Soil classification groups to quantify primary salinity, sodicity and alkalinity in South African soils

Abstract

This study aimed to quantify the primary salinity, sodicity and alkalinity of South African soils based on soil classification groups and rainfall classes, by using available analysed data. The 73 soil forms used by the Soil Classification Working Group were organised into 11 groups based on the occurrence of specific horizons. Soil analyses of 12 552 profiles for pH_water, 11 712 profiles for electrical conductivity, and 12 174 profiles for exchangeable sodium and cation exchange capacity were used. Median and not average values best represented salt-affected soil data on a national scale, because most data was strongly positively skewed. Primary soil alkalinity and sodicity were more common than primary salinity in terms of the different South African soil groups. None of the soil groups were saline, only the prismacutanic group was sodic, while the calcic, alluvial/aeolian, neocutanic and prismacutanic groups were alkaline when the median value or highest value in a profile was used as indicator. Sufficient data was not available, even for this relatively large dataset of 648 prismacutanic B-horizons, to proclaim a precise exchangeable sodium percentage value as a criterion for prismacutanic or solonetz soils in South Africa. Salinity, sodicity and alkalinity decreased as rainfall increased due to the depletion of cations and anions. The incidence of salt-affected soils in the soil classification groups decreased in the order: calcic ≥ alluvial/aeolian > prismacutanic > vertic > pedocutanic/red structured > neocutanic > hydromorphic ≥ lithosols > plinthic > apedal > podzolic. This information serves as an indicator of the primary sodicity and alkalinity of South African soils and can be utilised to deduce the expected sodicity and alkalinity of soils from soil survey data, to monitor anthropogenic sodicity and alkalinity impacts, and as an input to map the primary sodicity and alkalinity of South African soils.

Keywords:

• climate
• leaching
• rainfall
• salt-affected soil

Introduction

Salt-affected soils are represented differently in different soil classification systems and appear on different taxonomical levels (Szabolcs 1997). Although sodium is not used, the presence of free carbonates, calcium carbonate, or calcium-magnesium carbonate is used in the South African soil classification system (Soil Classification Working Group 1991) to define diagnostic horizons, materials and families.

Soils are inherently variable, both spatially and temporally, in their physical and chemical characteristics. Usually the variability is much greater vertically than horizontally, resulting from the variability in soil-forming processes. Soil variability, in turn, results in variability in the distribution of water and salts and in the ease with which they are transported within and removed from the soil. Salts move from the topsoil to the subsoil during wet periods and from the subsoil to the topsoil during dry periods (Nell and Lea 2004). This vertical fluctuation makes the quantification of salts per horizon complex, unless a large data set is used to lessen the effect of temporal variation.

Salt-affected soils generally occur in regions that receive salts from other areas and where water is the primary carrier (FAO 2001). Although the weathering of rocks and minerals is the source of all salts, the salt-affected soils are rarely formed from in situ accumulation of salts. A well-developed profile in low rainfall areas usually carries at some point (usually in the C-horizon) a calcium carbonate accumulation greater than that of its parent material (Brady 1990). Soluble salts occur in significant proportions in the soils of arid and semi-arid areas where they accumulate because annual precipitation is insufficient to leach the salts. According to Netterberg (1969), hardpan and boulder calcretes generally only occur in areas receiving less than 550 mm rain per year.

Van der Merwe (1940) concluded that saline soils in South Africa do not occur in extensive areas or a specific climatogenic soil zone, but are found in small to fairly big patches in several soil groups, due to localised factors. He did, however, classify and map ‘Solonetzic Soils‘ as one soil group in his study on the ‘Soil Groups and Subgroups of South Africa’. According to him, this soil group, occurring in the south-central Free State, covers a fairly extensive area and lies between the Prairie Soils in the east, the Kalahari Sand on Limestone in the northwest, and the Karoo or semi-desert soils in the west. The major portion of this soil group occurs in the Free State province, with two comparatively small areas in the Eastern Cape province. One area is confined to the Burgersdorp area and immediate surroundings, while the other has Queenstown as its centre. MacVicar (1972) produced a sketch map giving a tentative indication of the occurrence of salt-affected soils in South Africa, without quantifying the area or the intensity of the problem. According to him, it is not possible to make meaningful statements about the associated salt-affected soils in each map unit at this scale. No subsequent summary of salt-affected soils in South Africa has been done.

The aim of this study was therefore to quantify the primary salinity, sodicity and alkalinity of South African soils using available analysed data, based on soil classification groups and rainfall classes. A second paper (Nell and van Huyssteen 2014) discusses the relationship between primary soil salinity, sodicity and alkalinity with geology and groundwater regions.

Methodology

Soil samples (n = 21 252), collected by the Agricultural Research Council (ARC)–Institute for Soil, Climate and Water during various soil surveys for irrigation, environmental planning and from the national land type survey were used in this study. Minimum requirements for inclusion were (1) the profiles should have comprehensive chemical and physical analyses, (2) accurate profile location information should be available and (3) only data for primary (non-anthropogenic) profiles were used. A total of 11 712 profiles for electric conductivity, 12 552 profiles for pH_water and 12 174 profiles for exchangeable sodium percentage (ESP) were used due to differences in analyses.

Coordinates of the profiles were imported into ArcView 9.2 (ESRI, Redlands, CA, USA) to map the positions of the profiles. Profile locations were digitised from 1:50 000 topo cadastral maps (Chief Director of Surveys and Mapping 1993) published in the Clark 1880 ellipsoid (Cape datum), converted to the WGS84 ellipsoid and overlaid with other spatial data (Figure 1).

Figure 1: Distribution of the 12 552 soil profiles and the 550 mm rainfall isohyet

Soil samples were analysed for pH_water, electrical conductivity (EC) and cation exchange capacity according to the Non-Affiliated Soil Analysis Work Committee (1990). Soil samples in the database were predominantly analysed for CEC and extractable sodium using NH₄OAc (1 mol dm^–3, pH 7) as extractant. Less than 12% of samples used in this study were analysed by LiCl extraction. These samples were also predominantly from the higher rainfall areas in South Africa and therefore largely non-saline and non-sodic. Any discrepancies in the results between the two methods were therefore expected to be negligible for the purpose of this study and were therefore pooled. Sodium and electrical conductivity were also determined by suction filtration of the water-saturated soil paste. Soil pH was determined using a 1:2.5 soil to water suspension. Alkalinity and acidity was classified according to van der Watt and van Rooyen (1995).

All the values in the profile were used for calculation, because of the dynamic nature of salts in a profile due to climatic factors, especially rainfall. The soil profile was regarded as an array of layer attributes (Williams 1976), equivalent to the A (topsoil), B and C (subsoil) master horizons. The soil profile was further considered to be an isotropic body where only the highest value occurring in a profile was used. All A master horizons were considered as topsoil, while B, C and E master horizons were considered as subsoil.

The 73 soil forms of the Soil Classification Working Group (1991) were organised into 11 groups predominantly based on distinctive subsoil horizons or material (Table 1). The groups therefore primarily focused on the pedogenic accumulation of carbonates, structure development and wetness. Soil profiles classified according to MacVicar et al. (1977) were reclassified by the authors according to the Soil Classification Working Group (1991).

Table 1: Grouping of soil forms (Soil Classification Working Group 1991; Fey 2010) based on the presence of specific diagnostic horizons or materials

Soil Group	General description	Soil Form
Calcic	Soils characterised by the accumulation of secondary lime either as hardpan or as soft carbonate	Montagu, Augrabies, Brandvlei, Coega, Addo, Prieska, Trawal, Plooysburg, Etosha, Molopo, Askham, Kimberley, Kinkelbos, Steendal, Immerpan, Gamoep
Alluvial/Aeolian	Youthful soils formed in unconsolidated alluvium or aeolian sands	Dundee, Namib
Neocutanic	Youthful soils formed in unconsolidated sediments such as colluvium, alluvium or aeolian sands that are characterised by the absence of distinctive pedogenesis	Tukulu, Oudtshoorn, Oakleaf, Sweetwater, Vilafontes
Pedocutanic/Red Structured	Soils with distinct subsoil clay enrichment resulting in blocky or prismatic structure	Sepane, Valsrivier, Swartland, Lusiki, Bonheim, Klapmuts, Shortlands
Prismacutanic	Soils with distinct subsoil clay enrichment resulting in an abrupt transition with the overlying horizon and a strong prismatic structure	Sterkspruit, Estcourt
Vertic	Soils with swelling, high activity clay (with strong wedge structure and slickensides) that developed from basic parent materials in semi-arid to subhumid climates	Arcadia
Hydromorphic	Soils associated with intense reduction, either a fluctuating, perched water table or through stagnant water saturation	Rensburg, Katspruit, Kroonstad, Champagne, Pinedene, Willowbrook
Plinthic	Soils characterised by iron oxide enrichment (e.g. mottles, nodules, concretions and hardpan) through a fluctuating water table	Longlands, Wasbank, Avalon, Dresden, Glencoe, Bainsvlei, Bloemdal, Westleigh
Apedal	Well-drained, aerated soils characterised by a uniform red or yellow colour, apedal structure and the absence of lime	Griffin, Clovelly, Garies, Hutton, Kranskop, Magwa, Inanda, Constantia, Fernwood
Lithosols	Shallow soils with limited weathering, grading into underlying saprolite or hard rock	Mispah, Glenrosa, Mayo, Milkwood, Nomanci, Knersvlakte, Cartref
Podzolic	Soils characterised by a dark podzolic subsoil horizon, enriched with Fe, Al and organic matter, under a bleached horizon	Lamotte, Concordia, Houwhoek, Groenkop,Pinegrove, Witfontein

Data for EC, ESP and pH_water data were ranked from highest to lowest median values to establish which soil groups were most affected by salts. The soil groups with the highest median value were ranked 11th and those with the lowest median value were ranked first. The median and geometric mean was then calculated for each soil group, but because some soil groups had the same median ratings the geometric mean was used to rank the different soil groups (from the most likely to the least likely to be affected by salts). Primary soil alkalinity was taken as pH_water>7.4, sodicity at two values, namely ESP>6% and ESP>15%, and salinity as EC>400 mS m^-1 (United States Salinity Laboratory Staff 1954).

South Africa was divided into a high (>550 mm) and a low (< 550 mm) rainfall area. An average annual rainfall value for each profile was obtained by overlaying the profile points with the modelled 1 km × 1 km average annual rainfall grid of the ARC–Institute for Soil, Climate and Water (Figure 1).

Medians, averages (geometric mean), lower quartiles and upper quartiles were used to identify relationships between the soil groups (Statgraphics 2005). The 95% Bonferroni multiple comparison procedure was used to determine significance differences (p<0.05) of means between the rainfall classes. In this study outliers were considered to be ‘any observation that is not a realization from the target distribution’ rather than ‘any observation that appears surprising or discrepant to the investigator’ (Beckman and Cook 1983).

Results and discussion

The large differences between the median and average values for salinity (Tables 2–4) and sodicity (Tables 5–7) indicated the variability and skewness of the data. Division of the soil groups into high and low rainfall classes did not remove the skewness from the salinity and sodicity data. Outliers with values several-fold percent higher than the lowest value were not unusual. The small differences between the average and median pH_water values (Tables 8–10) indicated a normal distribution pattern.

Table 2: Statistics for electrical conductivity (mS m⁻¹) values in the topsoil horizons of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Vertic	<550 >550 All	159 143 150	59 55 59	230 233 230	228^a 197^a 205	310 274 284	62 167 229
Alluvial/Aeolian	<550 >550 All	59 35 51	25 23 24	215 62 158	543^a 69^a 397	1389 110 1173	115 52 167
Calcic	<550 >550 All	47 35 41	30 23 28	98 59 87	233^a 67^a 193	979 156 859	285 90 375
Neocutanic	<550 >550 All	45 26 36	30 17 23	112 41 70	278^a 42^a 188	970 64 772	547 336 883
Pedocutanic/Red structured	<550 >550 All	56 36 41	34 25 27	125 59 78	209^a 63^a 108	588 114 345	333 758 1091
Prismacutanic	<550 >550 All	41 28 28	21 19 19	80 41 50	111^a 47^a 69	223 74 146	134 261 395
Lithosols	<550 >550 All	38 26 28	26 18 18	68 38 42	120^a 38^a 56	512 88 254	310 1097 1407
Hydromorphic	<550 >550 All	26 25 26	15 15 15	137 49 53	522^a 68^a 158	2067 203 953	105 424 529
Plinthic	<550 >550 All	22 21 21	13 14 14	21 34 34	28^a 34^a 33	37 180 57	127 520 645
Apedal	<550 >550 All	25 17 20	16 12 12.8	41 26 31	78^a 24^a 43	349 33 207	1016 1962 2978
Podzolic	<550 >550 All	26 20 17	7 9 9	36 28 28	23^a 20^a 20	15 14 13	3 22 25

1 lues within the same soil group followed by different letters are significantly different at p < 0.05

Table 3: Statistics for electrical conductivity (mS m⁻¹) values in the subsoil horizons of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Calcic	<550	120	49	310	341^a	693	369
	>550	83	41	186	185^b	286	142
	All	107	43	275	298	612	511
Alluvial/ Aeolian	<550	115	49	489	680^a	1 370	107
	>550	30	11	48	51^b	76	47
	All	71	30	279	488	1178	154
Prismacutanic	<550	114	49	251	233^a	397	208
	>550	49	28	140	134^b	253	403
	All	70	33	178	168	313	611
Neocutanic	<550	100	45	350	405^a	855	930
	>550	26	15	49	61^b	116	463
	All	59	30	200	291	720	1 393
Pedocutanic/ Red structured	<550	105	49	238	298^a	679	479
	>550	41	26	87	98^b	199	1031
	All	54	32	128	161	426	1510
Hydromorphic	<550	39	13	231	599^a	2 087	155
	>550	26	12	59	63^b	116	463
	All	28	12	72	197	1 073	618
Lithosols	<550	67	36	147	382^a	1 566	121
	>550	21	15	38	37^b	54	424
	All	28	17	52	113	751	545
Plinthic	<550	29	16	49	53^a	140	230
	>550	18	11	32	33^b	72	943
	All	20	12	36	37	90	1 173
Apedal	<550	33	20	59	151^a	475	1 410
	>550	13	7	25	22^b	36	3 137
	All	18	9	36	62	273	4 547
Podzolic	<550	15	7	28	18^a	12	7
	>550	12	4	18	16^a	18	48
	All	12	4	18	16	17	55

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Table 4: Statistics for the highest electrical conductivity (mS m⁻¹) value in a profile of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Vertic	<550	190	99	300	280^a	346	53
	>	138	52	233	203^b	298	130
	All	159	59	252	225	313	183
Calcic	<550	108	41	315	412^a	1 063	306
	>550	87	41	224	206^b	311	97
	All	103	41	290	362	938	403
Prismacutanic	<550	173	90	354	323^a	478	136
	>550	59	34	165	157^b	283	272
	All	103	40	232	213	368	408
Alluvial/	<550	76	25	385	878^a	1 823	107
Aeolian	>550	38	23	88	88^b	132	44
	All	60	23	273	648	1 573	151
Neocutanic	<550	115	45	412	514^a	1 183	594
	>550	28	18	52	88^b	113	405
	All	54	28	203	331	937	999
Pedocutanic/	<550	124	54	274	347^a	750	346
Red structured	>550	41	28	87	96^b	188	971
	All	49	32	126	162	431	1 317
Hydromorphic	<550	84	26	323	706^a	2 330	104
	>550	30	17	64	80^b	205	451
	All	35	17	92	197	1 043	555
Lithosols	<550	42	28	89	203^a	1 013	318
	>550	27	18	41	41^b	91	1 129
	All	28	20	47	27	20	1 447
Plinthic	<550	36	20	59	66^a	168	137
	>550	22	15	40	41^a	87	785
	All	24	15	43	45	103	922
Apedal	<550	36	22	59	150^a	504	1 161
	>550	18	11	30	28^b	42	2 971
	All	22	13	39	62	275	4 132
Podzolic	<550	28	10	36	25^a	13	3
	>550	20	14	36	27^a	21	25
	All	20	13	36	26	20	28

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Table 5: Statistics for exchangeable sodium percentage values in the topsoil horizons of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Alluvial/ Aeolian	>550 All	4.7	1.7	12.5	15.6^a	34.5	121
	<550	1.8	0.8	3.4	2.4^b	2.3	63
	All	2.9	1.3	7.2	11.1	28.7	184
Prismacutanic	<550	3.5	1.5	7.7	6.9^a	11.1	137
	>550	2.8	1.6	5.0	4.5^b	4.3	256
	All	2.9	1.6	6.0	5.1	7.5	393
Hydromorphic	<550	4.9	1.9	10.0	27.6^a	158.6	105
	>550	2.4	1.0	4.9	4.0^b	6.3	416
	All	2.7	1.1	5.5	8.7	71.8	521
Vertic	<550	3.5	1.2	6.9	5.1^a	5.9	72
	>550	2.0	0.8	5.6	5.6^a	9.1	179
	All	2.3	0.9	6.0	5.5	8.3	251
Neocutanic	<550	2.6	1.3	7.0	10.5^a	30.7	637
	>550	1.2	0.7	2.6	3.1^b	7.1	396
	All	2.0	1.0	4.9	7.7	24.8	1 033
Pedocutanic/ Red structured	<550	2.2	1.0	6.2	7.4^a	17.6	409
	>550	1.5	0.8	3.3	3.4^b	6.8	712
	All	1.7	0.8	4.1	4.9	12.1	1 121
Calcic	<550	1.4	0.5	3.3	6.3^a	17.0	130
	>550	1.7	1.1	3.6	5.2^b	9.3	30
	All	1.6	0.7	3.3	6.1	15.9	160
Lithosols	<550	2.1	1.0	4.8	5.6^a	17.6	299
	>550	1.5	0.9	2.5	2.2^b	3.1	926
	All	1.6	0.9	2.9	3.1	9.2	1 225
Plinthic	<550	2.2	0.6	2.6	2.2^a	2.5	125
	>550	1.6	0.6	2.9	2.8^a	5.5	530
	All	1.6	0.6	2.8	2.6	5.1	655
Apedal	<550	2.0	0.9	3.8	5.1^a	20.6	1 047
	>550	0.9	0.4	1.6	1.4^b	2.7	1 980 3 027
	All	1.2	0.5	2.2	2.7	12.4
Podzolic	<550	1.8	0.8	2.8	1.8^a	1.2	4
	>550	2.9	1.4	3.9	4.0^b	4.7	23
	All	2.9	1.1	3.8	3.7	4.4	27

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Table 6: Statistics for exchangeable sodium percentage values in the subsoil horizons of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Prismacutanic	<550	9.1	4.7	19.2	15.1^a	18.3	211
	>550	8.0	4.2	15.8	12.2^b	12.6	437
	All	8.4	4.4	17.1	13.1	14.7	648
Calcic	<550	5.3	2.0	15.0	14.0^a	22.8	382
	>550	7.1	3.5	15.9	11.6^b	11.8	142
	All	6.1	2.3	15.2	13.3	13.3	524
Hydromorphic	<550	7.7	3.5	24.1	42.6^a	130.7	151
	>550	4.1	1.8	8.8	6.9^b	8.4	567
	All	4.8	2.0	10.0	14.4	61.9	718
Alluvial/ Aeolian	<550	6.1	2.5	30.8	32.7^a	94.7	114
	>550	1.9	1.0	4.1	2.9b	2.6	68
	All	4.0	1.5	10.3	21.6	76.2	182
Neocutanic	<550	6.5	2.3	18.4	17.9^a	32.2	913
	>550	2.0	1.1	3.9	4.8^b	11.0	466
	All	3.5	1.6	11.5	13.5	27.6	1 379
Lithosols	<550	3.6	1.7	10.6	15.9^a	45.2	123
	>550	2.8	1.5	4.7	4.2^b	5.3	469
	All	3.0	1.6	5.2	6.7	21.6	592
Vertic	<550	3.4	0.2	1.2	5.1^a	5.9	72
	>550	2.0	0.8	5.6	5.6^a	9.0	179
	All	2.6	0.9	6.0	5.5	8.3	251
Pedocutanic/ Red structured	<550	3.4	1.5	13.5	11.5^a	21.4	480
	>550	2.0	1.0	4.6	5.5^b	15.0	1 121
	All	2.2	1.1	6.5	7.3	17.4	1 601
Plinthic	<550	2.1	1.1	4.3	4.4^a	9.7	224
	>550	2.2	1.1	4.2	4.0^a	6.8	1 107
	All	2.2	1.1	4.2	4.1	7.4	1 331
Apedal	<550	2.5	1.3	5.9	9.1^a	22.8	1 431
	>550	1.2	0.6	2.2	2.2^b	19.5	3 551
	All	1.5	0.7	2.8	4.1	20.7	4 982
Podzolic	<550	1.5	0.4	2.3	1.5^a	1.1	8
	>550	4.2	2.6	6.8	5.9^b	5.7	50
	All	3.6	2.2	6.3	4.1	5.5	58

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Table 7: Statistics for the highest exchangeable sodium percentage values in a profile of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Prismacutanic	<550	10.9	4.7	21.8	17.3^a	18.9	135
	>550	9.6	4.3	18.0	18.0^b	13.3	267
	All	10.1	4.4	19.6	17.7	15.5	402
Podzolic	<550	2.0	1.1	2.8	2.0^a	1.1	4
	>550	5.8	3.6	12.5	8.1^b	7.3	26
	All	5.3	2.9	10.0	7.3	7.1	30
Hydromorphic	<550	10.0	5.0	30.2	56.5^a	198.0	105
	>550	4.6	2.2	10.0	8.0^b	10.0	414
	All	5.2	2.4	12.0	17.8	91.1	519
Calcic	<550	4.2	1.4	12.7	13.1^a	24.0	336
	>550	6.9	2.4	16.2	11.5^b	16.1	104
	All	4.9	1.7	13.7	12.7	21.8	440
Alluvial/ Aeolian	<550	5.2	1.9	16.8	31.0^a	96.8	114
	>550	2.1	1.2	4.2	3.5^b	3.1	55
	All	3.9	1.6	10.0	22.0	80.4	169
Vertic	<550	5.1	2.0	9.7	7.0^a	6.5	60
	>550	3.1	1.4	8.6	8.1^a	11.4	126
	All	3.8	1.6	9.6	7.8	10.1	186
Neocutanic	<550	7.5	2.6	23.0	22.2^a	43.3	566
	>550	2.1	1.2	4.2	5.2^b	11.6	395
	All	3.7	1.8	12.5	15.2	35.0	961
Plinthic	<550	3.1	1.6	5.9	5.7^a	1.6	136
	>550	2.4	1.3	4.4	4.6^a	8.2	792
	All	2.5	1.3	4.6	4.7	1.3	928
Pedocutanic/ Red structured	<550	3.5	1.6	14.5	12.8^a	23.5	345
	>550	2.0	1.1	4.3	5.5^b	16.1	960
	All	2.2	1.2	6.0	7.4	18.6	1 305
Lithosols	<550	2.3	1.1	5.0	8.9^a	31.1	308
	>550	1.9	1.1	3.6	3.1^b	4.3	946
	All	2.0	1.1	3.9	4.5	16.0	1 254
Apedal	<550	2.8	1.5	5.9	9.4^a	27.9	1 180
	>550	1.4	0.7	2.4	2.4^b	21.1	3 018
	All	1.7	0.8	3.0	4.4	23.4	4 198

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Table 8: Statistics for pH_water values in the topsoil horizons of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
	<550	8.5	8.2	8.8	8.4^a	0.72	132
Calcic	>550	8.4	7.8	8.8	8.2^a	0.90	30
	All	8.5	8.1	8.8	8.4	0.76	162
Alluvial/ Aeolian	<550	8.3	7.5	8.9	8.1^a	0.97	124
	>550	7.1	6.2	8.1	7.1^b	1.10	65
	All	8.0	7.1	8.6	7.8	1.13	189
Vertic	<550	8.1	7.9	8.6	8.1^a	0.61	62
	>550	7.7	7.0	8.3	7.6^a	0.90	212
	All	7.9	7.1	8.3	7.7	0.88	274
Neocutanic	<550	7.9	7.1	8.5	7.8^a	0.95	667
	>550	6.3	5.8	6.9	6.4^b	0.84	454
	All	7.2	6.3	8.1	7.2	1.12	1 121
Pedocutanic/ Red structured	<550	7.5	6.8	8.2	7.5^a	0.92	431
	>550	6.4	5.9	7.0	6.5^b	0.84	796
	All	6.7	6.1	7.5	6.9	1.00	1 227
Prismacutanic	<550	6.9	6.5	7.8	7.1^a	0.85	141
	>550	6.2	5.8	6.7	6.3^b	0.65	279
	All	6.4	6.0	7.0	6.6	0.82	420
Lithosols	<550	7.9	6.8	8.5	7.7^a	1.03	328
	>550	6.0	5.6	6.4	6.1^a	0.78	1 181
	All	6.2	5.7	6.9	6.4	1.07	1 509
Hydromorphic	<550	6.8	6.1	8.1	7.1^a	1.23	111
	>550	6.1	5.5	6.8	6.3^b	1.04	488
	All	6.2	5.6	7.1	6.4	1.12	599
Plinthic	<550	6.4	5.9	6.9	6.4^a	0.79	138
	>550	5.8	5.4	6.3	5.9^a	0.71	593
	All	5.9	5.5	6.4	6.0	0.76	731
Apedal	<550	7.2	6.3	8.1	7.2^a	1.01	1 087
	>550	5.5	5.1	6.1	5.7^b	0.74	2 176
	All	5.9	5.3	6.8	6.2	1.13	3 263
Podzolic	<550	5.5	5.2	7.0	6.1^a	1.63	4
	>550	5.6	5.2	6.0	5.6^b	0.69	24
	All	5.5	5.2	6.0	5.7	0.86	28

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Table 9: Statistics for pH_water values in the subsoil horizons of the different soil groups

Soil Group^Table Footnote1	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote2	Standard deviation	Sample size
	<550	8.3	7.9	8.7	8.3^a	0.66	388
Calcic	>550	8.3	8.0	8.7	8.4^a	0.64	148
	All	8.3	7.9	8.7	8.3	0.65	536
Alluvial/ Aeolian	<550	8.3	7.5	8.6	8.1^a	0.90	119
	>550	6.7	5.9	7.7	6.8^b	1.03	61
	All	8.0	6.9	8.5	7.7	1.12	180
Neocutanic	<550	8.3	7.8	8.7	8.2^a	0.90	932
	>550	6.8	6.0	7.6	6.8^b	1.06	498
	All	8.0	7.0	8.5	7.7	1.14	1 430
Prismacutanic	<550	7.9	7.1	8.4	7.8^a	1.02	215
	>550	7.1	6.4	8.0	7.2^a	1.04	423
	All	7.4	6.6	8.2	7.4	1.06	638
Pedocutanic/ Red structured	<550	7.6	6.9	8.3	7.6^a	0.94	490
	>550	6.6	6.1	7.5	6.8^b	0.96	1 071
	All	7.0	6.3	7.8	7.1	1.02	1 561
Lithosols	<550	7.4	6.7	8.2	7.4^a	1.07	126
	>550	6.4	5.8	7.0	6.4^b	0.89	454
	All	6.5	6.0	7.3	6.6	1.00	580
Hydromorphic	<550	7.0	6.2	8.0	7.1^a	1.18	161
	>550	6.5	5.7	7.4	6.6^a	1.16	527
	All	6.5	5.8	7.6	6.7	1.19	688
Plinthic	<550	6.4	5.8	6.9	6.4^a	0.92	239
	>550	5.9	5.4	6.4	6.0^a	0.82	1 012
	All	6.0	5.4	6.6	6.1	0.86	1 251
Apedal	<550	7.2	6.4	8.2	7.2^a	1.15	1 437
	>550	5.5	5.2	6.2	5.7^b	0.75	3 370
	All	5.9	5.3	6.8	6.2	1.13	4 807
Podzolic	<550	5.7	5.2	5.9	5.9^a	1.15	8
	>550	5.7	5.3	6.1	5.7^a	0.60	50
	All	5.7	5.3	6.1	5.7	0.68	58

1 The vertic soil group is excluded here because very few subsoil samples thereofwere available for this analysis

2 Values within the same soil group followed by different letters are significantlydifferent at p < 0.05

Table 10: Statistics for the highest pH_water value in a profile of the different soil groups

Soil Group	Rainfall (mm)	Median	Lower quartile	Upper quartile	Average^Table Footnote1	Standard deviation	Sample size
Calcic	<550	8.6	8.2	8.9	8.5^a	0.6	340
	>550	8.5	8.1	8.9	8.4^a	1.1	106
	All	8.6	8.2	8.9	8.5	0.7	446
Alluvial / Aeolian	<550	8.6	7.7	9.0	8.3^a	0.9	117
	>550	7.0	6.3	8.2	7.0^b	1.4	58
	All	8.2	7.1	8.8	8.2	1.3	175
Vertic	<550	8.3	7.9	8.7	8.2^a	0.6	52
	>550	8.0	7.1	8.4	7.8^a	0.9	160
	All	8.1	7.3	8.5	8.1	0.9	212
Neocutanic	<550	8.5	8.0	8.8	8.3^a	0.9	597
	>550	6.7	6.0	7.6	6.8^b	1.0	440
	All	7.9	6.8	8.6	7.7	1.2	1 037
Prismacutanic	<550	8.2	7.4	8.6	8.0^a	1.1	141
	>550	7.3	6.4	8.1	7.3^b	1.1	289
	All	7.6	6.8	8.4	7.5	1.2	430
Pedocutanic / Red structured	<550	7.7	7.0	8.4	7.7^a	0.9	362
	>550	6.5	6.1	7.3	6.7^b	1.2	1 033
	All	6.8	6.2	7.7	6.9	1.2	1 395
Hydromorphic	<550	7.5	6.5	8.5	7.5^a	1.2	111
	>550	6.4	5.7	7.5	6.6^b	1.3	508
	All	6.5	5.8	7.7	6.7	1.3	619
Lithosols	>550	8.0	6.9	8.5	7.7^a	1.0	334
	<550	6.1	5.6	6.7	6.2^b	0.9	1 206
	All	6.3	5.8	7.2	6.5	1.2	1 540
Plinthic	<550	6.7	6.1	7.4	6.8^a	0.9	147
	>550	5.9	5.4	6.2	6.0^b	1.1	855
	All	6.0	5.4	6.6	6.1	1.1	1 002
Apedal	<550	7.5	6.6	8.4	7.5^a	1.1	1 194
	>550	5.6	5.2	6.2	5.8^b	0.8	3 194
	All	6.0	5.3	6.9	6.2	1.2	4 388
Podzolic	<550	5.9	5.6	7.3	6.4^a	1.5	4
	>550	5.7	5.3	6.3	5.7^a	0.7	27
	All	5.7	5.4	6.3	5.8	0.8	31

1 Values within the same soil group followed by different letters are significantly different at p < 0.05

Electrical conductivity

Significant differences were found between rainfall classes for median EC based on both the topsoil and subsoil values, namely calcic, alluvial/aeolian, neocutanic, lithosols, pedocutanic/red structured, hydromorphic, prismacutanic, apedal and plinthic soil (Tables 2–4). This indicated the importance of rainfall class and therefore leaching on salt movement, even within the same soil group. No significant differences were recorded between rainfall classes for the topsoil and subsoil of vertic and podzolic, and for the topsoil of the plinthic group. Possible reasons for this inconsistency were that the leaching potential for vertic soils was much lower than that for other soil groups (because of the high clay content and resulting low infiltration rate). The podzolic soils were mostly found in areas where the annual rainfall was >550 mm. For the plinthic group, the capillary movement of water and salts from the fluctuating water table in the subsoil to the topsoil occurred both under low and high rainfall conditions to the same degree.

There was a decrease in median EC values for all the topsoil horizons in the order: vertic > alluvial/aeolian > calcic > neocutanic > pedocutanic/red structured > prismacutanic > lithosols > hydromorphic > plinthic > apedal > podzolic soil groups (Table 2). Eight of the 11 soil groups had significantly different median EC topsoil values between the rainfall classes.

Median EC values decreased for all the subsoil horizons (Table 3) in the order calcic > alluvial/aeolian > prismacutanic > neocutanic > pedocutanic/red structured > hydromorphic > lithosols > plinthic > apedal > podzolic. When the highest value in a profile was used, EC decreased in the order vertic > calcic > prismacutanic > alluvial/aeolian > neocutanic > pedocutanic/red structured > hydromorphic > lithosols > plinthic > apedal > podzolic. The higher salinity in the neocutanic subsoils than the pedocutanic/red structured subsoils was not expected. This irregularity could be ascribed to the large amount of samples from the high rainfall class (more leaching) for the pedocutanic/red structured soil group compared to the neocutanic soil group. If median values from the low rainfall class was used, the sequence changed to calcic > alluvial/aeolian > prismacutanic > pedocutanic/red structured > neocutanic > lithosols > hydromorphic > apedal > podzolic. This was more in agreement with the findings of Nell (1991) and Nell and Bennie (1991) where an increase in salt content accompanied an increased degree of structural development (neocutanic to pedocutanic to prismacutanic). The apparent anomaly of the relatively high salt content values for the alluvial/aeolian soil group was attributed to the position of the alluvial soil in lower terrain units and the subsequent accumulation of salts and because most of the aeolian soils were found in low rainfall areas. Pedogenic changes were minimal in transported alluvial and aeolian soils, which were reflected in the low degree of salt leaching. The lowest EC for the topsoil, subsoil and highest value in a profile were respectively found in the plinthic, apedal and podzolic soil groups. This was an indication of good salt leaching conditions out of these soils.

The EC of 31 rainfall classes of subsoil horizons differed significantly (p < 0.05). Soil of the alluvial/aeolian group differed significantly from all other soil groups, except for soils in the vertic group. When the highest value in a profile was used, 35 rainfall classes showed significant differences for EC and again the alluvial/aeolian was the only group that was significantly different from all other groups.

Only the alluvial/aeolian and hydromorphic groups tended to be saline for the topsoil and subsoil horizons under the following conditions: when a 400 mS m⁻¹ threshold value was used to separate saline from non-saline soils, < 550 mm rainfall class, and average values were used (Tables 2 and 3). When the highest average value in a profile was used, soils of the neocutanic and calcic groups (<550 mm rainfall class), as well as the alluvial/aeolian and hydromorphic groups tended to be saline (Table 4).

Exchangeable sodium percentage

Significant differences for median ESP in the topsoil and subsoil between rainfall classes occurred in the calcic, alluvial/aeolian, neocutanic, pedocutanic/red structured, hydromorphic, plinthic (subsoil), podzolic, prismacutanic, lithosols and apedal soil groups (Tables 5–7). This indicated the importance of rainfall and leaching on the movement of salts in general and on Na specifically, even within the same soil group. There were no significant median differences for the topsoil and subsoil between rainfall classes for vertic and plinthic soil groups or when the highest value in a profile was used, for the prismacutanic soil group. Possible reasons for this inconsistency was that the leaching potential for vertic soils was a great deal lower than for other soil groups, because of the high clay content and the swelling properties (low infiltration rate when wet and high water-holding capacity). According to Bresler (1981), swelling of soil clay particles causes the size of larger soil pores to decrease. Dispersion and movement of clay platelets further block soil pores. The strong structure of vertic and prismacutanic soils are likely to cause the flow of water to be confined to macro-pore flow, which results in low leaching. The reason for the anomaly in the plinthic group was probably because of the capillary movement of water and salts from the fluctuating watertable from the subsoil to the topsoil, which occurred under both low and high rainfall conditions.

There was a decrease in sodicity as measured by the median ESP values for all topsoil horizons, ranging from alluvial/aeolian > prismacutanic > hydromorphic > vertic > neocutanic > pedocutanic/red structured > calcic > lithosols > plinthic > apedal and podzolic (Table 5). The order for the subsoil horizons ranged from prismacutanic > calcic > hydromorphic > alluvial/aeolian > neocutanic > lithosols > vertic > pedocutanic/red structured > plinthic > apedal = podzolic (Table 6). For the highest value in a profile, sodicity ranged from prismacutanic > podzolic > hydromorphic > calcic > alluvial/aeolian > vertic > neocutanic > plinthic > structured > lithosols ≈ apedal (Table 7).

The alluvial/aeolian group differed significantly from all other groups except from the hydromorphic group (Table 6). For the highest ESP value in a profile, 22 rainfall classes (two rainfall classes for 11 soil groups) were significantly different (p < 0.05), with the alluvial/aeolian and neocutanic groups both significantly different from four other groups.

Soils of the plinthic, apedal and podzolic soil groups had the lowest median ESP values in the topsoil (Table 5) and subsoil (Table 6) for the <550 and >550 mm rainfall classes. When the highest value in a profile was used, the podzolic soil group had the second-highest median ESP value. This anomaly was probably because some podzolic C horizons had a strong hydromorphic tendency and probably mainly occured on binary parent material. The most likely scenario, however, was that the podzolic soils were mostly found near coastal areas where atmospheric deposits of sodium from the ocean occurred and together with the low cation exchange capacity of podzolic soils resulted in relatively high ESP values. Ellis and van Laar (1999) established for their study that soils with podzol B horizons have a higher reserve of sodium than soils without podzol B horizons and that trees planted closer to the sea accumulate more sodium in the leaves than those planted further away.

The prismacutanic soil group had the highest median ESP value (8.4%) for the subsoil horizons (Table 6), the highest median ESP value (10.1%) when the highest value in a profile was used (Table 7), and the joint highest median value (2.9%) for the topsoil horizons (Table 5). The classical concept of a solonetz therefore does not always apply for South African prismacutanic horizons. A reason for this is that certain prismacutanic horizons develop where the ESP values are less than 15% and in other circumstances soils with an ESP value of 15–20%, or even more, do not manifest prismacutanic morphological features. This is probably due other soil properties such as salinity, magnesium, mineralogy and anions also playing a role in clay dispersion (Emerson and Bakker 1973; Fey 2010).

When a median ESP value of 15% was used to separate sodic from non-sodic for both the topsoil and subsoil horizons, none of the soil groups were sodic (Tables 5 and 6). The topsoil of the hydromorphic and alluvial/aeolian soil groups (<550 mm rainfall class) were sodic (>15%) for average ESP values (Table 5). Similarly, the subsoil of the prismacutanic, hydromorphic, alluvial/aeolian, neocutanic and lithosols were sodic (>15%) when average ESP values were used (Table 6). For the highest average value in a profile, soils of the prismacutanic, hydromorphic, alluvial/ aeolian and neocutanic groups (<550 mm rainfall class) were sodic (>15%) when average ESP values were used, but not when median ESP values were used.

None of the topsoil group median values were sodic when a value of 6% was used to separate sodic from non-sodic soils. The subsoil horizons of the prismacutanic (both rainfall classes), calcic, hydromorphic, alluvial/aeolian and neocutanic soil groups were sodic (Table 6), when average values were used. For the average highest value in a profile (<550 mm rainfall class), soils of the prismacutanic, podzolic hydromorphic, calcic and neocutanic groups were sodic (Table 7). Sodicity was recorded for average ESP topsoil values (<550 mm rainfall class) in the hydromorphic, prismacutanic, neocutanic, calcic and alluvial/aeolian soil groups (Table 5). Most subsoil groups were sodic (<550 mm rainfall class), except for the podzolic and plinthic groups (Table 6). For the average highest value in a profile, most soil groups were sodic, except for podzolic (<550 mm rainfall class), as well as pedocutanic/red structured, lithosols and apedal when the annual rainfall was >550 mm (Table 7).

pH_water

The pH_water data was not skewed, with small differences between the median and average values (Tables 8–10), which was not the case for ESP and EC. There was a decrease in median pH_water for all topsoil horizons (Table 8), from the calcic > alluvial/aeolian > vertic > neocutanic > pedocutanic/red structured > prismacutanic > lithosols > hydromorphic > plinthic > apedal > podzolic soil group. For the subsoil horizons median from the calcic > alluvial/ aeolian > neocutanic > prismacutanic > pedocutanic/red structured > lithosols > hydromorphic > plinthic > apedal > podzolic soil group (Table 9). Highest value in a profile pH_water declined in the order calcic > alluvial/aeolian > vertic > neocutanic > prismacutanic > pedocutanic/red structured > hydromorphic > lithosols > plinthic > apedal > podzolic soil group (Table 10).

When the median pH_water in the topsoil was used (<550 mm rainfall class), the calcic soil group was strongly alkaline, alluvial/aeolian, vertic, lithosols and neocutanic soil groups moderately alkaline and soils of the pedocutanic/red structured groups were mildly alkaline (Table 8). When the median pH_water in the subsoil was used (<550 mm rainfall class), calcic, alluvial/aeolian, neocutanic and prismacutanic soil groups were moderately alkaline and pedocutanic/red and lithosols mildly alkaline (Table 9). When the median pH_water of the highest value in a profile was used (<550 mm rainfall class), calcic, alluvial/aeolian and neocutanic groups were strongly alkaline, vertic, prismacutanic and lithosols groups moderately alkaline, and pedocutanic/red structured and hydromorphic groups mildly alkaline (Table 10). Even under low rainfall conditions, the plinthic and podzolic soil groups were usually non-calcareous and when the median values were considered were classified as strongly acid to slightly acid.

Ranking salt-affected soils

To establish which soil classes were most affected by salts, the median EC, ESP and pH_water values were ranked from the highest to the lowest median value. Soil classes with the highest median value for each of the three variables were assigned 11 and the lowest median value was assigned one. The median and average were then calculated for each soil class (Table 11). The calcic soil group was the most salt-affected. There was, however, not much difference between the ranking for the calcic group and the alluvial/ aeolian group, probably due to the limited pedogenesis expected in these soils resulting in minimal leaching of salts. At the other extreme were the prismacutanic soils that were in a mature stage of pedogenesis, which were strongly affected by sodicity and alkalinity in the B-horizon. The apedal and podzolic soil groups, which as a rule had a sandy texture and where leaching of salts naturally occurred, also had the lowest ranking in terms of salt-affectedness.

Table 11: Soil groups affected by salts (average ranking for electrical conductivity, exchangeable sodium percentage and pH_water) in decreasing order

Soil Group	Average ranking
Calcic	9.30
Alluvial/Aeolian	9.24
Prismacutanic	7.95
Vertic	7.59
Pedocutanic/Red structured	7.68
Neocutanic	7.45
Hydromorphic	5.62
Lithosols	4.39
Plinthic	3.20
Apedal	1.85
Podzolic	1.29

Conclusions

Simple statements about salinity, sodicity and alkalinity in soils must be seen against the background of the ever-present variation observed between and within soil groups, which must be taken into account in all analyses. Median and not average values best represented soil data, because most data were strongly positively skewed with large differences between median and average values.

There was a strong relationship between rainfall, salt occurrence and salt movement. The >550 mm rainfall class had a reduced salinity, sodicity and alkalinity due to increased leaching of cations and anions. Significant differences in median EC, ESP and pH_water occurred in the topsoil and subsoil between rainfall classes within the calcic, alluvial/aeolian, neocutanic, pedocutanic/red structured, hydromorphic, plinthic (in the subsoil), prismacutanic, lithosols and apedal soil groups.

Primary alkalinity (pH_water>7.4) and sodicity (ESP>6%) were more common than primary salinity (EC>400 mS m⁻¹) in South Africa. None of the soil groups were saline, only the prismacutanic group was sodic, and the calcic, alluvial/ aeolian, neocutanic and prismacutanic groups were alkaline when the median value or highest value in a profile was used as indicator.

An ESP of 15% failed to differentiate the soil groups. The classical concept of a solonetz therefore did not apply to the prismacutanic group. A reason for this was that certain prismacutanic horizons developed where the ESP values were less than 15%, and in other circumstances soils with an ESP value of 15–20%, or even more, did not manifest prismacutanic morphological features. Sufficient data was not available, even for this relatively large data set of 648 prismacutanic horizons, to proclaim a precise ESP value as a criterion for prismacutanic or solonetz soils in South Africa.

The calcic and alluvial/aeolian soil groups were most affected by salts when the EC, ESP and pH_water were considered.

The information presented here should serve as an indicator for the background or natural sodicity and alkalinity of South African soils. As such it can be utilised to infer the expected sodicity and alkalinity of soils from soil survey data, to monitor the anthropogenic impact on sodicity and alkalinity, and as an input to map the primary sodicity and alkalinity of South African soils.

Acknowledgements

The endeavours of the ARC–Institute for Soil, Climate and Water staff in collecting and analysing the data as well as the assistance of Me Liesl van der Westhuizen in the preparation of this manuscript are gratefully acknowledged.