Cyclical signals are commonly used to temporally regulate periodic developmental events in animals. However, little is known about how cyclical signals act with other temporal signals to specify a stage-specific process. We study the temporal regulation of distal tip cell (DTC) migration during gonad development in C. elegans. The timing of DTC dorsal turn occurs in late L3 and is determined by a gene regulatory network, consisting of upstream cyclic LIN-42/PER signals, DRE-1/FBXO11-mediated protein degradation, transcriptional regulations via LIN-29, DAF-12 and BLMP-1, and the guidance receptor UNC-5. The cyclic LIN-42/PER signals, which peak in the mid-L2 and mid-L3 stages, inhibit expression of
lin-29 and
unc-5, while activating
blmp-1. By combining mathematical modeling and experiments, we have identified a missing gene regulation in which
blmp-1 can activate
lin-29 transcription in late L3 and found it important for the DTC to distinguish the first and second LIN-42 expression cycles. Our simulations show that bistability, contributed by mutual inhibition between
blmp-1 and
lin-29, and
blmp-1 autoactivation, keeps the DTC in a stable "
blmp-1-ON,
lin-29-OFF" state and insensitive to the drop of
lin-42 signal in late L2. Subsequent degradation of BLMP-1 by DRE-1 in mid-L3, reduces BLMP-1 to a level that, together with the new positive regulation of BLMP-1 to
lin-29, renders the DTC sensitive to the drop of LIN-42 and switching to a stable "
blmp-1-OFF,
lin-29-ON" state. This switch elevates UNC-5 above the threshold for dorsalward turning. Thus, UNC-5 in wild-type is induced by the drop of the second, but not the first,
lin-42 expression cycle. In support of our finding, loss of
blmp-1, lacking this
lin-42 signal differentiation mechanism, in a
daf-12 sensitized background causes variability in the timing of DTC dorsal turn. Our mathematical model shows that, in the
blmp-1;
daf-12 mutant, UNC-5 is expressed at a sub-threshold level and in a dynamic pattern complementary to that of the LIN-42 signals. Noise, or random fluctuations, in gene expression can bring the UNC-5 receptor level above or below the threshold needed for dorsalward turning, leading to a wide range of temporal phenotypes. Further, lowering the receptor threshold or modulating the receptor dynamics and noise shifts the heterogeneous phenotype distribution. Together, our work reveals a mechanism by which a gene regulatory network distinguishes different cyclic signals to precisely control a stage-specific developmental event.