Distinguishing distribution dynamics from temporary emigration using dynamic occupancy models

Abstract

Dynamic occupancy models are popular for estimating dynamic distribution rates (colonization and extinction) from repeated presence/absence surveys of unmarked animals. This approach assumes closure among repeated samples within primary periods, allowing estimation of dynamic rates between these periods. However, the impact of temporary emigration (TE; reversible changes in sampling availability) on dynamic rate estimates has not been tested. Using simulated data, we investigated the degree to which TE could mislead researchers interested in quantifying dynamics. We then compared results from three avian point count datasets to evaluate the likelihood that TE confounds estimates of dynamics for 19 species under a popular sampling protocol. Simulated experiments indicated that when secondary periods were open to TE, presence of dynamics was correctly identified ≥95.1% of the time, and dynamic rate estimates were accurate. However, dynamic rate estimates were biased when secondary periods were closed to TE. In empirical datasets, dynamic occupancy models had greater support than closed models for all species when secondary sampling periods occurred in immediate succession (i.e. 3 samples within 10 min); however, our results suggest that this is because dynamic estimates were heavily influenced by TE. When counts within a primary period were separated by 24–48 hr, we found evidence of dynamics for less than half of these species. We recommend an alternative sampling approach that allows accurate estimation of dynamic rates when TE is of no interest, and introduce a novel model for estimating both processes simultaneously in rare cases where they are both of biological interest. Concern for violating the occupancy modelling closure assumption has led to widespread recommendations that samples within primary periods be conducted extremely close in time. However, this may not be the best approach when interest is in quantifying dynamic rates. While dynamic occupancy models provide estimates of “colonization” and “extinction,” these values do not inherently represent dynamics unless TE has been explicitly modelled or accounted for with sampling design. Naiveté to this fact can result in incorrect conclusions about biological processes.

Seeking efficiency with carnivore survey methods: A case study with elusive martens

“…We can think of 2 obvious explanations. If the site was no longer occupied, then the animal was no longer present to be detected (Valente et al ). We do not expect this to be the case based on recent marten surveys with telemetry information, which indicated high fidelity of both social structure and movement (e.g., Moriarty et al , Linnell et al ).…”

“…The relationship between occupancy and animal abundance changes dramatically depending on whether the scale of inference is a point, an area smaller than a home range, or an area larger than a home range—with the most accurate relationship at the scale of a sample location (Steenweg et al ). Individuals may not be available for detection because of their movements, and we urge caution when interpreting results beyond the point location using occupancy models (Efford and Dawson , Hutto , Steenweg et al , Valente et al ). Characterizing the influence of effective survey distance versus animal behavior on detection probability in areas when the species is present will continue to be a central challenge of interpreting noninvasive surveys for elusive and rare carnivores; however, cameras and detection dog teams can be effective tools depending on the scale and scope of study objectives.…”

Removal pattern mitigates negative, short‐term effects of stepwise Russian olive eradication on breeding birds

“…In both models, we assumed sites were closed to changes in use over the two‐week sampling periods within a single year (likely reasonable for most species, Valente et al. ) and modeled detection probability as a function of standardized sampling time:…”

“…We conducted three or four avian point counts per year (depending on logistical constraints) at the center of each riparian plot during the peak of the breeding seasons of 2014-2018 (5 yr). All counts in a single year were conducted within a two-week window (31 May and 15 June) to minimize the probability of intra-annual changes in site use and were separated by 1-4 d to ensure availability for sampling was independent on each occasion (Valente et al 2017). Counts were only conducted in suitable weather conditions (i.e., low wind and no precipitation) between 04:30 and 09:00 hours to maximize bird detectability.…”

“…The values of y represent observed detections (1 if detected, 0 otherwise), Z represents true occupancy state (1 if occupied, 0 otherwise), p represents detection probability, and w indicates occupancy probability. In both models, we assumed sites were closed to changes in use over the two-week sampling periods within a single year (likely reasonable for most species, Valente et al 2017) and modeled detection probability as a function of standardized sampling time:…”