Eliciting and encoding expert knowledge on variable selection into classical or Bayesian species distribution models

The quality of species distribution models (SDMs) relies to a large degree on the quality of the input data, from bioclimatic indices to environmental and habitat descriptors (Austin, 2002). Recent reviews of SDM techniques, have sought to optimize predictive performance e.g. Elith et al., 2006. In general SDMs employ one of three approaches to variable selection. The simplest approach relies on the expert to select the variables, as in environmental niche models Nix, 1986 or a generalized linear model without variable selection (Miller and Franklin, 2002). A second approach explicitly incorporates variable selection into model fitting, which allows examination of particular combinations of variables. Examples include generalized linear or additive models with variable selection (Hastie et al. 2002); or classification trees with complexity or model based pruning (Breiman et al., 1984, Zeileis, 2008). A third approach uses model averaging, to summarize the overall contribution of a variable, without considering particular combinations. Examples include neural networks, boosted or bagged regression trees and Maximum Entropy as compared in Elith et al. 2006. Typically, users of SDMs will either consider a small number of variable sets, via the first approach, or else supply all of the candidate variables (often numbering more than a hundred) to the second or third approaches. Bayesian SDMs exist, with several methods for eliciting and encoding priors on model parameters (see review in Low Choy et al. 2010). However few methods have been published for informative variable selection; one example is Bayesian trees (O’Leary 2008). Here we report an elicitation protocol that helps makes explicit a priori expert judgements on the quality of candidate variables. This protocol can be flexibly applied to any of the three approaches to variable selection, described above, Bayesian or otherwise. We demonstrate how this information can be obtained then used to guide variable selection in classical or machine learning SDMs, or to define priors within Bayesian SDMs.

Familial aggregation of type 1 diabetes and diabetic nephropathy in Finland

Type 1 diabetes (T1D) is a common, multifactorial disease with strong familial clustering. In Finland, the incidence of T1D among children aged 14 years or under is the highest in the world. The increase in incidence has been approximately 2.4% per year. Although most new T1D cases are sporadic the first-degree relatives are at an increased risk of developing the same disease. This study was designed to examine the familial aggregation of T1D and one of its serious complications, diabetic nephropathy (DN). More specifically the study aimed (1) to determine the concordance rates of T1D in monozygotic (MZ) and dizygotic (DZ) twins and to estimate the relative contributions of genetic and environmental factors to the variability in liability to T1D as well as to study the age at onset of diabetes in twins; (2) to obtain long-term empirical estimates of the risk of T1D among siblings of T1D patients and the factors related to this risk, especially the effect of age at onset of diabetes in the proband and the birth cohort effect; (3) to establish if DN is aggregating in a Finnish population-based cohort of families with multiple cases of T1D, and to assess its magnitude and particularly to find out whether the risk of DN in siblings is varying according to the severity of DN in the proband and/or the age at onset of T1D: (4) to assess the recurrence risk of T1D in the offspring of a Finnish population-based cohort of patients with childhood onset T1D, and to investigate potential sex-related effects in the transmission of T1D from the diabetic parents to their offspring as well as to study whether there is a temporal trend in the incidence. The study population comprised of the Finnish Young Twin Cohort (22,650 twin pairs), a population-based cohort of patients with T1D diagnosed at the age of 17 years or earlier between 1965 and 1979 (n=5,144) and all their siblings (n=10,168) and offspring (n=5,291). A polygenic, multifactorial liability model was fitted to the twin data. Kaplan-Meier analyses were used to provide the cumulative incidence for the development of T1D and DN. Cox s proportional hazards models were fitted to the data. Poisson regression analysis was used to evaluate temporal trends in incidence. Standardized incidence ratios (SIRs) between the first-degree relatives of T1D patients and background population were determined. The twin study showed that the vast majority of affected MZ twin pairs remained discordant. Pairwise concordance for T1D was 27.3% in MZ and 3.8% in DZ twins. The probandwise concordance estimates were 42.9% and 7.4%, respectively. The model with additive genetic and individual environmental effects was the best-fitting liability model to T1D, with 88% of the phenotypic variance due to genetic factors. The second paper showed that the 50-year cumulative incidence of T1D in the siblings of diabetic probands was 6.9%. A young age at diagnosis in the probands considerably increased the risk. If the proband was diagnosed at the age of 0-4, 5-9, 10-14, 15 or more, the corresponding 40-year cumulative risks were 13.2%, 7.8%, 4.7% and 3.4%. The cumulative incidence increased with increasing birth year. However, SIR among children aged 14 years or under was approximately 12 throughout the follow-up. The third paper showed that diabetic siblings of the probands with nephropathy had a 2.3 times higher risk of DN compared with siblings of probands free of nephropathy. The presence of end stage renal disease (ESRD) in the proband increases the risk three-fold for diabetic siblings. Being diagnosed with diabetes during puberty (10-14) or a few years before (5-9) increased the susceptibility for DN in the siblings. The fourth paper revealed that of the offspring of male probands, 7.8% were affected by the age of 20 compared with 5.3% of the offspring of female probands. Offspring of fathers with T1D have 1.7 times greater risk to be affected with T1D than the offspring of mothers with T1D. The excess risk in the offspring of male fathers manifested itself through the higher risk the younger the father was when diagnosed with T1D. Young age at onset of diabetes in fathers increased the risk of T1D greatly in the offspring, but no such pattern was seen in the offspring of diabetic mothers. The SIR among offspring aged 14 years or under remained fairly constant throughout the follow-up, approximately 10. The present study has provided new knowledge on T1D recurrence risk in the first-degree relatives and the risk factors modifying the risk. Twin data demonstrated high genetic liability for T1D and increased heritability. The vast majority of affected MZ twin pairs, however, remain discordant for T1D. This study confirmed the drastic impact of the young age at onset of diabetes in the probands on the increased risk of T1D in the first-degree relatives. The only exception was the absence of this pattern in the offspring of T1D mothers. Both the sibling and the offspring recurrence risk studies revealed dynamic changes in the cumulative incidence of T1D in the first-degree relatives. SIRs among the first-degree relatives of T1D patients seems to remain fairly constant. The study demonstrates that the penetrance of the susceptibility genes for T1D may be low, although strongly influenced by the environmental factors. Presence of familial aggregation of DN was confirmed for the first time in a population-based study. Although the majority of the sibling pairs with T1D were discordant for DN, its presence in one sibling doubles and presence of ESRD triples the risk of DN in the other diabetic sibling. An encouraging observation was that although the proportion of children to be diagnosed with T1D at the age of 4 or under is increasing, they seem to have a decreased risk of DN or at least delayed onset.

Statistical analysis of the associations of hereditary factors with the risk of Type 1 diabetes and Diabetic Nephropathy based on the familial data collected through ascertaiment from population-based registers

In genetic epidemiology, population-based disease registries are commonly used to collect genotype or other risk factor information concerning affected subjects and their relatives. This work presents two new approaches for the statistical inference of ascertained data: a conditional and full likelihood approaches for the disease with variable age at onset phenotype using familial data obtained from population-based registry of incident cases. The aim is to obtain statistically reliable estimates of the general population parameters. The statistical analysis of familial data with variable age at onset becomes more complicated when some of the study subjects are non-susceptible, that is to say these subjects never get the disease. A statistical model for a variable age at onset with long-term survivors is proposed for studies of familial aggregation, using latent variable approach, as well as for prospective studies of genetic association studies with candidate genes. In addition, we explore the possibility of a genetic explanation of the observed increase in the incidence of Type 1 diabetes (T1D) in Finland in recent decades and the hypothesis of non-Mendelian transmission of T1D associated genes. Both classical and Bayesian statistical inference were used in the modelling and estimation. Despite the fact that this work contains five studies with different statistical models, they all concern data obtained from nationwide registries of T1D and genetics of T1D. In the analyses of T1D data, non-Mendelian transmission of T1D susceptibility alleles was not observed. In addition, non-Mendelian transmission of T1D susceptibility genes did not make a plausible explanation for the increase in T1D incidence in Finland. Instead, the Human Leucocyte Antigen associations with T1D were confirmed in the population-based analysis, which combines T1D registry information, reference sample of healthy subjects and birth cohort information of the Finnish population. Finally, a substantial familial variation in the susceptibility of T1D nephropathy was observed. The presented studies show the benefits of sophisticated statistical modelling to explore risk factors for complex diseases.

Carbon grazing: Managing carbon flows for resilient pasture in variable climate

Work with consultants to help graziers understand implications of carbon in their pastures.

A four-variable model for the pattern-forming mechanism in Hydra

A generalized Gierer-Meinhardt model has been used to account for the transplantation experiments in Hydra. In this model, a cross inhibition between the two organizing centres (namely, head and foot) are assumed to be the only mode of interaction in setting up a stable morphogen distribution for the pattern formation in Hydra.

Fluorescent alamethicin fragments A study of membrane activity and aqueous phase aggregation

The linear polypeptide antibiotic alamethicin is known to form channels in artificial lipid membranes. Synthetic 13- and 17-residue alamethicin fragments, labelled with a fluorescent dansyl group at the N-terminus, have been shown to translocate divalent cations across phospholipid membranes and to uncouple oxidative phosphorylation in rat liver mitochondria, in a manner analogous to the parent peptides. From studies of the aqueous phase aggregation behavior of the peptides, as well as their interaction with rat liver mitochondria, it is concluded that the interaction of the peptides with membranes is a complex process, probably involving both aqueous and membrane phase aggregation.

Sustainable management for rangelands in a variable climate: evidence and insights from northern Australia

Inter-annual rainfall variability is a major challenge to sustainable and productive grazing management on rangelands. In Australia, rainfall variability is particularly pronounced and failure to manage appropriately leads to major economic loss and environmental degradation. Recommended strategies to manage sustainably include stocking at long-term carrying capacity (LTCC) or varying stock numbers with forage availability. These strategies are conceptually simple but difficult to implement, given the scale and spatial heterogeneity of grazing properties and the uncertainty of the climate. This paper presents learnings and insights from northern Australia gained from research and modelling on managing for rainfall variability. A method to objectively estimate LTCC in large, heterogeneous paddocks is discussed, and guidelines and tools to tactically adjust stocking rates are presented. The possible use of seasonal climate forecasts (SCF) in management is also considered. Results from a 13-year grazing trial in Queensland show that constant stocking at LTCC was far more profitable and largely maintained land condition compared with heavy stocking (HSR). Variable stocking (VAR) with or without the use of SCF was marginally more profitable, but income variability was greater and land condition poorer than constant stocking at LTCC. Two commercial scale trials in the Northern Territory with breeder cows highlighted the practical difficulties of variable stocking and provided evidence that heavier pasture utilisation rates depress reproductive performance. Simulation modelling across a range of regions in northern Australia also showed a decline in resource condition and profitability under heavy stocking rates. Modelling further suggested that the relative value of variable v. constant stocking depends on stocking rate and land condition. Importantly, variable stocking may possibly allow slightly higher stocking rates without pasture degradation. Enterprise-level simulations run for breeder herds nevertheless show that poor economic performance can occur under constant stocking and even under variable stocking in some circumstances. Modelling and research results both suggest that a form of constrained flexible stocking should be applied to manage for climate variability. Active adaptive management and research will be required as future climate changes make managing for rainfall variability increasingly challenging.

X-Ray Studies On Crystalline Complexes Involving Amino-Acids And Peptides .11. Peptide Aggregation And Water-Structure In The Crystals Of A Hydrated 1-1 Complex Between L-Histidyl-L-Serine And Glycyl-L-Glutamic Acid

The hexahydrate of a 1:1 complex between L-histidyl-L-serine and glycyl-L-glutamic acid crystallizes in space group P1 with a = 4.706(1), b= 8.578(2), c= 16.521(3) ÅA; α= 85.9(1), β= 89.7(1)°, = 77.4(1). The crystal structure, solved by direct methods, has been refined to an R value of 0.046 for 2150 observed reflections. The two peptide molecules in the structure have somewhat extended conformations. The unlike molecules aggregate into separate alternating layers. Each layer is stabilized by hydrogen bonded head-to-tail sequences as well as sequences of hydrogen bonds involving peptide groups. The arrangement of molecules in each layer is similar to one of the plausible idealized arrangements of L-alanyl-L-alanine worked out from simple geometrical considerations. Adjacent layers in the structure are held together by interactions involving side chains as well as water molecules. The water structure observed in the complex provides a good model, at atomic resolution, for that in protein crystals. An interesting feature of the crystal structure is the existence of two water channels in the interfaces between adjacent peptide layers.

Design of single variable controllers for ammonia reactors

Abstract is not available.

Fluorescent probe and NMR studies of the aggregation of bile salts in aqueous solution

The binding of the fluorescent probes 1-anilino-8-naphthalene sulfonate and dansyl cadaverine to the sodium salts of cholic, deoxycholic and dehydrocholic acids has been investigated. Enhanced probe solubilisation accompanies aggregation. Monitoring of fluorescence intensities as a function of bile salt concentration permits the detection of primary micelle formation, as well as secondary association. The transition concentrations obtained by fluorescence are in good agreement with values determined for the critical micelle concentrations, by other methods. Differences in the behaviour of cholate and deoxycholate have been noted. Fluorescence polarisation studies of 1,6-diphenyl-1,3,5-hexatriene solubilised in bile salt micelles suggest a higher microviscosity for the interior of the deoxycholate micelle as compared to cholate. 1H NMR studies of deoxycholate over the range 1–100 mg/ml suggest that micelle formation leads to a greater immobilisation of the C18 and C19 methyl groups as compared to the C21 methyl group. Well resolved 13C resonances are observed for all three steroids even at high concentration. Both fluorescence and NMR studies confirm that dehydrocholate does not aggregate.

A new family of multivariate heavy-tailed distributions with variable marginal amounts of tailweight: Application to robust clustering

We propose a family of multivariate heavy-tailed distributions that allow variable marginal amounts of tailweight. The originality comes from introducing multidimensional instead of univariate scale variables for the mixture of scaled Gaussian family of distributions. In contrast to most existing approaches, the derived distributions can account for a variety of shapes and have a simple tractable form with a closed-form probability density function whatever the dimension. We examine a number of properties of these distributions and illustrate them in the particular case of Pearson type VII and t tails. For these latter cases, we provide maximum likelihood estimation of the parameters and illustrate their modelling flexibility on simulated and real data clustering examples.

Variable-number tandem repeats genotyping used to aid and inform management strategies for a bovine Johne's disease incursion in tropical and subtropical Australia

The application of variable-number tandem repeats (VNTR) genotyping of Mycobacterium avium subsp. paratuberculosis isolates to assist in investigating incidents of bovine Johne’s disease in a low-prevalence region of Australia is described in the current study. Isolates from a response to detection of bovine Johne’s disease in Queensland were compared with strains from national and international sources. The tandem application of mycobacterial interspersed repetitive unit (MIRU) and multilocus short sequence repeats (MLSSR) genotyping identified 2 strains, 1 that infected cattle on multiple properties with trace-forward histories from a common infected property, and 1 genotypically different strain recovered from a single property. The former strain showed an identical genotype to an isolate from India. Neither strain showed a genotypic link to regions of Australia with a higher prevalence of the disease. Genotyping has indicated incursions from 2 independent sources. This intelligence has informed investigations into potential routes of entry and the soundness of ongoing control measures, and supported strategy and policy decisions regarding management of Mycobacterium avium subsp. paratuberculosis incursions for Queensland.

Inelastic Postbuckling of Columns of Variable Flexural Rigidity

Effects of large deformation and inelasticity are considered in formulating the behavior of columns of variable cross section subjected to an axial compressive load. Simple, approximate methods are used to obtain numerical results. The combined effect of the nonlinearities is shown to be of a hardening type for small column deflections