Cumulative germination curves were recorded for carrot (Daucus carota L.) seeds at a range of constant temperatures (T) and water potentials (Ψ) in the laboratory and under variable soil conditions in 15 seed-bed environments in the field. A single base temperature (Tb), a distribution of base water potentials (Ψb(G)) for percentiles (G) of the population and the hydrothermal time constant (Hg) were determined from laboratory data. Although less effective at low Ψ, it was possible, using these germination parameters, to satisfactorily describe the effect of T and Ψ on germination rates under constant conditions according to the threshold models of thermal and hydrothermal time. These models were applied to field data with the condition that the germination process ceased if T<Tb for thermal time and additionally Ψ<Ψb(G) for hydrothermal time.

Neither model accurately predicted germination patterns in the field. However, the pattern of germination was adequately described in most situations by a modified threshold model in which the predicted progress towards germination was unaffected by soil Ψ, provided it remained above Ψb(G), and was therefore more rapid under variable seed bed conditions than hydrothermal time. In this modified threshold model, the condition Ψ<Ψb(G) had to be fulfilled at the initiation of radicle extension before germination occurred. This result implies that the initiation of radicle growth operates as a moisture-sensitive step that can determine germination and seedling emergence timing under variable soil-moisture conditions.

Seedling emergence was also recorded in the field and used to determine, separately, the impact of germination and post-germination growth on the variation in seedling emergence patterns. The analysis suggests that delays in seedling emergence occur largely in the germination phase, but that seedling losses and variation in the spread of seedling emergence times within the population occur largely during the post-germination growth phase.