This increase in accuracy is likely due to prolongation of movement time and a more complete specification of movement parameters (Hening et al., 1988; Ghez et al., 1997).
The relationship between learning how to better concatenate discrete movements into a sequence and learning to make more skilled continuous single movements is unclear, but the boundary is likely to be blurred. For example, for prehension one has to start with a proximal-muscle driven reach and end with a distal-muscle controlled grasp, in essence a sequence.
More fundamentally, even single movements
to a single target require sequences of activation
in agonist and antagonist muscles. Thus the learning
principles obtained from laboratory-based sequence
tasks may generalize to movements and tasks that are
not overtly sequential.