Files
Abstract
Fragmentation of hydrologic connectivity is one of many threats to aquatic biodiversity. Increasingly, culverts installed at road-stream crossings have been identified as a significant contributor in fragmentation. Culverts present an interesting challenge for researchers and those seeking to restore connectivity in that they are often not complete barriers to fish passage and are numerous on the landscape. The purposes of this study are to approach these problems with the use of spatial analysis and predictive modelling. This study utilizes random forest modelling and identifies a suite of environmental gradients that relate to impassable culverts. The necessity of rigorous classification of field data to be used to train models is highlighted. Predictions from the random forest models are used to explicitly state the cumulative effect of culverts on overall connectivity for the first time. Finally, this study guides managers by recommending that restoration activities focus on smaller, species-specific scales.