Combining hydrothermal models of fast‐ and slow‐germinating seeds predicts restoration outcomes for a widely seeded dryland grass species

Abstract Introduction Predicting suitable conditions for germination is critical for efficient restoration seeding. In dryland ecosystems, soil moisture and temperature conditions favoring germination may be brief and infrequent. Hydrothermal germination models can predict germination as a function of environmental conditions, but few studies have tested their ability to explain real‐world restoration outcomes. Objectives Our objective was to retrospectively determine if simulated soil moisture and temperature conditions that influence germination could explain restoration outcomes for a dryland grass species, Squirreltail ( Elymus elymoides ), that is widely seeded in the Western United States. Methods We manipulated temperature and osmotic potential of aqueous solutions to quantify hydrothermal germination requirements of Squirreltail in growth chambers. We then identified 383 locations where Squirreltail was seeded and the plant community was subsequently censused. We simulated temperature and moisture conditions at these locations to estimate the occurrence and timing of germination events and assessed how well these simulated events predicted the occurrence of Squirreltail in post‐treatment censuses. Results Hydrothermal time (HTT) models for fast‐germinating seeds had high specificity (good at predicting absence of Squirreltail from the census), while models for slow‐germinating seeds had high sensitivity (good at predicting presence). When combined, these models outperformed a retrospective climate‐informed model. Post‐germination conditions indicated effects of wetter soil and cooler temperature in improving seedling emergence and survival. Conclusions We found that HTT models can explain significant variation in restoration outcomes when models for seeds with different germination requirements are combined.