Integrated distance sampling models for simple point counts

Area of Science:

• Ecology
• Wildlife Biology
• Statistical Modeling

Background:

• Point counts (PCs) are common biodiversity survey methods but suffer from unknown detectability, leading to biased abundance estimates.
• Spatiotemporal variations in detectability and unknown survey areas hinder accurate density estimation and landscape-level scaling.
• Existing citizen-science data often lack information to correct for detection biases, limiting their ecological utility.

Purpose of the Study:

• Introduce integrated distance sampling (IDS) models to address limitations of traditional point count and detection/nondetection data.
• Enable accurate density estimation and landscape-level inference by combining multiple data types.
• Enhance the utility of citizen-science data by correcting for detection biases.

Main Methods:

• IDS models integrate distance sampling (DS) with point count (PC) or detection/nondetection (DND) data.
• Treats PC and DND data as aggregations of latent DS surveys to estimate separate detection functions and covariate effects.
• Utilizes repeat or time-removal surveys to estimate availability and perceptibility components of detectability.

Main Results:

• IDS models reconcile spatial and temporal mismatches between different data sets.
• Successfully address detectability and survey area issues inherent in simple PC and DND data.
• Provide JAGS code and an R package function ('IDS()') for fitting these models.

Conclusions:

• IDS models offer a robust solution for accurate density estimation in ecological surveys.
• Significantly extend the utility and reach of citizen-science data by correcting for detection biases.
• Applicable to hybrid survey designs combining DS with distance-free methods, with broad implications for ecology and conservation management.