Assessing Impacts of Climate Change on Vulnerability of Brook Trout in Lake Superior’s Tributary Streams of Minnesota

Abstract

Water temperature is generally considered one of the primary physical habitat parameter determining the suitability of stream habitat for fish species, with effects on the mortality, metabolism, growth, behavior, and reproduction of individuals. In this study we assessed the potential threats of climate change on stream temperatures and flow regimes in Lake Superior tributary streams in Minnesota, USA. The study included deterministic models for stream flow and temperature of three study streams (Amity Creek, Baptism River, Knife River), and regional (empirical) models for specific flow and temperature parameters to give better spatial coverage of the region. Information on stream flow, stream temperature, and land cover was used to develop a brook trout presence/absence model to understand the current pattern of distribution of brook trout and predict future distributions under future climate. The hydrology of north shore streams is mainly driven by air temperature and precipitation. Historical air temperatures in the region have a significant upward trend, particularly since 1980. Global climate model (GCM) outputs project a continued increasing trend in air temperature, with an increase in mean annual air temperature of 2 to 3 °C by 2089. The historical precipitation data shows an increasing trend for total annual precipitation at Duluth and Two Harbors between 1900 and 2010, whereas Grand Marais and Grand Portage do not have a clear trend. Based on an analysis of daily precipitation totals, there is some indication of an increasing trend in the number of days in summer with high precipitation (10-20 cm). Both the GENMOM and the ECHAM5 GCMs project overall increases in precipitation of about 15%, but differ with respect to the seasonal distribution of the precipitation changes. A significant and relatively certain impact of climate change is a projected shift in precipitation from snowfall to rainfall. While an increasing trend in precipitation leads to increasing streamflow, the increasing trend in spring and summer air temperature tends to reduce streamflow (by increasing evapotranspiration). Available streamflow records for north shore streams suggest there may be a decreasing trend in mean annual flow and summer low flow, but the trends are not statistically significant. Future projections of streamflow based on the GCM output were mixed, with the deterministic models projecting moderate increases in average stream flow and summer low flow, while the regression models for project a moderate decrease in low flow. Stream temperature analyses for the three study streams based on GCM climate output give the result of fairly uniform seasonal increases in stream temperature to 2089 ranging from 1.3 to 1.9 °C for the GENMOM model to 2.2 to 3.5°C for the ECHAM5 model. Application of the GENMOM climate data to the deterministic stream temperature models produced fairly similar stream temperature changes for the three study sites. The empirical stream temperature study found stream temperature in the north shore region to be influenced by air temperature, catchment size, percentage of woody wetlands, latitude, and soil permeability rate. In response to climate change projected by the GENMOM GCM, the regional stream temperature model projects July mean water temperature to rapidly increase by approximately 1.2oC from 1990s to 2060s, followed by a slight decrease to 2089. The temperature increase was predicted to be the largest in the coastal area of middle north shore region. The brook trout presence/absence model found water temperature to have the strongest influence on trout presence. Brook trout were predicted to be at risk for water temperatures above 18.7oC and be extirpated from streams for temperatures over 20oC. Stream flow was shown to have a negative effect on trout presence, though not as strong as water temperature. Overall, these data predict that brook trout may be extirpated from lower shore area, be exposed to increasing risk in middle shore region, and remain present in upper shore streams from the present to 2089. This work would benefit greatly from a number of modifications to the GCM’s, the spatial data used in the development of both the deterministic and empirical models, and implementation of a more detailed, spatially explicit, hydrologic model. Finally, additional fish data, including cool and warm water assemblage data, along with descriptors of landscape structure (i.e., connectivity) would allow us to assess the areas where cold water species may be threatened by the presence or potential presence of coolwater competitors.