Abstract The tectonic position of the Palaeozoic alkaline complexes containing phoscorites and carbonatites is determined by the combination of three factors. These are (1) the presence of rift systems; (2) the presence of deep fracture zones; (3) the powerful energetic excitement of lithosphere with an epicentre in the region of the Khibiny complex, i.e. the action of a mantle plume or ‘a hot spot’. The ages of alkaline magmatism in the eastern part of the Baltic Shield are in general synchronous with the manifestation of alkaline magmatism in the whole of the North Atlantic Alkaline Province. Early Palaeozoic alkaline magmatism on the Kola Peninsula is small-scale compared with that in western Fennoscandia. It consists of diatremes and dykes of olivine melilitite-alnöite-carbonatites. The formation of the carbonatitic ultramafite-foidolite complexes started in the early Palaeozoic. The peak of alkaline magmatism, of volcanic activity, large intrusions, and dykes, took place in the late Devonian, forming the Devonian Kola Alkaline Province (KAP). As a whole, the carbonatites in their various different forms and sizes are related to six petrogenetic types and varieties of alkaline complexes and dykes. They are: (1) Proterozoic complexes of ultramafic/mafic rocks-foidolites-foidosyenites; (2) dykes and diatremes of the early Palaeozoic series of the Kandalaksha Graben; (3) Kandagubian seriesdykes; (4) Turiy series dykes; (5) Palaeozoic plutons of alkaline-ultramafic rocks; (6) agpaite nepheline syenite complexes. A quantitative model proposed for the formation of the carbonatites involves liquid immiscibility from an evolved carbonated nepheline melilitite melt.

ABSTRACT A set of criteria has been established that we use to assess whether carbonatites are generated as primary mantle melts, or whether they are the products of magma differentiation (crystal fractionation, liquid immiscibility) of a parental, carbonated silicate melt. On the basis of these criteria, in conjunction with the vast amount of published field information and geochemical data from the Kola Alkaline Province (KAP), no clear-cut pattern emerges that favours any of these processes over the others. Any evidence for liquid immiscibility seems to be restricted to some members of the dyke swarms (Kandaguba, Turiy), while robust evidence for the generation of primary carbonatitic magmas seems to be lacking. Potential candidates for primary melts include the carbonatites from Turiy Mys and the older dyke swarms associated with the Kandalaksha Deep Fracture Zone. The spatial and temporal association of most carbonatites with silicate rocks at Kola, along with the presence of olivinites, clinopyroxenites and other cumulate rocks in some complexes favour crystal fractionation as an important process in generating some of the KAP rocks. We are left with the impression that all three processes may be responsible for carbonatite generation, even within the same complex. The isotopic evidence suggests the involvement of at least three distinct mantle sources, one of which is common to all of the complexes and perhaps indicative of a deep-seated, primitive mantle at least 3 Ga old. Overall, we propose an integrated plume-related model for the Devonian alkaline and carbonatitic magmatism that characterizes much of the KAP. Low-degree partial melting within the volatile-rich, and cooler parts of a plume head accompanied by the mixing of small-volume magma batches, their subsequent differentiation, and interaction with entrained materials and continental lithosphere may help explain some of the problems associated with unravelling the genesis of carbonatite magmas.