Assessing patterns and risk to Chilean freshwater fish distributions using multi-species occupancy models

To advance our understanding of freshwater biodiversity in data-limited systems, this study used multispecies occupancy models to predict species richness and individual species occupancy, providing critical insights for conservation of these rapidly declining ecosystems. We tested several model types and ultimately pursued latent spatial multi-species occupancy models, which gained popularity in wildlife ecology but are relatively underutilized in fisheries ecology. Advantages include simultaneously modeling multiple species to infer both species-specific and assemblage-level responses to hydro-geomorphological conditions while also accounting for imperfect species detections. Data used in this model include fish sampling data and hydro-geomorphological data.
, Data collection
The initial dataset used in this analysis included fish sampled from 2015 through 2023 using a variety of standard methods (backpack electrofishing, seines, fykes, and gill nets) with sampling efforts unequally distributed across the 11 river drainage basins. After collection, fish specimens were kept in containers filled with cold, oxygenated water while in the field. All specimens were sedated using BZ-20® (ethyl p-aminobenzoate), then measured for total length and weighed at the collection site. Following this, they were placed back in clean water containers. Once they fully recovered, all specimens were returned to their natural habitats (Orrell et al. 2025). Euthanasia was performed only when necessary, such as if a fish did not recover satisfactorily after sedation. In such cases, an overdose of the same anesthetic was used for euthanasia. Fish data were then subset to include only electrofishing data collected during the low flow period (Dec-April) of 2021 to meet..., , # Assessing patterns and risk to Chilean freshwater fish distributions using multi-species occupancy models

[https://doi.org/10.5061/dryad.7d7wm385c](https://doi.org/10.5061/dryad.7d7wm385c)

## Description of the data and file structure

Fishes were sampled from 2015 to 2023 using a variety of surveys unequally distributed across the 11 river basins. Fishes were collected using standard backpack electrofishing and netting methodologies (i.e., seine, fyke, and gill nets). A suite of abiotic data at the catchment (e.g., elevation, geology, rainfall), valley (e.g., valley and trough width and slope), and channel scales (e.g., belt width, sinuosity, wavelength) were also collated and analyzed. We used ArcGIS tools to extract geomorphic data from our study system at the catchment, valley, and channel scales at 2-5 km intervals. 

Because this dataset includes IUCN endangered, vulnerable, and data-deficient species, latitude and longitude are rounded to the nearest degree, which will not al...,