Results
Consistent with the flood of local observations of a die-off, we encountered the first carcasses floating mid-river nearly immediately upon leaving Hughes. In the ca. 315 km of survey, we counted 1,364 dead fish.
Sampled carcasses varied markedly in condition and state of decomposition. None of the carcasses had signs of trauma or post-mortem scavenging by animals, a pattern that was also widely noted by local residents. Every individual examined internally for sex had not completed maturation, indicated by firm testes in males and unovulated eggs in females (Fig. 3). Approximately 1/3 of the carcasses had signs of potential fungal growth, consistent with secondary infections following the growth of Flavobacterium columnare ( James Winton, USGS, personal communication).
The water temperature in the mainstem Koyukuk River during the survey averaged 17.1°C (range 16.8 – 17.4). Dissolved oxygen averaged 8.3 mg/L (Range 7.9 – 8.6). Given that DO saturation at 17.1°C occurs at 9.65 mg/L, our measurements indicate the water was on average only approximately 87% saturated with dissolved oxygen.
Consistent with size-selective mortality, individuals that died en route were ca. 4% smaller on average compared to individuals that survived to the spawning grounds (Table 1). Body size was significantly and inversely associated with probability of death (Fig. 4, p<0.0001) Estimates of standardized selection for males was 0.77, and 0.86 for females. These estimates exceed the 99th percentile of estimates reported by Siepielski et al. and is thus evidence of very strong selection. However, simulation analysis indicate that our observations of selection could have occurred approximately 25% of the time due to chance alone, which likely reflects the natural wide distribution of body sizes in summer Chum Salmon.