Local soil classification and comparison of indigenous and technical ...

3 downloads 31519 Views 3MB Size Report
The soil is recognized by farmers as a three-dimensional living body, composed of earth material, air and fluids, ... istic school', has been formalized by Berlin (1992) with his theory of ...... being with its own agency and auto-organization. Soil-.
Geoderma 135 (2006) 140 – 162 www.elsevier.com/locate/geoderma

Local soil classification and comparison of indigenous and technical soil maps in a Mesoamerican community using spatial analysis N. Barrera-Bassols a,⁎, J.A. Zinck b , E. Van Ranst c b

a Instituto de Geografía, UNAM, Aquíles Serdán 382, Morelia C.P. 58000, Michoacán, México International Institute for Geo-Information Science and Earth Observation (ITC), P.O. Box 6, Enschede, The Netherlands c Laboratory of Soil Science, Ghent University, Krijgslaan 281 (S8), 9000 Ghent, Belgium

Received 1 July 2005; received in revised form 21 November 2005; accepted 23 November 2005 Available online 26 January 2006

Abstract The soil classification system developed by a Purhépecha community in central Mexico was formalized, incorporating symbolic, cognitive and practical components. The soil is recognized by farmers as a three-dimensional living body, composed of earth material, air and fluids, organic matter, and living organisms including plants and animals, and organized in layers. The indigenous soil taxonomy is hierarchical, with four categorical levels, and soil classification is flexible enough to adapt to a changing social and environmental context. The local soil map units were compared to those provided by a technical soil classification of general scope (USDA soil taxonomy), using spatial analysis within a GIS environment to determine levels of cartographic correlation. The average spatial correlation at high taxonomic level, computed taking into account only the dominant soils in each map unit, is 74% for the technical–local comparison and 75% for local–technical comparison. At low taxonomic level, the average spatial correlation based on dominant soils only is 62% for the technical–local comparison and 61% for the local–technical comparison. The variable levels of spatial correlation between technical and local soil map units reflect differences in the ways both systems classify soils. However, similarities and discrepancies between making technical and local soil maps reveal complementarities. Critical is the evaluation of topsoil characteristics, as the understanding and monitoring of topsoil dynamics are fundamental for land use decision-making by farmers. Spatial correlation analysis of topsoil properties provides a good basis for collaboration between farmers and soil surveyors. Merging technical and local thinking is indispensable to formulate sustainable land management schemes. © 2005 Elsevier B.V. All rights reserved. Keywords: Local soil classification; Technical soil classification; GIS; Soil mapping comparison; Spatial correlation analysis; Mesoamerica; Mexico; Purhépecha people

1. Introduction Indigenous soil classifications have been analyzed using a variety of approaches from ethnographic to ⁎ Corresponding author. Tel.: +52 443 317 6888; fax: +52 443 317 9425. E-mail addresses: [email protected], [email protected] (N. Barrera-Bassols). 0016-7061/$ - see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.geoderma.2005.11.010

comparative to integral. Initially, classic ethnographic studies focused on the linguistic analysis of local soil and land classification systems, while the comparative approach aimed at establishing similarities and differences between local knowledge and scientific information. More recently, interest has shifted towards a more integral approach, which recognizes the relevance of the cultural context for understanding how farmers name and classify their soils.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Most ethnopedological studies have followed the comparative approach, influenced by two main research trends. One stream, related to the ‘cognitive universalistic school’, has been formalized by Berlin (1992) with his theory of universal principles of ‘folk’ biological classification systems. Much of this kind of research seeks to correlate local soil classification systems with the universal folk biological classification or with scientific soil taxonomies such as the USDA soil taxonomy (Soil Survey Staff, 1999) or the FAO soil legend (FAO/UNESCO, 1990). The other stream is rooted in modern soil science and focuses mainly on developing ‘natural’ or ‘objective’ universal soil taxonomies (Queiroz Neto, 1998). It is also concerned with the spatial and temporal patterns and genetic processes of the soil resource (Buol et al., 1997). Comparative studies tend to disregard fundamental elements belonging to the local environmental knowledge systems, such as symbolic meanings and values, as well as farmers' expertise, thus leaving out the practical implementation of local soil and land knowledge during the production process. However, they have contributed to highlighting similarities and complementarities between indigenous and scientific soil taxonomic systems showing potential synergism, especially for solving problems related with soil and land management (Talawar and Rhoades, 1998; Ericksen and Ardón, 2003). For instance, the combination of photo-interpretation and ethnopedological survey has revealed, in some cases, close correspondence between conventional soil map units and ethnopedological map units (Licona-Vargas et al., 1992; Payton et al., 2003). Similarly, statistical analysis of physical and chemical properties has been used in several studies to show that local soil classification systems match or reflect scientifically defined soil properties (Conklin, 1957; Carter, 1967; Bradley, 1983; Behrens, 1989; Bellón, 1990; Shah, 1993; Krogh and Paarup-Lauresen, 1997; Mikkelsen and Langhor, 1997). One of the main issues mentioned in several ethnopedological studies is the inconsistency of indigenous soil knowledge at regional scale. Indigenous soil and land classes are often named and characterized differently by members of the same ethnic group but from different villages, while technical soil surveys indicate a regional distribution of the same soil classes. This could result from the application of unsuitable research techniques or from real historical or cultural differences. Indeed, examples of ethnopedological research in Mexico reveal the existence of region-wide soil knowledge among Maya, Nahua, Otomi and Purhépecha peoples. The naming and characterization of soil and land classes are relatively homogeneous over

141

thousands of square kilometers, forming a regional ‘folk soil culture’ (Barrera-Bassols, 1988; Barrera-Bassols and Toledo, 2005). In this context, the objective of the present paper is to: (1) formalize the soil classification system developed and used by a local community, incorporating the three components of the local ethnoecological model, i.e. the symbolic, cognitive and practical components; and (2) use modern information technology for spatial analysis and correlation of soil map units. Toledo (1992a, 2002) was the first to acknowledge the inextricably links between symbolic (Kosmos), cognitive (Corpus) and practical (Praxis) dominions in relation to local environmental knowledge systems, inaugurating the Kosmos–Corpus–Praxis (K–C–P) or ethnoecological model. This article draws on this seminal approach, showing its suitability and potential for ethnopedological studies. We first analyze the body of soil knowledge in an indigenous community belonging to the Purhépecha culture of Mexico to understand how local people classify their soils. Soil classification was considered to be at the same time taxonomic and utilitarian because local people, in general, do not make distinction between soil and land. We then compare local and technical soil map units, using spatial analysis within a GIS environment to determine levels of cartographic correlation. 2. Materials and methods 2.1. Setting of the case study San Francisco Pichátaro, the selected study site, is located southwest of the Pátzcuaro lake, in the volcanic highlands of central Mexico (Fig. 1). The community territory extends between 2300 and 3200 m.a.s.l., along a bio-climatic gradient shifting from temperate sub-humid to cold humid as elevation increases (T = 16–12 °C; P = 1000–1500 mm). The configuration of the relief is controlled by a set of PlioQuaternary basalt cones covered by pyroclasts and separated by small fluvio-volcanic valleys. All soils are derived from volcanic materials, mainly ash and cinders and, to a lesser extent, basalt lava. Andisols cover about 75% of the study area; other soils have lost their andic properties because of time and/or climate effect. Soils include (1) Pachic Melanudands on the summits of the highest volcanoes, (2) Typic Haplustands on young volcano slopes, (3) Typic Haplustalfs on older volcano slopes, (4) Humic Haplustands in the higher valleys, and (5) Typic Haplustults in the lower valleys.

142

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 1. San Francisco Pichátaro location within the Pátzcuaro lake basin in Michoacán, Mexico.

Systematic land occupation in this area started as early as 3500 years ago (Schöndube, 1987). According to archaeological evidence, population substantially increased during the Classical Period (300–900 AD) and reached a peak during the Post-Classical Period (1200– 1523 AD), with 2400–3000 people living in the surroundings, mainly farmers (Gorenstein and Pollard, 1983). The presence of fertile volcanic soils and permanent springs at the foot of the volcanoes and lava flows contributed to making Pichátaro an early center of maize production. Pichátaro has certainly been affected by environmental changes after early human occupation took place, as some volcanoes have been active until recently (1393– 1453 AD), according to radiocarbon dating of buried ash layers in valley soils (Barrera-Bassols, 2003). Saporito (1975) found 15 tephra layers in a lake sediment core, ten of which were deposited throughout the Holocene within the last 10,000 years and seven dated between 3500 and 2500 BP. Thus, volcanic activity affected erosion and sedimentation rates in the Pátzcuaro lake when the basin was already inhabited. Over the last 500 years, volcanism and tectonism continued to be active until the eruption of the Paricutin volcano (1943–1952), located just 60 km from Pichátaro (Luhr and Smikin, 1993; Endfield and O'Hara, 1999; Fisher, 2000; Fisher et al., 2004; Israde-Alcántara et al., 2005). However,

there are no conspicuous soil erosion features visible on the landscape or recorded in historical documents. A short reference dated 1775 mentions that Pichátaro had fertile soils and large forest areas (Endfield and O'Hara, 1999). Another source from the 18th century refers to wood extraction and the production of irrigated maize, oranges and wheat in large family plots, together with the traditional milpa (maize–beans–squash) system (Toledo and Barrera-Bassols, 1984). This strategy of multiple land use, able to control soil erosion, continued during the 19th and 20th centuries (West, 1947; Toledo et al., 1980; Toledo, 1992b; Barrera-Bassols, 2003). Nowadays, Pichátaro is a community of some 4500 inhabitants that maintains indigenous structures and traditions, including local socio-political institutions, vivid Mesoamerican cultural elements in daily life, syncretic religious practices, communal land ownership, and multiple land use strategy. The present-time territory covers 10,000 ha, with 55% of them used for farming together with some cattle and lamb livestock; the rest is covered by pine and oak forest. 2.2. Approach and method The bulk of information was derived from primary sources, including findings from an extensive communication exchange with the local actors, although use

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

was also made of secondary sources, such as archival documents and statistical data (Barrera-Bassols, 2003). The first step was to determine the way farmers classify their soils, and how this can be used for soil mapping. Soil names are based on multiple criteria including texture, colour, consistence and stoniness. To determine the semantic ordering applied by farmers to the soil names collected in a free-listing exercise, several cross-checking techniques were used, including participatory cross-sections, open individual and group interviews, soil identification in correlation boxes as dialogical tool, pile sorting and triadic tests. Results were consistent, with minor differences among groups and individuals. A hierarchical, multi-level taxonomic ordering of all local soil names was established, and this was in agreement with earlier studies on Purhépecha soil names by Argueta (1988), Barrera-Bassols (1988), and Giunta (1998). However, soil names were grouped in different ways according to criteria used, such as geographic position, agricultural potential, crop adaptability, natural hazards and land management, showing that soil grouping depends on the context and questions asked. The local soil classification is multi-dimensional because farmers' soil conceptualization is polysemic and polyvalent. Technical soil mapping uses sophisticated tools and is conducted by professionals, while local soil mapping results from the integration of ‘mental documents’ that each individual acquires through experience using relatively simple tools, common sense and wisdom. To fill the gap between both knowledge systems, the ethnopedological approach promotes dialogue among farmers within a community, as well as between farmers and external experts. A semi-detailed geopedological survey, combining geomorphic and soil data, allowed to identify and classify soils at subgroup level according to USDA soil taxonomy (Soil Survey Staff, 1999) and map them in homogeneous terrain units (Zinck, 1989), using a quantitative approach based on technical profile descriptions and laboratory determinations (Fig. 2). Parallel to the former, an ethnopedological survey was carried out consisting of a set of activities that directly involved local farmers, including profile description, soil identification along participatory cross-sections, on-site characterization of soil properties, recognition of the distribution of soil classes on the landscape and drawing of the soil map. Local information was mainly based on topsoil (0–50 cm) assessment, using qualitative criteria evolved from farmers' experience of soil behaviour and land management. The ethnopedological survey mapped soil at ‘variety’ level, the lower hierarchical level of the local soil classification. Both soil maps,

143

the geopedological and the ethnopedological, are based on relief (terrain) units recognized according to scientific and local criteria, respectively (Fig. 3). In both cases, relief map units were considered instrumental not only for soil mapping, but also for understanding soil distribution patterns, soil forming processes and soil behaviour. They were further used as cartographic reference frames to identify similarities and differences between the indigenous and technical soil maps. Farmers possess accurate mental soil maps and are able to distinguish spatial variations according to relief, vegetation, land use and mesoclimate. They were, however, rather reluctant to determine soil classes and distinguish soil variations in locales (‘parajes’) distant from their own agricultural parcels or in unfamiliar places. In such conditions, during the participatory cross-sections, they left the assessment to other farmers with working experience in those locales. On the sites, discussions among farmers and between soil surveyors and farmers about soil spatial distribution took place, after carefully listening to experienced farmers. A preliminary version of the ethnopedological map was discussed in a threeday workshop with the farmers of the village (30 in total) until reaching a consensual version. At the end, farmers agreed that the local soil map was accurate enough, because it included all main soil types and associations, but it was considered of general scope because it did not include all soil varieties and soil mixtures. Discussions also led to statements about relations between relief, vegetation, land use, mesoclimate and the soil distribution patterns. A landscape model of soil distribution, soil behaviour and sediment transportation-accumulation was constructed synthesizing the local knowledge (Barrera-Bassols and Zinck, 2003) (Fig. 4). Once both the ethnopedological and geopedological soil maps were established, local and technical soil map units were spatially analyzed and correlated. Spatial analysis of soil patterns and soil correlation consisted in comparing technical and local soil knowledge systems, using GIS techniques to cross the soil maps resulting from the geopedological and ethnopedological surveys. Spatial analysis in itself does not allow making statements about the structure of and possible equivalencies between the two taxonomies, but is able to reveal similarities and differences between technical and local soil knowledge systems that might result in synergies aimed to resolve land management problems at local level. It also shows the consistency of each approach and the differences in perceiving, knowing and assessing soil and land resources between farmers and soil surveyors. Since farmers have a profound soillandscape understanding, spatial analysis with the

144

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 2. Spatial correlation analysis model. Source: Barrera-Bassols, 2003.

advantage of data processing and display in GIS was preferred to multivariate statistical analysis for comparing both knowledge systems. Spatial correlation was carried out at three levels, including high taxonomic level, low taxonomic level and soil property level (Fig. 2): (1) Spatial correlation analysis at high taxonomic level to compare soil map units composed of taxa at order level according to the USDA soil taxonomy (Soil Survey Staff, 1999) with soil map units based on soil types according to the Purhépecha soil classification; (2) Spatial correlation analysis at low taxonomic level to compare soil map units composed of

taxa at subgroup level according to the USDA soil taxonomy with soil map units based on soil subtypes and varieties according to the Purhépecha soil classification; (3) Spatial correlation analysis at soil property level, including texture, moist colour, organic matter content, consistence (dry, moist, wet), internal drainage condition, stoniness and rooting condition. Soil properties were chosen according to the importance given by farmers for soil quality assessment and land management. Spatial correlation of soil properties was used hereafter to explain similarities and differences emerging from the comparison of both classification and mapping approaches.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

145

Fig. 3. Comparison between technical and local relief maps. Source: Barrera-Bassols, 2003.

Profile characteristics and laboratory data were adjusted to topsoil depth (0–50 cm), as farmers assess soil properties mainly within surface horizons. The ILWIS 3.11 software package (ITC, 2002) was implemented to process the geopedological and ethnopedological data sets for spatial correlation of soil map units in several steps (Fig. 2): (1) Technical and indigenous soil maps were converted into raster format, using a common geor-

eference base, and a dependent attribute table was created for each map using the same domain as the raster maps. (2) The technical attribute table consisted of 22 columns, starting with relief units followed by soil map units according to USDA soil taxonomy, and 20 dependent soil properties, with 11 of them selected for spatial correlation. Profile characteristics and laboratory data were stored in the rows corresponding to each relief and soil map unit. Numerical values of soil properties, such as pH or organic matter content, were

146 N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 4. Purhépecha perception of soil-land distribution patterns. Sources: Barrera-Bassols, 2003; Barrera-Bassols and Zinck, 2003.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

(3)

(4)

(5)

(6)

(7)

converted into semantic classes according to conventional criteria (e.g. 6% organic matter was converted to class ‘very high’). Of note is that not all geopedological units had information on soil properties, as some of them were inferred during the survey, and were thus not correlated at soil property level. The indigenous attribute table consisted of 19 columns, starting with relief units followed by soil map units according to the Purhépecha soil classification, and 17 dependent soil properties assessed by farmers, with 11 of them selected for spatial correlation. All indigenous soil map units had soil property data that were stored in semantic information classes. The soil properties selected for spatial correlation are: texture, moist colour, organic matter content, structure, dry consistence, moist consistence, stickiness, plasticity, drainage condition, root abundance and stoniness. Thirteen attribute raster maps were created for each technical and indigenous soil map unit, totaling 26 attribute maps that were later overlaid using GIS cross operation. By creating an attribute map, the class name of each pixel in the original input map was replaced by the class found in a given column of the attribute table. The attribute output map uses the same domain as the specified attribute column and the same georeference base as the input map. Pairs of attribute raster maps, showing the same information at the three correlation levels (e.g. the technical and the indigenous topsoil moist colour maps), were overlaid using the GIS cross function. This operation allows comparing pixels at the same position in both attribute maps and store the combinations of class names of pixels in both input maps. Resulting combinations give an output cross map and cross table. The cross table includes the combinations of input classes, the number of pixels that occurs for each combination, and the surface area of each combination. A confusion matrix was established from the 13 cross maps and their related cross tables. The matrix determines the correlation percentage of similar technical and indigenous soil property classes (i.e. correlation % between the technical silty–clay–loam textural class and the tupuri– charandani (dusty–clayey soils) indigenous textural class). Spatial extent of the classes was given in numbers of pixels that were converted into

147

surface area and area percentage from known pixel size. The confusion matrix procedure gives a two-way possibility to assess spatial correlation accuracy, as it uses both attribute tables as first and second columns (i.e. the technical topsoil dry consistence table as first column with the indigenous dry consistence table as second column, against the indigenous dry consistence table as first column with the technical topsoil dry consistence table as second column), generating different results. Confusion matrix operation is generally used to assess the accuracy in image classification when compared to ground truth information and to identify the nature and magnitude of classification errors. In this case, similar images are correlated to find the degree of accuracy and reliability. In contrast, in our case, the same correlation matrix used soil maps derived from different procedures (quantitative against qualitative approaches) to find commonalities and differences between maps, instead of assessing accuracy and reliability. (8) For the purpose of spatial correlation, local soil cartographic units were taken as reference units. Information derived from cross maps and confusion matrices at high and low taxonomic levels was classified into two main domains: (1) strong spatial correlation and (2) moderate to weak spatial correlation. Spatial correlation was considered to be strong whenever one dominant technical soil class or two similar soil classes, one being at least 50%, occupied 75% or more of the extent of an indigenous soil cartographic unit; otherwise spatial correlation was considered to be moderate or weak. Good spatial correlation was expected to happen whenever cartographic delineations and taxonomic contents (i.e. soil classes, proportions and distribution patterns) of local and technical map units reasonably matched. Consociations with a dominant soil or with similar soils would provide higher spatial correlation than associations of dissimilar soils. (9) At soil property level, the percentage of spatial correlation between similar technical and indigenous soil property classes was taken as reference to assess which soil properties matched or did not. Spatial correlation was considered nil when two dissimilar classes occurred in the same correlation unit (i.e. when the technical ‘very high’ organic matter content class was spatially correlated with the indigenous ‘low’ organic matter content class).

148

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

3. Results 3.1. Indigenous soil nomenclature and classification 3.1.1. Soil and land concepts The word echeri (tierra in Spanish), used by Pichátaro people to designate the soil cover, is in fact polysemic and refers at the same time to soil, land, landscape, terrain and bio-climatic zone, in a fashion similar to the modern concepts of land (FAO, 1976) and landscape (Zonneveld, 1995). Thus, local people perceive soil-land as a multidimensional component of nature. When referring to soil types and properties, the farmer conceives soil as a three-dimensional body, similar to the technical concept of soil. When concerned with farming practices, the farmer uses echeri to designate a two-dimensional land surface, with variable management requirements according to local bio-climatic conditions. Beyond this practical relationship between farmer and soil-land as a resource, there is a symbolic relationship by means of which farmer's land care is rewarded by the land providing him with goods and services, including food, materials for construction and ceramics, as well as medical, ritual and magic uses. Land has thus a polyvalent meaning and remains as a primordial symbol, because land has supported life and the well-being of farmers since hundreds of years. This multi-criteria functionality shows that local soil knowledge is flexible and holistic, rather than static and closed as often stated by external agents, including soil scientists. The polysemic nature of the soil-land dominion also shows that the various landscape components are inextricably linked in structural, dynamic and relational terms. Connectivity stands as the central conceptual thinking about nature among local farmers. Farmers' holistic perception about nature connects water cycle, climate, meteorological phenomena and relief to understand and deal with soil behaviour and land functionality. This is similar to the modern soil forming factor model (Jenny, 1941, 1980). A variety of properties and characteristics of the soil cover is used by farmers to identify and classify local soils. Observed features of the soil material and soil formation processes are central to how the soil body is recognized, named and classified according to relevant attributes. Texture, colour, consistence and stoniness are diagnostic attributes at the higher levels of the classification system. Soil colour plays also an important role in the recognition of soil distribution patterns by farmers. In general, farmers relate soil colour to elevation, slope gradient, vegetation and relief. Dark soils occur

on high elevations, forestland and valleys. Yellowish soils occur on slopes and in plains, while reddish soils occur on washed steep slopes and in low-elevation valleys. Local soil classification is essentially fluid and flexible to cope with and account for the continuously changing soilscape in a changing social and environmental context. That is why local soil knowledge must be acknowledged as a ‘living resource that is constantly reinvented’, rather than being merely an achievement of the past (Röling and Browers, 1999; Mazzucato and Niemeijer, 2000). In Pichátaro, as in many other indigenous communities of Mesoamerica, soil and land symbolism, knowledge and practice changed through time when farmers adapted their skills, technologies and abilities through innovation. Farmers make use of Purhépecha, colonial and modern concepts and terms when discussing about natural resources and the problems they face nowadays. Farmers' agricultural knowledge is blended by empirical and scientific explanations, articulated around development programs, crop subsidies, market prices and local politics. The local theory of soil functioning and health is based on pre-Hispanic Mesoamerican beliefs and cosmovision, 16th century Spanish humoral theories about the human body functioning and health (Foster, 1953, 1960), and modern agronomic concepts. A relevant feature is that indigenous soil knowledge became bilingual after introduction of Spanish as the official national language. When talking about land and land management issues, Purhépecha, Spanish and technical terms are combined. 3.1.2. The soil as a living three-dimensional body Farmers recognize the soil as a three-dimensional body composed of (1) earth or hard matter, such as soil material and stones; (2) substances or soft matter, such as water, juices and air; (3) organic matter; and (4) living organisms, including plants and animals. It is also recognized that soil components are unevenly distributed in layers, according to soil type, soil variety and soil mixture. However, some shallow soils have no distinct layers with depth and are, therefore, considered ‘simple soils’, while the vast majority of soil classes is organized in layers because they are deep and considered ‘composite soils’. Soil components and soil properties are recognized throughout depth, although farmers mainly assess the first 50 cm of the soil body. The presence of microelements and organisms smaller than 1 mm is given less attention. Table 1 shows the local soil-land and soil-related nomenclature.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162 Table 1 Local soil and land nomenclature

Table 1 (continued)

Purhépecha

Spanish

English

Earth or hard matter Echeri Tzacapu uiramu Zacapurhu Tzacapu pupurash Charaki Poksinda Echeri kuatapiti Echeri choperi Echeri ietakata

Tierra Canto Piedra Piedra deleznable Grava Terrón Tierra suelta Tierra dura Tierra mixta

Earth Boulder Stone Brittle stone Gravel Ped Loose earth Hard earth Mixed earth

Substances or soft matter Itsi Agua Itsírhuky Jugo Tariat Aire Terendani Materia orgánica Iorhejpiti Fuerza, vitamina Tsipitícha Organismo viviente

Water Juice Air Organic matter Nutrient, vitamin Living organism

Plants and roots Plantáecha Siringua Siringua sahuápiti Siringua tepari

Planta Raíz Raíz delgada Raíz gruesa

Plant Root Fine root Coarse root

Fauna edáfica Mamífero Tuza Ratón Conejo Ardilla Reptil Lagartija Vibora Invertebrado

Soil fauna Mammal Mole Mouse Rabbit Squirrel Reptile Lizard Viper Invertebrate

Lombriz Escarabajo Hormiga Saltamonte Avispa

Worm Bettle Ant Grasshopper Wasp

Capa Todas las clases de tierras Tierra mixta Tierra simple o delgada Tierra gruesa o profunda Tierra compuesta Tierra seca Tierra húmeda

Layer All soil classes

Fauna Echeri animáliecha Itzúkua akuri Kumurhu Jeiákicha Auánicha Kuarákicha Antsíkurhiticha Tsákicha Akuítsecha Anháparikua no jukáricha Tsirákuecha Tenderápuecha Surúkicha Chochúecha Tsitsítsicha Soil classes Echéricha kurhúnda Echéricha Echeri ietakata Echeri sahuapiti Echeri jauámiti Echeri jauákurini Echeri kharíshi Echeri ukándeni

149

Soil mixture Simple or shallow soil Deep soil Composite soil Dry soil Humid soil

Purhépecha

Spanish

English

Soil function Itsúrhini Uekándeni Jiréjtani Apháreni Khúhni

Chupar Húmedo Respirar Sudar Hinchar

Water infiltration Moisture To breath To sweat To swell

Source: Barrera-Bassols, 2003.

Earth material is subdivided according to attributes such as structure, shape and size. The earth component is structured in peds, but earth can also be loose, hard and mixed. Coarse fragments are recognized according to colour, size, shape, porosity and hardness. Boulders, stones and gravels are classified according to hardness and weathering stage or susceptibility to weathering. Water infiltration is critical to crop growth. For this reason, farmers are permanently assessing the degree and distribution of moisture in the soil layers throughout the agricultural cycle. Water infiltration depends on soil location and soil type. It is believed that the soil ‘breathes’, ‘sweats’, ‘swells’, and requires air to balance its internal temperature. Although farmers would not explain soil ‘breathing’, they related ‘sweating’ to evaporation or the release of excessive heat and ‘swelling’ to moisture retention. These cause–effect relationships are part of the local theory of soil health. The presence of organic matter in the soil is considered critical for crop performance. For farmers, organic matter is clearly the main factor to sustain soil fertility, and soils with organic matter in the topsoil are potentially of good agricultural quality. Littered soils are blackish soils with high amount of decomposed organic matter in the first 50 cm of the soil profile. Litter of broad oak leaves is more appreciated, because of its ‘strength’, than litter of pine or fir needles. Oak litter is believed to possess more ‘vitamins’, retain moisture and be ‘temperate’, while pine-fir litter is considered ‘hot’, dry and with less ‘vitamins’, thus not commonly used to fertilize the land. Oak litter is collected and transported to agricultural fields or homegardens to be used as fertilizer. Ash from burnt dry maize stalks is also considered a vital organic material for crops. Parking cattle on agricultural parcels after harvest and during most part of the dry season promotes the addition of manure, considered having high strength but ‘hot’ temperature. Poultry dung is the most appreciated organic fertilizer from animal source, because it is very strong and ‘cold’ and helps balance the internal temperature of ‘hot’ soils, such as clayey soils.

150

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Farmers believe that earth material, substances, organic matter and living organisms are nutrient givers that ‘feed’ and ‘water’ the land. Water and air promote the decomposition of organic matter, which infiltrates into the soil body as ‘juices’ or ‘vitamins’. Soil fauna and roots provide organic matter and help litter decomposition, while permitting the soil to breath, sweat and swell. Behaviour of living organisms in the soil is well recognized, as it can be beneficial or, on the contrary, harmful to crops. Rooting condition is assessed to evaluate soil quality and crop performance. Roots are classified according to size and depth (Table 1). Distribution of fine and coarse roots through the soil is believed to be a consequence of differential moisture availability and retention capacity in the soil layers during dry and rainy seasons. Root development responds to soil ‘hardness’ or ‘softness’, according to local perception. Hard soils impede roots to grow because they are ‘dry’ soils, while ‘soft’ soils allow good root development because they are ‘humid’ soils. Soils that retain moisture in the topsoil are better fitted for short-root crops such as maize and wheat, while soils that retain moisture in the subsurface layers are

better fitted for long-root crops, such as beans, squash, pumpkins and fruit trees. Farmers recognize fauna living in the soil, such as (1) mammals like moles, mice, rabbits and squirrels; (2) reptiles like lizards and snakes; and (3) invertebrates, such as worms, beetles, ants, grasshoppers and wasps. Most of them are considered beneficial because they ‘feed’ the soil with ‘vitamins’, although they can become plagues when populations grow and damage crops. 3.1.3. The Purhépecha soil classification Purhépecha soil taxonomy is hierarchical with four levels, beginning with a cluster that includes all soils (Fig. 5). Above this level, soils are grouped together with animals, plants, mushrooms and human beings, showing that soils are ranked at the same level as all living beings in the Purhépecha worldview perception. This finding agrees with previous studies in the Pátzcuaro lake basin (Barrera-Bassols, 1988; Fisher, 2000) and the Purhépecha highlands (Giunta, 1998). The inclusion of the soil as a living being was confirmed when analyzing the local humoral theory about soil health, behaviour and agency.

Fig. 5. Purhépecha soil taxonomy according to farmers from San Francisco Pichátaro. Source: Barrera-Bassols, 2003.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

The second categorical level clusters the main soil types that are commonly found in Pichátaro, constituting the local soil world. In accordance with their soilland concepts and on the basis of the properties and components they attribute to the soil body, local farmers recognize five major soil types when referring to mutually exclusive taxonomic units (Fig. 5): (1) dusty soils (echeri tupuri or tierras polvillo), (2) clayey soils (echeri charanda or tierras arcillosas), (3) sandy soils (echeri kutzari or tierras arenosas), (4) stony soils (echeri tzacapendini or tierras pedregosas), and (5) hard soapy soils (echeri querekua or tierras duras y barrosas). Texture, consistence and stoniness are primary denominators used by farmers to group or discriminate main soil types at high taxonomic level (Table 2). The third category includes the largest number of soil taxa (15 subtypes), recognized using colour, organic matter content, abundance and size of stones, and geographic position. The dusty soil type has the greatest number of soil subtypes, reflecting the heterogeneous soilscape dominated by the dusty soils, which classify mainly as Andisols. The fourth and last level includes soil varieties, intergrades/extragrades and mixtures. Eight soil varieties are recognized on the basis of textural and colour intergrades and mixtures in the upper 50 cm. This is a tentative inventory and more soil varieties may exist, as soil intergrades and mixtures are recognized at microscale level, using an open and flexible nomenclature that reflects continuously varying features on the land-

151

scape. Central to this is the fluid use of adjectives as prefixes to designate minor changes in the soilscape. While soil units are mutually exclusive at the higher categories, boundaries between soils are more diffuse or fuzzy at the third and fourth levels. Additionally, farmers distinguish composite soils at plot level as textural or colour intergrades, as for instance dusty–clayey soils (echeri tupuri-charandani) or dusty black-yellow soils (echeri tupuri turipiti-spambiti). Extragrades are related to the soil position on the landscape, the adjacency to neighboring landscape units, the intensity of sediment transit and the volume of debris accumulation. Two main concepts lie at the center of the local soil classification, especially at the two lowest levels. One is the concept of soil purity, or the ideal condition of a given soil type. The former contrasts with the concept of composite soil, or the real condition of that type of soil on the landscape. Farmers agree that it is not common to find a pure soil type, but rather soil subtypes, varieties, intergrades/extragrades and mixtures. That is why the third and fourth levels of the classification system cluster more soil taxa than the second one. This was further shown when drawing the local soil distribution patterns. Farmers' mental soil maps were closely related to the ‘composite soil’ concept. Thus, farmers use ways of thinking similar to those applied by soil scientists when building up their own soil classification scheme and when drawing technical soil maps. Soil affinities according to farmers' perception are shown in Fig. 6.

Table 2 Properties of the main soil classes according to farmers' assessment Soil properties

Dusty soils (Echeri tupuri)

Clayey soils (Echeri charanda)

Sandy soils (Echeri kutzari)

Stony soils (Echeri tzacapendini)

Soapy, sticky and hard soils (Echeri querékua)

Texture Colour

Dusty Black to dark yellowish Weak, fine, blocky

Clayey Reddish to red yellowish Moderate, medium blocky Medium to low

Sandy Black, reddish and whitish Very weak, fine to coarse granular Low

Dusty Black to dark yellowish From granular to blocky

Clayey Dark brown to reddish brown Massive; hardpan

High

Medium

Loose to soft

Soft to slightly hard

Loose

Loose to soft

Hard

Friable, non sticky and non plastic Always moist

Firm, sticky and plastic Dry

Loose, non sticky and non plastic Dry

Loose to friable, non sticky and non plastic Dry to moist

Very firm, sticky and very plastic Dry

Well drained None Few to many Deep

Poorly drained Eventually Few to many Deep

Excessively drained None None Shallow

Well drained None Abundant Shallow

Poorly drained Annually Many Shallow

Structure Organic matter content Dry Consistence Moist Moisture condition Internal drainage Flooding Stoniness Depth

High to medium

Source: Barrera-Bassols, 2003.

152

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 6. Soil affinities according to farmers’ perception. Source: Barrera-Bassols, 2003.

Analysis of the local soil nomenclature and taxonomy reveals deep understanding of the soilscape and the soil world. Soil nomenclature is based on observed properties that change over time. Farmers combine and modify them to build up descriptive classes reflecting gradual variations or abrupt changes of the soil cover. The local soil knowledge is constantly reworked as an open cognitive template and transferred from generation to generation through the maintenance of a set of rules that sanction local land use practices, property rights, reciprocity and communal responsibilities. The local soil knowledge is explained symbolically and/or logically by recognition of cause–effect relationships, and it is conceptualized by formalizing practical experience into knowledge rules. The multi-dimensional nature of the soil taxonomy reveals the complexity of these knowledge rules. 3.2. Spatial analysis and soil correlation 3.2.1. Spatial correlation at high taxonomic level 3.2.1.1. Strongly correlated units. At high taxonomic level, strong spatial correlation between local and tech-

nical soil map units occurred in 25% of the study area, distributed over four out of eight cartographic units (Figs. 7 and 8; Tables 3 and 4). Sandy soils (echeri kutzari) coincide with Inceptisols in 81% of the sandy soil map unit. Sandy soils and Inceptisols are distributed along barrancos, gullies and ravines. Farmers classify these soils according to relief position, texture, structure and internal drainage, and consider them as non-agricultural soils. Texture is assessed by farmers as sandy because of the presence of fine and medium rounded peds with granular structure, cemented by volcanic ash. In contrast, according to the technical classification, the texture of the same material is silt loam and the structure is fine to medium subangular blocky. The local map unit of dusty and clayey soils (echeri tupuri ka echeri charanda) corresponds in 90% to an association of Andisols and Alfisols. Taxonomic heterogeneity is more apparent than real, as both Andisols and Alfisols strongly correlate with dusty soils (67% and 86%, respectively). Stony soils (echeri tzacapendini) are Andisols in 99% of the local soil map unit. In fact, stony soils are dusty soils confined to lava flows with large stone

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

153

Fig. 7. Spatial correlation of soil groups at high taxonomic level. Source: Barrera-Bassols, 2003.

outcrops, according to Purhépecha classification. Dusty and stony soils together correspond to 86% of the Andisols.

The association of dusty and sandy soils (echeri tupuri ka echeri kutzari) also strongly correlates with Andisols, which occupy 100% of the local soil map

154

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 8. Spatial correlation of local and technical soil map units at high taxonomic level. Barrera-Bassols, 2003.

unit. Dusty and sandy soils occur in small lava flow areas where cropping is possible. Farmers clearly distinguish dusty soils with relatively high percentage of volcanic sand in the topsoil from sandy soil inclusions. This is why the local soil group strongly correlates with Andisols. 3.2.1.2. Moderately and weakly correlated units. Looser spatial correlation between local and technical soil map units occurred in 75% of the study area, distributed in four cartographic units (Figs. 7 and 8;

Tables 3 and 4). However, some of these local map units can be considered taxonomically homogeneous because of the similarity between the taxa included in the compared technical soil map units. Such is the case of the dusty soils (echeri tupuri) cartographic unit that corresponds to Andisols in 65% of the unit extent, thus below the 75% threshold adopted for a map unit to be considered of high spatial correlation (Table 4). In fact, spatial correlation between dusty soils and Andisols is somewhat higher than 65%, as the dusty soil map unit includes also several smaller technical cartographic

Table 3 Correlation matrix between technical and local map units (%) Technical units/local units

Dusty soils

Sandy soils

Clayey and dusty soils

Clayey soils

Stony soils

Soapy, sticky and hard soils

Dusty and sandy soils

Dusty and clayey soils

Total

Andisols Andisols and Alfisols Ultisols and Andisols Alfisols Ultisols Alfisols and Andisols Alfisols and Inceptisol Inceptisols

67 90 54 86 100 44 0 16

3 2 2 0 0 0 0 77

0 0 37 0 0 47 0 5

4 0 0 14 0 9 59 2

19 0 0 0 0 0 0 0

5 0 5 0 0 0 41 0

2 0 0 0 0 0 0 0

0 8 2 0 0 0 0 0

100 100 100 100 100 100 100 100

Source: Barrera-Bassols, 2003.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

155

Table 4 Correlation matrix between local and technical map units (%) Local units/technical units

And.

And. and Alfi.

Ult. and And.

Alf.

Ult

Alf. and And.

Alf. and Incept.

Incept.

Total

Dusty soils Sandy soils Clayey and dusty soils Clayey soils Stony soils Soapy, sticky and hard soils Dusty and sandy soils Dusty and clayey soils

65 14 4 48 99 70 100 0

19 3 0 0 1 0 0 90

5 1 46 0 0 7 0 10

3 0 2 0 0 0 0 0

2 0 0 0 0 0 0 0

3 1 35 7 0 0 0 0

0 0 0 36 0 23 0 0

3 81 13 4 0 0 0 0

100 100 100 100 100 100 100 100

Source: Barrera-Bassols, 2003.

units where Andisols are mapped in association with Inceptisols, Alfisols and Ultisols, all derived from similar volcanic materials. Conversely, 67% of the Andisols map unit is composed of dusty soils (Table 3). Another example of apparent taxonomic heterogeneity is the soapy, sticky and hard soils (echeri querekua) cartographic unit that, with 70% of the area occupied by Andisols, is close to the threshold of 75% fixed for high spatial correlation. Farmers also single out the stony soils unit from related soils, although stony soils are in fact dusty soils with coarse fragments. The most heterogeneous cartographic units are the association clayey and dusty soils (echeri charanda ka echeri tupuri) and the clayey soils (echeri charanda), which show taxonomic dispersion with no clear dominant soil and reveal significant differences in classifying soils with relatively high clay content in the topsoil (Table 4). In general, spatial correlation analysis at high taxonomic level shows that both the local and the technical soil groups are taxonomically consistent. The average spatial correlation, computed taking into account only the dominant soil classes in each of the eight map units, is 74% for the technical–local comparison and 75% for local–technical comparison. This is quite remarkable if considering that the two soil classification procedures are significantly different. Spatial correlation of soil map units is controlled by the fact that both the local people and the soil surveyor used the same relief units as cartographic frames to delineate soil units, with 99% spatial matching between local and technical terrain units. Both implemented a similar landscape approach to soil mapping. 3.2.2. Spatial correlation at low taxonomic level 3.2.2.1. Strongly correlated units. At low taxonomic level, strong spatial correlation between local and technical soil map units occurred in 28% of the area, distributed over eight out of twenty cartographic units

(Figs. 9 and 10; Table 5). The map units composed of subtypes or varieties of dusty soils, dusty–clayey soils, dusty and clayey soils, and dusty alluvial soils correspond in 78–100% of their respective surface areas to Typic and Humic Haplustands. These soils occur on the backslope–footslope complex of the monogenic Plio-Quaternary volcanoes and in the valleys lying between volcanoes. Similarly, the map units of the stony humic soils and stony cliff soils, confined to a Holocene lava plateau with 50–80% rock outcrops, are dominantly covered by Typic and Lithic Haplustands. Medium- and fine-grained sandy soils with abundant (50%) rock outcrops, occurring along barrancos and gullies, strongly correlate with Lithic and Typic Haplustepts. Respective cartographic matching between local and technical soil groups was particularly high in the following map units: (1) Stony humic soils (echeri tzacapurhu terendani) with Typic and Lithic Haplustands, occupying 97% and 82% of their respective map units; (2) Stony cliff soils (echeri juskua karihiran) with Lithic Haplustands, occupying 81% and 89% of their respective map units; (3) Fine-grained sandy soils (echeri kutzari sahuápiti) with Typic and Lithic Haplustepts, occupying 78% and 80% of their respective map units; (4) Medium-grained sandy soils (echeri kutzari terókurhi) with Lithic Haplustepts, occupying 81% and 73% of their respective map units. All these soil groups correspond to relatively shallow soils, with abundant rock outcrops (50–80%) or lithic contact. They were thus technically classified according to diagnostic properties occurring within 0–50 cm depth, which is precisely the soil depth assessed by farmers to classify soils. Spatial correlation analysis at low taxonomic level shows less taxonomic consistence than at high

156 N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 9. Spatial correlation between technical and local soil groups at low taxonomic level. Source: Barrera-Bassols, 2003.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Fig. 10. Spatial correlation of local and technical soil map units at low taxonomic level. Source: Barrera-Bassols, 2003.

157

158

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Table 5 High and low spatial correlation map units at low taxonomic level Cartographic units

Area (km2)

%

High spatial correlation units Echeri tupuri turipiti ka echeri terendani with Typic Haplustults and Typic Haplustands Echeri tupuri-charandani tupiriti ka echeri tupuri-charandani tupiriti spambiti with Typic Haplustults and Typic Haplustands Echeri tupuri spambiti ka echeri charanda charapiti spambiti with Typic Haplustands and Typic Haplustalfs Echeri tzacapurhu terendani with Typic and Lithic Haplustands; many (40%) rock outcrops Echeri kutzari terókurhi with Lithic Haplustepts; abundant (50%) rock outcrops Echeri kutzari sahuápiti with Typic and Lithic Haplustepts; abundant (50%) rock outcrops Echeri tupuri iorhejpiti with humic Haplustands Echeri juskua karihiran with Lithic Haplustands; dominant (85%) rock outcrops

266 15 19 9 67 47 40 30 39

28 2 2 1 7 5 4 3 4

Low spatial correlation units Echeri charanda charapiti with 2 technical soil groups Echeri charanda charapiti spambiti ka echeri charanda spambiti charapiti with 5 technical soil groups Echeri tupuri charapiti turipiti with 7 technical soil groups Echeri tupuri charapiti turipiti ka echeri tupuri charapiti spambiti with 7 technical soil groups Echeri tupuri spambiti ka echeri tupuri turipiti spambiti with 4 technical soil groups Echeri tupuri spambiti ka echeri charakirhu with 10 technical soil groups Echeri tupuri terendani with 11 technical soil groups Echeri tupuri turipiti ka echeri kutzari charapiti with 2 technical soil groups Echeri tupuri turipiti ka echeri tupuri spambiti with 6 technical soil groups Echeri tupuri turipiti ka echeri tupuri turipiti spambiti with 4 technical soil groups Echeri tupuri-charandani charapiti ka echeri tupuri-terendani turipiti with 2 technical soil groups Echeri tupuri-charandani tupuri with 5 technical soil groups

671 29 68 115 51 129 34 54 11 97 41 17 25

71 3 7 12 5 14 4 6 1 10 4 2 2

13

1

Cartographic artifact Source: Barrera-Bassols, 2003.

taxonomic level. The average spatial correlation, computed taking into account only the dominant soil classes in each map unit, is 62% for the technical–local comparison and 61% for the local–technical comparison. Eight of the 20 local map units (40%) were highly correlated with seven of the 18 technical map units (39%). 3.2.2.2. Weakly correlated units. Weak spatial correlation between local and technical soil map units occurred in 71% of the study area, distributed over 12 cartographic units with high taxonomic heterogeneity (Figs. 9 and 10; Table 5). In general, low spatial correlation units were related to deep soils, especially Andisols in the technical map or dusty soils in the local map. This resulted from applying two contrasting classification procedures: the local one assessing only visible topsoil properties and the technical one assessing the overall soil profile properties, with the support of laboratory determination of non-visible properties, such as the andic properties. The example of the dark dusty litter soils (echeri tupuri terendani) illustrates the reasons of discrepancy. This local soil group clusters a variety of technical soil classes at subgroup level, belonging mainly to Andisols but also to Alfisols and Inceptisols. Dark dusty litter soils have humic, dark and dusty topsoils,

and are confined to mountain summits and a lava plateau with dense forest cover. The humic, dark and dusty diagnostic properties are visible and can be assessed by touching, feeling, smelling and tasting the soil material on the field, while the technical procedure relies on the assessment of andic properties to identify soil materials derived from volcanic ash as a requirement to classify Andisols. Andic properties are based on laboratory determination of the contents of extractable aluminum and iron, bulk density and phosphate retention, which are non-visible soil properties (Soil Survey Staff, 1999). Dark dusty litter soils, which are locally classified according to relief, vegetation cover and topsoil properties, including organic matter content, texture and colour, correspond to a large number of technical soil classes at order and subgroup levels, that have high organic carbon content, silty or loamy textures, and dark colour in the topsoil, but fail to meet the requirements for andic properties or those for mollic epipedon (Soil Survey Staff, 1999). 4. Discussion (1) Spatial correlation between local and technical soil map units is strong at high and low taxonomic levels in 25% and 28% of the study area,

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

(2)

(3)

(4)

(5)

respectively. This performance could be improved if the spatial correlation threshold were lowered from 75% to 60%, for instance, allowing more map units to qualify for high spatial correlation. In general, soil consociations provided stronger spatial correlation than soil associations. Spatial correlation is in fact higher than shown, because some apparently contrasting soils can be considered similar soils. For instance, stony soils are dusty soils with coarse fragments. Alfisols and Ultisols, mapped in association with Andisols, have formed from the same volcanic materials as the Andisols, but have lost part or all of their andic properties because of time or climate effect. Taxonomic consistence, assessed on the basis of the average area percentages of dominant soils per local and technical map units, is high (74– 75%) at high taxonomic level and moderate (61– 62%) at low taxonomic level. That allows concluding that both taxonomic systems are spatially robust at the two classification levels considered. Spatial distribution of both local and technical soil map units is to a certain extent similar, although the taxonomic systems are quite different. The technical soil taxonomy (USDA soil taxonomy) is based on the recognition of soil properties, diagnostic characteristics and pedogenic processes in all horizons of the soil mantle, using field descriptions and laboratory determinations, with emphasis on the subsoil characteristics that are relatively stable over time and reflect main soil formation processes. In contrast, the Purhépecha soil taxonomy is based on the recognition of observable topsoil (0–50 cm) properties and diagnostic characteristics, using longstanding farming experience to assess soil functionality and behaviour for practical purpose, specifically for the sustainability of rain-fed maize agriculture in mountain conditions. Farmers' monitoring of changes that affect topsoil attributes in space and time is critical for the maintenance of food production, soil management and soil conservation. Similarities between local and technical soil distribution patterns are related with the nature of the soils and the way these soils are classified by both systems. The example of Andisols and dusty soils is illustrative. Both map units cover more than 75% of the study area, despite differences in criteria applied to classify the soils. Sixty seven percent of Andisols are spatially correlated with

159

dusty soils and, conversely, 65% of dusty soils are spatially correlated with Andisols. (6) Criteria used to describe Andisols and dusty soils show commonalities with respect to properties such as texture, organic matter content, structure development, wet consistence, internal drainage condition and moisture retention capacity, all of them being field-observable properties. The main difference is in the emphasis given to the andic properties required to classify Andisols according to the technical approach, which are all non-observable properties determined in laboratory. In contrast, the Purhépecha approach emphasizes observable topsoil properties and agronomic qualities to identify dusty soils for maize cropping, but also technical determinations to assess the need of chemical fertilizers. Despite contrasting classification procedures, comparison shows a moderate spatial correlation between these soil map units. (7) The most contrasting factor between both classification procedures is soil depth. In general, technical soil classifications tend to ignore or downplay the diversity of topsoil characteristics, mainly because they can change fairly rapidly under human influence (FAO, 1998). In contrast, many local classifications, such as the Purhépecha system, are based on topsoil characteristics because the latter determine, to a large extent, soilrelated land qualities for food production, soil management and soil conservation. Accordingly, the comprehensive understanding of local soil classifications could contribute to reinforce technical topsoil classification efforts for sustainable land management, as recently proposed by FAO (FAO, 1998; see also Sanchez et al., 1982), or the soil quality assessment approach (Romig, 1995; Karlen et al., 1997; Lal, 1998). There is much need of fruitful dialogue between farmers, pedologists, extensionists and other specialists by applying multi-defined soil functions linking crop performance with soil properties and by using classifications that provide useful and practical information. (8) At high and low taxonomic levels, the geopedological and ethnopedological approaches, which both base soil mapping on relief configuration and variation, constitute synergic attempts mobilizing the convergence between geomorphology and local relief knowledge. This is supported by the very high spatial correlation (99%) between technical and local relief units. Soil-relief

160

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

relationship proved to be an outstanding factor for both Purhépecha farmers and soil surveyors with regard to soil classification and mapping. 5. Conclusion The way Pichátaro people classify their soils reflects the way they conceive and perceive soils. They have developed a complex knowledge about the soil and land resources, as part of their theory of nature. The local soil knowledge is based on the idea that soil-land is a polysemic dominion, a polyvalent resource and a living being with its own agency and auto-organization. Soilland is thought to be a dynamic entity that changes through time, with cyclic behaviour and functionality, independently of human intervention. However, connectivity explains relationships between humans and soil resources. Land is perceived as a primordial symbol, embedded in religious practices, that attaches local people to their place. The local soil knowledge is adapted to the heterogeneous mountain conditions of Pichátaro. It is bilingual (Spanish-Purhépecha) and mixes indigenous with scientific terms. In fact, the local ethnopedology reflects the history of Pichátaro, revealing landscape transformation, adaptation and maintenance through time and through accretion from generations of land users continually modifying their environment. The local soil classification is a mirror of this context, and this is to be taken into account when comparing indigenous and technical classifications. In this paper, indigenous and technical soil classifications were compared in terms of spatial correlation of map units, instead of the more common way of comparing the structure of taxonomic systems and correlating soil classes using multivariate statistics. GIS was used as a data integration tool, but also as a communication platform to compare diverse manners of perceiving, naming, classifying and mapping soils, and to highlight that contrasting knowledge systems can be in fact synergistic. The spatial correlation analysis allowed assessing the level of consistency between the USDA soil taxonomy and the Purhépecha soil classification as applied to the Pichátaro soils. The variable levels of spatial correlation between technical and local soil map units reflect differences in the ways both systems classify soils. However, similarities and discrepancies between making technical and local soil maps can reveal complementarities by comparing contrasting rationale in soil description, classification and mapping. This allows building communication bridges between farmers, soil surveyors, other scientists and extension officers. Relationships can also be explored to assess

soil performance to improve the indigenous precision agriculture by tailoring cropping and management practices to local soil condition and agro-ecologic variability. Critical to this is the evaluation of topsoil characteristics, behaviour and performance throughout the year and between years, as the evaluation of topsoil dynamics is fundamental for land use decision-making by farmers. Spatial correlation analysis at the level of topsoil properties provides a good basis for farmers and soil surveyors to collaborate in soil mapping and land evaluation. The establishment of a common language requires recognizing that all soil knowledge systems have limitations and that merging technical and local thinking is indispensable to formulate sustainable land management schemes. Ethnopedology helps validate scientific soil knowledge to assure that it is not only scientific but also relevant and functional. Acknowledgements We deeply acknowledge the substantial support from Heriberto Rodriguez (Takira), agronomist and local leader of the San Francisco Pichátaro community. Gratitude also goes to our local partners, to the Institute of Geography, UNAM in Mexico, and to ITC-Enschede in the Netherlands; special thanks to Abbas Farshad and Pedro S. Urquijo for their permanent technical assistance, and to Victor Toledo and the anonymous reviewers for their valuable comments. References Argueta, A., 1988. Etnozoología P'urhe. Historia, utilización y nomenclatura P'urhepecha de los animales. Professional thesis. Faculty of Science, UNAM, Mexico City. Barrera-Bassols, N., 1988. Etnoedafología Purhépecha. México Indígena 4 (24), 47–52. Barrera-Bassols, N., 2003. Symbolism, knowledge and management of soil and land resources in indigenous communities: ethnopedology at global, regional and local scales. PhD thesis. Laboratory of Soil Science, Ghent University, Belgium. Barrera-Bassols, N., Zinck, J.A., 2003. Land moves and behaves: indigenous discourse on sustainable land management in Pichátaro, Pátzcuaro Basin, Mexico. Geografiska Annaler 85A (3–4), 229–246. Barrera-Bassols, N., Toledo, V.M., 2005. Ethnoecology of the Yucatec Maya: symbolism, knowledge and management of natural resources. Special Issue: Ethnoecology: Journal of Latin American Geography, vol. 4 (1), pp. 9–42. Behrens, C.A., 1989. The scientific basis of Shipibo soil classification and land use: changes in soil-plant associations with cash cropping. American Anthropologist 91 (1), 83–100. Bellón, M.R., 1990. The ethnoecology of maize production under technological change. PhD thesis. University of California at Davis, USA.

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162 Berlin, B., 1992. Ethnobiological Classification: Principles of Categorization of Plants and Animals in Traditional Societies. Princeton University Press, New Jersey, USA. Buol, S.W., Hole, F.D., McCraken, R.J., Southard, R.J., 1997. Soil Genesis and Classification. Iowa State University Press, Ames, USA. Bradley, P., 1983. Peasants, Soils and Classification: An Investigation into Vernacular Soil Typology from Guidimaka of Mauritania. University of Newcastle Upon Tyne, Newcastle Upon Tyne, UK. Carter, W.E., 1967. New Land and Old Traditions: Kekchi Cultivators in Guatemala Lowlands. Latin American Monographs, 2nd Series, vol. 6. University of Florida Press, Gainesville, USA. Conklin, H.C., 1957. Hanunnó Agriculture: A Report on an Integrated System of Shifting Cultivation in the Philippines. FAO, Rome. Endfield, G.H., O'Hara, S.L., 1999. Perception or deception? Land degradation in post-conquest Michoacán, west central Mexico. Colonial Latin American Review 7 (2), 205–224. Ericksen, P.J., Ardón, M., 2003. Similarities and differences between farmer and scientist views on soil quality issues in central Honduras. Geoderma 111 (3–4), 233–248. FAO, 1976. A Framework for Land Evaluation. FAO Soils Bulletin, vol. 32. FAO, Rome. FAO, 1998. Topsoil Characterization for Sustainable Land Management. Land and Water Development Division, Soil Resources, Management and Conservation Service. FAO, Rome. FAO-UNESCO, 1990. Soil Map of the World. Revised Legend. FAO, Rome. Fisher, C.T., 2000. Landscapes of the Lake Pátzcuaro Basin. PhD thesis. University of Wisconsin, Madison, USA. Fisher, C.T., Pollard, H., Israde-Alcántara, I., Garduño-Monroy, V.H., Banerjee, S.K., 2004. A reexamination of human induced environmental change within the Lake Pátzcuaro Basin, Michoacán, Mexico. PNAS 100 (8), 4957–4962. Foster, G.M., 1953. Relationships between Spanish and Spanish– American folk medicine. Journal of American Folklore 62, 201–217. Foster, G.M., 1960. Culture and Conquest: America's Spanish Heritage. Publications in Anthropology, vol. 27. Viking Fund, USA. Giunta, I., 1998. La conscenza indigena e lo sviluppo agricolo. Uno studio di caso nella Meseta P'urhepecha-Messico. MA thesis. Universita ´Degli Studi di Roma´ “La Sapienza”. Roma. Gorenstein, S., Pollard, H.P., 1983. The Tarascan Civilization: A Late pre-Hispanic Cultural System. Vanderbilt University Press, Nashville, USA. ITC, 2002. ILWIS user's manual (version 3.11). ITC. Enschede, The Netherlands. Israde-Alcántara, I., Garduño-Monroy, V.H., Fisher, C.T., Pollard, D. H., Rodrı˙guez-Pascua, M.A., 2005. Lake level change, climate, and the impact of natural events: the role of seismic and volcanic events in the formation of the Lake Pátzcuaro Basin, Michoacán, Mexico. Quaternary International 135, 35–46. Jenny, H., 1941. Factors of Soil Formation. McGraw-Hill, New York, USA. Jenny, H., 1980. The Soil Resource: Origin and Behavior. Springer, New York, USA. Karlen, D.L., Mausbach, M.J., Doran, J.W., Cline, R.G., Harris, R.F., Schuman, G.E., 1997. Soil quality: a concept, definition, and framework for evaluation. Soil Science Society of America Journal 61, 4–10. Krogh, L., Paarup-Lauresen, B., 1997. Indigenous soil knowledge among the Fulani of northern Burkina Faso: linking soil science

161

and anthropology in analysis of natural resource management. GeoJournal (43), 189–197. Lal, R., 1998. Soil quality and agricultural sustainability. In: Lal, R. (Ed.), Soil Quality and Agricultural Sustainability. Ann Arbor Press, Chelsea, USA, pp. 3–13. Licona-Vargas, A.L., Ortı˙z-Solorio, C., Pájaro, H.D., 1992. Metodologı˙a para el levantamiento de tierras campesinas a nivel regional en ejidos del centro de Veracruz, México. Agrociencia, Serie Agua-Suelo-Clima 3 (4), 91–105. Luhr, J.F., Smikin, T., 1993. Paricutin. The Volcano Born in a Mexican Cornfield. Geoscience Press, Phoenix, USA. Mazzucato, V., Niemeijer, D., 2000. Rethinking soil and water conservation in a changing society: a case study in eastern Burkina Faso. PhD thesis. University of Wageningen, The Netherlands. Mikkelsen, J.H., Langhor, R., 1997. Comparison of international, national and farmers' classification systems, applied to soils of the western Dagomba district (northern Ghana). Geografisk Tidsskrift: Danish Journal of Geography 97, 47–57. Payton, R.W., Barr, J.J.F., Martin, A., Sillitoe, P., Deckers, J.F., Gowing, J.W., Hatibu, N., Naseem, S.B., Tenywa, M., Zuberi, M. I., 2003. Contrasting approaches to integrating indigenous knowledge about soils and scientific soil survey in East Africa and Bangladesh. Geoderma 111 (3–4), 355–386. Queiroz Neto, J.P., 1998. Soil science—its nature and the challenges it must face. Introductory Conferences and Debate. 16th World Congress of Soil Science. IUSS, Montpellier, France, pp. 25–40. Röling, N., Browers, H.A.M., 1999. Living local knowledge for sustainable development. In: Prain, G., Fujisaka, S., Warren, D.M. (Eds.), Biological and Cultural Diversity: The Role of Indigenous Agricultural Experimentation in Development. Intermediate Technology Publications, London, UK, pp. 147–157. Romig, D.E., 1995. Farmers knowledge of soil health and its role in quality assessment. MSc thesis. University of Wisconsin, Madison, USA. Sanchez, P.A., Couto, W., Buol, W.B., 1982. The fertility capability classification systems: interpretation, applicability and modification. Geoderma 27, 283–309. Saporito, M.S., 1975. Chemical and mineral studies of a core from Lake Pátzcuaro, Mexico. MSc thesis. University of Minnesota, USA. Schöndube, O., 1987. El occidente de México: algunas características y problemas. In: Homenaje a Ramón Piña Chan. IIA, Serie Antropológica 79. UNAM, Mexico, pp. 403–410. Shah, P.B., 1993. Local classification of agricultural land in the Jhiku Khola Watershed. In: Tamang, D. (Ed.), Indigenous Management of Natural Resources. HMG Ministry of Agriculture/Winrock International, Katmandu/London, pp. 159–163. Soil Survey Staff, 1999. Keys to Soil Taxonomy, Eight Edition. USDA Natural Resource Conservation Service, USA. Talawar, S., Rhoades, R.E., 1998. Scientific and local classification and management of soils. Agriculture and Human Values 15, 3–14. Toledo, V.M., 1992a. What is ethnoecology? Origins, scope and implications of a rising discipline. Etnoecológica 1, 5–21. Toledo, V.M., 1992b. Pátzcuaro's lesson: nature, production and culture in an indigenous region of Mexico. In: Oldfield, M.E., Alcorn, J. (Eds.), Biodiversity, Culture and Ecodevelopment. Westview Press, Boulder, CO, USA, pp. 147–171. Toledo, V.M., 2002. Ethnoecology: a conceptual framework for the study of indigenous knowledge of nature. In: Stepp, J.R., Wyndham, E.S., Zarger, R.S. (Eds.), Ethnobiology and Biocultural Diversity. International Society of Ethnobiology, Georgia, USA, pp. 511–522.

162

N. Barrera-Bassols et al. / Geoderma 135 (2006) 140–162

Toledo, V.M., Barrera-Bassols, N., 1984. Ecología y Desarrollo Rural en Pátzcuaro: un Modelo para el Análisis Interdisciplinario de Comunidades Campesinas. Instituto de Biología UNAM/Ford Foundation, México. Toledo, V.M., Caballero, J., Argueta, A., Mapes, C., Barrera-Bassols, N., Nuñez, M.A., 1980. Los Purhépecha de Pátzcuaro: una aproximación ecológica. América Indígena 40, 17–37. West, R., 1947. Cultural Geography of the Modern Tarascan Area. Institute of Social Anthropology Publication, vol. 7. Smithsonian Institution, Washington D.C., USA.

Zinck, J.A., 1989. Physiography and Soils 2. ITC, Enschede, The Netherlands. Zonneveld, I.S., 1995. Land Ecology: An Introduction to Landscape Ecology as a Base for Land Evaluation. SPB Academic Publishing, Amsterdam, The Netherlands.