Aust. J. Soil Res., 1984, 22, 413-31 Identification of Dispersive Behaviour and the Management of Red-brown Earths P. RengasamyA, R. S. B. GreeneB, G. W. Fordc, and A . H. MehanniB State Rivers and Water Supply Commission, Irrigation Research Institute, Tatura, Vic. 3616. Irrigation Research Institute, Tatura, Vic. 3616. CVictorian Crops Research Institute, Private Bag 260, Horsham, Vic. 3400. A Abstract A scheme is proposed, together with a procedure suitable for routine laboratory use, for the prediction of dispersive behaviour of surface layers of red-brown earths and their classification into one of six classes. Each class is defined on the basis of predictive relationships established between dispersion (spontaneous and mechanical), sodium adsorption ratio (SAR) and total cation concentration (TCC). These relationships were established experimentally using 138 samples representing both surface and subsurface layers from 69 red-brown earth profiles. Preliminary studies including samples from red clay and black earth profiles indicated that the proposed scheme is not suitable for these soils. Neither can it be used for soils containing free lime. The procedure proposed enables the prediction of the probable dispersive behaviour of the surface layer of red-brown earths, including exposed subsoils. It provides a rational basis for the formulation of appropriate management strategies for the manipulation of the surface structure of individual redbrown earths used for dryland or irrigated agriculture. Application of the proposed scheme to the estimation of the minimum level of residual gypsum required to maintain aggregate stability via the electrolyte effect is discussed, with special reference to low-sodic soils (i.e. with a SAR below 3, e.g. Classes 2a and 3c). Introduction Many soils currently used for irrigated or dryland agriculture are difficult to manage owing to their tendency to develop unsatisfactory structure, particularly in their surface and near-surface layers. Breakdown of soil aggregates leads to problems such as surface crusting, reduced water infiltration and restricted plant establishment and growth. Breakdown occurs due to slaking, and also in many soils due to dispersion of aggregates when wet by irrigation or rain. It is well known that aggregates susceptible to breakdown by these processes are sensitive to mechanical stress, particularly when moist. Thus, raindrop impact, cultivation or trampling by stock may aggravate the development of compacted layers in the upper portion of the soil profile, whilst the presence of a surface mulch or the use of reduced tillage techniques may minimize such problems. Although dispersion in surface aggregates is emphasized in this paper, it is pertinent to note that dispersion in the subsoil may impede water flow through preferential pathways, and may be important when subsoil becomes exposed through landforming. Dispersion of clay in soils is influenced by both the nature of exchangeable cations and the amount of electrolyte present (Quirk and Schofield 1955; Shainberg et al. 1981). A simple dispersion test developed by Emerson (1967), and modified 414 P. Rengasamy et al. by Loveday and Pyle (1973), is widely used to assess soil stability. These tests use only a few 3-5 mm air-dried aggregates, and no quantitative measurement is made of the exchangeable sodium levels and the electrolyte concentration. Another difficulty encountered with these tests is the variability in dispersion which arises due to differences in cation and electrolyte composition between individual aggregates. Non-uniform distribution of cations within aggregates may result from the phenomenon of salt sieving where salt is trapped within the microfabric of the aggregates (Blackmore 1976). In the above tests, the measured degree of dispersion cannot be related unambiguously to exchangeable sodium or electrolyte effects. However, in the management of dispersive soils, it may be necessary to distinguish between these two factors. For example, Loveday (1981) emphasized the importance of the electrolyte effect when using gypsum in ameliorating hard-setting soils. These problems may lead to incorrect conclusions relating to the gypsum requirement of soils. The probable dispersive behaviour of soils under different conditions may be estimated in various ways. Thus, the measurement of spontaneous dispersion in the absence of any imposed external force will reflect the behaviour of surface soils during the rainfall events when the soil surface is effectively protected by plant material such as stubble, pasture or established field crops. The extent to which soil particles disperse after gentle shaking may be used as a qualitative indication of their field behaviour when bare soil is subjected to raindrop impact. The more severe remoulding test as used by Emerson and others may simulate the effect of mechanical shearing during cultivation or trampling by stock when wet. The erodibility of dispersive clay soils (e.g. evaluation of the risk of pipe erosion of clay dams) may be estimated by inducing high flow rates through recompacted soil materials as in the pinhole test (Sherard et a[. 1976). The aims of the work reported here are: 1. To develop a general scheme for classifying the dispersive behaviour of redbrown earths based on a laboratory procedure suitable for routine use. 2. To define the soil solution parameters associated with the degree of dispersion in red-brown earths and to relate these results to management options. Materials and Methods Soils Used A representative range of red-brown earth profiles (Oades et al. 1981) from various locations in south-eastern Australia was used in the development of a dispersion test. In all, a total of 138 soil samples representing surface (0-15 cm) and subsurface (15-45 cm) layers from 69 red-brown earth profiles was used. Selected physical and chemical characteristics of some of the typical soils are given in Table 1. Thirty-two surface samples of red-brown earths from 16 sites in northern Victoria were used to evaluate the predicability of field response to gypsum by the models developed in our study. Dispersion Test Spontaneous dispersion Soil samples collected from representative locations were air dried, passed through a 2 mm sieve, and thoroughly mixed. Twenty grams were weighed into a 120 ml transparent jar (10 cm high), and 100 ml distilled water slowly added down the sides of the jar, care being taken to avoid disturbance of the soil sample. The mixture was left undisturbed for 12 h (overnight). The dispersed clay adjacent to the soil was then uniformly mixed, care again being taken to avoid disturbing the soil layer. This was Dispersive Behaviour of Red-brown Earths achieved by placing a mechanical stirrer (Dynamax Laboratory Stirrer) mid-way into the suspension and then stirring at a speed of 0.16 rev s-I for 30 s. After an appropriate sedimentation time (Loveday 1974a), the dispersed clay was estimated by pipetting 10 ml of the suspension from a depth of 5 cm. The clay was measured either gravimetrically or spectrophotometrically. The percentage of dispersed clay was expressed on an oven-dried soil basis. In gravimetric estimations, corrections were made for the weight contributed by any dissolved salts. Table 1. Selected chemical and physical properties of some surface (0-15 em) and subsurface (15-45 cm) soils used in this study - Depth (cm) pHA - OrganicB CaC0,B carbon ECA (70) ( 7 0 ) (dS m-I) TCCA (me I-]) SARA ESP TotalB clay (70) Mechanically dispersed clay (70) Lemnos loam (Kyabram) 0-15 15-45 nil 0.01 1.9 0.4 0.08 0.16 0.72 1.73 0.8 3.0 2.3 6.5 Shepparton jine sandy loam (Tatura) 0-15 15-45 nil 0.02 2.6 0.4 0.07 0.31 0.81 3.48 0.9 3.0 2.8 5.4 Goulburn clay (Tongala) 0-15 15-45 nil 0.02 2.1 0.2 0.52 0.75 4.56 6.98 3.7 6.8 7.9 13.3 Stat~sticsfor the surface (68) and subsurface (40) soils 0-15 range mean s.e. 15-45 range mean s.e. - - - - 0.06-1.4 0.48 0.32 0.1-1.3 0.69 0.41 0.3-12.5 3.56 0.36 0.9-12.0 4.86 0.38 0.1-12.4 1.0-21.6 3.1 0.38 3.0-12.4 6.5 0.43 0.42 5.4-22.2 13.8 0.51 6.7 * p H and EC determined in 1:5 soil water suspensions, and SAR and TCC in 1: 5 soil water extracts obtained after shaking. Determined using procedures described by Loveday (1974~). Standard error of the mean. Mechanical dispersion A duplicate sample prepared and treated as above was shaken for 1 h in an end-over-end shaker (0.5 rev s-I). After an appropriate sedimentation time, the dispersed clay was estimated as described above. Analysis of the soil solution composition After measurement of pH and electrical conductivity, the equilibrium solutions from the two preceding dispersions were separated from the soil suspensions. About 25 ml of the suspension was centrifuged for 10 min at 85 rev s-I. The cations Ca2+, Mg2+,K+ and Na+ in the supernatant were estimated by atomic absorption spectrophotometry. The sodium adsorption ratio (SAR) was calculated as ~ a / d { ( ~ Mg)/2) a+ (all cation concentrations in m.e. 1-1). TCC, the total cation concentration, was calculated by summing the concentrations (m.e. 1-I) of the cations N a + , K + , Ca2+ and Mg2+. Exchangeable Cation Analysis Exchangeable cations and cation exchange capacity of the soils were estimated by the method of Tucker (1974). Exchangeable cation percentages were based on the sum of four cations Nat , K + , Ca2+ and Mg2+,as these values were found to be better correlated with the soil physical properties than were the percentages based on CEC (Martin et al. 1964). Saturated Hydraulic Conductivity Measurements of the hydraulic conductivity of saturated soils were made using the falling head method of Klute (1965), with minor modifications as described by Mehanni (1973). P . Rengasamy et al. Residual Gypsum The amount of gypsum present was estimated from determining sulfate concentrations of aqueous 1 M NH,C1 extracts (Greene and Ford 1985). Flocculation Value Twenty g of soil were placed in a 120 ml transparent jar and 100 ml of CaCI, solution (from 0 to 6 m.e. I-') added. The mixture was shaken for 1 h in an end-over-end shaker (0.5 rev s-') and then allowed to stand for 6 h. The minimum concentration of CaCl, solution in which the suspension remained clear was then taken as the best estimate of the flocculation value or CCC (critical coagulation concentration) (Rengasamy and Oades 1977). Statistical Analysis The data were analysed using simple linear regression analysis and multiple linear stepwise regression analysis. Partial correlation coefficients and p-coefficients (standardized regression coefficients) were calculated as described by Nie et a[. (1975). Mitscherlich functions were fitted to the data relating to the residual gypsum, hydraulic conductivity and clay dispersion using the programme GENSTAT V (Rothamsted Experimental Station, U.K.). Results and Discussion During the development of our dispersion test, data on spontaneously dispersed clay, mechanically dispersed clay, SAR and TCC in 1 : 5 soil-water extracts were obtained for a range of soil profiles including 69 red-brown earths, 19 red clays and 7 black earths. The amount of dispersed clay in these samples was not significantly correlated with either SAR or TCC. However, when only data from the 108 redbrown earth samples which dispersed were considered, the amount of mechanically dispersed clay was found to be significantly correlated with both SAR and TCC (Table 2). As both the proportion of the soil clay fraction which disperses in water, and the rate at which dispersion occurs, vary with the clay mineralogy and the SAR and TCC levels in the soil-water suspension (Velasco-Molina et al. 1971), we decided to restrict our test, and the derivation of quantitative inter-relationships between the various soil parameters studied, to groups of soils having similar mineralogical composition. Red-brown earths are known to have a similar mineralogy in that their clay fraction is predominantly illite (Oades et al. 1981), and so are susceptible to dispersion even when weak mechanical forces are applied (Emerson 1983; Rengasamy 1983). Thus, we restricted subsequent statistical analysis of the results reported in this paper to samples of red-brown earths. Mechanical Dispersion Of the 138 samples from red-brown earth profiles, 30 samples showed no dispersion after mechanical shaking, while the remaining 108 samples dispersed. The difference in the dispersive behaviour between surface and subsoils is clearly brought out by the improvement in the various correlation coefficients obtained when the soils were separated into surface (0-15 cm) and subsurface (15-45 cm) samples (Table 2). The correlation coefficient was increased from r = 0 - 6 9 to R = 0.79 by including TCC with SAR. This was confirmed by stepwise regression analysis; the partial correlation coefficients (Table 2) indicate that SAR has a positive, and TCC a negative, influence on dispersion. The /3-coefficients (standardized regression coefficients) indicate that in surface soils SAR is 1.3 times more important than TCC in influencing dispersion, while in subsoils SAR is 3.6 times more important (equations 2.3 and 2.2, Table 2). 417 Dispersive Behaviour of Red-brown Earths For the 138 red-brown earth samples investigated, the SAR of the 1 : 5 extract obtained after shaking and the ESP of the soils were related as follows: ESP = 1.95SAR + 1.8 (R2=0.82). (1) This indicates that ESP is approximately twice the numerical value of the SAR (1 : 5 extract). Table 2. Correlation of SAR and TCC with per cent mechanically dispersed clay in some red-brown earths Data analysed by stepwise multiple linear regression using the model Y = a x , + bX, + c, where Y = % mechanically dispersed clay, and X I = SAR and X, = TCC of 1:5 w/v extract after shaking for 1 hour Multiple correlation coefficient Simple correlation coefficients '5 'XI RFXlX2 Regression equation with partial correlation coefficients [ ] and p-coefficients ( ) 2.1 All soils pooled (108 samples) Y 4 0.69** 0.37** 0.79** 0.85** 2.2 Subsoils, 15-45 cm (40 samples) Y 0.81** 0.47** 4 0.83** 0.74** 2.3 Surface soils, 0-15 cm (68 samples) Y 0.63** 0.31** 0.92** xi 0.92** 2.4 Surface soils with a SAR > 3 (28 samples) Y 0.45* - 0.02n.s. 0.95** x~ 0.87** 2.5 Surface soils with a SAR Y 0.35* -0.46** 4 < 3 (40 samples) 0.76** 0.44** ** P 6 0.01. * P < 0.05. n.s., not significant. The critical value for ESP above which a soil is sodic is controversial. For example, Northcote and Skene (1972) defined a soil as sodic if the ESP is 6 or more. However, this critical value may vary with soil type and with the basis on which the ESP is calculated, largely because of confusion relating to the use of CEC or the sum of the four major exchangeable cations (i.e. Na+ + K + + Ca2++ Mg2+)as the divisor. Moreover, because of the considerable experimental problems involved in an accurate determination of exchangeable cations, SAR of the soil extract is generally preferred as an index for sodic soils (Rhoades 1982). In our study, the surface soils with a SAR of less than 3 deviated from the general trend that SAR influences clay dispersion more than TCC (Table 2). The importance of TCC in controlling the dispersive behaviour of these soils is indicated by the higher simple correlation coefficient for TCC obtained when the surface soils with a SAR of less 418 P. Rengasarny et a[. than 3 were considered separately (equation 2.5, Table 2). The 0-coefficients show only a slightly greater influence of TCC. The importance of electrolyte concentration in soils with low levels of exchangeable sodium (i.e. ESP less than 6) in controlling dispersion, crusting and infiltration has been reported by many workers (e.g. Oster and Schroer 1979; Kazman et al. 1983). Further, it is known that in the absence of electrolytes, even calcium-saturated red-brown earths disperse due to weak mechanical forces (Rengasamy 1983). In our study, surface soils with a SAR > 3 would disperse spontaneously, whereas those with a SAR < 3 dispersed only after mechanical shaking. Therefore it seems appropriate to regard soils with SAR < 3 different from the soils with SAR > 3. Because sodium levels in soils with SAR c 3 can still influence dispersion, these soils can be considered as low-sodic instead of non-sodic. For surface soils with a SAR above 3, TCC is not significantly correlated with dispersed clay (Table 2). However, multiple linear regression analysis indicated that both SAR and TCC influence dispersion in these soils almost equally, as is shown by the @-coefficients of equation 2.4 (Table 2). Table 3. Correlation of SAR and TCC with per cent spontaneously dispersed clay in some red-brown earths Data analysed by stepwise multiple linear regression using the model Y = a x , + bX2 + c, where Y = % spontaneously dispersed clay, and X I = SAR of 1:5 w/v extract after shaking for 1 h and X 2 = TCC of undisturbed 1:5 w/v equilibrum solution Simple correlation coefficients rxl Multiple correlation coefficient 'xz - RY.XIx2 - - Regression equation with partial correlation coefficients [ ] and 6-coefficients ( ) - 3.1 AN soils pooled (28 samples) Y 0.43* - 0.14 n s . 4 0.82** 0.97** 3.2 Subsoils, 15-45 cm (18 samples) Y 0.49* - 0.16 n.s. 4 0.78** 0.99** 3.3 Surface soils, 0-15 cm (I0samples) Y 0.66** 0.39 n s . 0.90** XI 0.93 n s . ** P < 0.01. * P < 0.05. n.s., not significant. Spontaneous Dispersion Only 28 of the 138 soils tested dispersed spontaneously in the absence of mechanical stress. These soils had a SAR above 3. Multiple regression analysis of data from soils which spontaneously dispersed gave the results presented in Table 3. Simple correlation coefficients indicated that the percentage of spontaneously dispersed clay was not significantly correlated with either the SAR or the TCC of the undisturbed equilibrium solution. However, the percentage of spontaneously dispersed clay was significantly (r = 0.43, P < 0.05) correlated with the SAR of the extract obtained after shaking for 1 h. 419 Dispersive Behaviour of Red-brown Earths Multiple linear regression analysls showed that the correlation between percentage of clay and the SAR and TCC values measured in the extract obtained after 1 h shaking was low (R2 = 0.43, P < 0.05). However, when the SAR values measured after 1 h shaking were used, together with the TCC values measured in the undisturbed solution, the multiple correlation coefficient was higher (R2 = 0.94, P < 0.01). In our study the TCC of the undisturbed solution was very low compared to the TCC in solutions obtained after shaking. Probably the reason for this difference is the rate of diffusion of salts from the soil aggregates to the soil solution. The partial correlation coefficients and the &coefficients (equation 3.1, Table 3) suggest that the SAR (measured after 1 h shaking) and the TCC (measured in the undisturbed equilibrium solution) influence spontaneous dispersion to a similar extent. As these SAR values are correlated with the ESP of the soil (equation I), spontaneous dispersion may be affected by both exchangeable sodium levels and the rate of diffusion of salts from soil aggregates to the soil solution. The slow diffusion of salts would contribute to the marked variability in dispersion observed when large aggregates are used in the Emerson dispersion test. In contrast to soils which dispersed when subjected to mechanical stress, there was no further improvement in the prediction of dispersion by separating the spontaneously dispersed soils into surface v. subsoils (Table 3). The Influence of SAR and TCC on the State of Flocculation Using regression equation 3.1 (Table 3) obtained for spontaneously dispersed soils, i.e. Clay % = 0.49SAR-3.12TCC + 0.44 (2) and solving for nil dispersion, the relation between SAR and TCC which defines the state of flocculation can be expressed as TCC=O.l6SAR + 0.14, (3) where TCC is measured in undisturbed solutions and SAR is measured after shaking. In a similar way, the state of flocculation for the mechanically dispersed soils can be defined by: TCC = 1.46SAR + 1.44, (4) where both TCC and SAR are measured in solutions obtained after shaking. Equation 3 indicates that spontaneous dispersion occurs only when the diffused electrolyte level is below (O.16SAR + 0.14) m.e. I-'. This level is much lower than the minimum amount of electrolyte required to prevent mechanical dispersion (see equation 4). This supports the observation made by Rowel1 et al. (1969) that the type of treatment used to estimate the extent of clay dispersion could affect the apparent relationship between SAR (or ESP) and TCC corresponding to the state of flocculation or dispersion. They concluded that the application of mechanical stress to a clay could affect the electrolyte concentration at which dispersion occurred. By plotting the values of SAR and TCC for the 138 soils (Fig. I), measured in extracts obtained after shaking, we found that the line representing the equation (4) separated flocculated from dispersed soils. Only three of the 138 soils deviated P. Rengasamy et al. Dispersed Samples Total cation concentration (me. 1-l) Fig. 1. Dispersive behaviour of soil samples in relation to SAR and TCC measured in 1:5 soil-water extracts. ( O s t o r ot a l 1 9 8 0 ) Y P e r m o a b , l ~ t y( R h o a d e s 1 9 8 2 ) dlspersfon ( a u t h o r s M Spontaneous , 0 10 20 30 Total cation concentration ( m e . 1-1) . data) , , I 40 Fig. 2. Relationships between ESP and TCC defining a stable condition obtained in various studies. from this relationship. However, improved relationships could be obtained by separating the soils into different groups on the basis of their multiple linear correlation coefficients. The following relationships were obtained for different groups: TCC = 3.19SAR - 1 . 7 (subsoils) (5) TCC = 1.21SAR + 3.3 (surface soils) (6) Dispersive Behaviour of Red-brown Earths TCC = 1 18SAR TCC + 1-20SAR + + 42 1 3.1 (surface soils with an SAR > 3) (7) 3.9 (surface soils with an SAR < 3). (8) Equations 7 and 8 are not found to be significantly different from equation 6. Hence equation 6 can be used for all surface soils. The data obtained in soil permeability experiments, dispersion or coagulation studies on pure clays, and soil samples by different workers (Rowel1 et al. 1969; Quirk 1971; Arora and Coleman 1979; Oster et al. 1980; Rhoades 1982) were used to obtain the relationship between ESP and TCC which will define a stable soil condition, viz. unaffected permeability or a flocculated state. These lines are drawn in Figure 2, together with the lines obtained in our study for surface and subsurface soils. The SAR data of the 1: 5 extracts obtained after shaking were converted to ESP values using equation 1. Comparison on these different lines indicates that the relationship between ESP and TCC varies with the type of clay mineral, the amount of total clay, organic matter and the mechanical forces. Many other factors, not encountered in our study, may influence the degree of dispersion, e.g. iron and aluminium oxides, exchangeable aluminium or pH, presence of CaCO,, possibly exchangeable Ca:Mg ratios (Emerson 1983), organic anions acting as peptizing agents (Shanmuganathan and Oades 1983), absence of clay domain formation (Hardcastle and Mitchell 1976), the strength of edge-to-face attractions (Greene ef al. 1978), and the severity of drying (Collis-George and Smiles 1963). Practical Signi3cance of Spontaneous and Mechanical Dispersion The importance of mechanical stress in influencing dispersion has been clearly brought out in this study. As discussed earlier, red-brown earths in Australia are dominantly illitic, and so are susceptible to dispersion even when weak mechanical forces are applied. The practical significance of the equations obtained for spontaneous and mechanical dispersion is that they probably represent the two extreme types of disturbance experienced by a soil in the field, viz. zero tillage v. intensive cultivation. They may also represent the effects of raindrop impact when the soil surface is completely covered by plant material compared to that on bare soil. Some indication of the effects of these factors in the field is provided by observations made during a rainfall event (2.5 cm in 1 h) on 6 May, 1983 at the Irrigation Research Institute, Tatura. Pasture and bare cultivated plots, previously irrigated with saline waters of different total soluble salts, had different levels of ponded water and different turbidity of the undrained water. The pasture plots, with complete plant cover of the soil surface, were mainly well drained, although a few sodic plots had slightly turbid ponded water. In contrast, the water remained undrained on all the bare plots cultivated for cropping. In these plots, the amount of ponded water, and its turbidity, varied with the level of exchangeable sodium and the electrical conductivity of the soils. The water samples from these plots were collected and analysed for the amount of dispersed clay, SAR and TCC. It was found that the SAR-TCC relationship for the turbid samples from the bare cultivated plots correspond to the line derived for mechanical dispersion, and that the corresponding relationship for the samples from pasture plots corresponded to the line for spontaneous dispersion (Fig. 3). These results P. Rengasamy et al. 422 support the conclusion that the mechanical dispersion line represents the effect of raindrop impact on a bare soil, and the spontaneous dispersion line represents the effect of minimum mechanical disturbance, e.g. when the soil surface is protected by plants. The mechanical dispersion test described in this paper probably identifies the effect of weak to moderate mechanical stress, such as that caused by the flow of water across irrigation bays or rainfall of the intensity commonly experienced in south-eastern Australia. However, our test may not be suitable for identifying the effect of strong mechanical forces on aggregate stability, e.g. during cultivation of wet soil or during hydrodynamic flow in tunnel erosion. Probably tests such as severe remoulding when wet (Emerson 1967), or the pin-hole test (Sherard et al. 1976), may be more appropriate for such purposes. 0 I a / T u r b ~ dw a t e r j= Pasture plots 0 - Bare cult~vated Total cation concentration (m.e. 1-') Fig. 3. The effect of rainfall (2.5 cm in 1 h) on the stability of surface soils from pasture and bare cultivated plots at Tatura as related to the SAR and TCC of the ponded water. ClassiJication of Soils on the Basis of Dispersion and Soil Solution Composition The various relationships between SAR and TCC defining the susceptibility of soils to spontaneous and/or mechanical dispersion can be used to predict the probable behaviour of the surface layers of red-brown earths. The following classes, based upon the relationships derived in this study, characterize the surface soils which may disperse or flocculate under different electrolyte environments. It is important to note that the relationship derived for subsoils is applicable only when they are subjected to mechanical stress under a water regime similar to that of surface layers, e.g. when exposed after landforming. The key relationships for surface soils are diagramatically shown in Fig. 4. Details of the characteristics of each class are discussed below. Class I - dispersive soils Soils which disperse spontaneously and which have a TCC < (O.16SAR + 0 - 14) in the equilibrium solution obtained without shaking, will have severe problems Dispersive Behaviour of Red-bra-~n Earths associated with crusting, reduced porosity etc., even when subjected to minimum mechanical stress, e.g. under zero tillage. Such soils under pasture or with crop cover are highly dispersive. Red-brown earths with a SAR below 3 were not found to disperse spontaneously, as the cation concentrations in their equilibrium solutions were always above 0.6 m.e. 1- I. Soils which spontaneously disperse under field conditions will be non-saline but sodic. The stability of such soils is largely controlled by exchangeable sodium. Hence, amelioration with calcium compounds should aim to reduce sodium levels in the exchange complex of the soil, and also to maintain enough electrolyte to keep the clay flocculated. Class 3a Class 3b 0 2 4 6 8 10 12 14 16 Total cation concentration (m.e. 1-') Fig. 4. A classification scheme for the prediction of dispersive behaviour of A-horizon of red-brown earths. Class 2 - potentially dispersive soils Soils which disperse after mechanical shaking are potentially dispersive. Such soils are unstable if mechanically disturbed, e.g. by intensive cultivation or raindrop impact. The electrolyte concentration required to keep these soils flocculated varies with their SAR and their original profile position, i.e. A or B-horizon. Thus, three categories are possible. Class 2a Soils from the A-horizon of red-brown earths with a SAR of less than 3 and which mechanically disperse, require an electrolyte concentration of (1.21SAR + 3.3) m.e. 1- for structural stability. The small difference in correlation coefficients (equation 2.5, Table 2) may not warrant any separation of the soils with SAR less than 3 into a distinct class. However, in terms of management, reducing the exchangeable sodium levels to zero in these low-sodic soils will be inefficient (Loveday 1981), and probably impractical. The buffering effects due to mineral weathering and hydrolysis make it impossible to obtain a zero sodic soil (Rengasamy 1983). P. Rengasamy et al. In our study, even the samples with a SAR as low as 0.01 have been found to be potentially dispersive. Hence it appears that in these low-sodic soils, maintaining adequate electrolyte levels is the most practical management option. Problems of surface crusting and/or reduced surface permeability are likely to occur when rainfall leaches the soluble salts below the threshold electrolyte level, or when lowsalt waters are used for irrigation. Periodic surface application of an amendment, such as gypsum, should be used in such circumstances to avoid this potential problem (Rhoades 1982). Field studies of non-irrigated soils of this type in northcentral Victoria showed that the applied gypsum is leached at the rate of 1 t ha-l per 125 mm to 360 mm of rainfall, depending on the rate of gypsum application (Greene and Ford 1985). Hence, a periodic application of gypsum is necessary to maintain surface structure. Class 2a soils will have few structural problems if managed using minimum tillage techniques or if maintained under continuous pasture growth. Many (40) of the red-brown earths used in our study fall into this category. Despite their low sodium content, these soils when cultivated often crust following exposure to rainstorms or to flood irrigation. Similar soils are common throughout the agricultural areas of south-eastern Australia, and treatment with gypsum at the rate of 1 to 5 t ha-l has been reported as being beneficial (Sims and Rooney 1965; Matheson 1969). Class 2b Surface (A-horizon) soils with a SAR above 3 require an electrolyte level similar to class 2a soils in order to maintain flocculation. However, the addition of gypsum should aim to reduce exchangeable sodium levels and also to provide sufficient electrolyte. Reduction of exchangeable sodium from higher levels of sodium would be both efficient and practical. Once the exchangeable sodium is reduced to a practical limit (e.g. an ESP of 6 or less), then the management of these soils would be aimed at maintenance of a minimum level of electrolyte. Unlike class 2a soils, these soils become spontaneously dispersive (class 1) when leached without the addition of calcium compounds, and if there is no generation of electrolytes in the soils due to mineral weathering (Shainberg et al. 1981). Class 2c Subsoil (B-horizon) samples of red-brown earths, with similar SAR values (i.e. above 3), require higher electrolyte levels to prevent dispersion (not shown in Fig. 4). The amount required is equivalent to (3.19 SAR- 1.7) m.e. 1-l. It has been observed that the exposed subsoils in northern Victoria, where landforming for irrigation is widely practised, have more severe surface structure problems than corresponding areas where the topsoil is retained. Class 3 - jlocculated soils When soils have more than the minimum required electrolyte levels (as defined by equations 3 to 6), they remain flocculated when subjected to rainfall, irrigation or mechanical stress. However, it is important to recognize that excessive levels of soluble salts in soil water may reduce its availability to plants. Under conditions of excessive salinity, selection of a tolerant crop or leaching of salts from appropriate layers may become necessary. There are three possible categories in this class, differing in the management procedures required for their reclamation. Dispersive Behaviour of Red-brown Earths Class 3a If the SAR of a soil is above 3 and its TCC exceeds the flocculation value, then it is saline and sodic. Leaching with good quality water may change a saline-sodic soil to class 2b (e.g. Quirk 1971), or on extreme leaching to class 1. The soil may consequently disperse and cause severe crusting. Addition of calcium compounds in calculated amounts together with gradual leaching are essential for the reclamation of such soils (Quirk 1971). Many of the irrigated red-brown earths surveyed by Mehanni and Repsys (1978) in northern Victoria, which have been subjected to fluctuating water tables, fall into classes 2b and 3a. Class 3b When the SAR is less than 3 and the TCC is above the flocculation value, the soils are saline and dominated by non-sodium salts. Typically, such soils have a TCC of 7 m.e. 1-I or more (i.e. a saturation extract EC of approximately 4 dS m-I), and so would be classified as saline (Richards 1954). These soils have no physical problems, and their leaching requirements depend on the salt tolerance of the crops to be grown. Class 3c When the SAR is less than 3 and the TCC is ideally similar to the flocculation level, there are no dispersion or salinity problems. This situation should be the long-term aim of management strategies for the surface layer of red-brown earths. Validation of Dispersion Test Prediction of field response to gypsum The sensitivity of our dispersion test was assessed using samples from the control (nil gypsum) and gypsum (7.5 t ha-') treated plots from 16 sites on red-brown earths in northern Victoria (Ford 1978). Samples of the cultivated layer were collected after two fallow-wheat cycles. The results of the mechanical dispersion test on samples from control plots (Table 4) indicated that all soils were potentially dispersive and low-sodic (class 2a), and so would be expected to respond to gypsum in terms of maintaining electrolyte levels. Field observations indicated that on 12 sites the gypsum treatment improved the surface structure of the soils (Ford 1978). As this experiment was conducted during seasons in which there was little or no rainfall between sowing and crop emergence, there was no effect of gypsum on the establishment of either wheat crop. Any effect of gypsum on improving the surface structure of such potentially dispersive soils would probably operate via improved aggregate stability to raindrop impact. Loveday (1974b), in a study of laboratory methods for prediction of the probable response of a soil to application of gypsum, found that hydraulic conductivity was the most sensitive predictor. Analyses of samples from these 16 sites showed that treatment with gypsum had increased hydraulic conductivity (Fig. 5). The improvement, as indicated by the relative hydraulic conductivity (KJK,) values, was closely related to the amount of residual gypsum in the samples. The amount of dispersed clay in samples from gypsum-treated plots showed a reciprocal relationship with the amount of residual gypsum. The strong negative linear relationship between dispersed clay and relative hydraulic conductivity (r = 0.86, P < 0.05) suggests that clay dispersion may be the major mechanism controlling water movement in these soils. - P. Rengasamy et al. Table 4. Dispersion test results and flocculation values of surface soils from 16 red-brown earths (Ford 1978) Site No. Nil gypsum plots SAR TCC Dispersed (me. 1 clay (VQ) Gypsum treated plotsA SAR TCC Dispersed (me. 1 clay (70) Flocculation valuesB Predicted Actual (m.e. I-]) (m.e. 1-I) AGypsum was applied to the surface soil at the rate of 7.5 t ha-l of CaSO4.2H,O. Soil samples (0-10 cm) were collected from all plots after two fallow-wheat cycles. BFlocculation values either calculated from equation 6 or as experimentally determined. 10 20 30 40 50 60 Residual gypsum (m.e. kg- ') Fig. 5. Relationships between relative hydraulic conductivity (Y,, closed circles), dispersed clay % (Y,, open circles) and the residual gypsum ( X ) in 16 sites on red-brown earths (Ford 1978). The relative hydraulic conductivity was found to be related to residual gypsum by a function of the form (dy/dx) = k(A-Y), where A is the maximum possible value for the relative hydraulic conductivity (Y), X is the amount of residual gypsum and k is a constant (Fig. 5). At levels exceeding 20 m.e. gypsum kg-' soil there was no further increase in the relative hydraulic conductivity. By using an 427 Dispersive Behaviour of Red-brown Earths approach similar to that used in determining threshold concentration values (Quirk and Schofield 1955), the concentration of gypsum in soils that maintained 85% of the maximum possible relative hydraulic conductvity was 12 m.e. kg-' or approximately 1.5 t ha-l of gypsum per 10 cm depth. At this concentration of gypsum there was less than 1% of dispersed clay. The flocculation values of samples from these 16 sites were calculated using equation 6. The values varied from 4.2 to 5.6 m.e. I-', and were closely related to the flocculation values (critical coagulation concentrations) experimentally determined for these soils (Table 4). This indicates the ability of our dispersion test to predict gypsum response, and also supports our conclusion that potentially dispersive soils with a low sodium content (i.e. Class 2a) respond to gypsum via the electrolyte effect. Prediction of dispersive behaviour of red-brown earths The efficiency of equations 3-6 in predicting the dispersive behaviour of surface samples of red-brown earths was assessed as follows. Analytical data were obtained for 100 samples from our study, and for 250 samples analysed by the State Chemistry Laboratories, Melbourne (Mr. A. Brown, personal communication). The samples were classified using the appropriate equation and the measured values of SAR and TCC, and the numbers of samples in each class deviating from the predicted dispersive behaviour noted. The results are presented in Table 5, together with data for soil pH and organic carbon content. Table 5. Evaluation of equations 2-8 for prediction of dispersive behaviour of 350 samples of red-brown earths based on analytical data from various laboratories Soil class Total No. of soils/class No. of aberrant soils/classA Soil propertyB pH (1:s w/v) Organic carbon (70) Range Mean (s.e.) Range Mean (s.e.) Aberrant samples were those with dispersive behaviour differing from that predicted by the appropriate equation, i.e. those either not dispersing (although members of classes 1, 2a, 2b or 2c) or dispersing (although members of classes 3a, 3b or 3c). BData tabulated refer to all samples in a given class. Values given are for range, mean, and (in parenthesis) the standard error of mean. Corresponding mean values for aberrant samples are: Class 1: pH 5 SO, 2.9% organic carbon; Class 2a: pH 5.2, 1.8% organic carbon. Cn.d., not determined. A A chi-square test was used to assess the predictive accuracy of our dispersion test. The value of chi-square was 3.69 (6 d.f.), indicating no significant difference between the observed and predicted dispersive behaviour of these samples. It is however relevant to note that those samples in classes 1 and 2a which did not disperse as expected were characterized by having a lower pH and/or higher Pipette 10 ml from a depth of 5 cm and determine To clay I Pipette 10 mi from a depth of 5 cm and determine % clay Fig. 6 . Flow sheet of recommended procedure for classification of red-brown earths Centrifuge 25 ml of supernatant. Determine Na*, K + , Ca2+and MgZ' in supernatant. Calculate SAR, TCC. Measure pH and EC in supernatant without further stirring Centrifuge 25 ml of suspension. Determine Na+, K * , Ca2+and Mg2+ in supernatant. Calculate SAR, TCC. 1 Measure pH and EC in the suspension Shake for 1 h in an end-over-end shaker (0.5 rev s-I). Allow appropriate sedimentation time (e.g. 4 h at 20°C) Mix the dispersed clay without disturbing the soil using a stirrer at 0 - 16 rev s-' for 30 s. Allow appropriate sedimentation time (e.g. 4 h at 20°C) I No dispersion I 4, Spontaneous dispersion I Weigh 20 g air-dried soil (< 2 mm) into a transparent jar (in duplicate). Add 100 ml distilled water without disturbing the soil. Stand for 12 hours (overnight). Observe soil surface for zone of dispersed clay Dispersive Behaviour of Red-brown Earths 429 organic matter content than other members of these classes. Their failure to disperse would thus be consistent with the known effects of these factors on aggregate stability (Emerson 1983). Comments on Recommended Procedure The recommended experimental procedure for testing red-brown earth samples is summarized in Fig. 6. Further details of the various measurements are described in the Materials and Methods section of this paper. Our test can probably be simplified for routine use by using an infra-red nephelometer to measure turbidity (i.e. 70 clay), a conductivity meter to measure EC, and sodium-ion electrode to measure sodium ion concentrations in the equilibrium suspensions. We suggest, on the basis of the data obtained in this study, that TCC may be reliably estimated using the following relationship: TCC = 9.62EC + 0.14, (9) where TCC is in m.e. 1-I and EC in dS m-I (R2=0.97). Assuming that soluble K + is constant and negligible, SAR may then be estimated from the measured Na+ as follows: SAR = ~ a/ ~+( T C C- Na+). (10) The TCC required for flocculation may then be calculated using equations 3-6. Fig. 4 may be used to predict the probable dispersive behaviour of surface soils, and a decision can be made on gypsum requirement. Where the observed dispersive behaviour of a sample differs from that predicted from its SAR and TCC values, we recommend that the pH and organic carbon content also be determined. Our scheme will enable reasonably accurate prediction of the probable stability of red-brown earth aggregates in the field, particularly those in the surface soil. However, it should only be regarded as a guide to field behaviour, as many factors will influence the actual extent of dispersion of clay from aggregates. Further, the electrolyte environment of surface aggregates will be continuously changing owing to leaching by rainfall or irrigation, so that frequent testing of the soil is required when estimating gypsum application rates. The effects of exchangeable sodium (SAR) and electrolytes on soils in subsurface layers will probably be governed by swelling reactions rather than dispersion. Further work is needed to establish useful relationships between SAR, TCC and soil physical problems in subsurface soils. It should be noted that equations 3-6 cannot be applied to samples containing free lime, as Emerson (1983) has noted that the presence of lime affects the dispersive behaviour of soil aggregates. Thus our test probably cannot be applied to most samples from the deeper subsoil of red-brown earths, unless lime is known to be absent. Williams (1981) has summarized information illustrating the variability in lime content both within and between red-brown earth profiles. Conclusions 1. A scheme is proposed which enables the prediction of the probable dispersive behaviour of the surface layer of red-brown earths, including exposed subsoils. Evidence is presented for the ability of this scheme to adequately explain the observed field behaviour of both dryland and irrigated soils. 430 P. Rengasamy et a[. 2. A procedure for the routine laboratory testing of samples is described. Using this procedure, surface layers of red-brown earth may be classified into one of six classes based on relationships established between dispersion, SAR and TCC. The implications of these relationships for the development of strategies for the successful management of soils in each class are discussed. 3. Using the criterion of a SAR of 3 or less, a group of low-sodic soils was identified in which dispersive behaviour could be practically controlled by the electrolyte effect. Some of the implications for the prediction of the gypsum requirement of low-sodic soils are discussed. 4. Spontaneous dispersion occurred in 20% of the samples tested. These were found to have a SAR above 3, and it is suggested that in such soils spontaneous dispersion is controlled to a similar degree by both the level of exchangeable sodium and the rate of diffusion of salts from soil aggregates to the soil solution. 5. Most (nearly 80%) of the red-brown earth samples tested dispersed after mechanical shaking. A greater proportion of the clay fraction dispersed from subsoil than from surface aggregates, probably due to their higher clay content and lower organic matter levels. Surface soils released from 1 to 14% clay; a level of 1070 mechanically dispersed clay is proposed as desirable for minimum aggregate breakdown in the field. The application of this concept to the estimation of the minimum levels of residual gypsum required for the maintenance of satisfactory surface structure is discussed. Acknowledgments We acknowledge the statistical advice by Dr G . Robinson and the technical assistance by Mrs F. Robertson, Ms M. L. Mann and Mr 0. Dunne. References Arora, H. S., and Coleman, N. T. (1979). The influence of electrolyte concentration on flocculation of clay suspensions. Soil Sci. 127, 134-7. Blackmore, A. V. (1976). Salt sieving within clay soil aggregates. Aust. J. Soil Res. 14, 149-58. Collis-George, N . , and Smiles, D. E. (1963). An examination of cation balance and moisture characteristic methods of determining the stability of soil aggregates. J. Soil Sci. 14, 21-32. Emerson, W. W. (1967). A classification of soil aggregates based on their coherence in water. Aust. J. Soil Res. 5 , 47-57. Emerson, W. W. (1983). Inter-particle bonding. In 'Soils: an Australian Viewpoint'. pp. 477-98. (CSIRO: Melbourne/Academic Press: London.) Ford, G. W. (1978). Surface crusting and emergence problems in some dryland wheat soils of northern Victoria. Proc. Symposium on 'Soil Structure in Dry-land and Irrigated Agriculture', Aust. Soil Sci. Soc. Vict. Branch. Greene, R. S. B., and Ford, G. W. (1985). The effect of gypsum on cation exchange in two red duplex soils. Aust. J. Soil Res. 23 (in press). Greene, R. S. B., Posner, A. M., and Quirk, J. P. (1978). A study of the coagulation of montmorillonite and illite suspensions by CaC1, using the electron microscope. In 'Modification of Soil Structure'. (Eds. W. W. Emerson, R. D. Bond and A. R. Dexter.) pp. 35-40. (John Wiley & Sons: New York.) Hardcastle, J. H., and Mitchell, J. K. (1976). Water quality and aquitard permeability. J. Zrrig. Drainage Div. ASCE, 102, 205-20. Kazman, Z., Shainberg, I., and Gal, M. (1983). Effect of low levels of exchangeable sodium and applied phosphogypsum on the infiltration rate of various soils. Soil Sci. 135, 184-92. Klute, A. (1965). Laboratory measurement of hydraulic conductivity of saturated soil. In 'Methods of Soil Analysis, Part 1'. (Ed. C. A. Black.) pp. 210-21. Agronomy No. 9. (Am. Soc. Agron. Inc: Madison, Wisc.) Dispersive Behaviour of Red-brown Earths Loveday, J. (Ed.) (1974~).Methods for analysis of irrigated soils. Commw. Bur. Soils Tech. Commun. No. 54. Loveday, J. (19746). Recognition of gypsum-responsive soils. Aust. J. Soil Res. 12, 87-96. Loveday, J . (1981). Soil management and amelioration. In 'National Soils Conference 1980 Review Papers'. (Eds. T. S. Abbott, C. A. Hawkins and P . G. E. Searle.) pp. 39-57. (Aust. Soc. Soil Sci. Inc.: Glen Osmond, S.A.) Loveday, J., and Pyle, J. (1973). The Emerson dispersion test and its relationship to hydraulic conductivity. CSIRO Aust. Div. Soils Tech. Pap. 15. Martin, J . P., Richards, S. J., and Pratt, P . F. (1964). Relationship of exchangeable Na percentage at different soil pH levels to hydraulic conductivity. Soil Sci. Soc. Am. Proc. 28, 620-2. Matheson, W. W. (1969). Gypsum for hard setting soils. J. Agric. S.Aust. 72, 402-6. Mehanni, A. H. (1973). Laboratory method for measuring the hydraulic conductivity of heavy soils. J. Aust. Inst. Agric. Sci. 39, 262-4. Mehanni, A. H., and Repsys, A. (1978). Watertable conditions and soil salinity status in part of the irrigation areas of the Goulburn Valley. Vic. Dept. Agric. Res. Proj. Ser. 46. Nie, N. H., Hadlai Hull, C., Jenkins, J . G., Steinbrenner, K., and Bent, D. H. (1975). 'Statistical Package for the Social Sciences.' (McGraw Hill: New York.) Northcote, K. H., and Skene, J . K. M. (1972). Australian soils with saline and sodic properties. CSIRO Aust. Soil Publ. No. 27. Oades, J . M., Lewis, D. G., and Norrish, K. (Eds.) (1981). 'Red-Brown Earths of Australia.' (Waite Agricultural Research Institute: Adelaide.) Oster, J. D., and Schroer, F. W. (1979). Infiltration as influenced by irrigation water quality. Soil Sci. Soc. Am. J. 43, 444-7. Oster, J . D., Shainberg, I., and Wood, J. D. (1980). Flocculation value and gel structure of sodium/ calcium montmorillonite and illite suspensions. Soil Sci. Soc. Am. J. 44, 955-9. Quirk, J. P. (1971). Chemistry of saline soils and their physical properties. In 'Salinity and Water Use'. (Eds. T. Talsma and J. R. Philip.) pp. 79-94. (MacMillan: London.) Quirk, J . P., and Schofield, R. K. (1955). The effect of electrolyte concentration on soil permeability. J. Soil Sci. 6, 163-78. Rengasamy, P . (1983). Clay dispersion in relation to the changes in the electrolyte composition of dialysed Red-brown earths. J. Soil Sci. 34, 723-32. Rengasamy, P., and Oades, J. M. (1977). Interaction of monomeric and polymeric species of metal ions with clay surfaces. I. Adsorption of iron(II1) species. Aust. J. Soil Res. 15, 221-33. Rhoades, J. D. (1982). Reclamation and management of salt-affected soils after drainage. Proc. First Ann. Western Provincial Conf. Rationalization of Water and Soil Research and Management. pp. 123-97. Richards, L. A. (1954). Diagnosis and improvement of saline and alkali soils. U.S. Dep. Agric. Handb. No 60. ( U S . Govt Printing Office: Washington, D.C.) Rowell, D. L., Payne, D., and Ahmed, N. (1969). The effect of the concentration and movement of solutions on the swelling, dispersion and movement of clay in saline and alkali soils. J. Soil Sci. 20, 176-88. Shainberg, I., Rhoades, J. D., and Prather, R. J . (1981). Effect of low electrolyte concentration on clay dispersion and hydraulic conductivity of a sodic soil. Soil Sci. Soc. Am. J. 45, 273-7. Shanmuganathan, R. T., and Oades, J . M. (1983). Influence of anions on dispersion and physical properties of the A horizon of a red-brown earth. Geoderma 29, 257-77. Sherard, J. L., Dunnigan, L. P., Decker, R. S., and Steele, E. F. (1976). Pinhole test for identifying dispersive soils. J. Geotech. Eng. Div. ASCE. 102, 69-85. Sims, H. J., and Rooney, D. R. (1965). Gypsum for difficult clay wheat growing soils. J. Agric. Vict. 63, 401-9. Tucker, B. M. (1974). Laboratory procedures for cation exchange measurements on soils. CSIRO Aust. Div. Soils. Tech. Pap. No. 23. Velasco-Molina, H. A., Swoboda, A. R., and Godfrey, C. L. (1971). Dispersion of soils of different mineralogy in relation to sodium adsorption ratio and electrolyte concentration. Soil Sci. 111, 282-7. Williams, C. H. (1981). Chemical properties. In 'Red-Brown Earths of Australia'. (Eds. J. M. Oades, D. G. Lewis, and K. Norrish.) pp. 47-61. (Waite Agricultural Research Institute: Adelaide.) Manuscript received 22 February 1984, accepted 25 May 1984