In many rural areas of arid and semi-arid regions, balancing agricultural and environmental water needs is a key challenge facing resource managers. This is complicated by the tendency for the water needs of cultivated crops to be better understood than those of aquatic ecosystems. In particular, the timing and magnitude of flow needed to sustain key ecological functions remain poorly quantified in many regions. This work aims to quantify hydrologic conditions that support persistence of key ecosystem species using a functional flows framework. We use the coho ( Oncorhynchus kisutch) and Chinook ( Oncorhynchus tshawytscha) salmon run in Scott Valley, a 2,109 km ² undammed rural watershed in northern California, USA, as a case study. Taking advantage of a nearly two-decade ecological monitoring dataset and long-term stream gauge measurements, we used lasso regression to build predictive models of coho and Chinook salmon reproductive success based on hydrologic metrics. To control for cohort effects, we chose normalized ecological response metrics for coho and Chinook (number of outmigrating smolt per spawning adult or spawning adult female). For both species, we calculated optimal prediction models using a cross-validation bootstrapping approach to resample and test on unsampled observations. Lambda values, a key fitting parameter in the lasso models, were selected based on an average relative test error threshold of 1.0. Selected lambda values were used to calculate a final predictive model, or Hydrologic Benefit function, using the full dataset for each species. Hydrology could explain a greater degree of variance in relative coho reproduction than in Chinook. The hydrologic metrics that explain the greatest variance in coho reproduction values occur during the window of their parents’ spawning and, to a lesser extent, in the spring and fall of their year of rearing in freshwater. This supports an interpretation that spawning conditions may exert a significant influence on the mortality rates of the hatching juveniles. Robustness of the results indicate that this method for empirically deriving hydrologic metrics with the highest ecological benefit for a threatened species may be useful in other watersheds, where sufficient ecological data records are available, to evaluate trade-offs and support water management decisions in human-altered novel ecosystems.