Effects of temperature, flow, and climate factors on inter-annual variation of larval fish assemblages in a coastal watershed

Abstract

Predicting how populations will respond to future climate scenarios is of interest to ecologists and resource managers. Recent interest in the effect large-scale climate patterns have on local environmental conditions has identified relationships between climate, fish diversity, assemblage structure, abundance, and population fluctuations. I hypothesized that large-scale climate patterns could help explain the inter-annual variation in larval fish assemblage of the lower Roanoke River, North Carolina. Acting through temperature and precipitation, these large-scale climate indices can influence ecosystem response, and therefore should correlate with local weather variables. The North Atlantic Oscillation (NAO) and the Atlantic Multi-Decadal Oscillation (AMO) climate indices are known to influence temperature and precipitation in the North Atlantic Ocean basin and were chosen for their likely influence within the lower Roanoke River. Air temperature and precipitation data from Weldon, NC, and Roanoke Rapids, NC, weather stations were used to form a long-term combined (LTC) dataset from 1904 to 2011 to test their relationship with the large-scale climate indices. Relationships between the LTC, NAO, and AMO showed the winter NAO positively significant with four-month winter (December through March) mean maximum air temperatures (r = 0.41, p <0.001, n = 108), but not total winter precipitation (r = 0.03, p = 0.742, n = 108). The AMO was positively correlated with mean annual minimum air temperature (r = 0.39, p <0.0001, n = 108), but not with annual total precipitation (r = 0.02, p = 0.82, n = 108). Inter-annual variation in fish assemblage structure from 1984 to 1991 and 2000 to 2003 is examined in relation to the NAO, the AMO, winter and spring air temperature and precipitation, along with environmental variables collected during sampling using non-parametric multidimensional scaling (NMDS). A polar ordination was also conducted as a confirmatory analysis. A Mantel test was conducted to compare the two ordination methods, as well as to test the significance of correlations between environmental variables and fish family scores, the ordination scores from the NMDS. Mantel tests indicate that the NMDS and Polar ordination are comparable (r = 0.92, p <0.001). Mantel tests with environmental variables and fish family scores, indicate that winter minimum air temperature (r = 0.46, p = 0.017), the AMO (r =0.42, p = 0.009), pH (r = 0.4, p =0.022), and spring maximum air temperature (r = 0.37, p = 0.040) were the only significant environmental variables. The NAO was negatively correlated with the fish family scores (r = -0.29, p = 0.051). The AMO appears to influence freshwater fish families in the lower Roanoke River, while the NAO may influence anadromous fish families through minimum winter temperatures. Consideration of other climate indices, such as ENSO, that influence precipitation in the lower Roanoke River watershed may be warranted, particularly when precipitation influences fish abundance. Ideally, long-term datasets of larval or juvenile fish abundance from the lower Roanoke River would be valuable in defining the relationships between climate and the inter-annual variation in fish assemblages.