Aim
The factors that set range limits for animal populations can inform management plans aimed at maintaining regional biodiversity. We examine abiotic and biotic drivers of the distribution of finescale dace (Chrosomus neogaeus) in two Great Plains basins to identify limiting factors for a threatened freshwater fish population at the edge of their range.

Location
Great Plains, Nebraska, South Dakota, and Wyoming, USA

Methods
We investigated abiotic and biotic factors influencing the contemporary distribution of finescale dace in the Belle Fourche and Niobrara River basins with Random Forests classification models using fish surveys from multiple agencies spanning 2008 2019 and GIS-derived environmental data.

Results
In both basins, finescale dace occurrence exhibited a non-linear response to mean August water temperature. Abiotic covariates, including streamflow, water temperature, and channel slope, were important limiting factors in the final model fit with Belle Fourche River basin surveys (n = 131). In contrast, a biotic covariate, native minnow richness, was the most important predictor of finescale dace occurrence in the Niobrara River basin model (n = 27). In the Niobrara River native minnow richness was lower at sites with non-native northern pike (Esox lucius).

Main conclusions
Basin-specific analyses revealed context dependencies for species-environment relationships, which can inform targeted restoration actions. Similar relationships between water temperature and finescale dace occurrence across both basins suggests summer thermal habitat as a regional limiting factor. The importance of biotic interactions in the Niobrara River highlight an emergent threat from invasive predators to a distinct assemblage of native prairie fishes.

We compiled a dataset of fish surveys spanning 2008-2019 from multiple state agencies (Wyoming Game and Fish Department, South Dakota Natural Heritage Database and Department of Game, Fish, and Parks, and Nebraska Natural Heritage Program, Nebraska Game and Parks Commission) and field surveys conducted by the authors in 2018 and 2019. Fish collection methods by agencies were varied and included seining, single-pass electrofishing, and passive gears including minnow traps, fyke nets, and trammel nets. Given the variability in methods employed across multiple studies, we filtered these data to binary response metrics of finescale dace presence and pseudo-absence. To account for pseudo-replication of surveys within each sample unit reaches (National Hydrography Dataset (NHD) flow lines; range = 0.03 km – 7.64 km; mean = 1.94 ± 1.54 km (mean ± standard deviation)), we retained only the most recent observation of finescale dace presence / absence within each sample unit. 

We conducted targeted surveys for finescale dace during the summer seasons of 2018–2019. Fish collection methods included single-pass backpack electrofishing in 150-meter stream reaches and standardized gear sets with minnow traps and mini hoop nets in standing waters. Our aim in these surveys was to 1) update the status of finescale dace in historical localities and 2) characterize fish community and habitats under consideration for restoration activities by managers. Given the rarity of finescale dace in the study area and lack of data, we sampled nonrandom lentic and lotic sites in historical localities to refine distributions and evaluate limiting factors. We also randomly sampled lotic sites within drainages with known finescale dace occurrence, translocation history, or as recommended by managers for restoration potential. 

We jointly used these surveys and synthesized data from managers to assemble a dataset representative of contemporary basin-wide fish occurrence in the study area. These data were merged using table joins in Program R. The “DataID” in the provided column indicates the entity of origin. 

We used fish community data representing 36 species (25 native and 11 non-native) across 9 families to calculate native minnow richness and classify the occurrence of littoral predatory fish at study sites. Native minnow richness was calculated as the sum of unique occurrences of native, small bodied fishes. We then subtracted finescale dace from these totals at sites where they occurred. We also calculated a binary presence/absence variable for multiple non-native, predatory fish species that are known to forage in littoral, nearshore habitats and have negative effects on native cyprinid communities as introduced or invasive species. 

We retrieved data from multiple public sources to describe environmental factors with a known filtering effect on freshwater fishes at the basin scale (e.g., channel gradient, stream size, water temperature) or literature-derived relationship to finescale dace. All environmental data were extracted and joined to sites using ArcMap version 10.5 (Esri, Redlands, California) and package ‘rgeos’ in Program R (Bivand et al. 2019; R Core Team 2019). We extracted mean annual streamflow, a proxy for stream size, in cubic meters per second (m3/s) (1971-2000) and channel slope (%) for flow lines from the National Hydrography Dataset (NHDPlus High Resolution, https://www.usgs.gov/core-science-systems/ngp/national-hydrography/nhdplus-high-resolution), a digital database of surface water features in the United States. We retrieved baseflow index (%) values at the catchment scale (i.e., the local landscape that directly contributes water to a stream), a ratio representing an estimate of the amount of streamflow comprised of groundwater discharge, from the Environmental Protection Agency StreamCat dataset (Hill et al. 2016, https://www.epa.gov/national-aquatic-resource-surveys/streamcat). We used catchment-scale habitat condition index (HCI) scores from National Fish Habitat Partnership (Crawford et al., 2016, http://assessment.fishhabitat.org/), which assign degradation risk to catchments ranging from 1 (high risk) to 5 (low risk) by integrating multiple disturbance metrics (e.g., road length density, impervious surface, crop land use) into a metric of watershed disturbance. We calculated the percentage of open water and wetlands (%) within catchments from the 2011 National Land Cover Dataset (Multi-Resolution Land Characteristics Consortium, https://www.mrlc.gov/) using ArcMap version 10.5. 

We used spatial stream network models, a type of geostatistical network models, to develop predictions of mean August stream temperature (°C) for stream networks in each basin using continuous water temperature observations spanning 2004 – 2019. To account for interannual variability in climate, we fit these models with mean August values of air temperature (°C) (PRISM Climate Group, http://www.prism.oregonstate.edu/) and streamflow (m3/s) (U.S. Geological Survey [USGS], 2020, USGS water data for the Nation: U.S. Geological Survey National Water Information System database, accessed February 20, 2020, at https://doi.org/10.5066/F7P55KJN) that were cross-referenced to the matching year for each observation dataset. We then used a prediction dataset with mean air temperature and streamflow values from 1999-2019 to produce a composite 20-year mean stream temperature metric. 

Specific location data (spatial coordinates) of fish surveys are not being shared, per data use agreements with the Wyoming Game and Fish Department, South Dakota Natural Heritage Database and Department of Game, Fish, and Parks, and Nebraska Natural Heritage Program, Nebraska Game and Parks Commission. We have attached a readme file with descriptions of data and links to sources where applicable.