Incorporating antenna detections into abundance estimates of fish

Abstract

Autonomous passive integrated transponder (PIT) tag antennas are commonly used to detect fish marked with PIT tags but cannot detect unmarked fish, creating challenges for abundance estimation. Here we describe an approach to estimate abundance from paired physical capture and antenna detection data in closed and open mark–recapture models. Additionally, for open models, we develop an approach that incorporates uncertainty in fish size, because fish size changes through time (as fish grow bigger) but is unknown if fish are not physically captured (e.g., only detected on antennas). Incorporation of size uncertainty allows for estimation of size-specific abundances and demonstrates a generally useful method for obtaining state-specific abundances estimates under state uncertainty. Simulation studies comparing models with and without antenna detections illustrate that the benefit of our approach increases as a larger proportion of the population is marked. When applied to two field data sets, our approach to incorporating antenna detections reduced uncertainty in abundance substantially. We conclude that PIT antennas hold great potential for improving abundance estimation, despite the challenges they present.

Vital rates of a burgeoning population of Humpback Chub in western Grand Canyon

“…Capture probabilities were a logit-linear function of water temperature, and the means for movement parameters were season specific. Antenna detections from September 2016 and 2018-2021 were included as part of a robust design (RD) framework (one pass of physical captures and one pass of antennas) to help increase detection probabilities and inform availability (Dzul et al 2022).…”

“…For Humpback Chub in eastern Grand Canyon, the model included three spatial states (LCR, the Colorado River inside the MS–east reach, and the Colorado River outside the MS–east reach) and four size states that were based on TL measurements (small subadult: 100–149 mm TL; large subadult: 150–199 mm TL; small adult: 200–249 mm TL; large adult: >249 mm TL). This led to a total of 12 states that represented combinations of size and spatial location (Yackulic et al 2014; Dzul et al 2022); hence, the state transitions described both growth (transitions between sizes) and movement (transitions between sites). The state transition matrix allowed fish to move from the LCR into the Colorado River and from the Colorado River into the LCR.…”

“…Antenna detection probabilities were size and time dependent. Abundances were estimated as derived parameters using a method described by Dzul et al (2022), in which abundance is estimated using physical captures and PIT antenna detections. Priors for logit‐transformed means and coefficients were normal (μ = 0, standard deviation = 2), and standard deviations were uniform (0–5).…”

Estimating migration timing and abundance in partial migratory systems by integrating continuous antenna detections with physical captures

“…In the current study, abundance estimates displayed relatively high precision due to high array detection probabilities, where the precision of abundance estimates was positively correlated with the proportion marked (e.g. Conner et al, 2020;Dzul et al, 2022).…”

Translocation in a fragmented river provides demographic benefits for imperiled fishes

“…This higher survival estimate is likely because we accounted for movement, potential size differences between translocated and non‐translocated fish (Appendix S1: Figure S2), and due to differences in the temporal and spatial scope of data analyzed. Including capture and antenna detections also increased precision in our demographic parameter estimates (Dzul et al, 2022), especially for fishes above the waterfall.…”