Comparative and connected weight of evidence analyses to increase confidence in directions for Topeka Shiner conservation.

Abstract

Biodiversity monitoring data provides an important source of ecological evidence about native and invasive species trends. A high priority for many conservation agencies is to understand how habitat use, biotic interactions, and human impacts affect declining and uncommon fish, especially those listed as threatened or endangered species. However, quantitative analysis of biodiversity datasets at the state-wide or regional scales that are useful for management is a challenge. More specifically high variability, small sample sizes, uncontrolled underlying heterogeneity, and the violation of statistical assumptions can increase ambiguous results between studies, mask important trends, create spurious statistical patterns, or increase unassigned error. These quantitative difficulties can prevent conservation professionals from learning all they can about existing data. The purpose of this research is to modify and test a systematic approach to analyzing existing field monitoring data to extract insights, identify gaps, and guide future actions for research and management. For this purpose, we use the specific test case of Topeka Shiner (Notropis topeka), a native Great Plains fish species that is listed as federally-endangered and Kansas-threatened. Chapter 1 illustrates how comparing outputs of multiple statistical methodologies applied to the same dataset (multiple logistic regression vs random forest analyses) can corroborate consistent ecological trends on which professionals can build and diagnose inconsistent results that require further examination. Chapter 2 applies an existing 12-step adaptive management framework to Topeka Shiner data to advance understanding and guide future conservation actions. This iterative framework uses weight-of-evidence to integrate existing literature, visualizations, statistical analyses, and ecological interpretation, and guide future data and restoration action for a focused overarching question, taxa and scale. As a result of applying this framework to a rare taxon, useful insights into Topeka Shiner distributions emerged for ecological trends with previously described (stream depth, species richness, number of dams), and new (clay soil content, or clay) environmental variables. This research tested additional emerging questions through iterative application of the above-described framework. Comparing the outputs from models created to test these new data-driven questions underscored the importance of predator species richness, nonlinear (quadratic) relationships with baseflow index, and interactions between clay and baseflow index across river basins. By building upon the original framework in successive iterations, we identified useful and testable questions regarding the abiotic and biotic variables that impact Topeka Shiner presence. Information generated from our analyses can guide future data actions, collaborative agency review processes for fish protections (e.g., KDWP, EPA, USFWS, NMFS), and subsequent framework cycles. This research offers examples of practical pathways for researchers to leverage limited data and maximize accrual of knowledge for fish ecology and conservation.