A Stochastic Hourly Stream Temperature Model to Forecast Land-Use and Climate Change Effects on Temperature Threshold Exceedance Duration for Freshwater Mussels

Changes in stream temperature can have significant effects on freshwater ecosystems and aquatic organisms. Direct effects of temperature on aquatic organisms are often measured by acute thermal tolerance, or LT_%, over some time period (h). For example, LT₅₀ is an average median lethal temperature where 50% of individuals in a population die. Mean daily temperature above a given 24 h LT₅₀ does not necessarily indicate a 24 h exposure, and there is a need for hourly water temperature estimates to obtain exposure durations. We developed a stochastic hourly temperature model that provides forecasts for the probability of exceeding given threshold temperatures for specified durations (24 and 96 h). Daily mean stream temperatures from an existing model of the upper Tar River basin, North Carolina, USA, were used as input to our stochastic hourly temperature model for climate change and land-use change simulations for 2021-2030 and 2051-2060. Time series of hourly temperature data at 20 sites from July 2010 through November 2011, were used to parameterize autoregressive-moving average (ARMA) models. Parameterizations were done using site specific and basin-wide observations. The use of site-fitted parameters for ARMA simulations of hourly temperatures showed no significant differences with simulations using basin-fitted parameters. Stream temperature observations in 2010 revealed only 2 sites with temperatures above 30°C for > 24 h and temperatures were never > 31°C for more than 24 h at any site. Simulations suggest that higher temperature thresholds are likely to be exceeded for longer durations than have occurred in the past century. The probability that temperatures will exceed 32°C for at least 96 h in a given year increased from 0, at present, to 0.05 in 2021-2030 and to ~ 0.14 in 2051-2060. Simulations indicated that climate change had much greater affects on probabilities of temperature threshold exceedance for 24 and 96 h durations than land-use change. Increased probability of threshold exceedance occurred primarily in higher-order stream segments in the downstream reaches of the basin. Furthermore, simulations indicated that hourly stream temperatures responded more strongly to land-use change on the event scale (≤ 24 h) and did not affect stream temperatures for durations > 24 h. Climate change increased hourly stream temperature more uniformly throughout the basin at time scales > 24 h. These results seem to indicate that the response of stream temperature varies dependent on time scale. It is likely that the time scale of interaction between processes affects stream temperature response to land-use and climate change, and that climate change is likely to have a greater impact on temperature exceedance durations than land-use change. Hourly stream temperature simulations indicated that some of the known thermal thresholds for freshwater mussels could be exceeded within the next 50 years in many parts of the upper Tar River basin, and that exceedances could have negative consequences on the recruitment of freshwater mussels.