Latin Hexameter poetry, like Greek, is quantity based. When we consider the rhythm, we have ictus at the beginning of each of the six feet. However Latin is also a stress-based language which fact naturally adds another dimension to the poetry.
While tempting and not uncommon, the notion that the natural accent/stress was simply neglected and left the stage to the ictus alone (that is, always reciting stress on the ictus), the convincing evidence is rather that the natural stress/accent of the words was retained in poetry recitation.
The evidence is that Latin Hexameter poets deliberately created "harmony" (i.e., the coincidence between the ictus and the natural stress) in the last two feet. It was also maintained that the composers generally tried to avoid harmony in other feet, especially in the second and the third (maybe except Catullus that might have sought harmony in the fourth foot) - but, as the paper shows, the evidence for such argument is not strong.
However, apart from fixing the last two feet, it might be the case that sometimes the poets did deliberately used the stress (with or without relation to the ictus) to create an extra effect. For example, A. W. H. Adkins suggests that verse of Lucretius Propter egestatem linguae et rerum novitatem (DRN, 1.139) which deals with the difficulties of wring Greek Philosophy in Latin verse, was very carefully crafted in a manner that it contains 7 trochees (stress-based; and assuming secondary stress in the first syllable in 4-syllables-words where the third is heavy; and et without stress) in order to illustrate the conflict and the between the stress and the meter.
Though naturally hard to prove, Are there any other suggested uses of stress in Latin Poetry?