Mercury distribution in tropical soil profiles related to origin of mercury and soil processes.

The aim of the study is to improve our understanding of the vertical and lateral variations in the mercury content [Hg] of tropical soils. In addition to the distance to anthropogenic sources of Hg, the most frequently evoked determining factor is the abundance of Hg-bearing phases. Soil processes (weathering, mass or water transfer) determine the abundance of carrier phases. We assume that soil processes also have a direct impact on the distribution of Hg and that the impact is different according to the lithogenic or atmospheric origin of this element. We compare two types of soil (oxisol and ultisol) in the French Guiana forest, at localities a few tens of metres apart and exhibiting very different Hg contents. We show that vertical profiles of [Hg] are strongly related to the variations of [Hg(atmospheric)], whereas [Hg(lithogenic)] varies little. The penetration of Hg(atmospheric) from the surface deposits is favoured down to a depth of 3 m in the oxisol and limited to the upper horizons of the ultisol because of contrasted hydraulic conductivity between the two soils. Hg is primarily of lithogenic origin in the alteritic horizons of the ultisol. The relative accumulation of Hg(lithogenic) during the progressive weathering of parental material is limited near the soil surface by the disequilibrium of secondary minerals. Remobilization of Hg(atmospheric) or Hg(lithogenic) stored in the soil is a function of the chemical or particulate erosion of Hg-bearing phases, particularly active in the upper horizons of the ultisol, where lateral flow occurs during rain events. The correlations observed between the iron or clay contents and [Hg] can be caused by the affinity of Hg for these carrier phases, but may also reflect the weathering and the transfer processes which affect together the fate of Hg and the mineralogical and chemical composition of the soil.

Influence of hydroclimatic variations on solute concentration dynamics in nested subtropical catchments with heterogeneous landscapes.

Despite global efforts to monitor water quality in catchments worldwide, tropical and subtropical zones still lack data to study the influence of human activities and climate variations on solute dynamics. In this study, we monitored ten solutes every two weeks for six years (2010-2015) in three nested catchments (2 to30 km2), which contained heterogeneous landscapes composed of forests and agricultural land, and one small neighboring forested catchment (0.4 km2). Data analysis revealed that i) rainfall, discharge and solute concentrations displayed no clear seasonal patterns, unlike many catchments of the temperate zone; ii) solute concentrations in the agricultural area were higher than those in the forested area, but both areas displayed similar temporal patterns due to a common hydroclimatic driver; iii) all four catchments displayed a chemostatic export regime for most of the solutes, similar to catchments of the temperate zone; and iv) a positive correlation was observed between anion concentrations and ENSO (El Niño-Southern Oscillation) index. ENSO appeared to influence both hydroclimatic and anion dynamics in these subtropical catchments.

Geographical modeling of exposure risk to cyanobacteria for epidemiological purposes.

The cyanobacteria-derived neurotoxin ß-methylamino-L-alanine (BMAA) represents a plausible environmental trigger for amyotrophic lateral sclerosis (ALS), a debilitating and fatal neuromuscular disease. With the eutrophication of water bodies, cyanobacterial blooms and their toxins are becoming increasingly prevalent in France, especially in the Brittany region. Cyanobacteria are monitored at only a few recreational sites, preventing an estimation of exposure of the human population. By contrast, phosphorus, a limiting nutrient for cyanobacterial growth and thus considered a good proxy for cyanobacteria exposure, is monitored in many but not all surface water bodies. Our goal was to develop a geographic exposure indicator that could be used in epidemiological research. We considered the total phosphorus (TP) concentration (mg/L) of samples collected between October 2007 and September 2012 at 179 monitoring stations distributed throughout the Brittany region. Using readily available spatial data, we computed environmental descriptors at the watershed level with a Geographic Information System. Then, these descriptors were introduced into a backward stepwise linear regression model to predict the median TP concentration in unmonitored surface water bodies. TP concentrations in surface water follow an increasing gradient from West to East and inland to coast. The empirical concentration model included five predictor variables with a fair coefficient of determination (R(2) = 0.51). The specific total runoff and the watershed slope correlated negatively with the TP concentrations (p = 0.01 and p< 10(-9), respectively), whereas positive associations were found for the proportion of built-up area, the upstream presence of sewage treatment plants, and the algae volume as indicated by the Landsat red/green reflectance ratio (p < 0.01, p < 10(-6) and p < 0.01, respectively). Complementing the monitoring networks, this geographical modeling can help estimate TP concentrations at the watershed level, delivering a proxy for cyanobacteria exposure that can be used along with other risk factors in further ALS epidemiologic case-control studies.

Amazonian former gold mined soils as a source of methylmercury: evidence from a small scale watershed in French Guiana.

Total mercury (HgT) and monomethylmercury (MMHg) were investigated in a tropical head watershed (1 km(2)) of French Guiana. The watershed includes a pristine area on the hill slopes and a former gold mined flat in the bottomland. Concentrations of dissolved and particulate HgT and MMHg were measured in rain, throughfall, soil water and at three points along the stream. Samples were taken in-between and during 14 storm events at the beginning and middle of the 2005 and 2006 rainy seasons. Dissolved and particulate HgT concentrations in the stream slightly increased downstream, while dissolved and particulate MMHg concentrations were low at the pristine sub-watershed outlet (median = 0.006 ng L(-1) and 1.84 ng g(-1), respectively) and sharply increased at the gold mined flat outlet (median = 0.056 ng L(-1) and 6.80 ng g(-1), respectively). Oxisols, which are dominant in the pristine area act as a sink of HgT and MMHg from rain and throughfall inputs. Hydromorphic soils in the flat are strongly contaminated with Hg (including Hg(0) droplets) and their structure has been disturbed by former gold-mining processes, leading to multiple stagnant water areas where biogeochemical conditions are favorable for methylation. In the former gold mined flat high dissolved MMHg concentrations (up to 0.8 ng L(-1)) were measured in puddles or suboxic soil pore waters, whereas high dissolved HgT concentrations were found in lower Eh conditions. Iron-reducing bacteria were suggested as the main methylators since highest concentrations for dissolved MMHg were associated with high dissolved ferrous iron concentrations. The connection between saturated areas and stagnant waters with the hydrographic network during rain events leads to the export of dissolved MMHg and HgT in stream waters, especially at the beginning of the rainy season. As both legal and illegal gold-mining continues to expand in French Guiana, an increase in dissolved and particulate MMHg emissions in the hydrographic network is expected. This will enhance MMHg bio-amplification and present a threat to local populations, whose diet relies mainly on fish.