Mt or not Mt: Temporal variation in detection probability in spatial capture-recapture and occupancy modelsuse asterix (*) to get italics

Rahel SollmannPlease use the format "First name initials family name" as in "Marie S. Curie, Niels H. D. Bohr, Albert Einstein, John R. R. Tolkien, Donna T. Strickland"

2023

<p>State variables such as abundance and occurrence of species are central to many questions in ecology and conservation, but our ability to detect and enumerate species is imperfect and often varies across space and time. Accounting for imperfect and variable detection is important for obtaining unbiased estimates of state variables. Here, I investigate whether closed spatial capture-recapture (SCR) and single season occupancy models are robust to ignoring temporal variation in detection probability. Ignoring temporal variation allows collapsing detection data across repeated sampling occasions, speeding up computations, which can be important when analyzing large datasets with complex models. I simulated data under different scenarios of temporal and spatio-temporal variation in detection, analyzed data with the data-generating model and an alternative model ignoring temporal variation in detection, and compared estimates between these two models with respect to relative bias, coefficient of variation (CV) and relative root mean squared error (RMSE). SCR model estimates of abundance, the density-covariate coefficient β and the movement-related scale parameter of the detection function σ were robust to ignoring temporal variation in detection, with relative bias, CV and RMSE of the two models generally being within 4% of each other. An SCR case study for brown tree snakes showed identical estimates of density and σ under models accounting for or ignoring temporal variation in detection. Occupancy model estimates of the occupancy-covariate coefficient β and average occupancy were also largely robust to ignoring temporal variation in detection, and differences in occupancy predictions were mostly &lt;&lt;0.1. But there was a slight tendency for bias in β under the alternative model to increase when detection varied more strongly over time. Thus, when temporal variation in detection is extreme, it may be necessary to model that variation to avoid bias in parameter estimates in occupancy models. An occupancy case study for ten bird species with a more complex model structure showed considerable differences in occupancy parameter estimates under models accounting for or ignoring temporal variation in detection; but estimates and predictions from the latter were always within 95% confidence intervals of the former. There are cases where we cannot or may not want to ignore temporal variation in detection: a behavioral response to detection and certain SCR observation models do not allow collapsing data across sampling occasions; and temporal variation in detection may be informative of species phenology/behavior or for future study planning. But this study shows that it can be safely ignored under a range of conditions when analyzing SCR or occupancy data.&nbsp;</p>

https://doi.org/10.5281/zenodo.10053390You should fill this box only if you chose 'Scripts were used to obtain or analyze the results'. URL must start with http:// or https://

abundance, assumption violation, density, hierarchical statistical models, occurrence

NonePlease indicate the methods that may require specialised expertise during the peer review process (use a comma to separate various required expertises).

Euring Conference, Statistical ecology

Mark Kéry, marc.kery@vogelwarte.ch, Chris Sutherland, css6@st-andrews.ac.uk, Ehsan Moqanaki, ehsan.moqanaki@gmail.com, Olivier Gimenez suggested: Hi Ben,, Olivier Gimenez suggested: I wish I had time to review this preprint, but I'm in the middle of a period that I blocked for writing a book, and preparing lectures. As other referees, I would suggest approaching Rémy Fraysse <remi.fraysse@cefe.cnrs.fr> or Rachel McCrea <r.mccrea@lancaster.ac.uk>., Olivier Gimenez suggested: Cheers,, Olivier Gimenez suggested: Olivier, Dana Karelus suggested: Matthew Gould <mgould@usgs.gov>, Elise Zipkin [ezipkin@msu.edu] suggested: You caught me on a day when I got three new papers as AE so have to decline this one. How about Beth Gardner or someone in her lab for this?, John Fieberg [jfieberg@umn.edu] suggested: Sorry - this was mainly just bad timing (I've been away for work, have other reviews I've already committed to that more closely align with current research, and also overwhelmed with trying to prep for start of the semester). Some alternatives include:, John Fieberg [jfieberg@umn.edu] suggested: Mahdieh Tourani: mahdieh.tourani@gmail.com, John Fieberg [jfieberg@umn.edu] suggested: Ben Augustine: baugustine@usgs.gov, John Fieberg [jfieberg@umn.edu] suggested: Gabby Palomo: gpalomo@umd.edu, John Fieberg [jfieberg@umn.edu] suggested: Chris Sutherland: css6@st-andrews.ac.uk , John Fieberg [jfieberg@umn.edu] suggested: Alan Welsh: alan.welsh@anu.edu.au No need for them to be recommenders of PCIEcology. Please do not suggest reviewers for whom there might be a conflict of interest. Reviewers are not allowed to review preprints written by close colleagues (with whom they have published in the last four years, with whom they have received joint funding in the last four years, or with whom they are currently writing a manuscript, or submitting a grant proposal), or by family members, friends, or anyone for whom bias might affect the nature of the review - see the code of conduct

2023-08-10 09:18:56

Benjamin Bolker

Dana Karelus, Ben Augustine, Ben Augustine