Bayesian Methods for Regression Using Surrogate Variables

If a dependent variable in a regression analysis is exceptionally expensive or hard to obtain the overall sample size used to fit the model may be limited. To avoid this one may use a cheaper or more easily collected “surrogate” variable to supplement the expensive variable. The regression analysis will be enhanced to the degree the surrogate is associated with the costly dependent variable. We develop a Bayesian approach incorporating surrogate variables in regression based on a two‐stage experiment. Illustrative examples are given, along with comparisons to an existing frequentist method. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

Assessment of veld conditions with multivariate techniques in mixed bushveld,South Africa

The species composition of the herbaceous layer was surveyed in eight plant communities with a small-quadrat method. Ordination techniques were applied to these data to identify relative homogenous grazing areas, assess the veld condition of each unit and to identify indicator species, based on a grazing gradient. Five relative homogenous grazing areas were identified and should be treated as such for management purposes. The veld in the Nylsvlei Nature Reserve was generally in a good condition due to the low stocking densities of game (≈0.29 animals ha^?1) which have been maintained for many years. Nine indicator species, i.e. Cynodon dactylon, Digitaria eriantha, Ischaemum afrum, Leersia hexandra, Oryza longistaminata, Setaria sphacelata, Sporobolus ioclados, Sporobolus pyramidalis, and forbs, were identified which may be used for future monitoring of veld condition in the different vegetation units.     

A quantitative method to assess the ecological indicator system's effectiveness: a case study of the Ecological Province Construction Indicators of China

The ecological indicator system (EIS) is widely used. It is vital to develop a scientific and rational EIS for practical application. Assessment of an EIS's effectiveness is important to understand the gaps between practical indicators and indicators that are considered theoretically ideal. In this paper we develop a conceptual method to quantitatively assess the effectiveness of a practical EIS by comparing it with the ideal EIS that is considered perfect by using three assessment indices: linkage, coverage and representation degrees. Linkage degree reveals the relevance of individual indicators to the ideal system. Coverage degree shows the integrity of the practical indicator system compared with the ideal system. Representation degree is the ultimate goal of the quantitative assessment and can detect the gap and overlap between the practical indicator system and the ideal one. Case study on the Ecological Province Construction Indicators (EPCI) proved that our conceptual method is useful to assess the effectiveness of a practical EIS. Compared with an ideal EIS, the EPCI has a high linkage and coverage degree but biased representation. Although the practical EIS covers most of the objectives detailed in the Fujian eco-province construction documents, many social development indicators are overlooked, including ecological culture and green consumption. Using the quantitative method will improve the practical application of an EIS by detecting its redundancy and weakness.     

Testate amoebae colonizing a newly exposed land surface are of airborne origin

We hypothesized that at the very beginning of terrestrial ecosystem development, airborne testate amoebae play a pivotal role in facilitating organismic colonization and related soil processes. We, therefore, analyzed size and quantity of airborne testate amoebae and immigration and colonization success of airborne testate amoebae on a new land surface (experimental site “Chicken Creek”, artificial post-mining water catchment). Within an altogether 91-day exposure of 70 adhesive traps, 12 species of testate amoebae were identified to be of airborne origin. Phryganella acropodia (51% of all individuals found, diameter about 35–45 μm) and Centropyxis sphagnicola (23% of all individuals found, longest axis about 55–68 μm), occurred most frequently in the adhesive traps. We extrapolated an aerial amoeba deposition of 61 individuals d⁻¹ m⁻² (living and dead individuals combined). Although it would be necessary to have a longer sequence (some additional years), our analysis of the “target substrate” of aerial immigration (catchment site) may point to a shift from a stochastic (variable) beginning of community assembly to a more deterministic (stable) course. This shift was assigned to an age of seven years of initial soil development. Although experienced specialists are necessary to conduct these time-consuming studies, the presented data suggest that terrestrial amoebae are suitable indicators for initial ecosystem development and utilization.     

Modeling the biotic and abiotic factors that describe the number of active off‐host Amblyomma americanum larvae

Amblyomma americanum (L.) is a three‐host tick that spends most of its life off‐host and is an important vector of pathogens in the eastern United States. Our objectives were to develop a predictive statistical model that describes the number of active, off‐host larvae from 2007 to 2011 and to determine the environmental variables associated with this pattern. Data used in this study came from monitoring conducted in northeast Missouri in which off‐host ticks were collected from a permanent plot in a forest and an old field habitat every other week from approximately February to December. Variables examined were day length, degree days, total precipitation prior to sampling, wind speed, saturation deficit, number of adults prior to sampling, and collection site. Of the four regression models tested, the negative binomial model was selected. Fitted candidate models were compared relative to one another using values of eight model selection criteria and model averaging was used to develop a predictive model. The residual plots indicated that the weighted average model performs well in describing the number of larvae. Of the variables considered, the number of larvae was most strongly associated with increasing degree days, the number of active adults prior to sampling, and the forested site.     

Sex ratios and mental health: Evidence from China

While sex ratios (i.e., relative numbers of men and women) have been linked to various economic and social outcomes, how sex ratios affect mental health is underexplored. Using nationally representative data from the China Family Panel studies (CFPS) and Population Census, we evaluate the impact of sex ratios on mental health among Chinese men and explore potential mechanisms. Employing the instrumental variables (IV) approach where the One-Child Policy’s mandated fertility limits and implementation are used as exogenous variations in local sex ratios, we find that higher local sex ratios increase depressive symptoms and probability of depression among Chinese men. The impact is stronger for men with lower levels of education and living in rural areas. Analyses of potential mechanisms show that higher sex ratios increase the likelihood of marriage delay and unemployment for men, and prolong working hours for the employed men. The findings are of direct relevance to the health and population policy in China.     

Phenotyping of Partial Physiological Resistance to Rice Sheath Blight

Sheath blight, caused by Rhizoctonia solani, is one of the most important rice diseases worldwide especially under irrigated agro‐ecosystems. To date, no rice accession with complete resistance to sheath blight has been reported. However, a number of genotypes with varying levels of resistance have been reported. Twelve genotypes (including mega varieties) viz. Tetep, Jasmine 85, Te‐Qing, Duduruchi, Betichikon, Khatochalani, D‐6766, D‐256, Swarna, Sarju‐52, MTU‐1010 and Samba Mashuri were evaluated for quantitative measurement of partial physiological resistance to sheath blight under controlled conditions using detached tiller method. Three independent experiments, each involving three replications, were conducted. Seven days after inoculation, the following disease variables were measured: number of lesions, lesion length, vertical sheath colonization (VSC) on the tiller, disease severity, relative vertical sheath colonization (RVSC) and survival of the leaf blade. Variation between rice genotypes was observed for all the disease variables. Disease severity and VSC were the two most correlated variables, whereas the number of lesions and mean lesion length were the least correlated variables. The ranking of varieties often differed depending on the disease variable considered. Amongst the genotypes tested, D‐256, Tetep and Jasmin‐85 had the lowest number of lesions and disease severity. Similarly, Tetep and D‐256 showed the lowest levels of RVSC, whilst Jasmine‐85 was found to be intermediate. D‐6766, Samba Mashuri and Betichikon showed the highest levels of disease variables. The fraction of dead leaves ranged from 0.00 to 0.38. No dead leaves were observed in Te‐Qing, Swarna and MTU‐1010. The highest fraction of dead leaves was observed for Betichikon (0.38) followed by Duduruchi and D‐6766 (0.33). Our results suggest that this method in combination with other phenotyping methods could be used to quantify partial resistance to rice sheath blight.     

The roles of environment and space in shaping stream diatom communities

Diatoms are widely used in stream quality assessment due to their response to the local environment. Diatoms are also influenced by many large-scale processes and so the diatom communities of boreal streams incorporate a strong spatial component at a regional level. What is not properly known yet is whether the variation in diatom communities between regions is larger than the variation in measured environmental variables. We studied the roles of environment and space in accounting for variability in stream diatom communities across four regions in Finland. According to canonical correspondence analysis, geographical coordinates, nutrient concentrations (total N and P), and water conductivity were the most important factors affecting variation in diatom community composition. Of physical factors, depth and current velocity were also significant. According to Mantel tests, both environmental and geographical distances were related to dissimilarity in diatom community composition. Analysis of Similarities indicated that the regional differences in diatom community composition were larger than the regional differences in environmental variables. We also found many indicator species confined to certain regions. Our results suggest that the four study regions differ in their diatom species composition more than in their environmental features and that diatoms are structured not only by the local environment but also by large-scale processes, possibly related to history, climate and dispersal. These results imply that, while diatom species composition reflects well the environmental differences between regions, future bioassessments would benefit from regional stratification. Otherwise, relationships with environmental variables may be masked by trans-regional differences in species pools caused by the large-scale processes.