According to an earlier question, we do not know how stress was realized on classical Latin. It may have been dynamic (stressed syllables are louder), tonal (stress changes pitch), or a combination, or perhaps something yet different. While we do not know exactly how the stress was realized in prose (normal spoken language), do we know about the relations of stress in prose and poetry?
The way I pronounce Latin, prose stress is mainly dynamic. However, I also use the same kind of stress to indicate stressed syllables in verse, such as the first syllable of each of the six metrons in a hexameter. This means that prose stress is completely lost in verse. (Both kinds of stress have a tendency to fall on long syllables so they do coincide more than one would expect by random.) However, it occurred to me that this may be wrong. Perhaps prose stress is dynamic and metric stress is supposed to be tonal, allowing the two to coexist.
Do we know whether metric stress (the stressed syllables in a metric poem) replaced the usual stress pattern of classical Latin? Does prose stress have an audible role recited poetry? Do we know at all how the two stress types were related? It may also be that the answer depends on the kind of poetic meter.