Kerrie Mengersen

Eliciting Expert Knowledge in Conservation Science

Expert knowledge is used widely in the science and practice of conservation because of the complexity of problems, relative lack of data, and the imminent nature of many conservation decisions. Expert knowledge is substantive information on a particular topic that is not widely known by others. An expert is someone who holds this knowledge and who is often deferred to in its interpretation. We refer to predictions by experts of what may happen in a particular context as expert judgments. In general, an expert-elicitation approach consists of five steps: deciding how information will be used, determining what to elicit, designing the elicitation process, performing the elicitation, and translating the elicited information into quantitative statements that can be used in a model or directly to make decisions. This last step is known as encoding. Some of the considerations in eliciting expert knowledge include determining how to work with multiple experts and how to combine multiple judgments, minimizing bias in the elicited information, and verifying the accuracy of expert information. We highlight structured elicitation techniques that, if adopted, will improve the accuracy and information content of expert judgment and ensure uncertainty is captured accurately. We suggest four aspects of an expert elicitation exercise be examined to determine its comprehensiveness and effectiveness: study design and context, elicitation design, elicitation method, and elicitation output. Just as the reliability of empirical data depends on the rigor with which it was acquired so too does that of expert knowledge.

Bayesian modelling and inference on mixtures of distributions

Publisher Summary Mixture distributions comprise a finite or infinite number of components, possibly of different distributional types, that can describe different features of data. The Bayesian paradigm allows for probability statements to be made directly about the unknown parameters, prior or expert opinion to be included in the analysis, and hierarchical descriptions of both local-scale and global features of the model. This chapter aims to introduce the prior modeling, estimation, and evaluation of mixture distributions in a Bayesian paradigm. The chapter shows that mixture distributions provide a flexible, parametric framework for statistical modeling and analysis. Focus is on the methods rather than advanced examples, in the hope that an understanding of the practical aspects of such modeling can be carried into many disciplines. It also points out the fundamental difficulty in doing inference with such objects, along with a discussion about prior modeling, which is more restrictive than usual, and the constructions of estimators, which also is more involved than the standard posterior mean solution. Finally, this chapter gives some pointers to the related models and problems like mixtures of regressions and hidden Markov models as well as Dirichlet priors.

The power of expert opinion in ecological models using Bayesian methods: impact of grazing on birds

One of our greatest challenges as researchers is predicting impacts of land use on biota, and predicting the impact of livestock grazing on birds is no exception. Insufficient data and poor survey design often yield results that are not statistically sig- nificant or that are difficult to interpret because researchers cannot disentangle the effects of grazing from other disturbances. This has resulted in few publications on the impact of grazing on birds alone. Ecologists with extensive experience in bird ecology in grazed landscapes could inform an analysis when time and monetary constraints limit the amount of data that can be collected. Using responses from 20 well-recognized ecologists throughout Australia, we captured this expert knowledge and incorporated it into a statistical model using Bayesian methods. Although relatively new to ecology, Bayesian methods allow straightforward probability statements to be made about specific models or scenarios and the integration of different types of information, including scientific judgment, while formally accom- modating and incorporating the uncertainty in the information provided. Data on bird density were collected across three broad levels of grazing (no/low, mod- erate, and high) typical of subtropical Australia. These field data were used in conjunction with expert data to produce estimates of species persistence under grazing. The addition of expert data through priors in our model strengthened results under at least one grazing level for all but one bird species examined. When experts were in agreement credible intervals were tightened substantially, whereas, when experts were in disagreement, results were similar to those evaluated in the absence of expert information. In fields where there is extensive expert knowledge, yet little published data, the use of expert information as priors for ecological models is a cost-effective way of making more confident predictions about the effect of management on biodiversity.

Characterization of expiration air jets and droplet size distributions immediately at the mouth opening

Abstract Size distributions of expiratory droplets expelled during coughing and speaking and the velocities of the expiration air jets of healthy volunteers were measured. Droplet size was measured using the interferometric Mie imaging (IMI) technique while the particle image velocimetry (PIV) technique was used for measuring air velocity. These techniques allowed measurements in close proximity to the mouth and avoided air sampling losses. The average expiration air velocity was 11.7m/s for coughing and 3.9m/s for speaking. Under the experimental setting, evaporation and condensation effects had negligible impact on the measured droplet size. The geometric mean diameter of droplets from coughing was 13.5μm and it was 16.0μm for speaking (counting 1–100). The estimated total number of droplets expelled ranged from 947 to 2085 per cough and 112–6720 for speaking. The estimated droplet concentrations for coughing ranged from 2.4 to 5.2cm−3 per cough and 0.004–0.223cm−3 for speaking.

Differences in airborne particle and gaseous concentrations in urban air between weekdays and weekends

Airborne particle number concentrations and size distributions as well as CO and NOx concentrations monitored at a site within the central business district of Brisbane, Australia were correlated with the traffic flow rate on a nearby freeway with the aim of investigating differences between weekday and weekend pollutant characteristics. Observations over a 5-year monitoring period showed that the mean number particle concentration on weekdays was (8.8±0.1)×103 cm−3 and on weekends (5.9±0.2)×103 cm−3—a difference of 47%. The corresponding mean particle number median diameters during weekdays and weekends were 44.2±0.3 and 50.2±0.2 nm, respectively. The differences in mean particle number concentration and size between weekdays and weekends were found to be statistically significant at confidence levels of over 99%. During a 1-year period of observation, the mean traffic flow rate on the freeway was 14.2×104 and 9.6×104 vehicles per weekday and weekend day, respectively—a difference of 48%. The mean diurnal variations of the particle number and the gaseous concentrations closely followed the traffic flow rate on both weekdays and weekends (correlation coefficient of 0.86 for particles). The overall conclusion, as to the effect of traffic on concentration levels of pollutant concentration in the vicinity of a major road (about 100 m) carrying traffic of the order of 105 vehicles per day, is that about a 50% increase in traffic flow rate results in similar increases of CO and NOx concentrations and a higher increase of about 70% in particle number concentration.

Climate change and dengue: a critical and systematic review of quantitative modelling approaches

BackgroundMany studies have found associations between climatic conditions and dengue transmission. However, there is a debate about the future impacts of climate change on dengue transmission. This paper reviewed epidemiological evidence on the relationship between climate and dengue with a focus on quantitative methods for assessing the potential impacts of climate change on global dengue transmission.MethodsA literature search was conducted in October 2012, using the electronic databases PubMed, Scopus, ScienceDirect, ProQuest, and Web of Science. The search focused on peer-reviewed journal articles published in English from January 1991 through October 2012.ResultsSixteen studies met the inclusion criteria and most studies showed that the transmission of dengue is highly sensitive to climatic conditions, especially temperature, rainfall and relative humidity. Studies on the potential impacts of climate change on dengue indicate increased climatic suitability for transmission and an expansion of the geographic regions at risk during this century. A variety of quantitative modelling approaches were used in the studies. Several key methodological issues and current knowledge gaps were identified through this review.ConclusionsIt is important to assemble spatio-temporal patterns of dengue transmission compatible with long-term data on climate and other socio-ecological changes and this would advance projections of dengue risks associated with climate change.

Quantifying killing of Orangutans and Human-Orangutan conflict in Kalimantan, Indonesia

Human-orangutan conflict and hunting are thought to pose a serious threat to orangutan existence in Kalimantan, the Indonesian part of Borneo. No data existed prior to the present study to substantiate these threats. We investigated the rates, spatial distribution and causes of conflict and hunting through an interview-based survey in the orangutans range in Kalimantan, Indonesia. Between April 2008 and September 2009, we interviewed 6983 respondents in 687 villages to obtain socio-economic information, assess knowledge of local wildlife in general and orangutan encounters specifically, and to query respondents about their knowledge on orangutan conflicts and killing, and relevant laws. This survey revealed estimated killing rates of between 750 and 1800 animals killed in the last year, and between 1950 and 3100 animals killed per year on average within the lifetime of the survey respondents. These killing rates are higher than previously thought and are high enough to pose a serious threat to the continued existence of orangutans in Kalimantan. Importantly, the study contributes to our understanding of the spatial variation in threats, and the underlying causes of those threats, which can be used to facilitate the development of targeted conservation management.

Classification based upon gene expression data

MOTIVATION Gene expression data offer a large number of potentially useful predictors for the classification of tissue samples into classes, such as diseased and non-diseased. The predictive error rate of classifiers can be estimated using methods such as cross-validation. We have investigated issues of interpretation and potential bias in the reporting of error rate estimates. The issues considered here are optimization and selection biases, sampling effects, measures of misclassification rate, baseline error rates, two-level external cross-validation and a novel proposal for detection of bias using the permutation mean. RESULTS Reporting an optimal estimated error rate incurs an optimization bias. Downward bias of 3-5% was found in an existing study of classification based on gene expression data and may be endemic in similar studies. Using a simulated non-informative dataset and two example datasets from existing studies, we show how bias can be detected through the use of label permutations and avoided using two-level external cross-validation. Some studies avoid optimization bias by using single-level cross-validation and a test set, but error rates can be more accurately estimated via two-level cross-validation. In addition to estimating the simple overall error rate, we recommend reporting class error rates plus where possible the conditional risk incorporating prior class probabilities and a misclassification cost matrix. We also describe baseline error rates derived from three trivial classifiers which ignore the predictors. AVAILABILITY R code which implements two-level external cross-validation with the PAMR package, experiment code, dataset details and additional figures are freely available for non-commercial use from http://www.maths.qut.edu.au/profiles/wood/permr.jsp

Unmanned Aerial Vehicles (UAVs) and Artificial Intelligence Revolutionizing Wildlife Monitoring and Conservation

Surveying threatened and invasive species to obtain accurate population estimates is an important but challenging task that requires a considerable investment in time and resources. Estimates using existing ground-based monitoring techniques, such as camera traps and surveys performed on foot, are known to be resource intensive, potentially inaccurate and imprecise, and difficult to validate. Recent developments in unmanned aerial vehicles (UAV), artificial intelligence and miniaturized thermal imaging systems represent a new opportunity for wildlife experts to inexpensively survey relatively large areas. The system presented in this paper includes thermal image acquisition as well as a video processing pipeline to perform object detection, classification and tracking of wildlife in forest or open areas. The system is tested on thermal video data from ground based and test flight footage, and is found to be able to detect all the target wildlife located in the surveyed area. The system is flexible in that the user can readily define the types of objects to classify and the object characteristics that should be considered during classification.

Species Richness on Coral Reefs and the Pursuit of Convergent Global Estimates

Global species richness, whether estimated by taxon, habitat, or ecosystem, is a key biodiversity metric. Yet, despite the global importance of biodiversity and increasing threats to it (e.g., we are no better able to estimate global species richness now than we were six decades ago. Estimates of global species richness remain highly uncertain and are often logically inconsistent. They are also difficult to validate because estimation of global species richness requires extrapolation beyond the number of species known. Given that somewhere between 3% and >96% of species on Earth may remain undiscovered, depending on the methods used and the taxa considered, such extrapolations, especially from small percentages of known species, are likely to be highly uncertain. An alternative approach is to estimate all species, the known and unknown, directly. Using expert taxonomic knowledge of the species already described and named, those already discovered but not yet described and named, and those still awaiting discovery, we estimate there to be 830,000 (95% credible limits: 550,000-1,330,000) multi-cellular species on coral reefs worldwide, excluding fungi. Uncertainty surrounding this estimate and its components were often strongly skewed toward larger values, indicating that many more species on coral reefs is more plausible than many fewer. The uncertainties revealed here should guide future research toward achieving convergence in global species richness estimates for coral reefs and other ecosystems via adaptive learning protocols whereby such estimates can be tested and improved, and their uncertainties reduced, as new knowledge is acquired.