THE ARABIAN CONTINENTAL ALKALI BASALT PROVINCE .3. EVOLUTION OF HARRAT KISHB, KINGDOM OF SAUDI-ARABIA

Harrat Kishb is a 5,892 km2 lava field in western Saudi Arabia with an overall K-Ar age range from 2 Ma to prehistoric (at least 4,500 to 2, 000 yr B.P.). It contains three stratigraphic units: the Diakah, Nafra t, and Hil basalts. Harrat Kishb differs from the coeval, mildly alkal ine harrats along the 600-km-long Makkah-Madinah-Nafud volcanic line t o the west in that its lavas are nodule-bearing and considerably more silica-undersaturated. The nodules are most abundant in basanite from the central vent zone and include both Type I and Type II mantle xenol iths. Harrat Kishb is fundamentally a bimodal lava field that is domin ated, at one end, by alkali olivine basalt (AOB) and basanite, with su bordinate hawaiite and olivine transitional basalt (OTB), and at the o ther end, by phonolite. The few intermediate phonotephrites are compos itionally and texturally heterogeneous, and regarded as hybrid lavas o f basalt and phonolite. Although none of the basaltic lavas is a prima ry melt, their chemistry was controlled largely by partial melting. Fr om bottom to top, the stratigraphic units become less voluminous and i ncreasingly more undersaturated, reflecting decreasing degrees of part ial melting with time. Fractional crystallization of basanites (the sm allest-degree partial melts) probably occurred by the plating of pyrox ene and spinel (+/- olivine) along the walls of narrow conduits during their ascent through the subcontinental mantle lithosphere (flow crys tallization). Fractional crystallization of OTB and AOB (the larger-de gree partial melts) may have occurred in reservoirs at the crust-mantl e boundary, a density filter for rising magmas. The relatively higher volatile content of the basanites, a consequence of smaller degrees of partial melting, may have allowed many of them to accelerate through the crust-mantle density barrier carrying their load of mantle nodules rapidly to the surface. Major-element mass-balance calculations, and trace-element enrichment factors, demonstrate that the phonolites were probably derived from basanitic magmas via about 62% fractional cryst allization of a clinopyroxene-dominated, feldspar-bearing mineral asse mblage. There is a lack of evidence for fractional crystallization in static, high-level magma chambers, and a preferred model for the basan ite-to-phonolite link involves the variable removal of feldspar by con tinued "flow crystallization" during the ascent of basanitic magmas th rough the crust. Such a model requires that these parental basanites h ad slower flow rates than did the nodule-entrained basanites extruded at the surface. Field evidence for the ascent of basaltic and phonolit ic magmas in shared conduits is consistent with periodic magma recharg e during "flow crystallization." Although there is little evidence for the mixing of these diverse magma types at depth, they were mixed at the vent by clastogenic processes to produce hybrid lavas of phonoteph rite composition.