In nearly all large-scale terrestrial ecosystem models, soil respiration is represented as a function of soil temperature. However, the relationship between soil respiration and soil temperature is highly variable across sites and there is often a pronounced hysteresis in the soil respiration-temperature relationship over the course of the growing season. This phenomenon indicates the importance of biophysical factors beyond just temperature in controlling soil respiration. To identify the potential mechanisms of the seasonal soil respiration-temperature hysteresis, we developed a set of numerical models to demonstrate how photosynthesis, soil moisture, and soil temperature, alone and in combination, affect the hysteresis relationship. Then, we used a variant of the model informed by observations of soil respiration, soil temperature, photosynthesis, and soil moisture from multiple mesic and semi-arid ecosystems to quantify the frequency of hysteresis and identify its potential controls. We show that the hysteresis can result from the seasonal cycle of photosynthesis (which supplies carbon to rhizosphere respiration), and soil moisture (which limits heterotrophic respiration when too low or too high). Using field observations of soil respiration, we found evidence of seasonal hysteresis in 9 out of 15 site-years across 8 diverse biomes. Specifically, clockwise hysteresis occurred when photosynthesis preceded seasonal soil temperature and counterclockwise hysteresis occurred when photosynthesis lagged soil temperature. We found that across all sites, much of the respiration-temperature lag was explained by the decoupling of photosynthesis and temperature, highlighting the importance of recently assimilated carbon to soil respiration. An analysis of observations from 129 FLUXNET sites revealed that time lags between gross primary productivity (a proxy for canopy photosynthesis) and soil temperature were common phenomena, which would tend to drive counterclockwise hysteresis at low-latitude sites and clockwise hysteresis at high-latitude sites. Collectively, our results show that incorporating photosynthesis and soil moisture in the standard exponential soil respiration-temperature model (i.e., Q(10) model) improves the explanatory power of models at local scales.