Comparison between artificial neural networks and maximum likelihood classification in digital soil mapping

Soil surveys are the main source of spatial information on soils and have a range of different applications, mainly in agriculture. The continuity of this activity has however been severely compromised, mainly due to a lack of governmental funding. The purpose of this study was to evaluate the feasibility of two different classifiers (artificial neural networks and a maximum likelihood algorithm) in the prediction of soil classes in the northwest of the state of Rio de Janeiro. Terrain attributes such as elevation, slope, aspect, plan curvature and compound topographic index (CTI) and indices of clay minerals, iron oxide and Normalized Difference Vegetation Index (NDVI), derived from Landsat 7 ETM+ sensor imagery, were used as discriminating variables. The two classifiers were trained and validated for each soil class using 300 and 150 samples respectively, representing the characteristics of these classes in terms of the discriminating variables. According to the statistical tests, the accuracy of the classifier based on artificial neural networks (ANNs) was greater than of the classic Maximum Likelihood Classifier (MLC). Comparing the results with 126 points of reference showed that the resulting ANN map (73.81 %) was superior to the MLC map (57.94 %). The main errors when using the two classifiers were caused by: a) the geological heterogeneity of the area coupled with problems related to the geological map; b) the depth of lithic contact and/or rock exposure, and c) problems with the environmental correlation model used due to the polygenetic nature of the soils. This study confirms that the use of terrain attributes together with remote sensing data by an ANN approach can be a tool to facilitate soil mapping in Brazil, primarily due to the availability of low-cost remote sensing data and the ease by which terrain attributes can be obtained.

An appropriate data set size for digital soil mapping in Erechim, Rio Grande do Sul, Brazil

Digital information generates the possibility of a high degree of redundancy in the data available for fitting predictive models used for Digital Soil Mapping (DSM). Among these models, the Decision Tree (DT) technique has been increasingly applied due to its capacity of dealing with large datasets. The purpose of this study was to evaluate the impact of the data volume used to generate the DT models on the quality of soil maps. An area of 889.33 km² was chosen in the Northern region of the State of Rio Grande do Sul. The soil-landscape relationship was obtained from reambulation of the studied area and the alignment of the units in the 1:50,000 scale topographic mapping. Six predictive covariates linked to the factors soil formation, relief and organisms, together with data sets of 1, 3, 5, 10, 15, 20 and 25 % of the total data volume, were used to generate the predictive DT models in the data mining program Waikato Environment for Knowledge Analysis (WEKA). In this study, sample densities below 5 % resulted in models with lower power of capturing the complexity of the spatial distribution of the soil in the study area. The relation between the data volume to be handled and the predictive capacity of the models was best for samples between 5 and 15 %. For the models based on these sample densities, the collected field data indicated an accuracy of predictive mapping close to 70 %.

Seeds with high molybdenum concentration improved growth and nitrogen acquisition of rhizobium-inoculated and nitrogen-fertilized common bean plants

Seeds of common bean (Phaseolus vulgaris) with high molybdenum (Mo) concentration can supply Mo plant demands, but to date no studies have concomitantly evaluated the effects of Mo-enriched seeds on plants inoculated with rhizobia or treated with N fertilizer. This work evaluated the effects of seed Mo on growth and N acquisition of bean plants fertilized either by symbiotic N or mineral N, by measuring the activities of nitrogenase and nitrate reductase and the contribution of biological N2 fixation at different growth stages. Seeds enriched or not with Mo were sown with two N sources (inoculated with rhizobia or fertilized with N), in pots with 10 kg of soil. In experiment 1, an additional treatment consisted of Mo-enriched seeds with Mo applied to the soil. In experiment 2, the contribution of N2 fixation was estimated by 15N isotope dilution. Common bean plants grown from seeds with high Mo concentration flowered one day earlier. Seeds with high Mo concentration increased the leaf area, shoot mass and N accumulation, with both N sources. The absence of effects of Mo application to the soil indicated that Mo contents of Mo-enriched seeds were sufficient for plant growth. Seeds enriched with Mo increased nitrogenase activity at the vegetative stage of inoculated plants, and nitrate reductase activity at late growth stages with both N sources. The contribution of N2 fixation was 17 and 61 % in plants originating from low- or high-Mo seeds, respectively. The results demonstrate the benefits of sowing Mo-enriched seeds on growth and N nutrition of bean plants inoculated with rhizobia or fertilized with mineral N fertilizer.

Mapeamento pedológico digital da folha Botucatu (SF-22-Z-B-VI-3): treinamento de dados em mapa tradicional e validação de campo

O mapeamento digital de solos permite prever padrões de ocorrência de solos com base em áreas de referência e no uso de técnicas de mineração de dados para modelar associações solo-paisagem. Os objetivos deste trabalho foram produzir um mapa pedológico digital por meio de técnicas de mineração de dados aplicadas a variáveis geomorfométricas e de geologia, com base em áreas de referência; e testar a confiabilidade desse mapa por meio de validação em campo com diferentes sistemas de amostragem. O mapeamento foi realizado na folha Botucatu (SF-22-Z-B-VI-3), utilizando-se as folhas 1:50.000, Dois Córregos e São Pedro, como áreas de referência. Variáveis descritoras do relevo e de geologia associadas às unidades de mapeamento pedológico das áreas de referência compuseram a matriz de dados de treinamento. A matriz foi analisada pelo algoritmo PART de árvore de decisão, do aplicativo Weka (Waikato Environment for Knowledge Analysis), que cria regras de classificação. Essas regras foram aplicadas aos dados geomorfométricos e geológicos da folha Botucatu, para predição de unidades de mapeamento pedológico. A validação de campo dos mapas digitais deu-se por meio de amostragem por transectos em uma unidade de mapeamento da folha São Pedro e de forma aleatório-estratificada na folha Botucatu. A avaliação da unidade de mapeamento na folha São Pedro verificou confiabilidade, respectivamente, de 83 e 66 %, para os mapas pedológicos digital e tradicional com legenda simplificada. Apesar de terem sido geradas regras para todas as unidades de mapeamento pedológico das áreas de treinamento, nem todas as unidades de mapeamento foram preditas na folha Botucatu, o que resultou das diferenças de relevo e geologia entre as áreas de treinamento e de mapeamento. A validação de campo do mapa digital da folha Botucatu verificou exatidão global de 52 %, compatível com levantamentos em nível de reconhecimento de baixa intensidade, e kappa de 0,41, indicando qualidade Boa. Unidades de mapeamento mais extensas geraram mais regras, resultando melhor reprodução dos padrões solo-relevo na área a ser mapeada. A validação por transectos na folha São Pedro indicou compatibilidade do mapa digital com o nível de reconhecimento de alta intensidade e compatibilidade do mapa tradicional, após simplificação de sua legenda, com o nível de reconhecimento de baixa intensidade. O treinamento do algoritmo em mapas e não em observações pontuais reduziu em 14 % a exatidão do mapa pedológico digital da folha Botucatu. A amostragem aleatório-estratificada pelo hipercubo latino é apropriada a mapeamentos com extensa base de dados, o que permite avaliar o mapa como um todo, tornando os trabalhos de campo mais eficientes. A amostragem em transectos é compatível com a avaliação da pureza de unidades de mapeamento individualmente, não necessitando de base de dados detalhada e permitindo estudos de associações solo-paisagem em pedossequências.

Nutrient extraction and exportation by common bean cultivars under different fertilization levels: I - macronutrients

The use of cultivars with a higher yield potential and the adoption of new technology have achieved high grain yields in common bean, which probably changed the demand for nutrients in this crop. However, there is almost no information about the periods of the cycle in which nutrients are most demanded at which quantities by the main cultivars. The objective of this study was to evaluate the macronutrient extraction and exportation by the common bean cultivars Pérola and IAC Alvorada, under different levels of NPK fertilization, on a dystroferric Red Nitosol, in Botucatu, São Paulo State, Brazil. The experiment was arranged in a randomized complete block (split plot) design with four replications. The plots consisted of six treatments based on a 2 x 3 factorial model, represented by two cultivars and three NPK levels (PD0 - 'Pérola' without fertilization, PD1 - 'Pérola' with 50 % of recommended fertilization, PD2 - 'Pérola' with 100 % of recommended fertilization, AD0 - 'IAC Alvorada' without fertilization, AD1 - 'IAC Alvorada' with 50 % of recommended fertilization, and AD2 - 'IAC Alvorada' with 100 % of recommended fertilization) and subplots sampled seven times during the cycle. At higher levels of NPK fertilization, the grain yield and macronutrient extraction and exportation of both cultivars were higher, but without statistical differences. Macronutrient absorption was higher in the treatments with 100 % of recommended NPK fertilization (average amounts per hectare: 140 kg N, 16.5 kg P, 120 kg K, 69 kg Ca, 17.9 kg Mg, and 16.3 kg S). Regardless of the treatment, the demand for N, P, K, Ca, and Mg was highest from 45 to 55 days after emergence (DAE), i.e., in the R7 stage (pod formation), while the highest S absorption rates were concentrated between 55 and 65 DAE. More than 70 % of P, between 58 and 69 % of N, 40 and 52 % of S, 40 and 48 % of K, and 35 and 45 % of Mg absorbed during the cycle was exported with grains, whereas less than 15 % of Ca was exported.

Nutrient extraction and exportation by common bean cultivars under different fertilization levels: II - micronutrients

Where the level of agricultural technology is higher, common bean cultivars with a higher yield potential possibly require greater amounts of micronutrients. In Brazil however, there is a lack of information about the micronutrient extraction and exportation by the main grown cultivars. The objective of this study was to evaluate micronutrient (B, Cu, Fe, Mn, and Zn) extraction and exportation by common bean cultivars Pérola and IAC Alvorada, under different levels of NPK fertilization, on a dystroferric Red Nitosol, in Botucatu, São Paulo State, Brazil. The experiment was arranged in a randomized complete block (split plot) design with four replications. The plots consisted of six treatments based on a 2 x 3 factorial model, represented by two cultivars and three NPK levels (PD0 - 'Pérola' without fertilization, PD1 - 'Pérola' with 50 % of recommended fertilization, PD2 - 'Pérola' with 100 % of recommended fertilization, AD0 - 'IAC Alvorada' without fertilization, AD1 - 'IAC Alvorada' with 50 % of recommended fertilization, and AD2 - 'IAC Alvorada' with 100 % of recommended fertilization) and subplots sampled seven times during the cycle. Higher levels of NPK fertilization increased micronutrient extraction by both cultivars, and treatments with 100 % of recommended NPK fertilization extracted on average 167 g B, 58 g Cu, 1,405 g Fe, 1,213 g Mn and 211 g Zn per hectare. Regardless of the treatment, the highest demand period for B, Cu, Fe, Mn and Zn in both cultivars occurred at the R7 stage (pod formation), i.e. 42 to 55 days after emergence (DAE). The amount of B, Cu, Fe, Mn and Zn exported depended mainly on the level of NPK fertilization used, with values per hectare ranging from 38 to 90 g of B, 12 to 26 g of Cu, 222 to 568 g of Fe 234 to 467 g of Mn, and 40 to 96 g of Zn.

Comparison between detailed digital and conventional soil maps of an area with complex geology

Since different pedologists will draw different soil maps of a same area, it is important to compare the differences between mapping by specialists and mapping techniques, as for example currently intensively discussed Digital Soil Mapping. Four detailed soil maps (scale 1:10.000) of a 182-ha sugarcane farm in the county of Rafard, São Paulo State, Brazil, were compared. The area has a large variation of soil formation factors. The maps were drawn independently by four soil scientists and compared with a fifth map obtained by a digital soil mapping technique. All pedologists were given the same set of information. As many field expeditions and soil pits as required by each surveyor were provided to define the mapping units (MUs). For the Digital Soil Map (DSM), spectral data were extracted from Landsat 5 Thematic Mapper (TM) imagery as well as six terrain attributes from the topographic map of the area. These data were summarized by principal component analysis to generate the map designs of groups through Fuzzy K-means clustering. Field observations were made to identify the soils in the MUs and classify them according to the Brazilian Soil Classification System (BSCS). To compare the conventional and digital (DSM) soil maps, they were crossed pairwise to generate confusion matrices that were mapped. The categorical analysis at each classification level of the BSCS showed that the agreement between the maps decreased towards the lower levels of classification and the great influence of the surveyor on both the mapping and definition of MUs in the soil map. The average correspondence between the conventional and DSM maps was similar. Therefore, the method used to obtain the DSM yielded similar results to those obtained by the conventional technique, while providing additional information about the landscape of each soil, useful for applications in future surveys of similar areas.

The list of the chromosome races of the common shrew (Sorex araneus)

The list of chromosome races of the common shrew (Sorex araneus) was compiled, the vast literature has been scrutinized, and unpublished data have been added. Altogether, 50 chromosome races could be listed. The name and its synonyms, chromosomal constitution, author of the description, type locality, known distribution range, and additional information are reported for individual races. The present list should be considered a working document that will be regularly updated and supplemented.

Genotypic variability in seed accumulation of foliar-applied molybdenum to common bean

The genotypic variability in molybdenum (Mo) accumulation in common bean seeds has been demonstrated in cases in which soil is the main Mo source, but this variability is yet unknown when Mo is foliar-applied. Therefore, seed Mo concentrations (SMoCc) and seed Mo contents (SMoCt) of 12 genotypes were determined in four experiments in the Zona da Mata, Minas Gerais, Brazil, in which plants were sprayed with 600 g ha-1 Mo. For comparison, two additional experiments without external Mo were conducted. Without Mo application, the average SMoCc was undetectable or 2.83 µg g-1, without significant differences among genotypes. On average, with Mo applications, SMoCc ranged from 14.7 to 25.0 µg g-1 and SMoCt, from 3.94 to 6.84 µg. 'Majestoso' was among the genotypes with the highest SMoCc in the four experiments. However, the large-seeded 'Jalo MG-65' and 'Carnaval' generally had higher SMoCt than the small-seeded 'Majestoso'. 'Ouro Negro' and especially 'Valente' were among the genotypes with the lowest SMoCc and SMoCt. The values of these variables were 61 and 90 %, respectively, higher for 'Majestoso' than those for 'Valente'. Our results suggest that common bean genotypes differ in their capacity to accumulate foliar-applied Mo in the seeds. Mo-rich seeds of large-seeded genotypes or of small-seeded of small-seeded genotypes with good capacity to accumulate Mo in seeds can be produced with relatively less Mo fertilizer.

Método do hipercubo latino condicionado para a amostragem de solos na presença de covariáveis ambientais visando o mapeamento digital de solos

A amostragem é uma das etapas mais importantes dos levantamentos de solos. No entanto, os esquemas de amostragem utilizados nos levantamentos convencionais têm se evidenciado inadequados para o mapeamento digital de solos, pois podem comprometer os resultados e, além disso, não possibilitam a realização de análises estatísticas. Este estudo teve por objetivo avaliar o método de amostragem do hipercubo latino condicionado (cLHS, sigla em inglês), na presença de covariáveis ambientais (elevação, declividade, curvatura e mapa de uso e cobertura do solo), em comparação com a amostragem aleatória, na alocação de 100 pontos amostrais, buscando maior representatividade das características ambientais da bacia do rio Guapi-Macacu. O desempenho dos métodos foi avaliado pela análise qualitativa dos histogramas de frequência e das análises estatísticas pelos testes F, T de Student e Kolmogorov-Smirnov (K-S), para cada covariável. Os resultados apresentaram que os pontos selecionados pelo método cLHS possuíam distribuição geográfica mais adequada do que aqueles obtidos pela amostragem aleatória. Além disso, o método cLHS preservou mais a distribuição de frequência das covariáveis contínuas do que a amostragem aleatória; para covariável categórica uso e cobertura do solo os métodos foram equivalentes. Os testes estatísticos confirmaram o melhor desempenho do método cLHS, cujas amostras não diferiram estatisticamente da bacia. Entretanto, a amostragem aleatória apresentou diferença estatística para com a bacia, para todas as covariáveis contínuas para pelo menos um dos testes utilizados. Assim, o método cLHS pode ser considerado como um método satisfatório para seleção de locais de amostragem em áreas heterogêneas similares as deste estudo, visando a utilização no mapeamento digital de solos.

Evaluation of statistical and geostatistical models of digital soil properties mapping in tropical mountain regions

Soil properties have an enormous impact on economic and environmental aspects of agricultural production. Quantitative relationships between soil properties and the factors that influence their variability are the basis of digital soil mapping. The predictive models of soil properties evaluated in this work are statistical (multiple linear regression-MLR) and geostatistical (ordinary kriging and co-kriging). The study was conducted in the municipality of Bom Jardim, RJ, using a soil database with 208 sampling points. Predictive models were evaluated for sand, silt and clay fractions, pH in water and organic carbon at six depths according to the specifications of the consortium of digital soil mapping at the global level (GlobalSoilMap). Continuous covariates and categorical predictors were used and their contributions to the model assessed. Only the environmental covariates elevation, aspect, stream power index (SPI), soil wetness index (SWI), normalized difference vegetation index (NDVI), and b3/b2 band ratio were significantly correlated with soil properties. The predictive models had a mean coefficient of determination of 0.21. Best results were obtained with the geostatistical predictive models, where the highest coefficient of determination 0.43 was associated with sand properties between 60 to 100 cm deep. The use of a sparse data set of soil properties for digital mapping can explain only part of the spatial variation of these properties. The results may be related to the sampling density and the quantity and quality of the environmental covariates and predictive models used.

Common variants in mendelian kidney disease genes and their association with renal function.

Many common genetic variants identified by genome-wide association studies for complex traits map to genes previously linked to rare inherited Mendelian disorders. A systematic analysis of common single-nucleotide polymorphisms (SNPs) in genes responsible for Mendelian diseases with kidney phenotypes has not been performed. We thus developed a comprehensive database of genes for Mendelian kidney conditions and evaluated the association between common genetic variants within these genes and kidney function in the general population. Using the Online Mendelian Inheritance in Man database, we identified 731 unique disease entries related to specific renal search terms and confirmed a kidney phenotype in 218 of these entries, corresponding to mutations in 258 genes. We interrogated common SNPs (minor allele frequency >5%) within these genes for association with the estimated GFR in 74,354 European-ancestry participants from the CKDGen Consortium. However, the top four candidate SNPs (rs6433115 at LRP2, rs1050700 at TSC1, rs249942 at PALB2, and rs9827843 at ROBO2) did not achieve significance in a stage 2 meta-analysis performed in 56,246 additional independent individuals, indicating that these common SNPs are not associated with estimated GFR. The effect of less common or rare variants in these genes on kidney function in the general population and disease-specific cohorts requires further research.

Comparação do uso de modelos digitais de elevação em mapeamento digital de solos em Dois Irmãos, RS, Brasil

Os modelos digitais de elevação (MDEs) são fontes fundamentais para correlacionar a ocorrência e distribuição de solos com a paisagem pelo mapeamento digital de solos (MDS). A influência dos tipos e das resoluções dos MDEs na capacidade de predição dos modelos preditores de classes de solo ainda é pouco estudada. Neste estudo, foram avaliados e comparados os efeitos de diferentes MDEs na predição de ocorrência de unidades de mapeamento de solo (UM). Foram correlacionados 12 atributos do terreno derivados de diferentes MDEs com a ocorrência de UM. Os MDEs utilizados foram os oriundos dos projetos SRTM v4.1, ASTER GDEM v2, TOPODATA e Brasil em Relevo, e os MDEs gerados a partir de curvas de nível na escala de 1:50.000, com resoluções de 30 e 90 m. Os modelos preditores foram treinados por árvore de decisão (Simple Cart) com dados amostrados em 4.280 pontos aleatórios contendo informações dos solos extraídos de um mapa convencional de solos na escala 1:20.000 e 12 atributos do terreno derivados de seis MDEs com tamanhos de pixel de 30 e 90 m. A validação dos modelos preditores de UM foi realizada com a totalidade dos dados da área. Os atributos do terreno que melhor explicaram a ocorrência das UM foram elevação, declividade, comprimento de fluxo e orientação das vertentes. Os MDEs com tamanho de pixel de 30 m geraram correlações solo-paisagem menos acuradas. Os modelos preditores mais acurados e com maior número de UM estimadas foram os gerados a partir dos MDEs com resolução espacial de 90 m (SRTM v4.1 e CN90), sendo esses os MDEs mais indicados para o MDS, quando predominarem relevos plano e suave ondulado.