The upper Miocene Reef Complex of Mallorca is a 20-km prograding unit which crops out in sea cliffs along the southern side of the island. These vertical and exceptionally clean outcrops permit: (i) identification of different facies (lagoon, reef front, reef slope and open platform) and their geometries and boundaries at different scales, ranging from metre to kilometre, and (ii) construction of a 6-km-long high-resolution cross-section in the direction of reef progradation. This cross-section shows vertical shifts of the reefal facies and erosion surfaces linked to a general progradational pattern that defines the accretional units. Four hierarchical orders of magnitude (1-M to 4-M) of accretional units are identified by consideration of the vertical facies shifts and by which erosion surfaces are truncated by other erosion surfaces. All these orders show similar patterns: horizontal beds of lagoonal facies in the upper part (landward), reefal and slope facies with sigmoidal bedding in the central part, and open-platform facies with subhorizontal bedding in the lower part (basinwards). The boundaries are erosion surfaces, horizontal over the lagoon facies, dipping basinwards over the reef-front facies and connecting basinwards with their correlative conformities over the reef-slope and open-platform facies.

The four orders of accretional units are interpreted in terms of four (1-M to 4-M) hierarchies of sea-level cycles because (i) there is a close relation between the coral growth and the sea surface, (ii) there are vertical shifts in the reefal facies and their relation to the erosion surfaces, and (iii) there was very little tectonic subsidence in the study area during the late Miocene. Additionally, all these units can be described in terms of their position relative to the sea-level cycle: (i) the reefs prograde on the open-platform sediments during low stands of sea-level; (ii) aggradation of the lagoon, reef and open-platform facies dominates during sea-level rises, and the lagoonal beds onlap landwards upon the previous erosion surface; (iii) reefal progradation occurs during high stands of sea-level; and (iv) the 2-M sea-level fall produces an off-lapping reef and there is progradation with downward shifts of the reefal facies and erosion landward on the emerged (older) reefal units (A-erosion surfaces); the 3-M and 4-M sea-level falls produce only erosion (B-and C-erosion surfaces).

Although precise age data do not exist at present, some speculations upon the frequency of these Miocene relative sea-level cycles can be made by comparisons with Pleistocene cyclicity. There is a good correlation between the Miocene 2-M cycles and the 100-ka Pleistocene cycles. Consequently, the 1-M cycles can be assigned to a fourth order in relation to previously proposed global cycles and the 2-M to fifth-order cycles.

All these accretional units could be defined as ‘sequences’, according to the definition as commonly used in sequence stratigraphy. However, they represent higher than third-order sea-level cycles, but are not parasequences. The term subsequence, therefore, is suggested to define ‘a part of a sequence bounded by erosion surfaces (mostly subaerial) and their correlative conformities basinwards'. A hierarchy of subsequences can be established.