Dataset Information

Quantifying the degree of bias from using county-scale data in species distribution modeling: Can increasing sample size or using county-averaged environmental data reduce distributional overprediction?

ABSTRACT: Citizen-science databases have been used to develop species distribution models (SDMs), although many taxa may be only georeferenced to county. It is tacitly assumed that SDMs built from county-scale data should be less precise than those built with more accurate localities, but the extent of the bias is currently unknown. Our aims in this study were to illustrate the effects of using county-scale data on the spatial extent and accuracy of SDMs relative to true locality data and to compare potential compensatory methods (including increased sample size and using overall county environmental averages rather than point locality environmental data). To do so, we developed SDMs in maxent with PRISM-derived BIOCLIM parameters for 283 and 230 species of odonates (dragonflies and damselflies) and butterflies, respectively, for five subsets from the OdonataCentral and Butterflies and Moths of North America citizen-science databases: (1) a true locality dataset, (2) a corresponding sister dataset of county-centroid coordinates, (3) a dataset where the average environmental conditions within each county were assigned to each record, (4) a 50/50% mix of true localities and county-centroid coordinates, and (5) a 50/50% mix of true localities and records assigned the average environmental conditions within each county. These mixtures allowed us to quantify the degree of bias from county-scale data. Models developed with county centroids overpredicted the extent of suitable habitat by 15% on average compared to true locality models, although larger sample sizes (>100 locality records) reduced this disparity. Assigning county-averaged environmental conditions did not offer consistent improvement, however. Because county-level data are of limited value for developing SDMs except for species that are widespread and well collected or that inhabit regions where small, climatically uniform counties predominate, three means of encouraging more accurate georeferencing in citizen-science databases are provided.

SUBMITTER: Collins SD

PROVIDER: S-EPMC5551104 | biostudies-literature | 2017 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Quantifying the degree of bias from using county-scale data in species distribution modeling: Can increasing sample size or using county-averaged environmental data reduce distributional overprediction?

Collins Steven D SD Abbott John C JC McIntyre Nancy E NE

Ecology and evolution 20170628 15

Citizen-science databases have been used to develop species distribution models (SDMs), although many taxa may be only georeferenced to county. It is tacitly assumed that SDMs built from county-scale data should be less precise than those built with more accurate localities, but the extent of the bias is currently unknown. Our aims in this study were to illustrate the effects of using county-scale data on the spatial extent and accuracy of SDMs relative to true locality data and to compare poten ...[more]

PMID: 28808561

Similar Datasets

Project description:Importance:Analyses of female representation in clinical studies have been limited in scope and scale. Objective:To perform a large-scale analysis of global enrollment sex bias in clinical studies. Design, Setting, and Participants:In this cross-sectional study, clinical studies from published articles from PubMed from 1966 to 2018 and records from Aggregate Analysis of ClinicalTrials.gov from 1999 to 2018 were identified. Global disease prevalence was determined for male and female patients in 11 disease categories from the Global Burden of Disease database: cardiovascular, diabetes, digestive, hepatitis (types A, B, C, and E), HIV/AIDS, kidney (chronic), mental, musculoskeletal, neoplasms, neurological, and respiratory (chronic). Machine reading algorithms were developed that extracted sex data from tables in articles and records on December 31, 2018, at an artificial intelligence research institute. Male and female participants in 43 135 articles (792 004 915 participants) and 13 165 records (12 977 103 participants) were included. Main Outcomes and Measures:Sex bias was defined as the difference between the fraction of female participants in study participants minus prevalence fraction of female participants for each disease category. A total of 1000 bootstrap estimates of sex bias were computed by resampling individual studies with replacement. Sex bias was reported as mean and 95% bootstrap confidence intervals from articles and records in each disease category over time (before or during 1993 to 2018), with studies or participants as the measurement unit. Results:There were 792 004 915 participants, including 390 470 834 female participants (49%), in articles and 12 977 103 participants, including 6 351 619 female participants (49%), in records. With studies as measurement unit, substantial female underrepresentation (sex bias ≤ -0.05) was observed in 7 of 11 disease categories, especially HIV/AIDS (mean for articles, -0.17 [95% CI, -0.18 to -0.16]), chronic kidney diseases (mean, -0.17 [95% CI, -0.17 to -0.16]), and cardiovascular diseases (mean, -0.14 [95% CI, -0.14 to -0.13]). Sex bias in articles for all categories combined was unchanged over time with studies as measurement unit (range, -0.15 [95% CI, -0.16 to -0.13] to -0.10 [95% CI, -0.14 to -0.06]), but improved from before or during 1993 (mean, -0.11 [95% CI, -0.16 to -0.05]) to 2014 to 2018 (mean, -0.05 [95% CI, -0.09 to -0.02]) with participants as the measurement unit. Larger study size was associated with greater female representation. Conclusions and Relevance:Automated extraction of the number of participants in clinical reports provides an effective alternative to manual analysis of demographic bias. Despite legal and policy initiatives to increase female representation, sex bias against female participants in clinical studies persists. Studies with more participants have greater female representation. Differences between sex bias estimates with studies vs participants as measurement unit, and between articles vs records, suggest that sex bias with both measures and data sources should be reported.

Project description:UNLABELLED:With the advent of deep sequencing, genomic surveillance has become a popular method for detection of infectious disease, supplementing information gathered by classic clinical or serological techniques to identify host-determinant markers and trace the origin of transmission. However, two main factors complicate genomic surveillance. First, pathogens exhibiting high genetic diversity demand higher levels of scrutiny to obtain an accurate representation of the entire population. Second, current systems of detection are nonuniform, with significant gaps in certain geographic locations and animal reservoirs. Despite past unforeseen pandemics like the 2009 swine-origin H1N1 influenza virus, there is no standardized way of evaluating surveillance. A more complete surveillance system should capture a greater proportion of pathogen diversity. Here we present a novel quantitative method of assessing the completeness of genomic surveillance that incorporates the time of sequence collection, as well as the pathogen's evolutionary rate. We propose the q2 coefficient, which measures the proportion of sequenced isolates whose closest neighbor in the past is within a genetic distance equivalent to 2 years of evolution, roughly the median time of changing strain selection for influenza A vaccines. Easily interpretable and significantly faster than other methods, the q2 coefficient requires no full phylogenetic characterization or use of arbitrary clade definitions. Application of the q2 coefficient to influenza A virus confirmed poor sampling of swine and avian populations and identified regions with deficient surveillance. We demonstrate that the q2 coefficient can not only be applied to other pathogens, including dengue and West Nile viruses, but also used to describe surveillance dynamics, particularly the effects of different public health policies. IMPORTANCE:Surveillance programs have become key assets in determining the emergence or prevalence of pathogens circulating in human and animal populations. Genomic surveillance, in particular, provides comprehensive information on the history of isolates and potential molecular markers for infectivity and pathogenicity. Current techniques for evaluating genomic surveillance are inaccurate, ignoring the pathogen's evolutionary rate and biodiversity, as well as the timing of sequence collection. Using sequence data, we propose the q2 coefficient as a quantitative measure of surveillance completeness that combines elements of time and evolution without defining arbitrary criteria for clades or species. Through several case studies of influenza A, dengue, and West Nile viruses, we employed the q2 coefficient to identify sampling deficiencies in different host species and locations, as well as examine the effects of different public health policies through historical records of the q2 coefficient. These results can guide public health agencies to focus resource allocation and virus collection to bolster specific problems in surveillance.

Dataset Information

Quantifying the degree of bias from using county-scale data in species distribution modeling: Can increasing sample size or using county-averaged environmental data reduce distributional overprediction?

Publications

Quantifying the degree of bias from using county-scale data in species distribution modeling: Can increasing sample size or using county-averaged environmental data reduce distributional overprediction?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets