Discharge Anomalies Underlie Strong Negative Co-Variation Between Native and Non-Native Fishes

Discharge variation has long been known to mediate species interactions between native and non-native fishes. For example, brook trout (from eastern US rivers) outcompete rainbow trout (from western rivers) in Sierra Nevada streams, but only below dams where muted winter floods are not strong enough to export or kill their fall-emerging young of the year (YOY). Here, the natural hydrograph in the Sierra Nevada produces flows that are uncharacteristic in eastern rivers and these unfamiliar or 'anomalous' flows would naturally keep brook trout at bay. In desert rivers of Arizona, floods and droughts are common and assemblages of native and non-native fishes respond differently to this extreme variation. Here we study the relationship between discharge variation and covariation in abundance of native and non-native fishes in the Verde River of Arizona. We use the Fourier analysis to quantify the magnitude (maximum), frequency and timing of high- and low-flow anomalies in the daily discharge record from 1963-2012 (50 yr). We use wavelet analysis to quantify regime shifts in discharge anomalies between high- and low-flow periods over this record. We then link spectral anomalies and several summary statistics from the wavelet power spectrum to fish abundance in a Multivariate Autoregressive (MAR) framework. Results from MAR indicate that discharge anomalies determine the abundance of non-native (not native) fishes at lag zero (same year) but determine the abundance of native (but not non-native) fishes at a lag of one. Our results suggest that timing of high- and low-flow events (indexed by anomalies) is critical in determining the relationship between native and non-native fishes but that the time scale of response is different for each group of fishes.