To model or not to model: false positive detection error in camera surveys

Abstract

Occupancy models are commonly used with motion-sensitive camera data to estimate patterns of species occurrence while accounting for false negative detection error (i.e., the species is present but not detected). False positive detection error (i.e., the species is not present but is detected) is present in camera data sets, especially when morphologically similar species cooccur. Researchers use different approaches to address this problem: ignore the potential for false positive detections, remove all ambiguous detections and treat them as nondetections, or model false positive detection error by dividing detections into ambiguous detections (could be true or false positives) and unambiguous detections (true positives). We performed a simulation study to compare these 3 strategies. To implement these modeling strategies, detections must be classified as ambiguous or unambiguous, or all ambiguous detections must be re-classified as non-detections. We also performed a simulation study to assess the impact of researcher confidence in the designation of ambiguous and unambiguous detections. Ignoring false positive detection error resulted in biased parameter estimates, whereas removing ambiguous detections and modeling false positive detections resulted in similar estimates of occupancy probability (ψ) in most situations. Researcher over-confidence (i.e., the tendency for observers to overestimate their own ability) positively biased estimates of ψ. Moderate under-confidence did not increase bias or decrease precision in estimates of ψ.Consistent with the patterns observed in simulations, analysis of example data from a chipmunk (Neotamias minimus

Accounting for observation biases associated with counts of young when estimating fecundity: case study on the arboreal-nesting red kite (Milvus milvus)

“…In count based surveys, counting individuals more than once (i.e., a form of false positive error) is a common problem (e.g., in camera trapping - McKibben et al 2023; unmanned aerial surveys - Brack et al 2018). Yet, few modeling approaches explicitly account for such false positives (but see Kéry and Royle 2015).…”

“…False negatives have long been acknowledged as pervasive in wildlife surveys, and there is a wealth of statistical models dealing with false negatives: capture-recapture models (Otis et al 1978), N-mixture (Royle 2004), or distance sampling models (Buckland et al 2015) estimate abundance when the probability of detecting any individual is < 1; and occupancy models (MacKenzie et al 2002, Tyre et al 2003) estimate occurrence when species detection is < 1. Even though several studies have shown that false positives are also common in wildlife surveys (Clement et al 2014, Brack et al 2018, McKibben et al 2023) and that models accounting only for false negatives are very sensitive to the presence of false positives (Miller et al 2011, Lahoz-Monfort et al 2016, Link et al 2018, Nakashima 2020), the development of models accounting for false positives has lagged behind (Dénes et al 2015). Existing efforts to formally account for false positives have largely focused on occupancy models (Royle and Link 2006, Miller et al 2011, Chambert et al 2015), which require some form of additional information to be uniquely identifiable, including constraints on parameters (Royle and Link 2006) or detections with and without false positives (Miller et al 2011, Chambert et al 2015).…”

Using informative priors to account for identifiability issues in occupancy models with identification errors

“…For example, Miller et al, 2011 proposed a multiple detection state model in which certain and ambiguous data are used at each site. Then Chambert et al, 2015 built on to this proposal by considering “reference sites” exempt from detection error, and McKibben et al, 2023 returned to the notion of detection ambiguity introduced by Miller et al, 2011 by scoring the level of confidence of observers. These studies provide solutions to account for detection errors and in particular false positive ones but require to combine different sources of data, which represents a strong constraint that cannot always be met.…”