letters to nature

First, in some sectors and for some regions human-induced climate change may not have as great an impact on natural resources as might multi-decadal natural climate variability. Comparing present resources only with those simulated under future climate change may exaggerate the importance of climate change by ignoring the impacts of natural variability on these time-scales: the estimated impacts may occur even in the absence of human-induced climate change. Second, the results suggest that in many areas it will be very difcult to detect the impact of climate change, even on a multi-decadal time scale; the different spatial patterns of climate change and climate-variability impact, suggest that detection is best undertaken by looking over a large geographic area. Third, adapting our management systems to withstand multi-decadal natural climate variability (adequately dened) may, in some sectors and for some regions, be a sufcient medium-term response to the prospect of climate changealthough elsewhere it may not. Last, the results do not suggest that we can ignore the possibility that climate change will affect our natural resource base; what they do show is that some impacts of natural climate variability may be as great as, or greater than, the estimated impacts of human-induced climate change. This study shows that it is possible, and suggests that it is important, to compare the impacts of climate change alongside those of natural multi-decadal climate variability in order both to assess the importance of climate change and to help in the development of appropriate adaptation strategies. M
Received 30 April; accepted 30 November 1998. 1. Parry, M. L., Carter, T. R. & Hulme, M. What is a dangerous climate change? Glob. Environ. Change, 6, 16 (1996). 2. Wigley, T. M. L., Richels, R. & Edmonds, J. A. Economic and environmental choices in the stabilization of atmospheric CO2 concentrations. Nature 379, 240243 (1996). 3. IPCC (eds Watson, R. T., Zinyowera, M. C., Moss, R. H. & Dokken, D. J.) The Regional Impacts of Climate Change: an Assessment of Vulnerability (Cambridge Univ. Press, 1998). 4. Kittel, T. G. F., Giorgi, F. & Meehl, G. A. Intercomparison of regional biases and doubled CO2 sensitivity of coupled atmosphereocean general circulation model experiments. Clim. Dyn. 14, 115 (1998). 5. Shugart, H. H. & Smith, T. M. A review of forest patch models and their applications to global change research. Clim. Change 34, 131153 (1996). 6. Dowlatabadi, H. Assessing the health impacts of climate change. Clim. Change 35, 137144 (1997). 7. Watson, R. T., Zinyowera, M. C. & Moss, R. H. (eds) Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change: ScienticTechnical Analyses (Cambridge Univ. Press, 1996). 8. Mitchell, J. F. B. & Johns, T. C. On the modication of global warming by sulphate aerosols. J .Clim. 10, 245267 (1997). 9. Arnell, N. W. The effect of climate change on hydrological regimes in Europe: a continental perspective. Glob. Environ. Change 9, 523 (1999). 10. Harrison, P. A. & Buttereld, R. E. Effects of climate change on Europe-wide winter wheat and sunower productivity. Clim. Res. 7, 225241 (1996). 11. Carter, T. R., Parry, M. L., Harasawa, H. & Nishioka, S. IPCC Technical Guidelines for Assessing Climate Change Impacts and Adaptations (UCL/CGER, London/Tsukuba, 1994). 12. Parry, M. L. & Carter, T. Climate Impact and Adaptation Assessment (Earthscan, London, 1998). 13. Hope, C., Anderson, J. & Wenman, P. Policy analysis of the greenhouse effect. Energy Policy 21, 327 338 (1993). 14. Nordhaus, W. D. Optimal greenhouse gas reductions & tax policy in the 'DICE' model. Am. Econ. Rev. Pap. Proc. 83, 313317 (1993). 15. Fankhauser, S. Valuing Climate Change (Earthscan, London, 1995). 16. Tol, R. S. J. The damage costs of climate change: toward more comprehensive calculations. Environ. Resource Econ. 5, 353374 (1995). 17. Mitchell, J. F. B., Johns, T. C., Eagles, M., Ingram, W. J. & Davis, R. A. Towards the construction of climate change scenarios. Clim. Change (in the press). 18. Hulme, M. et al. Construction of a 196190 climatology for Europe for climate change impacts and modelling applications. Int. J. Climatol. 15, 13331363 (1995). 19. Wilby, R. & Wigley, T. M. L. Downscaling general circulation model output: a review of methods and limitations. Prog. Phys. Geogr. 21, 530548 (1997). 20. Richie, J. & Otter, S. in ARS Wheat Yield Project (ed.Willis, W. O.) 159175 (Dept of Agriculture, Agriculture Research Service, ARS-38, Washington DC, 1985). 21. Godwin, D., Ritchie, J., Singh, U. & Hunt, L. A User's Guide to CERES-Wheat - V2.10 (Simulation manual IFDC-SM-2, Int. Fertilizer Development Center, Muscle Shoals, AL, 1990). 22. Weir, A. H., Bragg, P. L., Porter, J. R. & Rayner, J. H. A winter wheat crop simulation model without water or nutrient limitations. J. Agric. Sci. 102, 371382 (1984). 23. Porter, J. R. AFRCWHEAT2: a model of the growth and development of wheat incorporating responses to water and nitrogen Eur. J. Agronomy 2, 6982 (1993). 24. Harrison, P. A. Modelling the effects of climate change on wheat productivity in Europe. Aspects Appl. Biol. 45, 4148 (1996). 25. Harrison, P. A., Buttereld, R. E. & Gawith, M. J. in Climate Change and Agriculture in Europe: Assessment of Impacts and Adaptations (eds Harrison, P. A., Buttereld, R. E. & Downing, T. E.) 370 379 (Res. Rep. No. 9, Environmental Change Unit, Univ. Oxford, 1995). 26. Cure, J. D. & Acock, B. Crop responses to carbon dioxide doubling: a literature survey. Agric. Forest Meteorol. 38, 127145 (1986). 27. Kimball, B. A. et al. Productivity and water-use of winter wheat under free-air CO2 enrichment. Glob. Change Biol. 1, 429442 (1995). 28. Batts, G. R. et al. Yield and partitioning in crops of contrasting cultivars of winter wheat in response to CO2 and temperature in eld studies using temperature gradient tunnels. J. Agric. Sci. 130, 1727 (1998). 29. Tett, S. F. B., Johns, T. C. & Mitchell, J. F. B. Global and regional variability in a coupled AOGCM. Clim. Dyn. 13, 303323 (1997). 30. Leggett, J., Pepper, W. J. & Swart, R. J. in Climate Change 1992: the Supplementary Report to the IPCC Scientic Assessment (eds Houghton, J. T., Callander, B. A. & Varney, S. K.) 7595 (Cambridge Univ. Press, 1992). Acknowledgements. Model data were obtained through the Climate Impacts LINK Project. This work was supported by DGXII of the Commission of the European Community. Correspondence and requests for materials should be addressed to M.H. (e-mail: m.hulme@uea.ac.uk).

The origin of spinifex texture in komatiites

Mark Shore & Anthony D. Fowler
Ottawa-Carleton Geoscience Centre and Department of Earth Sciences, University of Ottawa, 140 Louis Pasteur, PO Box 450, Station A, Ottawa, Ontario K1N 6N5, Canada

.........................................................................................................................

Komatiites are high-temperature, uid, magnesium-rich lavas typically of Archaean age. A striking characteristic feature of such lavas is `spinifex' textureplate-like crystals of olivine ((Mg,Fe)2SiO4), millimetres to decimetres long, in a ne-grained matrix of spherulitic clinopyroxene (Ca(Mg,Fe,Al)(Si,Al)2O6), dendritic chromite ((Mg,Fe)(Cr,Al,Fe)2O4) and altered glass14. Sheaves of olivine crystals can reach lengths exceeding one metre, even in komatiite ows less than 10 metres thick, in sharp contrast to the millimetre-scale post-eruption growth of crystals in more common volcanic rocks. Crystal growth of this magnitude might be a consequence of the high content of the constituent elements of olivine in komatiitic liquid, combined with the low viscosity and high chemical diffusivity of the lavas. But ows lacking spinifex texture are not uncommon, and those with such texture often contain substantial amounts of submillimetre olivine crystals of unremarkable appearance, so chemical considerations alone do not appear to provide a sufcient explanation. Here we present evidence that spinifex texture develops as a result of large thermal gradients, coupled with conductive and radiative heat transfer within olivine crystals xed in the cool upper layers of the lava ows. This mode of growth has features in common with the high-temperature techniques used to grow large synthetic single crystals, but is rarely considered in geological contexts. Many komatiites, including the exceptionally fresh and well exposed ows at Pyke hill in Munro township, northeastern Ontario, erupted as subaqueous lava ows1,4. The thermal effect of seawater inltration into the fractured upper crust of the ows has not previously been considered in komatiite cooling models5,6. Thermal contraction of the solidied sheet-like lava ows (below ,1,000 8C) caused extensive fracturing on a centimetre to decimetre scale (Fig. 1). The permeability of a rock body can be estimated from the number and mean width of its fractures7; that of the illustrated Pyke hill lava tube is ,10-11 m-2. This is close to the calculated bulk permeability of young oceanic crust (,10-12 m-2)8; by contrast, unfractured igneous rocks have typical in situ permeabilities of the order of 10-16 to 10-18 m-2. Thus we expect a substantial heat ux to have been transported by water circulation through cracks. Semiquantitative measurements of hydrothermal heat uxes from basalt ows include ,40 kWm-2 at Heimaey9, 100 kWm-2 (with transient uxes approaching 1 MWm-2) at Kilauea10, and ,1 MWm-2 at Vatnajokull11 (estimated from data in that reference). Such heat transfer rates cannot be achieved by conduction through more than a few centimetres of solid rock. Because fracturing leads to further ingress of sea water, hydrothermal cooling/cracking fronts are probably self-propagating (at least in thin ows) and would migrate rapidly downwards. Observations911 and modelling12 of subaqueous basalt ows have demonstrated that cooling fronts could move at rates as high as several decimetres per hour, resulting in thermal gradients .104 K m-1 and
691

NATURE | VOL 397 | 25 FEBRUARY 1999 | www.nature.com

1999 Macmillan Magazines Ltd

letters to nature
much faster cooling of ow interiors than would occur via purely conductive cooling. Rapid cooling, coupled with the absence of pre-existing nuclei of clinopyroxene and plagioclase (komatiite lavas were commonly superheated with respect to these minerals3), can explain the spherulitic habit and aluminium-rich composition of clinopyroxene (high pressure, high temperature and rapid growth are some factors promoting coupled substitution of 2 Al for Mg Si in clinopyroxene13,14), as well as the delayed crystallization of plagioclase in many komatiites. However, rapid cooling and high thermal gradients offer only a partial explanation for spinifex texture, as komatiite ows typically contain a substantial proportion of millimetre-sized olivine crystals of similar morphology to those of basalt and other common rock types. Textural evidence, including the preferred orientation of the olivine sheaves perpendicular to cooling contacts and their thickening away from the ow surface (Fig. 2), indicates that downward growth of olivine was a competitive process2,3. At the microscopic scale, we observe (Fig. 3) that the growth of plate-like olivine crystals was commonly blocked by earlier-formed crystals. Growth of the later crystals proceeded only on faces away from the blocking crystal owing to a lack of available space and material (primarily Mg). The nucleation and growth of olivine occur readily in ultramac liquids15,16, to the point that it is difcult to quench even strongly superheated liquids to crystal-free glasses. Chilled margins of Pyke hill ows, originally glassy, now contain sparse millimetre-sized euhedral olivine crystals as well as hundreds to thousands of micrometre-sized olivine crystallites per cubic centimetre (ref. 4). Thus, even minor differences in olivine growth rates, due to favourable orientation with respect to thermal gradients, gravity or crystallographic axes, could produce a strongly anisotropic fabric over the hours to days required to solidify a thin komatiite ow. An important factor is the large difference in the thermal properties of olivine and molten komatiite at the temperatures of interest (,1,3001,550 8C). The mean lattice (or phonon) thermal conductivity kL of olivine is estimated to be 23 Wm-1 K-1, as extrapolated from lower-temperature measurements (T 20 1;200 8C)1720 assuming kL 1=a bT, where a and b are constants. This relationship is not exact, but is reasonably accurate at ambient to high temperatures for olivine and other ionic crystals with moderately complex structures19. Inconsistencies between different data sets are largely due to the variable effects of crystallographic orientation in single-crystal studies, grain boundaries in polycrystalline material, geometrically dependent radiative heat transfer, oxidation/reduction of samples at high temperature, and overall experimental difculties of measurements at high pressures. Because of the high temperature of molten komatiites, we must also consider radiative heat transfer in olivine, which has previously been studied as a mechanism of heat ow within the Earth's mantle. The radiative thermal conductivity (kR) within an optically thick body (such that heat transfer can be considered a diffusive process) at thermal equilibrium, with temperature varying in one direction, is described by the expression21 kR 20jT 3 p4

` 0

lx

ex x 4 n 2 x dx ex 2 12

where x hv=kT, lx and nx are the photon mean free path and refractive index at frequency v and mean temperature T, h is Plank's constant (6:626 3 10 2 34 J s), k is Boltzmann's constant (1:38 3 10 2 23 J K 2 1 ), and j is the StefanBoltzmann constant (5:67 3 10 2 8 W m 2 2 K 2 4 ). Where scattering is minimal, as in a single crystal, the mean free path approximately equals the reciprocal of the absorption coefcient. Calculations21 using equation (1) and the measured visible to mid-infrared absorption spectra of magnesian olivine Fe=Mg Fe 0:08, similar to komatiitic olivines) show that the effective radiative thermal conductivity within single crystals is 2.02.2 Wm-1 K-1 at 1,400 8C. Although blackbody emissivity increases by 40% from 1,400 to 1,550 8C, temperature-induced broadening of absorption bands largely offsets the increased emissivity21. Assuming an effective radiative thermal conductivity for komatiitic olivine of ,2 Wm-1 K-1, we arrive at a total thermal conductivity of 45 Wm-1 K-1 for olivine, calculated as the sum of the lattice and effective radiative thermal conductivities. Similar results have been obtained by direct measurements of combined lattice/radiative conductivity of single-crystal specimens several millimetres in size17,19. Owing to experimental difculties, measurements of the thermal conductivity of molten silicates vary greatly. We assume a value of ,1 Wm-1 K-1 for the thermal conductivity (including transmission of thermal radiation) of molten komatiite, consistent with experimental measurements on iron-bearing silicate liquids (0.5 1.5 Wm-1 K-1)22,23, as other measurements on iron-free silicate melts

Figure 1 Fractures within a komatiite lava tube, Pyke hill, Ontario. Although the tube lacks spinifex texture, nearby spinifex-textured ows have similar fracture densities. Most fractures are unrelated to the conjugate set (orientation shown above) associated with eastwest regional folding. Measurements of average crack width (,0.5 mm) and fracture length per unit area (28 m-1) lead to an estimated permeability of 10-11 m-2. Figure 2 Well-developed spinifex texture in a komatiite ow at Pyke hill. Large sheaves of subparallel olivine crystals broaden towards the base of the ow. The plate-like olivine crystals in the central sheaf have their largest faces parallel to the outcrop surface.

692

1999 Macmillan Magazines Ltd

NATURE | VOL 397 | 25 FEBRUARY 1999 | www.nature.com

letters to nature
showing very low thermal conductivities at high temperatures (0.04 Wm-1 K-1 at 1,600 8C) may be awed24. Even using the higher value for the thermal conductivity of the molten silicate, a robust conclusion is that the thermal conductivity of olivine is approximately three to ve times greater than that of its growth liquid. We therefore propose that olivine, rooted in the hydrothermally cooled upper layer of ows, grew by (thermally) constrained crystallization25; this is a process wherein crystals grow into (and transport heat from) a hotter liquid. This differs from most magmatic crystallization wherein the latent heat of crystallization is dissipated in cooler liquid surrounding the crystals. In effect, despite the release of latent heat, the growing olivine crystals would act as heat sinks for the surrounding komatiitic liquid, not as heat sources. In contrast, olivine crystals freely suspended in liquid (as in a owing komatiite lava) would grow normally, and would not be expected to show unusual morphological features. The most pronounced effect of the different thermal and optical properties of liquid and olivine would occur at the leading tip of a growing crystal. The result is to increase local thermal gradients and make the crystalmelt interface more convex (that is, sharpen crystal tips), as shown by observations26 and nite-element modelling27,28 of Czochralski and Bridgman growth of oxide single crystals. We now examine evidence supporting this mode of growth, considering rst the crystallographic properties of olivine in greater detail. The thermal conductivity of olivine is strongly anisotropic, with values (at ambient temperature) of 5.9 (60.1), 3.4 (60.06) and 5.1 (60.1) Wm-1 K-1 parallel to the a, b and c crystallographic axes (space group Pbnm), respectively29. This anisotropy persists at higher temperatures (M. D. Collins, personal communication). In addition, olivine is strongly pleochroic (its optical absorption varies with crystallographic orientation of light) in the near-infrared range from 0.8 to 1.2 mm (refs 21, 30), the spectral region of maximum blackbody emission from a partially molten komatiite. At these wavelengths, absorption is lowest for non-polarized light propagating parallel to the a crystallographic axis. The total thermal conductivity at high temperatures is expected to vary as a . c q b, with conductivity parallel to a being 1020% greater than that parallel to c. Downward growth parallel to the a axis is predicted if heat transfer from the leading tips of crystals is a dominant growth-rate factor. Accordingly, we examined olivine within coarse spinifex-textured lavas from Pyke hill for evidence of a preferred crystallographic orientation (in addition to the previously noted attening perpendicular to the b axis2). Orientated thin sections were cut from 14 different olivine sheaves that were perpendicular (6208) to the upper surface of ows. In every case, the olivine crystals were orientated with a perpendicular to the ow tops, parallel to the inferred maximum thermal gradient. The same orientation has been found in synthesized comb-layered olivine, a texture also attributed to constrained crystallization31. This growth mechanism might be applicable to the vertically orientated needle-like clinopyroxene crystals found in some komatiitic basalts. But the lower temperatures involved (,1,150 1,300 8C) would reduce the role of radiative heat transfer, while the greater effect of superheating on clinopyroxene nucleation (versus that of olivine) and the role of the polymerized chain structure of pyroxene parallel to the c axis are complicating factors. In addition, the requisite high-temperature thermal conductivity and infrared spectral data are not available. Thermally constrained crystallization of olivine within spinifextextured layers of komatiite ows would have been favoured by specic conditions present (kinetically easy growth of olivine, elevated temperatures, large and consistently orientated thermal gradients, and signicant differences in the thermal transport properties of olivine and molten komatiite). Preserved textural and mineralogical evidence is consistent with this geologically unusual mode of growth. Other processes that probably occurred within cooling ows (for example, compositionally and thermally driven melt convection5,6) may have been of less importance to the development of spinifex texture. M
Received 16 February; accepted 14 December 1998. 1. Pyke, D. R., Naldrett, A. J. & Eckstrand, O. R. Archean ultramac ows in Munro Township, Ontario. Bull. Geol. Soc. Am. 84, 955978 (1973). 2. Donaldson, C. H. in Komatiites (eds Arndt, N. T. & Nisbet, E. G.) 213244 (Allen & Unwin, London, 1982). 3. Arndt, N. T. in Archean Crustal Evolution (ed. Condie, K. C.) 1144 (Elsevier, Amsterdam, 1994). 4. Shore, M. Cooling and Crystallization of Komatiite Flows Thesis, Univ. Ottawa (1996). 5. Huppert, H. E. & Sparks, R. S. J. Komatiites I: Eruption and ow. J. Petrol. 26, 694725 (1985). 6. Turner, J. S., Huppert, H. E. & Sparks, R. S. J. Komatiites II: Experimental and theoretical investigations of post-emplacement cooling and crystallization. J. Petrol. 27, 397437 (1986). 7. Nehlig, P. Fracture and permeability analysis in magma-hydrothermal transition zones in the Samail ophiolite (Oman). J. Geophys. Res. 99, 589601 (1994). 8. Davis, E. E., Chapman, D. S. & Forster, C. B. Observations concerning the vigor of hydrothermal circulation in young oceanic crust. J. Geophys. Res. 101, 29272942 (1996). 9. Bjornsson, H., Bjornsson, S. & Sigurgeirsson, Th. Penetration of water into hot rock boundaries of magma at Grmsvotn. Nature 295, 580581 (1982). 10. Hardee, H. C., Dunn, J. C. & Hills, R. G. Probing the melt zone of Kilauea Iki lava lake, Kilauea volcano, Hawaii. Geophys. Res. Lett. 8, 12111214 (1981). 11. Gudmundsson, M. T., Sigmundsson, F. & Bjornsson, H. Icevolcano interaction of the 1996 Gjalp subglacial eruption, Vatnajokull, Iceland. Nature 389, 954957 (1997). 12. Lowell, R. P. & Germanovich, L. N. On the temporal evolution of high-temperature hydrothermal systems at ocean ridge crests. J. Geophys. Res. 99, 565575 (1994). 13. Kinzler, R. T. & Grove, T. L. Crystallization and differentiation of Archean komatiite lavas from northeast Ontario: phase equilibrium and kinetic studies. Am. Mineral. 70, 4051 (1985). 14. Parman, S. W., Dann, J. C., Grove, T. L. & de Wit, M. J. Emplacement conditions of komatiite magmas from the 3.49 Ga Komati Formation, Barberton Greenstone Belt, South Africa. Earth Planet. Science Lett. 150, 303323 (1997). 15. Donaldson, C. H. An experimental investigation of olivine morphology. Contrib. Mineral. Petrol. 57, 187213 (1976). 16. Donaldson, C. H. An experimental investigation of the delay in nucleation of olivine in mac magmas. Contrib. Mineral. Petrol. 69, 2132 (1979). 17. Kanamori, H., Fujii, N. & Mizutani, H. Thermal diffusivity measurement of rock-forming minerals from 3008 to 1100 8K. J. Geophys. Res. 73, 595605 (1968).

Figure 3 Cross-section of plate-like olivine crystals (largely replaced by serpentine and magnetite) in a spinifex-textured komatiite. The crystal to the left has interfered with the growth of its neighbour, which has formed a series of stepped-out segments to the right. Gently curved olivine crystals are common in komatiites. In such crystals the crystallographic orientation does not vary; rather, the curvature arises from numerous small growth offsets, all in the same direction. NATURE | VOL 397 | 25 FEBRUARY 1999 | www.nature.com

1999 Macmillan Magazines Ltd

693

letters to nature
18. Fujisawa, H. et al. Thermal diffusivity of Mg2SiO4, Fe2SiO4, and NaCl at high pressures and temperatures. J. Geophys. Res. 73, 47274733 (1968). 19. Schatz, J. F. & Simmons, G. Thermal conductivity of earth materials at high temperatures. J. Geophys. Res. 77, 69666983 (1972). 20. Katsura, T. Thermal diffusivity of olivine under upper mantle conditions. Geophys. J. Int. 122, 6369 (1995). 21. Shankland, T. J., Nitsan, U. & Duba, A. G. Optical absorption and radiative heat transport in olivine at high temperature. J. Geophys. Res. 84, 16031610 (1979). 22. Murase, T. & McBirney, A. R. Properties of some common igneous rocks and their melts at high temperatures. Bull. Geol. Soc. Am. 84, 35633592 (1973). 23. Buttner, R., Zimanowski, B., Blumm, J. & Hagemann, L. Thermal conductivity of a volcanic rock material (olivine-melilitite) in the temperature range between 288 and 1470 K. J. Volcanol. Geotherm. Res. 80, 293302 (1998). 24. Shore, M. Comment on ``Experimental determination of the thermal conductivity of molten CaMgSi2O6 and the transport of heat through magmas''. J. Geophys. Res. 100, 2240122402 (1995). 25. Tiller, W. A. The Science of Crystallization: Macroscopic Phenomena and Defect Generation (Cambridge Univ. Press, 1991). 26. Cockayne, B., Chesswas, M. & Gasson, D. B. Facetting and optical perfection in Czochralski grown garnets and rubies. J. Mater. Sci. 4, 450456 (1969). 27. Brandon, S. & Derby, J. J. Heat transfer in vertical Bridgman growth of oxides: effects of conduction, convection, and internal radiation. J. Cryst. Growth 121, 473494 (1992). 28. Tsukada, T., Kakinoki, K. & Hozawa, M. Effect of internal radiation within crystal and melt on Czochralski crystal growth of oxide. Int. J. Heat Mass Transfer 38, 27072714 (1995). 29. Chai, M., Brown, J. M. & Slutsky, L. J. Thermal diffusivity of mantle minerals. Phys. Chem. Miner. 23, 470475 (1996). 30. Burns, R. G. Mineralogical Applications of Crystal Field Theory 2nd edn (Cambridge Univ. Press, 1993). 31. Donaldson, C. H. Laboratory duplication of comb layering in the Rhum pluton. Mineral. Mag. 41, 323336 (1977). Acknowledgements. We thank A. M. Hofmeister for comments on the manuscript. This work was supported by NSERC. Correspondence and requests for materials should be addressed to A.D.F. (e-mail: afowler@uottawa.ca).

A ferric-chelate reductase for iron uptake from soils

Nigel J. Robinson*, Catherine M. Procter*, Erin L. Connolly & Mary Lou Guerinot
* Department of Biochemistry and Genetics, The Medical School, University of Newcastle, Newcastle NE2 4HH, UK Department of Biological Sciences, Dartmouth College, Hanover, New Hampshire 03755-3576, USA
.........................................................................................................................

Iron deciency aficts more than three billion people worldwide1, and plants are the principal source of iron in most diets. Low availability of iron often limits plant growth because iron forms insoluble ferric oxides, leaving only a small, organically complexed fraction in soil solutions2. The enzyme ferric-chelate reductase is required for most plants to acquire soluble iron. Here we report the isolation of the FRO2 gene, which is expressed in iron-decient roots of Arabidopsis. FRO2 belongs to a superfamily of avocytochromes that transport electrons across membranes. It possesses intramembranous binding sites for haem and cytoplasmic binding sites for nucleotide cofactors that donate and transfer electrons. We show that FRO2 is allelic to the frd1 mutations that impair the activity of ferric-chelate reductase3. There is a nonsense mutation within the rst exon of FRO2 in frd1-1 and a missense mutation within FRO2 in frd1-3. Introduction of functional FRO2 complements the frd1-1 phenotype in transgenic plants. The isolation of FRO2 has implications for the generation of crops with improved nutritional quality and increased growth in iron-decient soils. Arabidopsis is an `iron-efcient' plant, able to acquire iron from soils of low iron availability, but the genetic basis of iron efciency is unclear. In such plants, the activity of ferric-chelate reductase at the plasma membrane of root epidermal cells increases when iron is decient4,5. Organic compounds of either plant or microbial origin that retain Fe3+ in the soil solution have a lesser afnity for Fe2+. It is therefore likely that ferric-chelate reductase activity releases iron from organic compounds, generating free iron for uptake6. Because the ferric-chelate reductase of plant roots has some functional similarity to the human phagocytic NADPH oxidase gp91phox
694

(ref. 7) and yeast ferric-chelate reductases such as FRE1 (ref. 8) and FRP1 (ref. 9), we speculated that it may share elements of structural similarity with these enzymes. The enzymes are involved in the transfer of electrons from cytosolic donors to FAD and then, through two consecutive haem groups, to single electron acceptors on the opposing face of a membrane. The haem groups are coordinated to four conserved histidine residues located on two transmembrane a-helices10. However, gp91phox and the yeast ferric-chelate reductases have in common only a few short sequence motifs, most notably associated with the cofactor-binding sites. Two approaches were initiated to obtain a ferric-chelate reductase gene: generating reductase-decient plant mutants3, and cloning candidate plant DNA sequences. Degenerate oligonucleotide primers were designed to anneal to sequences encoding the only tetrapeptide motif (the FAD-binding site) common to gp91phox and the yeast ferric-chelate reductases, and to a second partly conserved motif associated with the NADPH-binding sites. Several polymerase chain reaction (PCR) products were amplied from Arabidopsis genomic DNA using a low annealing temperature. Seven products were cloned and sequenced and one, which encodes a serine-rich polypeptide, was selected for further analysis because FRP1 is serine rich in the region between the cofactor-binding sites. This PCR product was used to screen genomic and complementary DNA libraries, and a genomic fragment containing two closely related genes, designated FRO1 and FRO2, was fully sequenced (Fig. 1a). The FRO2 intronexon boundaries were identied from the cDNA sequence; the FRO1 intronexon boundaries were deduced by analogy to FRO2 by using predictions11 of 59 donor and 39 acceptor sites and by reverse transcriptase PCR (RT-PCR). FRO2 and FRO1 have 61.9% sequence identity (90.5% similarity). Flanking the 39 end of FRO1 is a synvergently transcribed gene encoding a cytochrome P450, designated CYP86A4 (Fig. 1a), followed by a convergently transcribed unknown gene with similarity to the expressed sequence tags Z17619, N96903 and H76348. Flanking the 59 end of FRO2 is a synvergently transcribed gene related to MKP4 that encodes a deduced mitogen-activated protein (MAP) kinase (not shown on Fig. 1a). The sequence and predicted structure of FRO2 is shown in Fig.1, FRO2 is composed of 725 amino acids, has a predicted pI of 9.37 and relative molecular mass (Mr) of 81.50, although the presence of glycosylation motifs suggest that the native protein may have a greater Mr. FRO2 contains sequences identical to the FAD-binding site and the conserved region adjacent to the NADPH-binding site of FRE1 (underlined in Fig. 1b) but no other identical sequence motifs of an equivalent length. However, predictions of secondary structure indicate further similarity. The amino-terminal regions of the yeast ferric-chelate reductases and gp91phox form several transmembrane a-helices that are apparent in hydrophobicity plots. Six hydrophobic domains are identied within the N-terminal regions of FRO2 along with two carboxy-terminal hydrophobic domains, all of which are predicted12 to form transmembrane ahelices. This suggests that an intracellular region of FRO2, containing the deduced cofactor-binding sites, is anchored at both ends by membrane-spanning regions (Fig. 1c). In contrast, the gp91phox and yeast ferric-chelate reductase cofactor-binding sites and associated C-terminal regions are all predicted to be cytosolic. To form an active complex, gp91phox requires a second membrane protein, p22phox. The total number of predicted transmembrane a-helices in FRO2 is equivalent to the number in gp91phox (six) and p22phox (two) combined, and it is feasible that the C-terminal domain, including the two additional transmembrane a-helices in FRO2, performs an analogous function to p22phox (p22phox shows 10% identity and 50% sequence similarity to the C-terminal domain of FRO2). Two pairs of histidine residues in FRO2 that lie on two predicted, similarly orientated, transmembrane a-helices are in equivalent locations (Fig. 1b) to the histidine residues in FRE1 and gp91phox that coordinate haem10 (Fig. 1c).
NATURE | VOL 397 | 25 FEBRUARY 1999 | www.nature.com

1999 Macmillan Magazines Ltd