Application of mineral magnetism to describe profile development of toposequences of a sedimentary soil in south-eastern Australia.
Mineral magnetic properties of soils have been increasingly studied in recent decades. Thompson and Oldfield (1986) summarised much of the earlier work and pointed to a wide range of applications. Several authors, notably Maher (1988), have compared the magnetic properties of synthesised fine-grained magnetic materials with those within soil profiles and weathering horizons. Magnetoviscous properties of ash soils have been used for dating (Bagina et al. 1989) and mineral magnetism is being increasingly used in the study of pedogenesis (Alekseyev et al. 1988; Fine et al. 1989, 1993). Singer and Fine (1989) considered the pedogenic factors affecting the magnetic susceptibility of a soil. Zheng et al. (1991) showed pedogenesis to be more important than detrital processes, and Singer et al. (1996) presented a conceptual model for enhancement of magnetic susceptibility in soils. Zhou et al. (1990) presented information suggesting a partial pedogenic origin for magnetic assemblages in Chinese Loess. It seems therefore that both colluvial/alluvial fill and pedogenesis can contribute to the magnetic properties of soil profiles.
The original purpose of this study was to assess mineral magnetism as a method for identifying the source of sediments in farm dams within the Clare catchment, NSW, Australia. If the sediments are from gullies, the section of the gully through which the water flows can differ through time; and if the magnetic properties of the exposed sections differ, misleading results will be acquired. A preliminary examination of the magnetic properties of the gully walls showed a remarkably varied pattern down the wall and among the various particle sizes tested.
In order to understand the reasons for the pattern of magnetic properties in the gully wall, a more extensive survey was done on soil profiles up-slope from the gully. The main features studied were the down-profile changes in concentration of magnetic minerals, the down-profile changes in magnetic ratios, and the relationships between panicle size, magnetic grain size (MGS), and magnetic susceptibility. MGS is a term used to describe the volume of the magnetic domains (Maher 1984). The term as commonly used refers to the mean of a mixture of magnetic grain sizes in the magnetic minerals within a sample. These properties were studied through a wide range of particle sizes. In addition, relationships between magnetic properties and concentrations of extractable iron and aluminium were established.
Materials and methods
Description of study area and sampling procedure
The area studied was a shallow gullied valley located 30 km NNE of Canberra at 35 [degrees] 13'S and 149 [degrees] 22'E. The soil is classified as a Yellow Dermosol (Isbell 1996), derived from Pittmann Formation of middle- to late-Ordovician (Abell 1991). Figure 1 is a map of the Clare Valley showing its location and the location of the sampled transect.
Surface to bedrock profiles of the transect across the mid-point of the valley were examined. The width and depth of the gully in the middle of the transect were 10 m and 3.4 m, respectively. The profiles sampled were at Sites 1 to 8, and 21, on the west and east sides (Site 1 being the gully wall). Sites 1 to 8 were 5 m apart; and sites 21 were 100 m up-slope (Fig. 1). Profile 4 on the west slope (called W4) was exposed with a mechanical excavator whereas the other profiles were sampled with a soil auger. A detailed description of profile W4 and brief descriptions of the other profiles are presented in Crockford and Willett (1997).
After profile W4 had been exposed, the surface bulk magnetic susceptibility of the profile was measured with a susceptibility probe. The profile was then sampled according to the susceptibility values and features such as colour and texture. For the other profiles the susceptibility probe was used to measure bulk susceptibility of the horizontal surfaces exposed by the auger. The bulk susceptibility of the samples was also measured after they had been put in plastic bags.
Sample preparation and measurement of mineral magnetic properties
Samples were wet sieved to the following sizes: -A2 [is greater than] 3.36 mm, -A1 2.0-3.36 mm, A 1.4-2.0 mm, B 500 [micro]m-1.4 mm, C 250-500 [micro]m, D 125-250 [micro]m, E 63-125 [micro]m, and F [is less than] 63 [micro]m. Size F was subdivided into G 40-63 [micro]m, H 20-40 [micro]m, I 10-20 [micro]m, J 2-10 [micro]m, and K [is less than] 2 [micro]m by settling in water. Dispersant was not used because it is diamagnetic.
For measurement of magnetic properties the samples were placed in plastic cuvettes (20 mm by 20 mm by 20 mm). When the weight of a sample was small, it was wrapped in thin plastic sheet and centred in the cuvette by putting cotton wool below and above the sample. The geometry of placement was important, particularly for measurement of magnetic susceptibility. As the cuvettes and cotton wool were diamagnetic, appropriate blank measurements were made and the [chi] values adjusted. This was very important for the small samples, particularly those with low concentrations of magnetic minerals.
Magnetic susceptibility ([chi]) was measured at low frequency (0.47 kHz; [chi]lf) and high frequency (4.7 kHz; [chi]hf) using a Bartington dual frequency sensor. This enabled frequency dependence to be determined [[chi]fd% = ([chi]lf - [chi]hf)*100/[chi]lf]. Isothermal remanent magnetisation parameters (IRM20, IRM200) and saturated isothermal remanent magnetisation (SIRM) were measured in a Molspin flux-gate magnetometer after magnetisation in a Molspin pulse magnetiser (induced at 20 mT, 200 mT, and 850 mT). For convenience, 850 mT is termed SIRM, although the term more strictly applies to induction at 1000 mT. The maximum field our pulse magnetiser could apply was 850 mT. Anhysteretic magnetisation (ARM) was also measured in the Molspin magnetometer after magnetisation with an anhysteretic attachment. The peak AF field used was 100 mT and the DC bias was 0.04 mT. Details of the methods may be found in Thompson and Oldfield (1986). Throughout the paper, susceptibility ([chi]) is expressed as [10.sup.-8] [m.sup.3]/kg, and remanence as mA [m.sup.2]/kg.
The properties were measured in the order [chi]lf, [chi]hf, ARM, IRM20, IRM200, and SIRM. To avoid the problem of variations in viscous loss of IRM between induction and measurement, the measurements were made 15 min after induction; by this time viscous loss was minimal. Also, because we have found that short periods between inductions can enhance subsequent measurements, inductions were done at 1-h intervals. These procedures minimised measurement variability.
The magnetic parameter IRMh (IRM hard) was calculated as (SIRM - IRM200) as a percentage of SIRM. This was used in order to relate to our previous work. SIRM - IRM300 is more commonly used now, but tests on various samples in our projects have shown little difference in the relative values. It is used as a measure of the canted anti-ferromagnetic minerals haematite and goethite (Thompson and Oldfield 1986). The ratios were calculated thus; for SIRM/[chi] and IRM20/[chi] remanence is multiplied by [10.sup.2], and for ARM/[chi] the factor is [10.sup.5]. Factors are not applied to the other ratios.
The magnetic properties of particles [is greater than] 2 mm were not routinely measured for all samples because concentrations of magnetic minerals were very variable and difficult to measure. In many samples of this size most of magnetic susceptibility was carried by just a few highly magnetic particles, and their position in the cuvette during measurement had a great influence on the value, e.g. if the very magnetic particles were near the side of the cuvette the reading was 20-50% greater than if they were near the middle. As these particles are often only a small proportion of the total sample weight, the representativeness of the sample becomes an issue. However, a study was made of the properties of some of the larger particles from samples in the S and H layers after separation into 3 magnetic classes, as described below.
The magnetic separation of particles
Samples were separated by wet sieving into the following sizes: 1, [is greater than] 3.6 mm; 2, 2.0-3.36 mm; 3,500 [micro]m-2.0 mm; 4, 250-500 [micro]m. Size 4 is the size C discussed in the other sections of this paper. Particles were then separated into 3 magnetic classes with a hand magnet, using the following procedure. (i) The sample was spread out on paper between wooden bars 8 mm thick. (ii) A hand magnet covered in plastic was placed across the bars and moved slowly forwards and backwards, stopping frequently to remove particles attracted to the magnet. This was achieved by removing the plastic cover. The process was repeated after mixing the residual sample. These particles were termed VM (very magnetic). (iii) When no more particles were picked up, the spacer bars were removed and the plastic covered magnet was moved through the sample with frequent removal of attached particles until no more could be separated. These were termed M (moderately magnetic). The remaining particles were termed L (low to weakly magnetic). The sample was thus divided into the groups, VM, M, and L.
Samples of particle sizes B-K were extracted with acidic ammonium oxalate, which extracts ferrous compounds, magnetite, and poorly ordered ferric hydrous oxides (McKeague et al. 1971). Total free iron oxides and aluminium were extracted by citrate-dithionite bicarbonate (CDB, Mehra and Jackson 1958). Iron and aluminium in the extracts were determined by atomic absorption spectrophotometry.
Down-profile changes in concentration of magnetic minerals
Figure 2a shows the bulk susceptibility values of profiles on the west slope and the results for the east slope profiles are shown in Fig. 2b. The field results for profiles W4-W8 and E1-E8 showed similar general down-profile patterns in bulk susceptibility. The surface layers (S) had high to intermediate values of susceptibility. Values then decreased withdepth to a minimum layer (termed layer P), and then increased to maxima (layer H). The values then diminished in the base layer (B). The surface soils of profiles nearer the western side of the gully wall (W1-W3) did not show the high values of susceptibility (Fig. 2a).
Figure 3 shows the down-profile values of mass [chi] for all particle sizes of profile W4. The changes down the profile show the patterns of high [chi] values in layer S, low values in P, high values in H, and low values in B. The pattern is similar for each size class. Because the profile pattern for sites W6, W8, and W21 were very similar to W4 for particle sizes A to F ([is less than] 63 [micro]m), sizes G (40-63 [micro]m) to K ([is less than] 2 [micro]m) were separated only from samples representing depths S, P, B, and H for these profiles. These data are shown in Table 1 and Figs 4-6.
Table 1. Mass magnetic susceptibility of particle sizes in profiles Size fractions are -A2, >3.36 mm; -A1, 2.0-3.36 mm; A, 2.0-1.4 mm; B, 1.4 mm-500 [micro]m; C, 250 [micro]m; D, 250-125 [micro]m; E, 125-63 [micro]m; G, 63-40 [micro]m; H, 40-20 [micro]m; I, 20-10 [micro]m; J, 10-2 [micro]m; K, <2 [micro]m. Letters S, P, H, and B correspond to depths shown in Fig. 3 Size S P H B Profile W4 -A2 862.0 62.0 915.0 4.0 -A1 1550.0 113.0 839.0 29.0 A 1393.0 68.5 1139.0 16.5 B 1138.0 113.0 763.0 18.2 C 531.0 29.2 361.0 16.1 D 118.0 10.5 154.0 9.7 E 39.2 11.0 93.3 9.5 G 29.3 7.8 94.3 11.6 H 33.4 7.4 109.0 11.7 I 48.0 7.4 118.0 12.1 J 55.4 8.0 135.3 12.9 K 33.1 9.7 210.1 18.3 Profile W8 -A2 580.0 20.0 625.0 982.0 -A1 1146.0 143.0 762.0 530.0 A 1452.0 348.0 465.0 482.0 B 1124.0 272.0 394.0 98.0 C 332.0 76.5 108.0 22.7 D 99.0 21.8 45.3 12.8 E 53.9 13.4 34.6 12.3 G 26.1 11.3 21.9 11.0 H 36.3 9.7 15.0 13.4 I 60.8 11.9 26.7 10.5 J 120.7 17.6 36.9 12.8 K 141.6 24.8 48.1 16.7 Size S P H B Profile W6 -A2 623.0 195.0 490.0 8.0 -A1 1636.0 354.0 885.0 21.0 A 1497.0 474.0 944.0 23.2 B 1307.0 319.0 613.0 15.1 C 338.0 108.7 226.8 13.0 D 103.0 31.5 89.4 10.3 E 44.1 16.9 62.3 9.7 G 19.6 7.7 27.7 8.8 H 18.6 8.2 25.2 4.9 I 30.9 8.2 36.8 5.7 J 63.7 13.2 65.6 10.3 K 93.5 18.5 90.8 14.0 Profile W21 -A2 2132.0 11.0 -A1 543.0 16.0 A 430.0 34.3 B 361.0 26.9 C 168.0 24.2 D 76.0 18.7 E 45.5 15.5 G 30.0 15.8 H 35.8 12.0 I 38.0 12.1 J 61.0 17.9 K 31.8 107.0
The down-profile changes were different through the particle size range, with the largest differences being shown by the larger particles (note the differing x-axis scales between parts (a) and (b) of each figure). The differences with depth were fairly similar for sizes E-J. For all particle sizes the value of [chi] in layer H became smaller (relative to the S value) up-slope from W4 to W8 (Figs 3-5 and Table 1).
The basic S-P-H-B pattern was quite consistent through the entire size range, even though the magnitude of the changes varied somewhat. However, the pattern did not persist beyond W8, which was 40 m up-slope from the gully. Even 5 m further up-slope (W9.5) the S-P-H-B pattern was not shown (Fig. 2a), and the W9.5 profile was very similar to that of W21, at the top of the slope. Size K ([is less than] 2 [micro]m) of profile W21 (Fig. 6b) is the only exception to the particle size-susceptibility and susceptibility-depth relationships.
In profile W4 (Table 1) the ratio of [chi] in the H layer to that in the S layer increased through the particle size range (B-K), showing that fine particles with high [chi] were present in H, but almost absent in S. For the coarser sizes B and C, the S layer values were higher than in the H layer samples. Possible explanations of these observations are: (a) fine particles with high concentrations of magnetic material were formed in S but have been either removed by erosion or destroyed by pedogenic processes; (b) those formed in H when it was possibly a surface layer have been preserved; and (c) the processes did not produce highly magnetic fine particles in the present S layer.
The same trends were evident in profile W6 (Table 1), but to a lesser extent. They were not evident in W8.
Down-profile changes in magnetic ratios
The ratios of magnetic properties reflect the MGS and detect the presence of paramagnetic and canted anti-ferromagnetic material. Figure 7 is a diagrammatic representation of changes in magnetic properties in relation to magnetic grain size. It was derived from data presented by Maher (1988).
Initially the results for profiles W4, W6, and W8 will be presented, with later reference to W21. The high frequency dependence ([chi]fd%) values of the samples from these 3 profiles, up to 12% (see Table 1a-d in Crockford and Willett 1997), suggest that the dominant MGS is in the 0.02-0.03 [micro]m area of Fig. 7.
As [chi]fd% is difficult to measure, particularly for samples with low [chi] values, the ratio most sensitive to changes in MGS in the 0.02-0.03 [micro]m area of Fig. 7 is IRM20/[chi]. The down-profile changes in this ratio, for all profiles and all particle sizes, are shown in Figs 8-11. With the exception of size G for profile W6 (Fig. 9b), size A for profile W8 (Fig. 10a), and size K for profile W21 (Fig. 11b), there is a consistent down-profile pattern: a decrease from S to P, increase from P to H, and a decrease from H to B. The exceptional S to B increase in the IRM20/[chi] value of size K of profile W21 (Fig. 11) suggests that its MGS is near the IRM20 peak in Fig. 7. The ratio SIRM/[chi] shows a similar pattern to IRM20/[chi], but is slightly more variable through the size range (Crockford and Willett 1997).
The proportions of the magnetic minerals represented by the canted anti-ferromagnetic minerals (goethite or haematite) are shown by the ratio IRHh%. For profiles W4, W6, and W8 this property increased from S to P, decreased from P to H, then increased slightly from H to B, for all particle sizes (see Tables 2 and 3 in Crockford and Willett 1997). This pattern is consistent with the P and B layers being more hydromorphic than the other layers.
Table 2. Particle size as per cent of total for profile depths Size fractions are -A2, >3.36 mm; -A1, 2.0-3.36 mm; A, 2.0-1.4 mm; B, 1.4 mm-500 [micro]m; C, 250 [micro]m; D, 250-125 [micro]m; E, 125-63 [micro]m; G, 63-40 [micro]m; H, 40-20 [micro]m; I, 20-10 [micro]m; J, 10-2 [micro]m; K, <2 [micro]m. Letters S, P, H, and B correspond to depths shown in Fig. 3 Size S P H B S P H B Profile W4 Profile W6 -A2 5.6 1.2 3.7 16.7 5.3 2.2 10.1 31.4 -A1 5.6 1.2 5.6 9.0 9.3 2.2 11.5 7.0 A 3.5 1.8 3.7 4.4 2.3 1.3 5.1 3.3 B 13.8 2.0 8.6 11.6 12.9 4.8 9.9 6.2 C 6.3 1.6 4.1 8.6 5.5 3.5 5.4 3.1 D 7.7 1.7 4.9 4.9 6.6 5.2 7.7 3.1 E 11.7 0.2 5.8 6.3 12.2 8.3 7.0 3.0 G 10.1 8.6 8.0 1.9 9.1 9.9 6.5 2.4 H 9.0 9.9 9.9 3.1 7.5 9.1 5.4 3.2 I 8.5 13.1 9.9 5.8 8.4 7.7 5.7 4.5 J 12.2 28.9 18.1 12.2 12.6 20.4 12.0 10.5 K 6.0 29.8 17.8 15.5 8.0 25.5 13.7 22.3 Profile W8 Profile W21 -A2 6.7 4.2 12.0 3.4 22.4 38.9 -A1 12.0 2.8 8.0 4.1 10.6 4.3 A 4.3 2.2 3.6 1.5 3.2 1.7 B 12.0 6.3 7.4 8.2 8.7 3.6 C 7.7 4.2 5.7 5.5 6.2 2.4 D 10.4 6.5 6.7 5.9 5.9 2.3 E 12.2 6.8 10.4 6.2 6.2 3.3 G 8.8 8.0 3.9 1.6 4.4 1.3 H 6.0 6.4 7.2 5.9 4.8 3.3 I 5.4 6.7 5.1 7.5 6.6 5.7 J 9.0 18.1 12.9 19.2 13.8 13.7 K 5.5 27.8 17.1 30.9 7.2 19.3 Table 3. Percentage of total magnetic susceptibility of particle sizes in profiles Size fractions are -A2, >3.36 mm; -A1, 2.0-3.36 mm; A, 2.0-1.4 mm; B, 1.4 mm-500 [micro]m; C, 250 [micro]m; D, 250-125 [micro]m; E, 125-63 [micro]m; G, 63-40 [micro]m; H, 40-20 [micro]m; I, 20-10 [micro]m; J, 10-2 [micro]m; K, <2 [micro]m. Letters S, P, H, and B correspond to depths shown in Fig. 3 Size S P H B S P H B Profile W4 Profile W6 -A2 12.0 6.7 11.3 4.8 7.5 8.4 15.9 21.8 -A1 21.6 9.2 15.9 18.6 34.6 15.3 32.6 12.8 A 12.0 8.6 14.3 5.1 7.9 12.3 15.6 6.6 B 38.8 16.0 22.0 15.1 38.3 30.9 19.6 8.2 C 8.3 3.2 5.0 9.9 4.2 7.5 4.0 3.4 D 2.2 1.3 2.5 3.4 1.5 3.3 2.2 2.8 E 1.1 0.1 1.8 4.3 1.2 2.8 1.4 2.5 G 0.87 5.2 3.4 2.0 0.93 2.6 1.3 2.3 H 0.77 6.0 4.2 3.1 0.77 2.4 1.1 3.1 I 0.73 8.0 4.2 5.9 0.86 2.0 1.1 4.4 J 1.1 17.6 7.8 12.3 1.3 5.4 2.4 10.3 K 0.52 18.1 7.6 15.6 0.82 6.8 2.7 21.8 Profile W8 Profile W21 -A2 8.7 1.8 35.5 40.7 25.2 24.8 -A1 31.1 8.5 28.8 26.3 30.3 4.0 A 14.0 16.6 7.9 8.8 7.2 3.4 B 30.5 36.5 13.7 9.7 16.6 5.7 C 5.8 6.8 2.9 1.5 5.5 3.4 D 2.3 3.0 1.4 0.9 2.3 2.5 E 1.5 2.0 1.7 0.9 1.5 3.0 G 1.6 3.0 0.7 0.3 1.4 1.6 H 1.1 2.4 1.3 1.0 1.5 4.0 I 0.90 2.5 0.9 1.3 2.0 6.9 J 1.6 6.7 2.3 3.3 4.2 16.8 K 1.0 10.3 3.0 5.3 2.2 23.7
Variation of mass magnetic susceptibility through the range of particle sizes
In all 4 profiles (W4, W6, W8, and W21) and at all depths, [chi] (which represents the concentration of ferrimagnetic minerals) decreased from size A to size G or H then increased slightly to size J or K (Table 1). The small increase from G or H to J or K occurs at all depths for all profiles, and is much less than the decreases from size A to size G or H. The greatest increase from size G to K (a factor of 4) occurs in the S layer of profile W8.
As stated earlier, the values of [chi] in the largest sizes, -A2 and -A1, were difficult to measure because of the uneven distribution of magnetic material within the samples in the cuvettes, and therefore were not representative. For this reason, subsequent discussion of the results will concentrate on sizes A to K. The largest size A or B to G or H ratio, 73, is
shown in the surface samples of profile W6 (Table 1), i.e. the magnetic concentration of the large particle sizes, A and B, is up to 73 times that of the smaller sizes G and H (63-20 [micro]m). With the exception of some of the smaller sizes the differences in [Chi] through the range of particle sizes followed the pattern S [is greater than] H [is greater than] P [is greater than] B. The properties ARM, IRM20, and SIRM followed the same trends, but with some differences due to variations in MGS.
The particle size distributions for each profile are shown in Table 2, and the proportions of total [Chi] contained in each particle size are shown in Table 3. The coarser sizes (-A2 to B) generally contained the largest proportion of [Chi], particularly in the S and H layers. For the P and B layer samples the results varied; for example, size K ([is less than] 2 [micro]m) for W21 contained 23.7% of [Chi]. The values depend, of course, on the weight proportion (Table 2) and the [Chi] value (Table 1).
Variation of magnetic ratios and magnetic grain size through the range of particle sizes
Figure 12 shows the relationship between [Chi]fd% and IRM20/[Chi] for the depths in profile W4 where [Chi]fd% was determined on almost all particle sizes. All depths shown in Fig. 12 showed a consistent inverse relationship between these parameters, i.e. the [Chi]fd% values increased and IRM20/[Chi] decreased as the particle size became smaller.
These relationships suggest that change in the MGS is the main cause of the change in IRM20/[Chi] for the samples tested, i.e. paramagnetic material had little influence. As the particle sizes are very much greater than the magnetic grain sizes it seems unlikely that the physical dimensions of particles could influence the MGS. As an example, particle size H is 20-40 [micro]m, whereas the dominant MGS in the fraction are [is less than] 0.05 [micro]m. The response through the particle size range may depend on whether the magnetic minerals are within particles, or in coatings where they would be more accessible to soil chemical processes.
Similar relationships between particle size and magnetic properties were found by Crockford and Olley (1998) in a sedimentary soil near Wagga, in the Murrumbidgee catchment, NSW, and by Crockford and Fleming (1998) in river sediments of sedimentary origin.
The magnetic properties of magnetically separated size fractions
The reason for this part of the study is that, as stated earlier, magnetic susceptibility was difficult to measure on the coarser particle sizes ([is greater than] 2 mm). In this experiment the problem was overcome by measuring susceptibility repeatedly, after packing and repacking the material 7 times.
Table 4 shows the weight proportions of the magnetic classes and the weight proportions of susceptibility and the magnetic properties of these classes. With such a method of separation large differences in concentration of magnetic minerals between VM, M, and L are to be expected. The [Chi] values of the VM and M samples for sizes 2 and 3 (Table 4) are substantially greater than for these magnetic classes of size 1, for both soil depths.
Table 4. Distribution of magnetic susceptibility in magnetically separated fractions and their magnetic properties (S and H layers of profile W4) MC, magnetic classification; VM, very magnetic; M, moderately magnetic; L, low (weakly magnetic); WDM, weight distributed mean Size(A) MC Weight, Weight of [Chi] % % [Chi] S H S H S H 1 VM 12 26 40 77 2378 3651 1 M 25 24 30 17 860 848 1 L 63 50 31 7 348 162 WDM 720 2 VM 10 20 50 80 4482 6536 2 M 25 22 35 16 1252 1167 2 L 65 58 15 5 198 132 WDM 890 1641 3 VM 5 4 58 55 6882 6030 3 M 16 11 33 28 1240 1098 3 L 79 85 9 17 68 89 WDM 596 437 4 VM 0 0 0 0 4 M 4 5 79 44 2150 1060 4 L 96 95 21 56 24 72 Size(A) MC Xfd% SIRM/[Chi] ARM/[Chi] S H S H S H 1 VM 4.1 4.1 32.2 27.5 45.4 32.3 1 M 6.1 5.8 21.2 20.0 34.9 25.8 1 L 9.7 11.7 6.2 6.0 12.4 15.4 WDM 1234 8.1 8.3 21.0 24.9 32.2 2 VM 6.1 4.3 21.5 28.6 31.7 28.0 2 M 9.2 8.6 10.5 15.1 19.6 26.7 2 L 9.8 10.4 6.8 8.7 15.7 20.5 WDM 9.3 8.8 15.5 25.6 25.2 27.4 3 VM 5.1 4.6 18.0 22.7 23.8 26.5 3 M 8.2 7.4 9.2 14.8 19.1 26.0 3 L 9.2 12.5 7.3 6.5 16.2 19.1 WDM 8.8 11.6 14.1 17.7 21.6 25.2 4 VM 4 M 7.8 6.4 10.5 14.8 19.1 32.0 4 L 10.7 13.7 7.9 4.6 20.0 18.2 Size(A) MC SIRM/ARM IRMh% S H S H 1 VM 71 85 0.05 0.45 1 M 61 78 0.55 4.71 1 L 50 39 1.85 2.06 WDM 30.1 65 82 1.3 2 VM 68 102 0.04 0.08 2 M 53 56 1.70 0.20 2 L 43 43 2.99 2.18 WDM 61 93 2.4 1.3 3 VM 75 86 0.02 0.30 3 M 48 57 1.60 0.60 3 L 45 34 4.26 3.45 WDM 65 70 3.6 3.0 4 VM 4 M 55 46 1.41 1.27 4 L 40 25 9.47 6.97 (A) Sizes: 1, >3.36 [micro]m; 2, 2.0-3.3.6 [micro]m; 3,500-1400 [micro]m; 4,250-500 [micro]m.
The proportional weight data in Table 4 are the weights after magnetic separation. For both S and H layer samples the proportion of VM material diminished through the size range, and the H layer samples had about double the amount of VM material in the S layer for sizes 1 and 2. This causes the weight distributed mean values for [Chi] to be much higher for the H soil than the S soil for sizes 1 and 2 (1234 and 1641 compared with 720 and 890 for S).
From class VM to M to L for all sizes and both soil depths there was a very consistent trend towards smaller magnetic grain sizes. This is shown by all remanence/[Chi] ratios, SIRM/ ARM, and [Chi]fd% (Table 4). This is identical to the trend of the mass magnetic values. The regressions for mass values versus any of the above ratios had [R.sup.2] values [is greater than] 0.85 for the individual sizes. This means that the smaller the concentration of ferrimagnetic minerals the smaller the MGS, regardless of particle size. Even where the 3 classes are combined, the SIRM/ARM v. [Chi] relationships are positive and substantial ([R.sup.2] values of 0.73 for S, and 0.76 for H). In summary, at both soil depths, the higher the concentrations of magnetic materials, the larger the MGS. This is identical to the results for the other sizes shown in the previous sections. These results suggest that the magnetic material was formed by similar processes at both depths (S and H).
Relationship between the concentrations of iron, aluminium, and magnetic minerals
Figure 13 shows [Chi], dithionite-extractable Fe and Al, and IRMh% plotted against depth for size I (10-20 [micro]m). The Fe and Al curves show a marked inverse relationship with [Chi], i.e. they peak in the weakly magnetic P layer and are low across the highly magnetic H layer. They are also very similar to the IRMh% curves. This figure shows low concentrations of ferrimagnetic minerals in the presence of high IRMh% values and high concentrations of total free iron (Fe CDB).
The soil profiles of the western slope showed a distinct pattern in magnetic susceptibility with depth (S [much greater than] P [is less than] H [is greater than] B). The results are consistent with a 2-soil layer explanation that layer H is a buried previous surface layer. However, the S-P-H-B pattern for susceptibility became compressed up-slope from the gully, as shown diagrammatically in Fig. 14. A similar pattern for bulk susceptibility is shown in the profiles of the eastern slope (Fig. 2b). The 2-layer explanation could also cover the similarity in the concentrations of magnetic minerals and magnetic grain sizes (Figs 4-6 and 8-11) through the particle size range in the S-P and H-B sections. It is likely that H (an earlier S) was originally at a slightly higher level, a layer that has become diluted by movement of material from P.
High susceptibility values in the S layer are shown by profiles E1-E3 (Fig. 2b) on the eastern side of the gully. In contrast, those of the western profiles (W1-W3) did not show high values of susceptibility (Fig. 2a). This could be due to more dynamic erosion/ deposition occurring in this lower part of the slope. There is no forest on the western slope as it was cleared for grazing some 110 years ago, whereas the eastern slope is a primary dry sclerophyll forest coming within 5 m of the gully. This is consistent with the suggestion that the low values of [Chi] in the S layer of profiles W1-W3 may be due to erosion.
Singer et al. (1996) suggested another process that may be responsible for high concentrations of magnetic minerals at depth. Neoformation of secondary magnetite with maghemite resulting from the precipitation of iron from the soil solution--a process which can also occur in surface horizons. Given the very large difference in depths of profiles W4-W8 (3.3 and 0.7 m, respectively) it is unlikely that such a process would create the consistent but compressed pattern as the profiles became shallower from W4 to W8 at our site (Fig. 14).
The formation of maghemite in surface soils has frequently been attributed to the dehydration of goethite or lepidocrocite in the presence of organic matter under high temperatures (Mullins 1977). Le Borgne (1955, 1960) studied the effects of fire by heating soil to 700 [degrees] C, and found that the most pronounced change occurred in a reducing atmosphere. He suggested that haematite, [Alpha]-[Fe.sub.2][O.sub.3], was being reduced to magnetite, [Fe.sub.3][O.sub.4], which was subsequently reoxidised to maghemite, [Gamma]-[Fe.sub.2][O.sub.3]. Sandgren and Thompson (1990) considered that fire enhancement was responsible for the magnetic minerals in surface podzolic soils in central Poland. As fires were common in this area of Australia (Flannery 1994) fire enhancement of susceptibility is a strong possibility.
A possible reason for the decrease in concentration of magnetic minerals from the surface to layer P in these profiles is that magnetic minerals were formed in the surface by fire enhancement. If magnetic material was moved down the profile by bioturbation it may then be destroyed by reductive dissolution. Singer et al. (1996) suggested that, where the concentration of magnetic minerals was initially constant through a profile, a down-profile decrease in [Chi] could be caused by dissolution of maghemite if the soil layer is frequently anaerobic, as iron is removed by leaching. Schwertmann (1988) also suggested that maghemite and haematite could be dissolved under the influence of organic matter and anaerobic conditions. Dissolution would reduce the concentration of maghemite as well as perhaps decreasing the magnetic grain size. This matches Schwertmann's (1988) proposal that maghemite and haematite may be transformed via solution to goethite. Synthesis of goethite in the P layer is also consistent with the higher proportions of canted anti-ferromagnetic minerals (goethite and haematite) in this layer (Crockford and Willett 1997). The P layer depths of profiles W4, W6, and W8 are much finer in texture than the S and H layers [the sums of the J and K sizes ([is less than] 10 [micro]m) are between 45 and 59% of the totals, Table 2], and could therefore become anaerobic during wet periods.
In contrast to the H layer, the surface 40 cm of profile W4 had no mottles. It was a uniform light grey colour. One notable difference between the S and H samples of the magnetically separated material, however, is that the H samples, for sizes 1 and 2, had about double the weight proportion of VM material of the S samples (Table 4). The magnetic ratios change between the VM and L material in a similar manner for both S and H. This is further evidence that similar processes formed the magnetic minerals in S and H, and that H may have been an earlier surface soil. The fact that there are no VM particles in size 4 (250-500 [micro]m) suggests that the processes causing formation of ferrimagnetic material are biased toward larger particle sizes. Perhaps the larger particles provide a more stable chemical and physical environment for production of ferrimagnetic material.
Concerning the possible `burning history' of such buried layers in Australia, Schwertmann (1988) stated that `buried soils that were previously heated by fires may be recognised in the field by an abrupt increase in bulk soil magnetism'. This is consistent with the `2-layer' explanation. Truswell and Harris (1982) noted the extensive role that fires played in pre-Quaternary periods in the development of the modern vegetation. Idnurm and Senior (1978) used paleomagnetic stratigraphy to establish Paleocene/Eocene and Miocene ages for soil profiles in south-west Queensland.
Fire enhancement (high temperatures with reducing conditions) could be expected to favour smaller particles because they have higher surface area to volume ratios. If it is assumed that significant proportions of ferrimagnetic material in the surface soils in this study resulted from fire enhancement, what accounts for our findings that the reverse is true? Where are fine the particles with very high [Chi] values, particularly in the surface soils? It is very unlikely that they have been washed away by erosion, because had this happened some of the high [Chi] fines from, say, W21 would appear at sites down the slope (W4, W6, and W8), but there was no sign of them (Table 1).
Possible explanations of the magnetics/particle size relationship are as follows:
(1) Formation of maghemite by low temperature fires, with subsequent higher temperature fires oxidising the maghemite present in smaller particles, possibly to haematite.
(2) The clay and fine silt content of these soil samples (Table 2) is sufficient to promote aggregation, and if the smaller particles were present as aggregates during the initial high temperature reduction, and maghemite and other iron oxides were in coatings (common in sedimentary soils), the outside layer would be most enhanced. As aggregates disintegrate through time, the smaller particles from within the aggregates would carry less of the ferrimagnetic material than the aggregate coating. The extent of dilution would depend on the thickness of the aggregate coating and the aggregate volume.
The latter suggestion is supported by the work of Taylor and Schwertmann (1974a, 1974b), who found highly magnetic concretions in a wide range of Australian soils. The magnetic material in the concretions was predominantly maghemite. Varying proportions of haematite were also evident, which would be responsible for the red colour of the concretions. Many concretions were composed of smaller discrete concretions that had been cemented. The iron oxides in non-magnetic concretions were goethite and/or haematite. The magnetic and non-magnetic concretions together constituted about 60% the total soil in some horizons of some soils, and the proportion of the concretions that were magnetic was in some cases as high as 98% (Taylor and Schwertmann 1974a, 1974b).
The concretions that remained intact after size separation, and those that contained large particles, could be responsible for the larger particle sizes being more magnetic. The more weakly magnetic smaller particles could result from the non-concretion material and from breakdown of the magnetic concretions from which the disaggregated smaller particles within the concretions would be less magnetic than those at or near the surface. The relatively small increases in [Chi] shown by size classes H-K for most profiles (Table 1) could possibly be fine particles broken off the highly magnetic larger particles during size separation. Crockford and Olley (1998) found that breakage and abrasion increased the [Chi] of sizes H-K of a sedimentary soil.
Red/brown concretions were not evident in the S layer samples, even though the magnetics/particle size relationships were the same as for the H level. This could suggest preferential dissolution of haematite.
Another possible explanation for the particle size/magnetics relationships is provided by Schwertmann (1988), who suggested that `maghemite and/or haematite may be wholly or partly dissolved in soils (depending on the effective surface area) under the influence of organic matter and/or hydromorphy, and may be translated via solution to goethite.' The effect of this process would be greater as the particle size diminished, which matches our reported values. This is also consistent with the high IRMh% values recorded in the P layers (Crockford and Willett 1997). This explanation also fits with the positive relationship between MGS and particle size noted earlier. Such dissolution would reduce the crystal size of maghemite, thus influencing the MGS, and in the smaller particle sizes maghemite would be more readily dissolved
Zheng et al. (1991) examined the magnetic properties of a range of particle sizes ([is less than] 1-125 [micro]m) from certain depths of loess and paleosols at Luo Chuan, China. For the paleosols, the concentration of magnetic minerals increased as the particle size diminished, which is the reverse of our study. Also, the magnetic ratios reflecting MGS did not show the same smooth trends through the particle size range. For the loess, the mass magnetic properties changed almost randomly through the particle size range.
In podzolic soils developed on sand dunes in the Lake Gosciaz catchment, central Poland, magnetic susceptibility varied through the particle size range (Sandgren and Thompson 1990), with the highest values shown by the [is less than] 63 [micro]m size. It was also found that the down-profile trend in [Chi] values was different for the different sizes. Both of these features are in marked contrast to our study.
The particle size-[Chi] relationship for granite-derived soils and their sediments in this area of New South Wales (Crockford and Starr 1996; Crockford and Olley 1998) is in marked contrast to that of sedimentary soils, in that the magnetic susceptibility in granitic soils increases as the particle size becomes smaller.
The relationships between particle size and magnetic grain size were an interesting feature of the results. Figure 12 shows the relationships between [Chi]fd% and IRM20/[Chi] for S, P, H, and B samples from profile W4, which show that MGS decreased as the particle size became smaller. The increase in [Chi]fd% is consistent with a diminishing MGS, as is the decrease in IRM20/[Chi]. These parameters behave this way in the 0.03-0.02 [micro]m area of the magnetic response/MGS spectrum (Fig. 7). For these samples, similar relationships exist between [Chi]fd% and IRM20/ARM, with regression coefficients between 0.84 and 0.95 (Crockford and Willett 1997). In addition the values of IRM20/ ARM (Table 1a in Crockford and Willett 1997) diminished from size A to size K, for all depths. This is further evidence for the decrease of MGS from size A to size K. Similar relationships between particle size and frequency dependence to those in this study were shown by Sandgren and Thompson (1990) for 4 horizons of a podzolic soil in central Poland.
As the particle sizes are much greater than the magnetic grain sizes it seems unlikely that the physical dimensions of particles could influence the MGS. As an example, particle size H is 20-40 [micro]m, whereas the dominant MGS in the fraction are [is less than] 0.05 [micro]m. The response through the particle size range may depend on whether the magnetic minerals are within particles, or in coatings where they would be more accessible to soil chemical processes.
Similar relationships between particle size and magnetic properties were found by Crockford and Olley (1998) in a sedimentary soil near Wagga Wagga, in the Murrumbidgee catchment of NSW, and by Crockford and Fleming (1998) in river sediments of sedimentary origin. The apparent changes in MGS with depth, the S to P decrease in IRM20/[Chi] for almost all sizes (Figs 8-11), could be caused by either a smaller MGS or the presence of paramagnetic material--material that has susceptibility but no remanence (Fig. 7). In this part of the magnetic grain size spectrum ([is less than] 0.03 gm), a smaller MGS would increase the [Chi]fd% values and decrease the SIRM/ARM values. For profile W4, the profile most extensively analysed, there was actually a decrease in [Chi]fd% for the sizes measured (Table 1a in Crockford and Willett 1997). Paramagnetic material dilutes the [Chi]fd% value because it does not respond differently to low and high frequency [Chi] measurements, and therefore does not exhibit [Chi]fd.
There was a small decrease in SIRM/ARM for sizes A, B, and C but substantial increases for the other sizes (Crockford and Willett 1997; Figs 12 and 13), i.e. the MGS is larger for most sizes, and cannot be responsible for the decrease in IRM20/[Chi]. It seems therefore that the S to P and H to B decreases in IRM20/[Chi] are due to the presence of paramagnetic material in P and B. The S to P decrease in the ratio SIRM/[Chi] (Crockford and Willett 1997) is also consistent with the presence of paramagnetic material
For profiles W6 and W8 the ratio SIRM/ARM showed no consistent trends with depth (Crockford and Willett 1997), which suggests that for these profiles also there was little change in MGS with depth. For profile W21, however, the MGS decreased with depth as shown by decreases in the ratios IRM20/[Chi] and SIRM/ARM.
Figure 13 shows low concentrations of ferrimagnetic minerals in the presence of high IRMh% values and high concentrations of total free iron (Fe CDB); and there is a very noticeable presence of paramagnetic material. These results could be due to the concentrations of Fe and Al (and possibly other elements) and the physical conditions in the P layer. The presence of Al in an Fe(II) system causes the formation of alpha oxides (goethite and haematite) via ferrihydrite, and the goethite can be Al substituted, while lower concentrations of Al cause production of the gamma oxides lepidocrocite and maghemite (Taylor 1987). The high concentration of Al in the P layer may minimise the concentration of maghemite and favour the production of goethite and ferrihydrite. Ferrihydrite is probably the paramagnetic compound referred to earlier and goethite the canted anti-ferromagnetic compound responsible for the high IRMh% value in the P layer (Fig. 13).
In the H layer the IRMh (percentages) are much lower than in the P layer but the IRMh mass values are greater. The much redder colouration in the H layers suggests that a larger proportion of the canted anti-ferromagnetic material is haematite. It is possible therefore that the high level of Al in the P layer is responsible for the small amount of ferrimagnetic material and the larger amounts of paramagnetic and canted anti-ferromagnetic material. The P layers of all profiles have high proportions of clay and size J (2-10 [micro]m, Table 2). They therefore have high Fe, high clay, and fine silt fractions and little ferrimagnetic material, i.e. most of the Fe is in a paramagnetic or non-magnetic form, the latter being ferrous iron or non-magnetic iron oxy-hydroxides.
The magnetic properties of the subsurface (H layer) were similar to those of the present surface (S) in the higher part of the west transect and all of the east transect. It was concluded that the H layer represents an earlier soil above which a new soil has developed.
On the basis of the susceptibility and the remanence magnetic properties it was concluded that fire enhancement was the origin of the magnetic material in both the present surface (S) and former surface (H) layers, and that maghemite is the dominant magnetic mineral.
The results showed consistent variations of magnetic properties, both mass magnetic values and magnetic grain size, through the range of particle sizes. In contrast to results for soils derived from granitic parent materials in Australia (Crockford and Fleming 1998; Crockford and Olley 1998) and elsewhere (Sandgren and Thompson 1990; Zheng et al. 1991), the magnetic susceptibility of soil from our study (derived from sedimentary material) decreased with decreasing particle size. Magnetic grain size also decreased as particle size became smaller.
The coarser size fractions ([is greater than] 3.36 mm to 250-500 [micro]m), separated by a magnet into grades of different magnetic concentrations, showed the same relationships between particle size and magnetic concentration and magnetic grain size as the finer fractions. This suggests that the magnetic concentrations through the particle size range may be due to the proportion of highly magnetic particles in each size class.
Our findings show that it is unlikely that mineral magnetic properties can be used to source sediments in such gullied soils. In addition, the results show that magnetic susceptibility values, as frequently applied in pedological and sedimentological studies, cannot be used for sourcing sediments without taking particle size into account. And if bulk susceptibility is measured with a probe sensor, the results are strongly influenced by the dominant particle size immediately below the sensor. Thus, the values derived may not be representative. A further constraint is breakage and abrasion of particles during transport, which can substantially alter the magnetic properties (Crockford and Olley 1998). In addition, if sediments are stored in dams, chemically induced changes in magnetic properties can occur (Crockford and Willett 1995a, 1995b). However, these changes should be of little influence when the method is used to quantify the stream sediment inputs at river confluences because the changes will have already occurred during movement from the source.
We thank Professor Frank Oldfield (International Geosphere-Biosphere Programme, Berne, Switzerland) for his early involvement in this project and his helpful advice, and Professor R. J. Wasson (Department of Geography, Australian National University, Canberra, Australia) for his continued support.
Abell RS (1991) Geology of the Canberra 1:100 000 Sheet Area. Bureau of Mineral Resources, Bulletin 233.
Alekseyev AO, Kovalevskaya IS, Morgun YG, Samoylova YM (1988) Magnetic susceptibility of soils in a catena Polvovedeniye 8, 27-35.
Bagina OL, Karpachevskity LO, Kashinskaya IV, Milinov PA, Trukhin VI (1989) Magnetoviscous properties of ash soils of Kamchatka and their use for dating. Polvovedeniye 12, 69-78.
Crockford RH, Fleming PM (1998) Environmental magnetism as a sediment tracer; an interpretation of the methodology and some case studies. Australian Journal of Soil Research 36, 167-184.
Crockford RH, Olley JM (1998) The effects of breakage and abrasion on the magnetic properties of two soils. Hydrological Processes 12, 1495-1505.
Crockford RH, Starr BJ (1996) Sediment tracing using mineral magnetism in the Molonglo/Queanbeyan river system. Technical Memorandum 96.19. CSIRO Division of Water Resources.
Crockford RH, Willett IR (1995a) Drying and oxidation effects on the magnetic properties of sulfidic material during oxidation. Australian Journal of Soil Research 33, 19-29.
Crockford RH, Willett IR (1995b) Magnetic properties of two soils during reduction, drying and reoxidation. Australian Journal of Soil Research 33, 597-609.
Crockford RH, Willett IR (1997) The magnetic properties of a stratographic section of a sedimentary soil in New South Wales, Australia. Technical Report 22.97 (available at http://www.clw.csiro.au/ publications/technical/tr22-97.pdf)
Fine P, Singer M J, Laven R, Verosub K, Southard RJ (1989) Role of pedogenesis in distribution of magnetic minerals in two California chronosequences. Geoderma 44, 287-306.
Fine P, Singer MJ, Laven R, Verosub K, Tenpass J (1993) New evidence for the origin of ferrimagnetic minerals in Loess from China. Soil Science Society of America 57, 1537-1542.
Flannery TF (1994) `The Future Eaters.' Chapter 21. (Reed Books: Melbourne)
Idnurm M, Senior BR (1978) Palaeomagnetic ages of late Cretaceous and Tertiary weathered profiles in the Eromanga Basin, Queensland. Palaeogeography Palaeoclimatology, Palaeoecology, 24, 263-277.
Isbell RF (1996) `The Australian soil classification.' (CSIRO Publishing: Melbourne)
Le Borgne E (1955) Susceptibilitie magnetique anormal de sol superficial. Annals of Geophysics 11, 399-419.
Le Borgne E (1960) Influence du feu sur les proprieties magnetique du sol de du granite. Annals of Geophysics 16, 159-195.
Maher BA (1984) Origins and transformations of magnetic minerals in soils. PhD thesis, University of Liverpool, UK.
Maher BA (1988) Magnetic properties of some synthetic submicron magnetites. Geophysical Journal, Oxford 94, 83-93.
McKeague JA, Brydon JE, Miles NM (1971) Differentiation of extractable forms of iron and aluminum in soil. Soil Science Society of America Proceedings 35, 33-38.
Mehra OP, Jackson ML (1958) Iron oxide removal from soils and clays by a dithionite-citrate system buffered with sodium bicarbonate. Clays and Clay Minerals. In `Seventh National Conference of the Clay and Clay Mineral Society'. Washington, D.C. pp. 317-327.
Mullins CE (1977) Magnetic susceptibility of the soil and its significance in soil science--a review. Journal of Soil Science 28, 223-246.
Sandgren P, Thompson R (1990) Mineral magnetic characteristics of podzolic soils developed on sand dunes in the Lake Gosciaz catchment, central Poland. Physics of the Earth and Planetary Interiors 60, 297-313.
Schwertmann U (1988) In `Iron in soils and clay minerals.' (Eds JW Stucki, BA Goodman, U Schwertmann) p. 386. (D. Reidel Publishing Company)
Singer MJ, Fine P (1989) Pedogenic factors affecting magnetic susceptibility of northern California soils. Soil Science Society of America 53, 1119-1127.
Singer MJ, Verosub KL, Fine P, TenPas J (1996) A conceptual model for the enhancement of magnetic susceptibility in soils. Quaternary International 34-36, 243-248.
Taylor RM (1987) Non silicate oxides and hydroxides. In `Chemistry of clays and clay minerals'. (Ed. ADC Newman) Mineralogical Society Monograph No. 6.
Taylor RM, Schwertmann U (1974a) The association of phosphorus with iron ferruginous soil concretions. Australian Journal of Soil Research 12, 133-145.
Taylor RM, Schwertmann U (1974b) Maghemite in Soils and its origins. Properties and observations on soil maghemites. Clay Minerals 10, 289-298.
Thompson R, Oldfield F (1986) `Environmental magnetism.' (Allan and Unwin: London)
Truswell EM, Harris WK (1982) The Cainozoic paleo-botanical record in arid Australia: fossil evidence for the origins of arid-adapted flora. In `Evolution of the flora and fauna of arid Australia'. (Eds WR Barker, PJM Greenslade) pp. 67-76. (Peacock Publications in Association with the Australian Systematic Botanical Society: Frewville, S. Aust.)
Zheng H, Oldfield F, Yu L, Shaw J, An Z (1991) The magnetic properties of particle-sized samples from the Luo Chuan loess section; evidence of pedogenesis. Physics of the Earth and Planetary Interiors 68, 250-258.
Zhou LP, Oldfield F, Wintle AG, Robinson SG, Wang JT (1990) Partly pedogenic origin of magnetic variations in Chinese Loess. Nature 346, 737-739.
R. H. Crockford(A) and I. R. Willett(B)
(A) CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601, Australia.
(B) ACIAR, GPO Box 1571, Canberra, ACT 2601, Australia.
Manuscript received 15 September 2000, accepted 15 March 2001
|Printer friendly Cite/link Email Feedback|
|Author:||Crockford, R. H.; Willett, I. R.|
|Publication:||Australian Journal of Soil Research|
|Article Type:||Statistical Data Included|
|Date:||Sep 1, 2001|
|Previous Article:||Brazilian latosols and their B horizon microstructure as long-term biotic constructs.|
|Next Article:||Fractionation of cobalt and manganese in New Zealand soils.|