In an attempt to quantify the extent of geochemical heterogeneity within a restricted and well dated portion of the upper mantle, 27 chromite separates from the 90 My old chromite deposits in the Mayarí–Baracoa ophiolite belt in eastern Cuba have been investigated for platinum group element (PGE) concentrations and Re–Os isotopic systematics. The samples are characterized by systematically subchondritic initial 187Os/188Os ratios and substantial heterogeneity. The initial 187Os/188Os ratios vary with chromite chemistry and with geographical distribution, reflecting differences in the Os isotopic evolution for the different upper mantle sections represented by the ophiolite. Accordingly, the Os isotope data might be divided into three groups. In the Moa–Baracoa district, where the chromite bodies are located in the mantle–crust transition zone, the calculated initial γOs values average − 0.97 ± 0.69 (n = 13). In the Sagua de Tanamo district, where chromite chemistry is highly variable and their location in relation the mantle sequence is less clear, the initial γOs values are intermediate, with an average of − 1.77 ± 0.80 (n = 7). In the Mayarí district, where the chromite bodies are located in the lower part of the mantle sequence, initial γOs values average − 2.66 ± 0.29 (n = 7). These subchondritic (i.e. negative) initial γOs values are most simply explained by Re depletion during ancient partial melting and/or melt percolation events. The Os isotope heterogeneity documented here indicates a high degree of geochemical complexity on small to intermediate length scales in the upper mantle. Our results, in combination with data on chromites from the literature, show that an “average present-day Os isotopic composition” for the hypothetical depleted MORB mantle (DMM) reservoir cannot be precisely established beyond the statement that it is “broadly chondritic”. Indeed, the upper mantle cannot be considered a sufficiently homogeneous geochemical “reservoir” to serve meaningfully as a baseline against which geochemical “anomalies” are evaluated. On the other hand, our findings are consistent with the “Statistical Upper Mantle Assemblage” or “SUMA”-concept, according to which a high level of geochemical heterogeneity is maintained in the upper mantle at all relevant length scales, as a result of the plate-tectonic cycle and intra-mantle processes such as melt-migration and metasomatism.

We thank Toby Leeper for his insightful technical assistance with the mass spectrometer at the Geological Institute. Peter Venslev is thanked for his help in mineral separation. Britta Munch helped with drafting. Two anonymous reviewers provided valuable and critical comments which led to a significant improvement of the manuscript. RF is grateful for financial support through SNF (Statens Naturvidenskabelig Forskningsra°d; Danish Research Agency) grant number 21-01- 0492 56493. This paper is also a contribution to the Spanish research project GL2004-00622.

11 pages, 3 figures, 1 table.

Peer reviewed