Handling covariates subject to limits of detection in regression

In the environmental health sciences, measurements of toxic exposures are often constrained by a lower limit called the limit of detection (LOD), with observations below this limit called non-detects. Although valid inference may be obtained by excluding non-detects in the estimation of exposure effects, this practice can lead to substantial reduction in power to detect a significant effect, depending on the proportion of censoring and the closeness of the effect size to the null value. Therefore, a variety of methods have been commonly used in the environmental science literature to substitute values for the non-detects for the purpose of estimating exposure effects, including ad hoc values such as ${LOD/2, LOD/\sqrt{2}}$ and LOD. Another method substitutes the expected value of the non-detects, i.e., E[X|X ≤?LOD] but this requires that the inference be robust to mild miss-specifications in the distribution of the exposure variable. In this paper, we demonstrate that the estimate of the exposure effect is extremely sensitive to ad-hoc substitutions and moderate distribution miss-specifications under the conditions of large sample sizes and moderate effect size, potentially leading to biased estimates. We propose instead the use of the generalized gamma distribution to estimate imputed values for the non-detects, and show that this method avoids the risk of distribution miss-specification among the class of distributions represented by the generalized gamma distribution. A multiple imputation-based procedure is employed to estimate the regression parameters. Compared to the method of excluding non-detects, the proposed method can substantially increase the power to detect a significant effect when the effect size is close to the null value in small samples with moderate levels of censoring (?≤ 50%), without compromising the coverage and relative bias of the estimates.     

Population structure of the planktonic copepod Calanus pacificus in the North Pacific Ocean

The pelagic copepod Calanus pacificus ranges nearly continuously across temperate-boreal regions of the North Pacific Ocean and is currently divided into three subspecies—C. pacificus oceanicus, C. pacificus californicus, C. pacificus pacificus—based on subtle morphological differences and geographic location. The relation between geography and genetic differentiation was examined for 398 C. pacificus individuals sampled from six widely distributed locations across the North Pacific, including an open ocean site and coastal sites on both sides of the North Pacific basin. For each individual copepod, the DNA sequence was determined for a 421-bp region of the mitochondrial coxI gene (mtCOI). A total of sixty-three different mtCOI sequences, or haplotypes, were detected, with a sequence divergence between haplotypes of 0.2–3.1%. The number and distribution of haplotypes varied with sampling location; 12 haplotypes were distributed across multiple sampling locations, and 51 occurred at only one location. Five genetically distinct populations were detected based on F _ST values. Haplotype minimum spanning networks, nucleotide divergence and F _ST values indicated that individuals from coastal sites in the North Pacific Ocean were more closely related to each other than to individuals from the open ocean site at Station P. These results provide genetic support for the designation of two subspecies—a coastal subspecies that consists of what is currently referred to as C. p. pacificus and C. p. californicus and an open ocean subspecies C. p. oceanicus. This work also indicates that planktonic copepods with potentially high dispersal capacity can develop genetically structured populations in the absence of obvious geographic barriers between proximate locales within an ocean basin.     

Variation along the shy–bold continuum in extremophile fishes (<Emphasis Type="BoldItalic">Poecilia mexicana</Emphasis>, <Emphasis Type="BoldItalic">Poecilia sulphuraria</Emphasis>)

One potential trade-off that bold individuals face is between increased predation risks and gains in resources. Individuals experiencing high predation and hungry individuals (or individuals with low body condition) are predicted to show increased boldness. We examined one behavioral trait previously reported to be associated with boldness (the time individual fish needed to emerge from shelter) in various populations of mollies (Poecilia spp.). Our study system included several southern Mexican surface streams with high piscine predation and high food availability, sulfidic surface streams with high avian predation, in which the inhabiting fish show reduced body condition, and a sulfidic cave, where predation and body condition are low. Our comparison revealed very short times to emerge from the start box in populations from non-sulfidic streams. In sulfidic habitats (whether surface or cave), it took individual Poecilia mexicana considerably longer to emerge from the start box, and the same difference was also found in an independent comparison between P. mexicana and the closely related, highly sulfide-adapted Poecilia sulphuraria. Fish reared under common garden conditions (in the absence of predators and hydrogen sulfide) showed intermediate boldness scores to the extremes observed in the field. Our data thus indicate that (a) boldness is shaped by environmental conditions/experiential effects, but is not heritable, (b) predation affects boldness in the predicted direction, but (c) low body condition leads to reduced boldness. Extremophile Poecilia spp. spend most of their time surfacing to survive under sulfidic and hypoxic conditions, which exposes them to increased levels of predations, but the fish forage on the bottom. Hence, in this system, increased boldness does not increase foraging success. We argue that energy limitation favors reducing energetically costly behaviors, and exploring novel environments may be just one of them.     

Identifying hydrologically sensitive areas: bridging the gap between science and application

Researchers have noted that current water quality protection strategies, like nutrient management plans, lack a sound hydrological underpinning for pollutant transport processes. This is especially true for areas like the northeastern U.S. where copious research has shown that variable source area hydrology largely governs runoff generation. The goal of this study was to develop a scientifically justified method to identify the locations that generate overland flow. Furthermore, this methodology must be computationally simple enough that it can be utilized or incorporated into nutrient management plans and other established water quality tools. We specifically tested the reliability of the 'distance from a stream,'D(s), and the 'topographic index,'lambda, to predict areas with a high propensity for generating overland flow, i.e. hydrologically sensitive areas (HSA). HSAs were defined by their probability of generating runoff, P(sat), based on 30 year simulations using a physically based hydrological model. Using GIS, each location's P(sat) was correlated with D(s) and lambda. We used three Delaware Co., NY watersheds in the New York City watershed system with areas varying in size from 1.6 to 37 km2 and with forested and agricultural land uses. The topographic index gave stronger, more regionally consistent correlations with P(sat) than did D(s). Equations correlating lambda and P(sat) for each month are presented and can be used to estimate hydrological sensitivity in the region surrounding our study watersheds, i.e. in Delaware Co. This work is currently being incorporated into an Internet Mapping System to facilitate user-friendly, on-line identification of HSAs.     

Organochlorine pesticides in feathers of three raptor species in southern Brazil

Environmental Science and Pollution Research - In this study, we determined the presence of organochlorine pesticides (OCPs) in back feathers from three raptor species, Phalcoboenus chimango,...     

Concentrations of particulate matter emitted from large cattle feedlots in Kansas

Particulate matter (PM) emitted from cattle feedlots are thought to affect air quality in rural communities, yet little is known about factors controlling their emissions. The concentrations of PM (i.e., PM2.5, PM10, and total suspended particulates or TSP) upwind and downwind at two large cattle feedlots (KS1, KS2) in Kansas were measured with gravimetric samplers from May 2006 to October 2009 (at KS1) and from September 2007 to April 2008 (at KS2). The mean downwind and net (i.e., downwind - upwind) mass concentrations of PM2.5, PM10, and TSP varied seasonally, indicating the need for multiple-day, seasonal sampling. The downwind and net concentrations were closely related to the moisture content of the pen surface. The PM2.5/PM10 and PM2.5/TSP ratios at the downwind sampling location were also related to the moisture content of the pen surface, humidity, and temperature. Measurement of the particle size distribution downwind of the feedlot with a cascade impactor showed geometric mean diameter ranging from 7 to 18 microm, indicating that particles that were emitted from the feedlots were generally large in size.     

Determination of vapor pressure-temperature relationships of current–use pesticides and transformation products

Sub-cooled liquid vapor pressures (P_L ⁰) of current–use organochlorine and organophosphate pesticides (chlorothalonil, chlorpyrifos methyl, diazinon, fipronil) and selected transformation products (chlorpyrifos oxon, heptachlor epoxide, oxychlordane, 3,5,6-trichloro-2-pyridinol) were determined at multiple temperatures using the gas chromatography retention time technique. Results were utilized to determine vapor pressure-temperature relationships and to calculate enthalpies of vaporization (ΔH_vap). While results for chlorothalonil and diazinon were comparable with published values, the measured value for fipronil (1.82 × 10^{? 6} Pa) is almost an order of magnitude higher than the reported literature value (3.7 × 10^{? 7} Pa). The availability of vapor pressure temperature relationships for these chemicals will aid in pesticide risk assessment development and improve the effectiveness of mitigation and remediation efforts.     

Towards the selection of a produced water enrichment for biological gas hydrate inhibitors

Economic concerns associated with the recovery of non-conventional hydrocarbon reserves include unexpected ice as well as ice-like gas hydrate formation. Antifreeze proteins (AFPs) inhibit ice growth, and experiments with fish, plant, and insect AFPs have shown promise of effective gas hydrate inhibition in lab-scale experiments. If produced on an industrial scale, AFPs could provide a more environmentally friendly alternative to kinetic inhibitors, but a large-scale production of these AFPs is not currently feasible. We believe that these difficulties could be surmounted by the production of microbial AFPs, but to date, only a few such proteins have been identified and purified, and none of these are associated with hydrocarbon reserves. Here, we have used ice-affinity and freeze-thaw stress to select microbes derived from oil and gas formation water, or produced water, as a source of anaerobic microbial communities. Ice-affinity successfully incorporated anaerobic bacteria under aerobic conditions, and the mixed culture had ice-associating properties. Under these conditions, ice-affinity selection does not result in cultivatable isolates, but similar, cultivable microbes were obtained following freeze-thaw selection under anaerobic conditions. Since these mixed cultures inhibited the growth of ice crystals, they also have the potential to inhibit hydrate growth. Overall, freeze-thaw selection provides a promising first step towards the isolation of microbes capable of the inhibition of ice and gas hydrate growth, for possible application for energy exploration and recovery at high-latitudes and in-deep, cold waters.