Aim: Recurrent species assemblages integrate important biotic interactions and joint responses to environmental and spatial filters that enable local coexistence. Here, we applied a bipartite (site-species) network approach to develop a natural typology of lakes sharing distinct fish faunas and provide a detailed, hierarchical view of their bioregions. We then compared the roles of key biogeographic factors to evaluate alternative hypotheses about how fish communities are assembled from the regional species pool.

Location: Ontario, Canada and the Upper Midwest, USA.

Time period: 1957–2017.

Major taxa studied: Freshwater fishes.

Methods: Bipartite modularity analysis was performed on 90 taxa from 10,016 inland lakes in the Southwestern Hudson Bay, Mississippi River, and St. Lawrence River drainages, uncovering bioregionalization of North American fishes at a large, subcontinental scale. We then used a latent variable approach, pairing non-metric partial least-squares structural equation modelling with multiple logistic regression, to show differences in the biogeographic templates of each community type. Indicators of contemporary and historical connectivity, climate, and habitat constructs were estimated using a geographic information system.

Results: Fish assemblages reflected broad, overlapping patterns of postglacial colonization, climate, and geological setting, but community differentiation was most linked to temperature, precipitation, and, for certain groups, lake area and water quality. Bioregions were also marked by non-native species, showing broad-scale impacts of introductions to the Great Lakes and surrounding basins.

Main conclusions: The dominant effects of climate across broad spatial gradients indicate differing sensitivities of fish communities to rapidly accelerating climate change and opportunities for targeted conservation strategies. By assessing biological variation at the level of recurrent assemblages, we accounted for the non-stationarity of macroecological processes structuring different sets of species on the landscape, and thus offer novel inference on the assembly of inland fish communities.

Fish occurrence records in portions of the Southwestern Hudson Bay, St. Lawrence, and Mississippi River drainages were compiled from existing datasets in Ontario (Aquatic Habitat Inventory; Dodge, Goodchild, Tilt, Waldriff, & MacRitchie, 1987), Michigan (Status and Trends; Wehrly, Carter, & Breck, 2021), and Minnesota (Index of Biotic Integrity; Drake & Pereira, 2002). Fishes were sampled from lakes, ponds, reservoirs, and embayments during the open-water season over a 60-year period (1957–2017) using a combination of netting, trapping, and electrofishing. Records of 132 fish taxa were initially compiled across 11,112 lakes, but several species and lakes were removed (see manuscript for selection details). Our final dataset included 90 taxa across 10,016 lakes (mean species richness = 8).

References

Dodge, D.P., Goodchild, G.A., Tilt, J.C., Waldriff, D.G., & MacRitchie, I. (1987). Manual of instructions: Aquatic habitat inventory surveys. Ontario Ministry of Natural Resources, Toronto, Canada.

Drake, M.T., & Pereira, D.L. (2002). Development of a fish-based index of biotic integrity for small inland lakes in Central Minnesota. North American Journal of Fisheries Management, 22, 1105–1123.

Wehrly, K.E., Carter, G.S., & Breck, J.E. (2021). Standardized sampling methods for the inland lakes status and trends program. Fisheries Special Report. Lansing, MI: Michigan Department of Natural Resources.

Species at risk have been removed for their protection and those not used in analysis have also been excluded (see manuscript for selection details). As contractual requirements prevent us from releasing detailed sampling records for Ontario directly, these data are provided with approximate location coordinates. The full set of fish data (including species at risk) are readily available for non-commercial purposes from the Ministry of Northern Development, Mines, Natural Resources, and Forestry in Ontario (https://geohub.lio.gov.on.ca/datasets/aquatic-resource-area-polygon-segment-) and Departments of Natural Resources in Michigan (https://www.michigan.gov/dnr/0,4570,7-350-79137_79765_84005_84013---,00.html) and Minnesota (https://www.dnr.state.mn.us/waters/surfacewater_section/lake_ibi/index.html) through their standard data-sharing agreements.