Models and data used to predict the abundance and distribution of Ixodes scapularis (blacklegged tick) in North America: a scoping review

Abstract

Tick-borne diseases (TBD) remain prevalent worldwide, and risk assessment of tick habitat suitability is crucial to prevent or reduce their burden. This scoping review provides a comprehensive survey of models and data used to predict I. scapularis distribution and abundance in North America. We identified 4661 relevant primary research articles published in English between January 1st, 2012, and July 18th, 2022, and selected 41 articles following full-text review. Models used data-driven and mechanistic modelling frameworks informed by diverse tick, hydroclimatic, and ecological variables. Predictions captured tick abundance (n = 14, 34.1%), distribution (n = 22, 53.6%) and both (n = 5, 12.1%). All studies used tick data, and many incorporated both hydroclimatic and ecological variables. Minimal host- and human-specific data were utilized. Biases related to data collection, protocols, and tick data quality affect completeness and representativeness of prediction models. Further research and collaboration are needed to improve prediction accuracy and develop effective strategies to reduce TBD.

Keywords: Data-driven; Ixodes scapularis; Mathematical; Mechanistic; Models; North America; Ticks.

Fig 1

PRISMA Flowchart Illustrating the Scoping Review screening and selection process. The figure was generated by the PRISMA flow chart online tool (https://estech.shinyapps.io/prisma_flowdiagram/).

Fig 2

Alluvial Diagram depicting relationships among studies, modelling methods, tick data, hydroclimatic variables, and ecological variables.

Fig 3

Timeline of included studies from the literature in North America.

Fig 4

Tree diagram representing the hierarchical structure of the modelling frameworks and general model types used in the 41 included studies according to modelling output/response variable.

Fig 5

(a) Total number of included studies of all model types according to the resulting model output; (b) Total number of included studies of all model types according to the overarching model framework and associated model output.

Fig 6

Types of regression models employed in the included studies according to the model output/response variable. Abbreviations: Simple logistic regression (LgR); Simple linear regression (LR); Simple mixed-effect NB regression (Mixed-effect NB); Mixed-effect multivariable logistic regression (Mmv LgR); Multivariable boosted regression tree (Mv BRT); Multivariable Count Generalized Linear Model with forward and backward stepwise selection (Mv Count GLM + F & B); Multivariable Generalized Estimating Equation negative binomial regression (Mv GEE NB); Multivariable Generalized Estimating Equation (GEE) Poisson regression (Mv GEE PR); Multivariable Generalized Linear Model (Mv GLM); Multivariable GLM with NB (Mv GLM + NB); Multivariable logistic regression (Mv LgR); Multivariable linear regression (Mv LR); Mixed-effect Multivariable Poisson regression (Mixed-effect Mv PR); Multivariable multinomial regression (Mv Mn R); Multivariable multivariate adaptive spline regression (Mv Mv ASR); Multivariable NB regression (Mv NB); Multivariable Poisson regression (Mv PR); Multivariable zero-inflated negative binomial regression (Mv ZI NB); Multivariable zero-inflated Poisson regression (Mv ZI PR); Simple negative binomial regression (NB); Simple Poisson regression (PR); Weighted multivariable linear regression (Weighted Mv LR); Multivariable survival analysis (Mv Survival Analysis∗). ∗includes parametric (n = 1) and Cox regression (n = 1) methods.

Fig 7

Hydroclimatic and ecological predictor variable categories used across all studies to model tick distribution, tick abundance, and both tick distribution and abundance. Specific predictor variables used in each study are found in Appendix B.