I don't think there is an obvious contradiction here.
(1) Mapping neuronal function to the perception of threat (or emotions in general) is often advertised as well-understood, but in fact it is not. Serotonin-concentration-based models (such as the monoaminergic theory of depression) are unavoidably imprecise because ”the brain is not soup”. Additionally, it is also impossible with current technology to measure actual serotonergic signalling in vivo with whole-brain comparability (though progress is being made, e.g. via: fMRI contrast-agents and stimulus-evoked opto-fMRI for monoaminergic systems).
(2) Numerous pharmacological interventions which increase signalling at the serotonergic synapse have been documented to cause widely varying effects (compare e.g. monoamine reuptake inhibitors, monoamine release agents, and monoamine receptor agonists). While selective serotonin reuptake inhibitors (SSRIs), such as fluoxetine and citalopram, are commonly reported to decrease anxiety, serotonin receptor agonists, such as LSD, are reported to induce acute anxiety (though interestingly small numbers of LSD administrations, in conjunction with psychotherapy, have also been reported to reduce anxiety in the long term).
The conclusions thus far would be: (1) we can't be entirely sure about how serotonin mediates or counterbalances anxiety (2) but the specific kinetics of activity at the serotonergic synapse seem to be just as relevant as the actual amount of signalling. You may think, however, that the effects of SSRIs are kinetically similar to reduced expression of the serotonin transporter, SERT. This is however not the case, as under SSRI treatment it is generally believed that chronic downregulation of 5-HT1A inhibitory autoreceptors is responsible for the therapeutic effect, rather than the acute increase of serotonin concentration in the synaptic cleft (this being in line with the fact that therapeutic SSRI effects are observed chronically rather than acutely). Further evidence corroborating the notion that constitutively reduced expression of the serotonin transporter increases anxiety can be found from studies on the 5-HTTLPR polymorphism. In these studies the short version of the allele, which leads to reduced expression of SERT, has also been reported (though afaik this effect is not very robust) to correlate with increased activity in the amygdala.