Practical Bias Correction in Aerial Surveys of Large Mammals: Validation of Hybrid Double-Observer with Sightability Method against Known Abundance of Feral Horse (Equus caballus) Populations.
Authors: Lubow Bruce C, Ransom Jason I
Journal: PloS one
Summary
# Editorial Summary Accurate population estimates are essential for evidence-based management of feral horses, yet aerial surveys are notoriously prone to detection bias that can seriously skew abundance figures. Lubow and Ransom tested a hybrid bias-correction method combining double-observer and sightability techniques against feral horse populations of known size in the western United States—a rare validation opportunity afforded by long-term individual identification studies. Across six tests on populations of known abundance, the method produced estimates within −8.5% to +13.7% of actual values with tight confidence intervals (all <0.7 standard errors), though coefficient of variation ranged considerably from 6.1% to 25.0%. The approach proved less reliable when helicopter disturbance altered horse behaviour during surveys, demonstrating that statistical correction cannot fully compensate for observer-induced detection heterogeneity. For equine professionals involved in population monitoring or advising on feral horse management, this work validates the rigour of modern aerial survey methodology whilst underscoring the critical importance of standardised survey protocols and careful consideration of disturbance factors—particularly observer platform choice—when designing monitoring programmes that will inform culling, contraception, or relocation decisions.
Read the full abstract on PubMed
Practical Takeaways
- •Aerial surveys using hybrid double-observer and sightability correction can reliably estimate feral horse populations with reasonable accuracy (within ~10% of actual numbers) when properly designed
- •Survey method and aircraft choice significantly impact data quality—fixed-wing aircraft or unobtrusive platforms may outperform helicopters that disturb horses and introduce detection errors
- •Precision of population estimates varies substantially by conditions; plan for 6-25% coefficient of variation depending on terrain, visibility, and population behavior
Key Findings
- •Hybrid double-observer with sightability bias correction method produced population estimates within -8.5% to +13.7% of known abundance across six validation tests
- •Precision estimates ranged from 6.1% to 25.0% coefficient of variation across different survey conditions
- •Helicopter use during surveys introduced detection error and heterogeneity that could not be corrected by statistical models
- •Identified multiple covariates across surveys that explained and corrected for detection biases in aerial survey estimates