Great Soil Groups Research Articles

A fuzzy allocation scheme for the Australian Great Soil Groups Classification system

SummaryConventional methods of soil classification are based largely on hierarchies and treat soil attribute value ranges as exactly specifiable quantities. Misclassification may result from ignoring the continuous complex nature of soil variation, and the inability to sample and measure every aspect of soil. The Australian Great Soil Groups (GSG) system of classification is essentially a fuzzy classification, and for the most part central concepts have not been defined explicitly.Centroids were generated, together with fuzzy group membership, using data from the 147 soil profile descriptions in The Handbook of Australian Soils. Some of the GSG, such as Siliceous Sands and Red and Brown Hardpan Soils, were divided into their component parts for better and easier quantification and allocation. The centroids of GSG were examined, and the method of fuzzy κ‐means was then used to allocate unknown profiles to the GSG. The results show that the system is intuitively reasonable.

Distribution and fractionation of soil phosphorus in particle‐size separates in soils of Western Turkey

Abstract Evaluation of soil phosphorus (P) status is not possible without considering the various forms of P present and distribution of these forms in the soil fractions. Distribution of organic and total P on the whole soil and soil separates were examined in the selected three soil profiles representative to three great soil groups found in Izmir province in the Aegean region of Turkey. The results obtained indicated that total and organic P decrease, to a certain depth, as the depths of the soil profile increases. However, there is an increase in total P content in the lower parts of the B or in the upper parts of the C horizons in Profile I. In all the profiles other than Profile I, two‐thirds or more of the total and organic P contents of the whole soil were concentrated in the clay plus silt fractions. The total P contents of the silt and sand fractions in Profile I were much more higher than the clay fraction throughout the profile. In all the profiles, aluminum‐(Al), iron‐(Fe), and reductant‐(Red...

Flow and fate of soil nitrogen in an annual grassland and a young mixed-conifer forest

A comparative study of N dynamics in an annual grassland and a young mixed-conifer forest, that occur on the same soil great group in California, was made by adding small amounts of 15NH 4 + to mineral soils within field microplots and following changes in 15N recoveries in soil and plant pools over 12–16 months. In annual grassland microplots, the aboveground plant biomass contained 5–6% of the added 15N at the end of the growing season. One month after 15N addition almost 70% of the added 15N was recovered in the soil organic matter (SOM) pool, which included surface litter and roots but not microbial biomass. Laboratory incubation of sterilized soil cores indicated that as much as half of the 15N recovered in the SOM pool may have been fixed abiotically. Nevertheless, substantial and rapid incorporation of 15N in the SOM pool indicates a high potential for N immobilization and a rapid turnover of the microbial biomass-N pool in this soil. Recovery of 15N in microbial biomass ranged from 9 to 15%; recovery in the soil solution was lower than in the microbial biomass, but showed a similar seasonal trend. Total 15N recovery ranged from 72 to 85% over the study period. In forest microplots (which excluded plants), recovery of added 15N was generally lower than in the grassland for all soil pools examined. Recovery of 15N in SOM and microbial pools showed reciprocal trends seasonally, suggesting that there was considerable movement of N between these pools in the forest soil. Total 15N recovery was only about 54% 6 months after 15N addition, indicating a high potential for rapid N loss from the surface soil of this forest. Substantially lower recovery of 15N in the forest soil cannot be attributed entirely to the exclusion of plant uptake because plant 15N recovery in the grassland was relatively low. The low 15N recovery in the forest soil might be due to the low capacity of this soil to immobilize N and protect immobilized N from being remineralized and lost from the soil via leaching or denitrification. In both ecosystems estimates of N flow from the mineral soil to decomposing surface residues (presumably via fungal translocation) was significant (0.29 and 0.02 g-N m −2 yr −1 for the grassland and forest, respectively) relative to the net internal dynamics of N within the surface detritus. Comparison of net changes in litter N and accumulation of 15N during decomposition suggests that N was mineralized from litter concurrently as N was immobilized from the mineral soil in both sites.

Overwinter nitrate loss and denitrification potential of cultivated soils in Alberta

Studies comparing overwinter field losses of NO 3-N and denitrification potential using laboratory methods were conducted on the dominant soil Great Groups in Alberta. Recovery in spring of 15N-labelled NO 3-N incorporated into the 0–15 cm depth during winter was lower in the Solod, Black and Gray Luvisolic soils in northern and central Alberta (3–30%) compared to the Dark Brown, Brown and Thin Black soils in southern Alberta (68–95%). The northern and central Alberta sites retained a melting snowpack as thawing of top soil began. Nitrate loss was measured in the laboratory on the 0–15 cm samples from 8 soils amended with 150mg N kg −1 as KNO 3. After 7 days incubation, losses ranged from 62 to 90 mg of added NO 3-N kg −1 (i.e. 8.9–12.9 mg N kg −1 d −1). In an additional experiment, cumulative emissions of N 2O ranged from 19 to 58 mg N kg −1 after 48 h incubation (i.e. 9.5–29.5 mg N kg −1 d −1). A Brown, Black and Gray Luvisol soil amended with 100mg N kg −1 as KNO 3 was used to compare two laboratory methods of measuring denitrification potential. Nitrate loss after 3 days ranged from 53 to 60 mg N kg −1 using a method developed in our laboratory compared to 2–17 mg N kg −1 using a published method. The laboratory losses of NO 3-N did not correlate with losses in the field. These findings suggest that denitrification in the field was controlled by climatic rather than inherent soil chemical or biological variables.

A brief history of national soil classification in Australia since the 1920s

Modern soil science concepts and ideas relating to classification were introduced into Australia in the late 1920's by J. A. Prescott who, in 1931 and 1944, also produced the first maps of Australia showing the broad soil zones. This was followed up in the fifties by C. G. Stephens who formalised the Great Soil Group concept in his Manual of Australian Soils (1953) and in 1961 produced a map and publication titled The Soil Landscapes of Australia. Around this time however, other ideas were being put forward, notably by E. G. Hallsworth and colleagues, and especially by G. W. Leeper, whose original ideas on classification were to provide the foundation for the Factual Key of K. H. Northcote (1960a), which was used as the basis of the Atlas of Australian Soils project (1960-68). The Great Soil Group concept of Stephens was amplified in 1968 in A Handbook of Australian Soils (Stace et al. 1968) which was produced for the Adelaide International Society of Soil Science Congress. This review also considers the role of numerical methods and of Soil Taxonomy in Australia and concludes that while neither are likely to provide the most suitable scheme for Australia, the use of the latter to identify our soils must be encouraged so that the rest of the world is able to relate to our published soil research. Currently, the Factual Key and Handbook classifications are both used in Australia. Both are obsolescent as they date from the early sixties and the vast amount of soils knowledge accumulated since then, particularly in tropical Australia, has not been incorporated into either system. Their deficiencies have led over the past seven years to the development of a new Australian Classification System (a five-level hierarchial general purpose scheme with mutually exclusive classes identified by keys). This scheme is now being tested throughout Australia.

Phosphate sorption approach for determining phosphorus requirements of wheat in calcareous soils

Five field experiments involving P application rates from 0 to 66 kg P ha−1 were conducted on irrigated wheat at Tandojam, Pakistan. The soils belonged to two great soil groups, Torrifluvent and Camborthid. All soils were calcareous. Olsen-P contents ranged from 3.5 to 6.3 mg P kg−1. Phosphate sorption curves were developed for soils from control (no P) plots at each site. Concentrations of P in solution established by fertilization in the field as estimated from the sorption curves ranged from 0.008 to 0.16mg P L−1. Actual grain yields were converted to relative grain yields and plotted against corresponding concentrations of P in solution. Yield response to P application was obtained in each experiment. Control plot yields ranged from 57 to 89% of maximum yield of respective experiments. Phosphorus requirements of wheat were 0.032 mg L−1 for 95% yield as determined from a composite yield response curve. Predicted quantities of P required to attain 0.032 mg P L−1 ranged from 18 to 29 kg P ha−1. The results of the study suggest that the P sorption approach can be used as a rational basis for making P fertilizer recommendations for various soil-crop combinations.

Soil and geomorphic evolution within the rolling red plains using pleistocene volcanic ash deposits

Pleistocene volcanic ash deposits are found within alluvium from the Arkansas river south to the Brazos river. This drainage area includes tributaries originating in the High Plains, the Raton volcanic field and the Rocky Mountain Front Range within the states of Colorado, New Mexico, Kansas, Texas, and Oklahoma. Thirteen ash deposits are dated from within the High Plains of Kansas and Texas eastward into central Oklahoma to understand the geomorphic history and improve soil and geologic mapping. Within the study area unsolidated Tertiary and Quaternary sediments deposited in a west to east direction overlie Triassic, Permian, and Pennsylvanian bedrock. Volcanic ash deposits are predominantly Early to Middle Pleistocene age. Volcanic ash deposits were dated by the fission track method on shards. The ash deposits are contained within four land resource regions, the Southern and Central High Plains, the High Plains Breaks, the Rolling Red Plains, and the Reddish Prairies. Extensive Middle Pleistocene constructional stream terrace surfaces occur within the Rolling Red Plains. Multiple stream terrace surfaces were recognized across the study area with the highest level being dated Early Pleistocene to Pliocene and the lowest bordering the Holocene floodplains. Topographic cross-sections (100 km long at 1:24,000 scale) transecting dated ash deposits and perpendicular to major river systems were used to distinguish terrace levels. Constructional terrace surfaces dated by ash deposits range from 21 to 100 m above and 1 to 16 km distance from present river channels. Soil orders formed in Quaternary alluvium are Entisols, Inceptisols, Mollisols, Alfisols and Vertisols. Ustic and udic soil moisture regimes and a thermic (15 to 22°C mean annual temperature) soil temperature regime dominate the study area. The same soil series is often mapped on terrace surfaces spanning Early to Middle Pleistocene age because current classification does not recognize differences in deeply weathered soils (greater than 2 m) and lacks recognition of loess. Soil great groups found on Early and Middle Pleistocene surfaces were Paleustolls and Argiustolls.

A soil technical classification system for Pinus radiata (D. Don) plantations. II. A basis for estimation of crop yield

The Technical Classification is a system for encoding soil attributes according to parent rock; texture profile; depth to and type of impeding layer; texture, condition, and weathering of surface horizons; and nature and colour of the subsoil. The encoded data may then be used as a basis for soil-based silvicultural management decisions, such as prediction of wood yield, fertilizer prescription, thinning schedules, trafficking for off-road equipment, and placement of roads. The objective of this paper is to examine the suitability of the classification as a basis from which to estimate the yield of wood from Pinus radiata plantations. Data were collected from trees and soil profiles on 181 sites in New South Wales, South Australia, Victoria and Tasmania. All but one of the attributes (texture profile) of the Technical Classification were significantly related to wood volume as single factors at the 5% level of significance. The combined effect of soil attributes in explaining the variation in wood volume across the sites was analysed by backward stepwise elimination of variables from a multiple analysis of variance model. When summed, the parameter estimates for each soil attribute gave an estimate of wood volume produced at age 11 years. The model included the attributes: parent rock, soil depth, impeding layer, texture of surface soil, surface condition, and an interaction between texture profile and weathering of surface soil horizons. The attributes subsoil texture and subsoil colour were not significant in explaining the variation in wood volume in the data set under test. The model accounted for 75% of the variation in wood volume. The addition of rainfall did not improve the variation accounted for by the model. By comparison, the Great Soil Group classification accounted for 32% of variation in wood volume. The Factual Key classification at the level of principal profile form accounted for 53% of the variation in wood volume. We conclude that the Technical Classification is suitable as a basis from which to estimate the yield of wood from Pinus radiata plantations.

Potential wind and water erodibility of four land-use systems in south-eastern Montana, U.S.A.

As part of an assessment of a 1930s federal government land purchase/management program, a study was undertaken to compare the potential wind and water erodibility (as determined by soil properties) of reseeded cropland (purchased, retired and reseeded to crested wheatgrass) with abandoned cropland, cropland and native range land-use systems on series of Haploboroll, Camborthid and Argiboroll soil great groups in south-eastern Montana. The percentage of non-wind erodible (>0.84 mm diameter) aggregates (sampled in spring) of the surface 30 mm was greatest in cropland (85.0%), with lower values in abandoned and reseeded cropland, and native range (66.3, 66.1 and 61.4%, respectively). Permanent vegetation on the non-cropland systems probably prevented significant wind erosion. The universal soil loss equation (USLE) soil erodibility factor, K, was higher (and thus more erodible) in cropland (0.33) and lower in the other systems (0.26, 0.25 and 0.26 for native range, abandoned and reseeded cropland, respectively). Lower organic matter and platier soil structure were primarily responsible for the higher cropland K value. Comparison of reseeded cropland and cropland suggests reseeded cropland was more wind erodible but less water erodible (based only on a one-time soil sampling). Similarly managed (except for reseeding) abandoned cropland and reseeded cropland were not different in any property measured suggesting no present-day effect of seeding on erosion susceptibility. The federal purchase, retirement and revegetation of these lands almost certainly reduced erosion as compared with continually cropped land, although reseeding the lands did not affect potential wind/water erodibility as compared with naturally revegetated (abandoned) cropland.

Soil evolution: Evidence from southern australia

Soil evolution in Southern Australia has been studied by stratigraphic analysis of the regolith. The sedimentary record shows that tectonics is followed by marine, transitional and continental deposition and, finally, soil formation. These episodes are repeated on a time scale shortening geometrically. Soil is formed in the upper layered part of the regolith where materials and processes are involved in and with the biosphere. Both physical and chemical layers are included in the concept. Where a break in soil formation is followed by the development of younger material—either sediment or soil—a paleosol can be recognised. The layer in the paleosol of prime importance from a stratigraphic point of view is the horizon of accumulation which characterises the soil of the time. Paleosols record climates of the past which prevailed over wide areas. Soil formation is episodic. Assemblages of paleosols constitute soil profiles and record interrupted pedogenesis. Soil zonation begins with the development of paleosols at particular times in the past. There is general distribution of great soil groups in southern Australia from older soils with fossil silcrete pans and fossil lateritic soils of the inland to younger poorly differentiated coastal sands and solonchaks of the coastal margin. This is due partly to uplifting of the older basin margins inland and downwarping near the coast, and to cyclic incision of exoreic drainage on the continental margin as a result of the interplay between climate and first and second order geomorphic structures. A similar, but local, relationship occurs between older soils in the ranges and younger soils in saline depressions inland or estuarine plains on the coastal margin.

Utilization of non-exchangeable soil potassium in relation to soil type, plant species and stage of growth

The stage of growth at which three plant species, a grass, a legume and a cereal, absorbed initially non-exchangeable potassium (Kn) from two soils, a red podzolic and a krasnozem, was examined by sequential harvesting in a glasshouse experiment. Large amounts of Kn were removed from the red podzolic soil, which contained greater amounts of micaceous material, in the order setaria > maize > lucerne. Uptake of Kn appeared insignificant after exchangeable Kn as measured on the bulk soil, had been depleted by about 60%, and its contribution to total plant uptake thereafter was independent of species. For setaria and maize, plant K concentrations at this time were above the critical level, so that Kn represents a potassium resource that can be exploited during normal growth of an appropriate species. Uptake of Kn from the krasnozem was relatively small. The red podzolic soil yielded much more Kn than a similar soil in earlier experiments. This was attributed to differences in clay mineralogy, possibly associated with differences in the weathering stages reached by the two soils. Thus, it is not possible to infer the pattern of release of Kn simply from a knowledge of Great Soil Group and parent material.

Cowpea and Pearl Millet response to organic amendments in semiarid alfisols of telengana, Andhra Pradesh, India

Abstract Experiments were established in a loamy sand and a sandy loam (belonging to the family of Typic Haplustalfs and Typic Rhodustalfs of the great soil group Alfisols) with pearl millet and cowpea crop residues, cattle manure, and compost at 20 kg N ha−1 equivalence grown over an 8‐year period. These two crops were grown in rotation. Pearl millet received additionally 20 kg N ha−1, while 8.7 kg P ha−1 was applied to cowpea in this rotation. It was found that incorporation of crop residues at 2.0 tons • ha−1 (20 kg N ha−1) of pearl millet or cowpea gave increased yields of 18 and 13%, respectively. This was accompanied by improvement in the hydraulic conductivity, available N and K, and a tendency for increased mean weight.

Plant community pattern analysis: a cartographic approach applied in the Lac des Deux-Montagnes area (Québec)

Following quantitative sampling of riparian and forest plant communities bordering on Lac des Deux-Montagnes, a classification and a plant community pattern analysis were made. First a community classification was established by using an hierarchic agglomeration cluster analysis technique. These communities were mapped at a 1:10 000 scale. Second, by superimposing on this map, soil and topographic maps of the same region, a systematic sampling of the maps was done. The variables considered were elevation, drainage, aspect, slope, soil great groups, soil series, deposit, soil texture, stoniness, parent rock, and topography. The discrimination potential of these abiotic variables over the classification was investigated by means of information theory. The variables which most controlled the plant community pattern distribution in the studied area were soil texture, slope, elevation, and aspect. To study relationships among the abiotic variables and their relation to the plant communities, contingency tables, based on the map sampling, were prepared. By analysing each box of these tables it was possible to characterize each of the plant communities and establish the toposequence for nine communities. These communities were dominated respectively by Acer saccharum, A. saccharinum – Cephalanthus occidentalis, A. saccharinum – Fraxinus pennsylvanica, A. saccharinum – Matteuccia struthiopteris, A. saccharinum – Phalaris arundinacea, Pinus strobus, Quercus bicolor, Q. rubra, and Salix sp.

Effect of Fertilizer Phosphorus Placement Depth on Winter Wheat Yield

AbstractField studies were conducted with winter wheat (Triticum aestivum L. ‘Centurk 78’) at nine locations in southwestern, central, and eastern Nebraska in 1982 and 1983 to determine the effect of phosphorus (P) placement depth (0, 5, 10, and 15 cm). As a comparison to these depths, seed placed P, soil surface placed P, and no P treatments were used. A rate of 11 kg P ha−1 was applied. Soil great groups were either Argiustolls or Argiudolls with surface soil Bray and Kurtz no. 1 P levels ranging from 4 to 9 mg kg−1 and pH values ranging from 5.5 to 7.5. Grain yield was increased by added P in both years though magnitude of response varied at different locations. A linear and quadratic response to depth was evident in both years, though variation occurred between locations. Optium depths of P placement for maximum grain yield occurred at 11.9 cm in 1982 and 10.4 cm in 1983. No differences between either the best depth (10 cm) and seed treatments or the surface and seed treatments were found, indicating that, depending on management situations, these methods of P application could be used interchangeably. Total P uptake was increased by P applications in a manner similar to grain yield. Phosphorus uptake of the seed treatment and the best depth treatment at several growth stages (Feekes' stages 2.0, 6.0, 10.0, 10.5, and 11.4) was compared. An early advantage (stage 2.0) of the seed treatment was not evident at harvest, indicating that early P uptake is not necessary for maximum yield. The number of heads and seed weight were both increased due to applied P in both years and seed weight responded linearly to depth in 1982. No effect on the seed number per head was found.

An investigation of the base status of soils from the Lockyer Valley uplands, and some implications for land use

Samples were examined from 129 profiles representing the major soils on all of the parent material types occurring in the Lockyer Valley Uplands. The base status and CEC were determined by an unbuffered silver thiourea method and pH, and conductivity of the 1:5 soil/water suspensions measured. A factor analysis of the data showed that 62.5% of the variation could be accounted for by three factors, each of which described a different facet of soil development. Factor 1 was related to the balance between Ca and Na+Mg. Factor 2 was mainly concerned with salinity, pH and cation saturation, whereas factor 3 was related to the balance between Ca and exchangeable Al. The soils were classified into seven groups according to the factor 1 scores for the Al horizon, and the factor 2 and 3 scores for the B horizon. Parent material and Great Soil Group classification proved only to be a general guide to base status. Nearly all the soils had Ca-dominant surface soils with low base saturation and salinity. They became more Na + Mg dominant, and more saline, with increasing depth, and many of the podzolic soils had acid B horizons with high exchangeable Al. The best agricultural soils are the yellow-brown earthy soils, and sands developed on Ma Ma Creek Sandstones, and these tend to be Ca dominant and non-saline at least in the upper solum. Of the clay soils, although most possessed Ca-dominant surface soils, they became Mg dominant at a depth greater than 10-20 cm, and may be intractable for that reason. This includes many of the soils developed on basalt, or material derived from basalt.

The place of 'genesis' in the classification of soils

The purpose of classification is to indicate relationships and differences between objects and to establish groups at various levels of generalization; i.e. to produce simplified models based on experience. In soils the object to be classified is often the soil profile, which may be considered to be analogous to the hand specimen in geology and to the individual organism in biology. For purposes of discussion a profile is considered to be a vertical column, up to 20 cm square, and extending to c. 2 m from the surface, to rock or to some distinct soil layer underlying the profile, across which physical, chemical and mineralogical variations are small. There is no conceptual continuum of soil profiles, even though there is a physical continuum across the surface of the earth. Rather there are a number of common forms each of which may be represented by a 'central concept' with intergrades between them. Such a central concept can be recognized at each level of a hierarchy. Soils are natural objects with a complex structure and history, as are organisms, rock specimens and even landforms. They should be considered in the same way as the other natural objects when a classification is being constructed. Their 'history', as conceived at present, should be involved in classification, even though major emphasis is placed on physical, chemical and mineralogical properties. To illustrate these ideas, the Australian Great Soil Groups are re-examined and suggestions are made about possible subgroups.

WATER RETENTION EQUATIONS AND THEIR RELATIONSHIP TO SOIL ORGANIC MATTER AND PARTICLE SIZE DISTRIBUTION FOR DISTURBED SAMPLES

Functional relationships between soil water content and water suction were examined and related to textural and organic carbon content data. Soil water retention curves between 5 and 10 000 kPa were determined on disturbed samples of 18 soils representing various soil Great Groups in the Canadian prairies. The best fit was obtained with a two-straight-line regression model. Correlation and regression analysis showed that texture was the main soil property influencing the shape and position of the water retention curve. Organic matter influenced primarily the water content at which a break in the curve occurred. Soil zone and cultivation history had little effect on water retention. Key words: Water retention, texture, organic matter, two-straight-line regression

Effects of lime on phosphate sorption characteristics, and exchangeable and soluble phosphate in fifteen acid soils

Fifteen acid soils (pH <5.2 in 0.02 M KCl) from eight Great Soil Groups were used to examine the effects of lime on exchangeable and soluble phosphate in relation to changes in soil phosphate sorptivity, affinity and sorption capacity. Soils varied widely in their reactions to liming, which raised the pH to as high as 7.0 in a sandy solodized soil but only to 4.85 in a yellow earth. Liming to a pH of about 4.8 decreased sorptivity and sorption capacity in four soils whose initial pH was <4.5. Higher sorptivity in these unlimed soils was attributed to precipitation by soluble aluminium. Liming to pH values of about 4.8 to 5.8 minimized soluble aluminium, but increased sorption and greatly increased sorptivity and affinity in most soils. As the pH rose above 5.8 sorptivity and affinity decreased, while sorption capacity remained constant or increased. The results suggested that in most soils sorption was dominated by adsorption, but in a sandy solodized soil liming appeared to change the dominant sorption process from precipitation to adsorption. Soluble phosphate was increased by liming in most soils, but these increases were often inconsistent, with small or negative increases in exchangeable and extractable phosphate and large increases in sorptivity. These results suggest that phosphate solubility in some acid soils is controlled more by the dissolution of iron and aluminium phosphates than by the desorption of exchangeable phosphate.

Comparative study of phosphorus and potassium levels of basaltic soils associated with scrub and forest communities on the Darling Downs

Morphologically similar scrub and forest basalt soils were investigated for differences in soil phosphorus and potassium levels. Surface samples (0-10 cm) from 181 scrub and 111 forest sites were analysed for five soil tests: 0.05 M H2SO4-extractable phosphorus, 0.5 M NaHCO3- extractable phosphorus, 0.05 M HCl-extractable potassium, total phosphorus and total potassium by X-ray fluorescence. Highly significant differences (P <0.001) between scrub and forest soils were found for each soil test when all sites were assessed in terms of regional differences based on vegetation. These differences were then investigated further for each of the three common Great Soil Groups - black earths, euchrozems and lithosols - and then for six individual soil series common to both the scrub and forest areas. An unexplained geographic trend in phosphorus values previously reported (1972) in the same area could be accounted for by this difference between scrub and forest soils. Similar differences in soil phosphorus and potassium for scrub and forest soils on other parent materials are also discussed. It was clearly shown that the scrub means were always greater than the forest means for all five soil tests for all soils combined, each Great Soil Group and each soil series, and that there is a strong association between vegetation distribution and these soil differences.

SPATIAL ASSOCIATION BETWEEN AGRICULTURE AND THE SOIL RESOURCE IN CANADA

A study was undertaken to characterize the spatial relation between agriculture and the soil resource in Canada. Cluster analysis was used to develop crop groups, using data from a large, computerized soils, climate and census data base. The crop groups were interpreted spatially and relative to soil Great Groups. Good spatial grouping was found with land-based agricultural activities such as crop type, crop acreage profiles and extent of cultivation. Poor association was found with livestock and amount of off-farm work. The study demonstrates the regional specialization in Canadian agriculture, and shows that particular physical environments are best suited to certain kinds of agriculture.