Multitask learning of environmental spatial data

The present research deals with an application of artificial neural networks for multitask learning from spatial environmental data. The real case study (sediments contamination of Geneva Lake) consists of 8 pollutants. There are different relationships between these variables, from linear correlations to strong nonlinear dependencies. The main idea is to construct a subsets of pollutants which can be efficiently modeled together within the multitask framework. The proposed two-step approach is based on: 1) the criterion of nonlinear predictability of each variable ?k? by analyzing all possible models composed from the rest of the variables by using a General Regression Neural Network (GRNN) as a model; 2) a multitask learning of the best model using multilayer perceptron and spatial predictions. The results of the study are analyzed using both machine learning and geostatistical tools.

SYMBIOTIC EFFECTIVENESS AND COMPETITIVENESS OF CALOPO RHIZOBIAL ISOLATES IN AN ARGISSOLO VERMELHO-AMARELO UNDER THREE VEGETATION COVERS IN THE DRY FOREST ZONE OF PERNAMBUCO

Biological N fixation in forage legumes is an important alternative to reduce pasture degradation, and is strongly influenced by the inoculant symbiotic capability. This paper evaluates the effectiveness of Calopo (Calopogonium mucunoides) rhizobial isolated from soil under three vegetation covers of an Argissolo Vermelho-Amarelo of the Dry Forest Zone of Pernambuco. An experiment was conducted evaluating 25 isolates, aside from 5 uninoculated controls with 0; 309; 60; 90 and 120 kg ha-1 N, and a treatment inoculated with the SEMIA 6152 strain. The first cut was performed 45 days after inoculation and a second and third cut after 45-day-intervals. Shoot N content was quantified at all cuts. Shoot dry mass was affected by N rates at all cuts. Shoot dry mass increased from the first to the second cut in inoculated plants. There was no difference between rhizobial isolates from the different plant covers for any of the variables. Most variables were significantly and positively correlated.

SOIL PHOSPHORUS THRESHOLDS IN EVALUATING RISK OF ENVIRONMENTAL TRANSFER TO SURFACE WATERS IN SANTA CATARINA, BRAZIL

The State of Santa Catarina, Brazil, has agricultural and livestock activities, such as pig farming, that are responsible for adding large amounts of phosphorus (P) to soils. However, a method is required to evaluate the environmental risk of these high soil P levels. One possible method for evaluating the environmental risk of P fertilization, whether organic or mineral, is to establish threshold levels of soil available P, measured by Mehlich-1 extractions, below which there is not a high risk of P transfer from the soil to surface waters. However, the Mehlich-1 extractant is sensitive to soil clay content, and that factor should be considered when establishing such P-thresholds. The objective of this study was to determine P-thresholds using the Mehlich-1 extractant for soils with different clay contents in the State of Santa Catarina, Brazil. Soil from the B-horizon of an Oxisol with 800 g kg-1 clay was mixed with different amounts of sand to prepare artificial soils with 200, 400, 600, and 800 g kg-1 clay. The artificial soils were incubated for 30 days with moisture content at 80 % of field capacity to stabilize their physicochemical properties, followed by additional incubation for 30 days after liming to raise the pH(H2O) to 6.0. Soil P sorption curves were produced, and the maximum sorption (Pmax) was determined using the Langmuir model for each soil texture evaluated. Based on the Pmax values, seven rates of P were added to four replicates of each soil, and incubated for 20 days more. Following incubation, available P contents (P-Mehlich-1) and P dissolved in the soil solution (P-water) were determined. A change-point value (the P-Mehlich-1 value above which P-water starts increasing sharply) was calculated through the use of segmented equations. The maximum level of P that a soil might safely adsorb (P-threshold) was defined as 80 % of the change-point value to maintain a margin for environmental safety. The P-threshold value, in mg dm-3, was dependent on the soil clay content according to the model P-threshold = 40 + Clay, where the soil clay content is expressed as a percentage. The model was tested in 82 diverse soil samples from the State of Santa Catarina and was able to distinguish samples with high and low environmental risk.

Field Evaluation of Timber Preservation Treatments for Highway Applications Final Report, December 2007

Timber material repair and replacement cost for timber bridges is a considerable expense to highway agencies in Iowa, especially to county road departments. To address these needs, the objectives of this investigation was to study the field effectiveness of various treatment alternatives used on Iowa roadway projects and to determine if the current specifications and testing are adequate for providing proper wood preservation. To satisfy the research needs, the project scope involved a literature review, identification of metrics, questionnaire survey of Iowa counties, onsite inspections, and a review of current specifications and testing procedures. Based on the preservative information obtained, the following general conclusions were made: Copper naphthenate is recommended as the plant-applied preservative treatment for timber bridges. Best Management Practices should be followed to ensure quality treatment of timber materials. Bridge maintenance programs need to be developed and implemented. The Iowa Department of Transportation specifications for preservative treatment are the regulating specification for bridges constructed with state or federal funding in Iowa and are also recommended for all other bridges.

Antibiotic resistance genes in the bacteriophage DNA fraction of environmental samples

Antibiotic resistance is an increasing global problem resulting from the pressure of antibiotic usage, greater mobility of the population, and industrialization. Many antibiotic resistance genes are believed to have originated in microorganisms in the environment, and to have been transferred to other bacteria through mobile genetic elements. Among others, ß-lactam antibiotics show clinical efficacy and low toxicity, and they are thus widely used as antimicrobials. Resistance to ß-lactam antibiotics is conferred by ß-lactamase genes and penicillin-binding proteins, which are chromosomal- or plasmid-encoded, although there is little information available on the contribution of other mobile genetic elements, such as phages. This study is focused on three genes that confer resistance to ß-lactam antibiotics, namely two ß-lactamase genes (blaTEM and blaCTX-M9) and one encoding a penicillin-binding protein (mecA) in bacteriophage DNA isolated from environmental water samples. The three genes were quantified in the DNA isolated from bacteriophages collected from 30 urban sewage and river water samples, using quantitative PCR amplification. All three genes were detected in the DNA of phages from all the samples tested, in some cases reaching 104 gene copies (GC) of blaTEM or 102 GC of blaCTX-M and mecA. These values are consistent with the amount of fecal pollution in the sample, except for mecA, which showed a higher number of copies in river water samples than in urban sewage. The bla genes from phage DNA were transferred by electroporation to sensitive host bacteria, which became resistant to ampicillin. blaTEM and blaCTX were detected in the DNA of the resistant clones after transfection. This study indicates that phages are reservoirs of resistance genes in the environment.

Groundwater fluctuations and footslope ferricrete soils in the humid tropical zone of southern Cameroon

This paper discusses the relationship between the differentiation of ferruginous accumulations and the variable water saturation of footslope soil patterns. An analysis of the slope morphology of a typical hill in the forest zone of southern Cameroon and a seasonal survey of the levels of groundwaters, springs and rivers were considered in relation to the petrology of different soil patterns. The study site is a tabular hillock whose slopes present a progressive development from steep to gentle slopes. The variable residence time of water within the soil, creating an alternation of reducing and oxidizing conditions, affects oil chemistry, structure and lateral extension of the soil patterns. The ferruginous soil patterns, being formed on the footslopes, gradually increase in extent with decreasing slope angle and the relative rise of the groundwater level. The steep footslopes, where groundwater has a shorter residence time, show a soft mottled clay pattern, restricted to the bottom part of the slope. The moderate footslopes exhibit a deep permanent and a temporary perched groundwater table. The latter, with its regular capillary fringe, contributes to more reducing conditions within isolated domains in the soil patterns, and thus to the alternation with oxidizing conditions, generating a continuous hard soil pattern (massive carapace). The more gently dipping footslopes exhibit groundwater levels near the surface and also a significant amplitude of groundwater fluctuation. Iron, previously accumulated in moderate footslope patterns, is reduced, remobilized, and leached. The soil patterns formed develop into a variegated carapace, more extended along the slope, containing less iron, but nevertheless more hardened, due to the important fluctuations of the groundwater table. These patterns are limited to the zone of groundwater fluctuation and deteriorate as the water fluctuation zone recedes. Copyright (c) 2005 John Wiley & Sons, Ltd.

Multiple Kernel Learning of Environmental Data. Case study: Analysis and Mapping of Wind Fields

The paper presents the Multiple Kernel Learning (MKL) approach as a modelling and data exploratory tool and applies it to the problem of wind speed mapping. Support Vector Regression (SVR) is used to predict spatial variations of the mean wind speed from terrain features (slopes, terrain curvature, directional derivatives) generated at different spatial scales. Multiple Kernel Learning is applied to learn kernels for individual features and thematic feature subsets, both in the context of feature selection and optimal parameters determination. An empirical study on real-life data confirms the usefulness of MKL as a tool that enhances the interpretability of data-driven models.

Induction of brain aquaporin 9 (AQP9) in catecholaminergic neurons in diabetic rats.

Aquaporin 9 facilitates the diffusion of water but also glycerol and monocarboxylates, known as brain energy substrates. AQP9 was recently observed in catecholaminergic neurons that are implicated in energy homeostasis and also possibly in neuroendocrine effects of diabetes. Recently it has been observed that the level of AQP9 expression in hepatocytes is sensitive to the blood concentration of insulin. Furthermore, insulin injection in the brain is known to be related to the energy homeostasis. Based on these observations, we investigated if the concentration of insulin affects the level of brain AQP9 expression and if so, in which cell types. This study has been carried out, in a model of the diabetic rat generated by streptozotocin injection and on brainstem slices. In diabetic rats showing a decrease in systemic insulin concentration, AQP9 is only increased in brain areas containing catecholaminergic neurons. In contrast, no significant change is detected in the cerebral cortex and the cerebellum. Using immunocytochemistry, we are able to show that the increase in AQP9 expression is specifically present in catecholaminergic neurons. In brainstem slice cultures, 2 microM insulin induces a significant decrease in AQP9 protein levels 6 h after application, suggesting that brain AQP9 is also regulated by the insulin. These results show that the level of expression of brain AQP9 is affected by variations of the concentration of insulin in a diabetic model and in vitro.

The influence of environmental spatial structure on the life history traits and diversity of species in a metacommunity

Several models have been proposed to understand how so many species can coexist in ecosystems. Despite evidence showing that natural habitats are often patchy and fragmented, these models rarely take into account environmental spatial structure. In this study we investigated the influence of spatial structure in habitat and disturbance regime upon species' traits and species' coexistence in a metacommunity. We used a population-based model to simulate competing species in spatially explicit landscapes. The species traits we focused on were dispersal ability, competitiveness, reproductive investment and survival rate. Communities were characterized by their species richness and by the four life-history traits averaged over all the surviving species. Our results show that spatial structure and disturbance have a strong influence on the equilibrium life-history traits within a metacommunity. In the absence of disturbance, spatially structured landscapes favour species investing more in reproduction, but less in dispersal and survival. However, this influence is strongly dependent on the disturbance rate, pointing to an important interaction between spatial structure and disturbance. This interaction also plays a role in species coexistence. While spatial structure tends to reduce diversity in the absence of disturbance, the tendency is reversed when disturbance occurs. In conclusion, the spatial structure of communities is an important determinant of their diversity and characteristic traits. These traits are likely to influence important ecological properties such as resistance to invasion or response to climate change, which in turn will determine the fate of ecosystems facing the current global ecological crisis.

Carbonic anhydrase activity in primary sensory neurons. II. Influence of environmental factors on the phenotypic expression of the enzyme in dissociated cultures of chicken dorsal root ganglion cells.

Neuronal subpopulations of dorsal root ganglion (DRG) cells in the chicken exhibit carbonic anhydrase (CA) activity. To determine whether CA activity is expressed by DRG cells maintained in in vitro cultures, dissociated DRG cells from 10-day-old chick embryos were cultured on a collagen substrate. The influence exerted by environmental factors on the enzyme expression was tested under various conditions of culture. Neuron-enriched cell cultures and mixed DRG-cell cultures (including numerous non-neuronal cells) were performed either in a defined medium or in a horse serum-supplemented medium. In all the tested conditions, subpopulations of cultured sensory neurons expressed CA activity in their cell bodies, while their neurites were rarely stained; in each case, the percentage of CA-positive neurons declined with the age of the cultures. The number and the persistence of neurons possessing CA activity as well as the intensity of the reaction were enhanced by addition of horse serum. In contrast, the expression of the neuronal CA activity was not affected by the presence of non-neuronal cells or by the rise of CO2 concentration. Thus, the appearance and disappearance of neuronal subpopulations expressing CA activity may be decisively influenced by factors contained in the horse serum. The loss of CA-positive neurons with time could result from a cell selection or from genetic repression. Analysis of the time curves does not support a preferential cell death of CA-positive neurons but suggests that the eventual conversion of CA-positive neurons into CA-negative neurons results from a loss of the enzyme activity. These results indicate that the phenotypic expression of cultured sensory neurons is dependent on defined environmental factors.

Influence of environmental factors on the expression of phenotypic characters by chicken dorsal root ganglion cells in culture.

Primary sensory neurons were grown under four conditions of culture. The influence of nonneuronal cells, horse serum or both was studied on the phenotypic expression of certain neuronal subpopulations. The number of neurons expressing acetylcholinesterase, alpha-bungarotoxin-binding sites or a high uptake capacity for glutamine was enhanced by nonneuronal cells. The horse serum increases the neuronal subpopulation exhibiting a carbonic anhydrase activity. Certain phenotypic changes fit conditions consistent with an epigenetic induction rather than a cell selection.

The importance of environmental exposures to physical, mental and social well-being

"Health is a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity" states the WHO. However, the current focus in this important area seems to be on reducing diseases, while less attention is paid on aspects how to increase the well-being of populations. This paper reviews three examples where well-being has drawn attention of the public and policy makers, and compares the policies of two wealthy countries. The first example is noise. Noise can reduce sleep quality and cause physiological, mental, and social effects. In Switzerland, noise receives a lot of attention by the public. Swiss laws are extensive, e.g., they prohibit trucks and planes from traveling at night. In the USA, there is little public attention and no national strategy against environmental noise. The second example is aesthetics and recreation. Many humans seek contact with the beauty of nature. The USA and Switzerland have similar strategies for achieving clear waters, while the protection of scenic views is approached very differently. Lifestyle is the last example. In the USA, the desire for individual freedom is a leading cause for suburban sprawl, a car-dependent sedentary lifestyle resulting in obesity, asthma and loss of community spirit. In Switzerland, a strict land use planning seeks to balance individual and public interests and stresses public transportation, which seems to be a more promising approach. Paying attention to aspects of well-being while developing political strategies might be a promising model to tackle environmental problems. Successful strategies employed so far seem to include the public, local authorities, politicians and scientists in this process, which might have been a key for their success. [Authors]