# Systematic Review of Variable Selection Bias in Species Distribution Models for Aedes vexans (Diptera: Culicidae)

## TL;DR

This study reviews how well mosquito models for Aedes vexans include important environmental factors, finding that flood-related variables are often missing, which could affect prediction accuracy.

## Contribution

The paper systematically identifies variable selection bias in Aedes vexans distribution models, emphasizing the omission of flood-related factors.

## Key findings

- Most models for Aedes vexans lack flood-related variables despite their ecological importance.
- Urban-landscape variables are frequently used but may reflect sampling bias rather than true habitat preferences.
- Including flood dynamics could improve the accuracy of disease risk predictions for Aedes vexans.

## Abstract

The mosquito Aedes vexans is common in Europe and often appears in large numbers after floods. Because it can transmit diseases, scientists use computer models to predict where it might occur. These models rely on environmental factors such as temperature, rainfall, and land use to describe the conditions that make habitats suitable for the species. We reviewed all published studies that used such models for Aedes vexans to check whether they include the environmental factors that really matter for this species. We found 28 studies and analysed nearly 500 variables used to describe weather, land use, water, and human activity. Surprisingly, most models did not include information about flooding or temporary water bodies, even though these are essential for the mosquito’s reproduction. Instead, many studies focused on urban features, a choice that may reflect where people collect samples rather than where the mosquito actually lives. Our findings highlight that understanding how and where data are collected is as important as the modelling technique itself. Taking flood dynamics into account could make predictions more realistic and help to make management of potential disease risks more effective.

We conducted a systematic literature review, following PRISMA guidelines, to assess whether existing species distribution models for Aedes vexans reflect its known ecological requirements. This mosquito is closely associated with temporary floodwaters, making hydrological dynamics a critical factor for accurate modelling. From 28 peer-reviewed studies, we extracted 477 environmental and ecological variables and organized them into a hierarchical classification scheme with four main categories: weather, land characteristics, water characteristics, and population. We analysed patterns in variable usage and the reported importance of each variable. Our results show that flood-related variables were largely absent, despite the species’ reliance on ephemeral water bodies for reproduction. This may possibly reduce the predictive utility of existing Aedes vexans species distribution models. In contrast, urban-landscape variables were frequently used and often ranked as highly predictive, but such results were primarily found in studies that did not account for sampling bias.