). The relatively small variation in N concentration in the surface layer of most soils suggests that it is reasonable to use a constant value of N concentration for the sediment, so that measured or predicted sediment loss can be used to estimate N loss by erosion. 4.3. Implications for N management of forest soils N is a key nutrient element maintaining growth of trees in forests and plantations and is often applied in plantations as fertilizer. Much of the soil management efforts in forestry are directed towards reducing N loss through leaching and volatilization and increasing availability through mineralization. In agroecosystembased studies, it has been shown that the loss of N by erosion poses the greatest threat to decline in soil fertility (e.g. Rose and Dalal, 1988; White, 1986). For forest ecosystems and plantations, such information is limited. The loss of N by erosion presented here probably gives an upper limit for N loss by erosion from Australian forest soils. This is because the soils used for the erosion studies were temperate forest soils with high organic-N concentrations and the simulated erosion events used represented severe storm events (average rainfall rate of 116 mm h
1) on bare soils. The loss of N in 5 min of erosion ranged from 1 to 15 kg ha
1 depending on the soil and erosion treatments used (Fig. 7) and was similar to the rate of N loss reported by Palis et al. (1990). The highest value was similar to N uptake by 10 month-old

216

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217

Eucalyptus trees in a plantation (11.9 1.8 kg N ha
1 per year) reported by Misra et al. (1998) and close to the lowest rate of net N mineralization in 1–2 year-old plantations in Tasmania (18–91 kg ha
1 per year) reported by Wang et al. (1998). At first, these comparisons may seem unrealistic because our erosion plots were small (0.8 m2). However, most measurements of N uptake and mineralization are also based on small areas; for example, Wang et al. (1998) used 0.05 m diameter tubes for N mineralization studies and Misra et al. (1998) measured N uptake by 6 trees only, so the comparisons are probably realistic. These results suggest that losses of N from forest soils by erosion are as significant as those reported for agricultural landscapes.

5. Conclusions Among the soils studied and erosion treatments imposed, N loss by erosion was high for those soils (D and M) and erosion treatments (3 and 4) where the sediment loss was high. Evidence for N enrichment of sediment (ER) through raindrop stripping was limited because there was no consistent relationship between concentration ratio (CR) and the size of sediment. Nevertheless, both ER and CR were generally >1 indicating the ease with which erosion removes the N-rich component of the soil. Although predictions of ER using the method of Menzel (1980) did not agree with the measured ER, predicted N loss using ER obtained by Menzel’s method agreed reasonably with measured N loss. The similarity in the variation between sediment loss and N loss of various soils and erosion treatments suggests that sediment loss can be used to estimate N loss. A linear regression between N loss and sediment loss obtained with data from the literature predicted N loss well for the soils and erosion treatments used in this study. However, as the method of predicting N loss proposed is empirical, it needs further testing.

Acknowledgement We thank an anonymous reviewer for meticulous comments to improve this manuscript.

References Alberts, E., Moldenhauer, W.C., 1981. Nitrogen and phosphorus transported by eroded soil aggregates. Soil Sci. Soc. Am. J. 45, 391–396. Barrows, H.L., Kilmer, V.J., 1963. Plant nutrient losses from soils by water erosion. Adv. Agron. 15, 303–316. Bedell, G.D., Kohnke, H., Hickok, R.B., 1946. The effects of two farming systems on erosion from cropland. Soil Sci. Soc. Proc. 11, 522–526. Bremner, J.M., 1965. Organic nitrogen in soils. In: Bartholomew, W.V., Clark, F.E. (Eds.), Soil Nitrogen. Agronomy, vol. 10. American Society of Agronomy, Madison, pp. 93–150. Catt, J.A., Quinton, J.N., Rickson, R.J., Styles, P., 1994. Nutrient losses and crop yields in the Woburn erosion reference experiment. In: Rickson, R.J. (Ed.), Conserving Soil Resources: European Perspective.. CAB International, Oxford, pp. 94–104. DeBano, L.F., Conrad, C.E., 1976. Nutrients lost in debris and runoff water from a burned chaparral watershed. In: Proceedings of the Third Federal Inter-Agency Sedimentation Conference, Denver, Colorado, Water Resources Council, USA, pp. 3.13–3.27. Flanagan, D.C., Foster, G.R., 1989. Storm pattern effect on nitrogen and phosphorus losses in surface runoff. Trans. ASAE 32, 535– 544. Foth, H.D., Ellis, B.G., 1988. Soil Fertility. Wiley, New York. Gachene, C.K.K., Jarvis, N.J., Linner, H., Mbuvi, J.P., 1997. Soil erosion effects on soil properties in a highland area of Central Kenya. Soil Sci. Soc. Am. J. 61, 559–564. Ghadiri, H., Rose, C.W., 1991a. Sorbed chemical transport in overland flow I. A nutrient and pesticide enrichment mechanism. J. Environ. Qual. 20, 628–633. Ghadiri, H., Rose, C.W., 1991b. Sorbed chemical transport in overland flow II. Enrichment ratio variation with erosion processes. J. Environ. Qual. 20, 634–641. Grierson, I.T., Oades, J.M., 1977. A rainfall simulator for field studies of run-off and soil erosion. J. Agric. Eng. Res. 22, 37–44. Hansen, A.C., Nielsen, J.D., 1995. Runoff and loss of soil and nutrients. In: Correll, A. (Ed.), Surface Runoff Erosion and Loss of Phosphorus at Two Agricultural Soils in Denmark. (SP report number 14, Danish Institute of Plant and Soil Science Tjele), pp. 149–188. Hargrave, A.P., Shaykewich, C.F., 1997. Rainfall induced nitrogen and phosphorus losses from Manitoba soils. Can. J. Soil Sci. 77, 59–65. Isbell, R.F., 1996. The Australian Soil Classification. CSIRO Publishing, Collingwood, Australia. Knisel, W.G., 1980. CREAMS: A Field Scale Model for Chemicals, Runoff, and Erosion from Agricultural Management Systems. Conservation Research Report no. 26, USDA. Lane, L.J., Nearing, M.A., 1989. USDA–Water Erosion Prediction Project: Hillslope Profile Model Documentation. National Soil Erosion Research Laboratory, West Lafayette, Indiana (NSERL Report number 2, USDA-ARS). Loch, R.J., Donnollan, T.E., 1982. Field rainfall simulator studies on two clay soils of the Darling Downs, Queensland. II. Aggregate breakdown, sediment properties and soil erodibility. Aust. J. Soil Res. 21, 47–58.

P.C. Teixeira, R.K. Misra / Soil & Tillage Research 83 (2005) 204–217 Massey, H.F., Jackson, M.L., 1952. Selective erosion of soil fertility constituents. Soil Sci. Soc. Proc. 16, 353–356. McDowell, L.L., McGregor, K.C., 1980. Nitrogen and phosphorus losses in runoff from no-till soybeans. Trans. ASAE 23, 643–648. McIsaac, G.F., Hirschi, M.C., Mitchell, J.K., 1989. Nitrogen and phosphorus in eroded sediment from corn and soybean tillage systems. In: Sediment and the Environment, Proceedings of the Baltimore Symposium, May 1989. IAHS Publication number 184, pp 3–10. McIsaac, G.F., Hirschi, M.C., Mitchell, J.K., 1991. Nitrogen and phosphorus in eroded sediment from corn and soybean tillage systems. J. Environ. Qual. 20, 663–670. Menzel, R.G., 1980. Enrichment ratios for water quality modeling. In: Knisel, W.G. (Ed.), CREAMS: A Field Scale Model for Chemicals, Runoff, and Erosion from Agricultural Management Systems. Conservation Research Report no. 26, USDA, pp. 486– 492. Misra, R.K., Rose, C.W., 1995. An examination of the relationship between erodibility parameters and soil strength. Aust. J. Soil Res. 33, 715–732. Misra, R.K., Rose, C.W., 1996. Application and sensitivity analysis of process-based erosion model GUEST. Eur. J. Soil Sci. 47, 593–604. Misra, R.K., Turnbull, C.R.A., Cromer, R.N., Gibbons, A.K., LaSala, A.V., 1998. Below- and above-ground growth of Eucalyptus nitens in a young plantation. II Nitrogen and phosphorus. Forest Ecol. Manage. 106, 295–305. Palis, R.G., Okwach, G., Rose, C.W., Saffigna, P.G., 1990. Soil erosion processes and nutrient loss. I. The interpretation of enrichment ratio and nitrogen loss in runoff sediment. Aust. J. Soil Res. 28, 623–639. Proffitt, A.P.B., Hairsine, P.B., Rose, C.W., 1993. Modeling soil erosion by overland flow: application over a range of hydraulic conditions. Trans. ASAE. 36, 1743–1753. Rayment, G.E., Higginson, F.R., 1992. Australian Laboratory Handbook of Soil and Water Chemical Methods. Inkata Press, Melbourne.

217

Renard, K.G., Foster, G.R., Weesies, G.A., Porter, J.P., 1991. RUSLE: Revised Universal Soil Loss Equation. J. Soil Water Cons. 46, 30–33. Richardson, C.W., King, K.W., 1995. Erosion and nutrient losses from zero tillage on a clay soil. J. Agric. Eng. Res. 61, 81–86. Rose, C.W., Dalal, R.C., 1988. Erosion and runoff of nitrogen. In: Wilson, J.R. (Ed.), Advances in Nitrogen Cycling in Agricultural Ecosystems. CAB International, Wellington, pp. 212–235. Sharpley, A.N., 1980. The enrichment of soil phosphorus in runoff sediments. J. Environ. Qual. 9, 521–526. Sharpley, A.N., 1985. The selective erosion of plant nutrients in runoff. Soil Sci. Soc. Am. J. 49, 1527–1534. Sombatpanit, S., Rose, C.W., Ciesiolka, C.A., Coughlan, K.J., 1995. Soil and nutrient loss under rozelle (Hibiscus subdariffa L. var. altissima) at Khon Kaen, Thailand. Soil Technol. 8, 235–241. Stoltenberg, N.L., White, J.L., 1953. Selective loss of plant nutrients by erosion. Soil Sci. Soc. Proc. 17, 406–410. Teixeira, P.C., Misra, R.K., 1997. Erosion and sediment characteristics of cultivated forest soils as affected by the mechanical stability of aggregates. Catena 30, 119–134. Torbert, H.A., Potter, K.N., Morrison Jr., J.E., 1996. Management effects on nitrogen and phosphorus losses in runoff on expansive clay soils. Trans. ASAE 39, 161–166. Wan, Y., El-Swaify, S.A., 1998. Sediment enrichment mechanisms of organic carbon and phosphorus in a well-aggregated Oxisol. J. Environ. Qual. 27, 132–138. Wang, X.J., Smethurst, P.J., Holz, G.K., 1998. Nitrogen fluxes in surface soils of 1–2-year-old eucalypt plantations in Tasmania. Aust. J. Soil Res. 36, 17–29. White, P.J., 1986. A review of soil erosion and agricultural productivity with particular reference to grain crop production in Queensland. J. Aust. Inst. Agric. Sci. 52, 12–22. Wischmeier, W.H., Smith, D.D., 1978. Predicting rainfall erosion losses: a guide to conservation planning, agricultural Handbook, number 537, USDA, Washington.